Interaction of fibronectin-coated beads with CHO cells

Experimental Cell Research 152 (1984) 302-312

Interaction

of Fibronectin-Coated with CHO Cells

Beads

M. A. SCHWARZ and R. L. JULIAN0 Department of Pharmacology and the Graduate School of Biomedical Sciences, University of Texas Medical School Houston, TX 77025, USA

One of the earliest events in the adhesion of fibroblasts to a substratum is the recognition by the cells of macromolecular adhesive factors, such as tibronectin. This early event is followed by a complex series of cell alterations leading to adhesion and spreading. To identify cell surface components involved in the initial cell-tibronectin recognition step, we have employed an assay involving latex particles coated with radiolabelled plasma Fibronectin (Fn). In previous studies from this laboratory (Harper & Juliano, J cell bio187 (1980) 755) [28], we demonstrated that Fn-mediated adhesion of CHO cells is temperaturedependent, cation-dependent and sensitive to cytoskeletat disrupting agents; by contrast, binding of ‘H-Fn beads was unaffected by these factors, indicating that this process reflects binding and recognition events at the cell surface which are independent of cytoskeletal and metabolic activity. Biological specificity of 3H-Fn bead-to-cell binding was confirmed by the ability of anti-Fn antisera to completely block the process. To examine surface components which may mediate binding we treated Fn beads with purified glycosaminoglycans (GAGS) or glycolipids prior to incubation with cells. Among the GAGS tested, heparin, heparan sulfate and dermatan sulfate blocked bead binding in a doserelated fashion with heparin being most potent. The gangliosides GTI, and GMI, also inhibited bead binding. However, treatment of cells with neuramimdase had no effect on bead binding while subsequent analysis of treated cells by thin layer chromatography revealed a drastic reduction in the amount of GM3, the predominant CHO cell ganglioside. CHO cells were also incubated with a panel of proteolytic enzymes to study the possible role of cell surface proteins or glycoproteins in Fn bead binding. We found 3H-Fn bead binding to be quite sensitive to pretreatment with thermolysin, pronase, and papain but only moderately sensitive to treatment with trypsin. From our findings we suggest: (1) binding of Fn beads to CHO cells reflects an early step in the adhesion process; (2) glycolipids may block bead binding but are probably not the endogenous binding site for Fn; (3) protease sensitive components (glycoproteins or proteoglycans) may be more likely candidates as cell surface-binding sites for Fn.

Although cell adhesion is a fundamentally important process which has been implicated in morphogenesis, metastasis, and wound healing, the biochemical determinants involved in adhesion remain to be elucidated. It has become clear that cell interaction with certain extracellular ‘linking elements’ is vital to the establishment of adhesive bonds with a substratum. Fibronectin, a large glycoprotein present in body fluids, as a component of the extracellular matrix and on the cell membranes, plays the role of a linker protein in promoting tibroblast adhesion and spreading [l-5]. Thus the interaction of tibroblast membrane components with substratum-immobilized tibronectin is a critical early step in the adhesion process. Recently considerable effort has been directed toward characCopyright 0 1984 by Academic Press, Inc. All rights of reproduction in any fom reserved oow4s27md so3.00

Znteraction of fibronectin-coated

beads with CHO cells 303

terizing cell surface components involved in fibronectin-specific adhesion. Although certain classes of cell surface macromolecules have been suggested to play a role in cell-fibronectin recognition (e.g., sulfated proteoglycans [6-%J, glycoproteins [l, 9, lo], and sialylated glycolipids [II, 121)definitive identification of cell surface ‘fibronectin receptors’ remains problematical. Cell adhesion is a very complex process involving cytoskeletal action and energy metabolism as well as binding of cell surface receptors to adhesiveligands [13, 141.In order to study the earliest events of adhesion, namely those involved in the binding of adhesion ligands to the cell surface, we have employed an assay which is essentially independent of complex cellular activities, but which may simply reflect the existence of adhesive ligand ‘receptor’ sites. This assay is adapted from that described by Grinnell [lo] and involves the binding of fibronectin-coated latex beads to Chinese hamster ovary (CHO) fibroblasts in suspension. In this report we demonstrate that the binding of beads coated with bovine plasma fibronectin (Fn beads) is not affected by metabolic or cytoskeletal inhibitors, occurs about equally well at 4 and 37’C, and can even occur after fixation of the cells. The binding of Fn beads is, however, inhibited by the presence of anti Fn antibody, by certain types of glycosaminoglycans, and by sialyated glycolipids. Perhaps most importantly, Fn bead binding is ablated by treatment of the cells with low doses of certain proteolytic enzymes (thermolysin, papain). MATERIALS AND METHODS Cell Culture CHO cells were routinely maintained in suspension culture in a minimal essential medium (aMEM) plus 5 % fetal calf serum (FCS) at 37°C and 5% CO2 [13, 141.

Preparation of Bovine Plasma Fibronectin (Fn) Fn was prepared from bovine plasma by modification of the procedures described by McDonald & Kelley [15] and Vuento & Vaheri [16]. Fresh-frozen acid citrate dextrose plasma was thawed at 37°C and soybean trypsin inhibitor (100 mg/l) was added. Vitamin-K-dependent clotting factors were removed by addition of IS g/i of BaClr. The precipitate, consisting of barium citrate and absorbed protein, was removed by centrifugation at 5 000 g for 20 min at 4°C. Excess barium was precipitated from the supematant by adding 9.5 g/l of (NH,), SO4 followed by centrifugation as above. Phenylmethyl sulfonyl fluoride (PMSF) and benzamidine HCl were added to final concentrations, 1 and 3 mM, respectively. At this point the plasma was passed sequentially over columns of underivatized Sepharose 4B, lysine Sepharose, and gelatin Sepharose. All columns were equilibrated with phosphate-buffered saline (PBS), benzamidine-HCl (3 mM) and NaNs(O.05 %). The gelatin column was washed successively with PBS plus 1 M NaCl, and PBS plus 1 M urea. Plasma fibronectin (Fn) was eluted with PBS plus 4 M urea. The 4 M urea eluate was immediately dialyzed into 50 mM Tris HCl pH 7.5 (1 m PMSF and 0.05 % NaN& The dialysed urea eluate was then passed over a column of arginine Sepharose equilibrated with 50 mM ‘Itis-HCl, pH 7.5. Bound tibronectin was eluted with 0.1 M NaCl in 50 mM Tris-HCl, pH 7.5. Fibronectin was stored in plastic microfuge tubes at 1.0-3.0 mg/ml at -20°C. Fibronectin prepared in this manner was judged 95% pure by SDS-PAGE.

Radiolabelling Procedure Radiolabelled tibronectin (3H-Fn) and Bovine Serum Albumin OH-BSA) were prepared by reductive alkylation [17]. Retention of adhesion-promoting activity of 3H-Fn preparations was determined E.xp Cell Res 152 (1984)

304 Schwarz and Juliano in adhesion assays as described previously [13, 14, 181. All preparations used were fully active as compared with equivalent amounts of unlabelled bovine plasma Fn.

Preparation of Fn-coated Latex Particles Polystyrene latex particles (Polysciences, 0.75 urn 0) were coated with 3H-Fn as follows: 1 ml of latex beads (1x10” beads/ml) was suspended in cold phosphate-buffered saline pH 7.4 (PBS) and centrifuged at 1OOOO g for 5 min in a Beckman microfuge. Routinely 0.200 ml of 3H-Fn (lOO-200 ILL), 0.200 ml of unlabelled Fn (100-200 ug), and 0.100 ml of washed latex beads (1~10’~ beads) were incubated for 3060 min at room temperature. Following centrifugation and washing, the 3H-Fn-bead pellet was resuspended in 0.400 ml of PBS plus 1% antibiotics and stored at 4°C. Alternatively, it was found that ‘H-Fn beads could be stored in 50 % glycerol in PBS at -20°C with no appreciable loss of biological activity. This procedure for preparing Fn-coated beads resulted in the saturation of the bead surface with ‘H-Fn, with excess material remaining in the supematant. We estimate that each bead was coated with approx. 12500 molecules of Fn.

Bead-Binding Assay To 1x10’ beads (briefly sonicated immediately prior to assay) in 0.300 ml PBS+1 % BSA were added various numbers (O-10’) CHO cells in 0.300 ml PBS+ 1% BSA. After 60-90 min the incubation mixture was layered onto a 2 ml cushion of 20% sucrose in PBS and centrifuged at 1800 g for 3 min. After removing non-cell-associated beads at the 20 % interface by aspiration, the resulting cell pellet was solubilized with 1 ml of 1% SDS in HzO. Cell-associated radioactivity was determined by mixing SDS-solubilized cell pellets with 16 ml of Aquasol- (New England Nuclear) and counting in a scintillation counter. Preliminary experiments with either labelled Fn beads or labelled cells demonstrated that this procedure gave essentially 100% recovery of cells in the pellet, that 95 % of non-cellassociated beads (either as individual beads or as aggregates) remained at the 20 % sucrose interface and that cell-associated beads were not stripped from the cells during centrifugation.

Effects of Metabolic or Cytoskeletal Inhibitors To determine the role of active cell processes in bead binding, cells were pretreated for 10 min at room temperature with the following inhibitors: cytochalasin B (10 @ml), n-dansylcadaverine (3X lO-4 M), or tetracaine (lo-‘-lo-’ M), prior to incubation with ‘H-Fn beads. In all cases, cell viability following drug treatment exceeded 9O%, as determined by trypan blue dye exclusion. The effects of varying divalent cation concentration, and effects of fixation with formaldehyde and glutaraldehyde were also studied.

Binding Inhibition Experiments To test the ability of isolated glycolipids (Supelco) and/or glycosaminoglycans (GAGS) (generous gift of Drs Cifonelli and Mathews, University of Chicago) to block Fn bead-binding to CHO cells, the following experimental protocol was employed. Fn beads (5x10s) were preincubated with varying amounts of either glycolipids or GAGS in PBS+1 % BSA for 30 min at room temperature. Subsequently, 0.3 ml of CHO cells (5~ 10”) were added to each tube and the incubation was continued for 60-90 mitt and then the samples processed as described above. In other experiments, Fn beads pretreated with GAGS were centrifuged and resuspended in GAG-free PBS plus 1% BSA, prior to incubation with cells.

Analysis of Gangliosides in Control and Neuraminidase Treated CHO Cells CHO cells were washed by centrifugation and resuspended in cold phosphate-buffered saline (PBS). Cells were treated with V. cholerae neuraminidase as described by Markwell et al. [19]. Briefly, washed cells were resuspended in 10 mM Tris-Cl 150 mM NaCl, pH 7.5, supplemented with bovine serum albumin (BSA) (10 mg/ml) and CaC12 (1 mM). Cells were then incubated with neuraminidase at varying concentrations for 2 h at 37°C. After washing twice with cold PBS, glycolipids were extracted according to the method of Stanley et al. [20]. The cell pellet was resuspended in 10 ml of 10 mM Tris-Cl, pH 7.4 (cell concentration was approx. 2x 10’ per ml). After centrifugation, the cell pellet was extracted with 20 vol chloroform : methanol (CHClJMEOH) (2 : 1) Exp Ceil Res 152 (1984)

Znteraction of fibronectin-coated

beads with CHO cells 305

Fig. 1. Kinetics of ‘H-Fn bead-binding. 3H-Fn beads (1 x lo9 beads) were incubated at the indicated temperatures 0,4; 0, 24; A, 37°C with 5x lo6 CHO cells (0.3 ml total volume). At intervals samples were layered onto a cushion of 20% sucrose and centrifuged at 1800 g for 5 min and the bead radioactivity associated with the cell pellet was determined. Data represented is the mean of triplicate determinations. (lOO%, bead binding at 25”C, 60 min.) Fig. 2. Effects of glycosaminoglycans (GAGS) upon 3H-Fn-bead-cell binding 3H-Fn beads (1 x 109 were preincubated with l , heparin; 0, heparan sulfate; A, dermatan sulfate; A, chondroitin-6sulfate; Cl, dextran sulfate for 30 min prior to addition of SX lo6 CHO cells. Bead binding assays were then performed as described in Methods.

and then filtered through a sintered glass filter. Following re-extraction of the residue with 20 ml CHC13: Methanol (1: 4), the filtrates were dried under nitrogen. The extract was redissolved in 1 ml of 2: 1 CHC13/MEOH and the volume reduced to 0.10-0.20 ml. Thin layer chromatograms of extracted glycolipids were developed by employing a solvent system CHCl,/MEOH 2.5 N NH3 (55 : 35 : 8) (v/v). Routinely 0.25 ml samples (5x 10’ cell equivalents) were applied to 10x20 cm silica gel G plates (Analtech). Total glycolipids were stained with alpha-naphthol. In some cases gangliosides were visualized with VIS ganglioside stain (Supelco).

Protease Sensitivity of Fn Bead-Binding to CHO Ceils The protease sensitivity of CHO cell surface sites necessary for Fn bead-binding was determined in the following manner: 1-2x 10’ cells in 2 ml were incubated with varying concentrations of proteolytic enzymes for 10 min at room temperature. Proteolysis was terminated by addition of cold PBS plus 10 mg/ml bovine serum albumin or the specific enzyme inhibitor (e.g. 10 mM EDTA, in the case of thermolysin). Cells were then washed by centrifugation and were suspended at 1.SX 10’ cells/ml in PBS+10 mg/ml BSA. Cells pretreated with appropriately inactivated proteases served as controls. Cell viability was greater than 90% in these experiments and cell recovery (as determined by cells labelled with [‘4C]leucine) was greater than 95 %.

RESULTS Fn Bead-to-Cell Binding, Basic Characteristics As seen in fig. 1, the rate of binding of Fn beads to CHO cells in suspension is only modestly affected by temperature. Although binding is somewhat slower, experiments at 4°C result in similar total accumulation of beads as experiments at 24” or 37°C. Microscopic observations indicate that CHO cells internalize some of the beads by 90 min at 37°C but not during incubation at 4 or 24°C. Bead-to-cell Exp Cell Res 152 (1984)

306 Schwarz and Juliano Table 1. Binding of latex beads coated with 3H-Fn or 3H-BSA to CHO cells % input cpm in cell pellet Cell no.

‘H-Fn beads

‘H-BSA beads

1x10’ 5x106 2.5x 106 1x106 5x105 1x 10’ cell + anti-fibronectin antisera

45.2k8.7 35.3t2.9 27.6k4.6 15.3k2.5 2.6kO.4 3.551.2

1.41+0.11 1.75+0.11 1.78kO.17 1.71+0.16 1.66kO.30 1.8kO.57

CHO cells were incubated with 1x lo9 latex beads coated with ‘H-Fn or 3H-BSA for 30 min at 4°C. Bead binding was determined as described in Methods. Specific activities of fibronectin and BSA were 4 200 cpm/pg and 10320 cpmlpg, respectively.

binding is linear with bead number up to a ratio of 40 beads per cell and thereafter begins to saturate (data not shown). It is not possible to obtain an estimate of the number of Fn-binding sites from these studies since (a) the beads are polyvalent and it is not clear how many binding sites are necessary to stabilize a single bead-cell interaction; (b) a certain amount of bead-bead association occurs. The binding of Fn beads to CHO cells increases monotonically with cell number and is biologically specific in that beads coated with control proteins, such as albumin or gelatin, did not bind to cells. Treatment of Fn beads with anti-Fn serum, but not with pre-immune serum, inhibited bead-cell binding; neutralization of the antiserum with excess Fn prevented the inhibition (table 1). It should be noted that some aggregation of Fn beads is produced by the antiserum; thus both blocking of sites on the Fn molecule and depletion of beads by aggregate formation may contribute to the observed inhibition. The attachment of Fn beads to CHO cells probably reflects an association between Fn on the bead and surface ‘receptors’ or binding sites for Fn on the cell. This association, in contrast to the full process of cell adhesion, apparently does not require cellular metabolic or cytoskeletal processes. A number of agents which completely block CHO cell adhesion, failed to block Fn bead-cell attachment (table 2). Cytoskeletal inhibitors such as cytochalasin B, local anesthetics and dansylcadaverine are without effect. Fn bead binding does not require exogenous divalent metals, as does cell adhesion [13]. In addition, even pretixation of the cells with paraformaldehyde or glutaraldehyde did not block Fnbead-cell binding. Agents which block Fn-bead-cell binding We have determined that a number of macromolecular ligands can interfere with Fn-bead-cell binding. For example as shown in fg 2, glycosamino@ycans (GAGS) including heparin and to a lesser extent, heparan sulfate and dermatan Exp CeNRes I52 (1984)

Interaction of fibronectin-coated

beads with CHO cells

307

Table 2. Effect of adhesion modulating agents upon Fn bead binding to CHO cells

Cytochalasin B m-Dansylcadaverine Tetracaine Glutaraldehyde Paraformaldehyde Divalent cations”

Treatment concentration

Fn bead binding (% untreated control)

10 ug/ml 50 uM 10-3M 1% 1%

116k15.7 97520.1 85.7k3.01 70.6kO.36 91.2k3.07 100

CHO cells were incubated with the indicated drug in PBS for 10 min before addition of 1x lo9 3H Fn beads. In the cases of fixation with ghrtaraldehyde and pamformaldehyde, cells were incubated with fixative in PBS for 10 min. After washing with in PBS then 0.1 M glycine in PBS, cells were resuspended in PBS plus 1% bovine serum albumin. Fn bead-binding to treated samples and untreated controls was determined as described above. Data represents the mean and standard errors of triplicate determinations. a CaClz or MgC12, O-10 mM.

sulfate, can inhibit Fn-beadseli binding. The order of potency in these studies is the same as that obtained by direct studies on GAG-fibronectin association [21] and is probably a reflection of the existence on Fn of specific binding sites for GAGS. The association between Fn and heparin is quite stable and the inhibitory effect on bead-cell binding is maintained even when the heparin-treated Fn beads are washed to remove excess free heparin. Thus the concentration of heparin

Fig. 3. Effect of glycolipids on ‘H-Fn-bead-binding. ‘H-Fn beads were pretreated with l , GT,; 0, GD,,; A, mixed brain gangliosides; or A, galactocerebroside for 30 min and then incubated with CHO cells for a further 60 min and bead-binding determinated as described in Methods. Glycolipids in chloroform : methanol (1: 1) were concentrated by rotary evaporation and dissolved in PBS + 1% BSA at 1 mg/ml. Fig. 5. Sensitivity of ‘H-Fn-bead binding to proteolytic digestion of CHO cells. The cells (2x 10’) were incubated with 0, trypsin; 0, thermolysin; Cl, papain; A, plasmin; or A, thrombin for 10 min at room temperature. Cell viabilities (determined via trypan blue dye exclusion) at all enzyme concentrations were at least 90%. Excess enzyme was removed by washing and the binding of lo9 ‘H-Fn bead to 5x 106cells was assessed as described in Methods. Exp Cell Res 152 (1984)

308

Schwarz and Juliano

a

b

Fig. 4. Thin-layer chromatogram of lipids of

GM3

neuraminidase-treated and untreated CHO cells. CHO cells were treated with V. cholerae neuraminidase, as described in Materials and Methods. Glycolipid fractions were extracted as described by Stanley et al. [20]. 25 pl samples (5x10’ cell equivalents) were applied in the case of cell extracts. (a) buffer-treated cells; (b) cells incubated 100 units V. cholerue neuraminidase. It should be noted that alpha naphthol stains neutral glycolipids as well as gangliosides, but these can be distinguished on the basis of color. In agreement with Stanley et al. [20] the only detectable ganglioside in CHO cells was GM3. Other bands visible in the TLC represent non-sialic acid contained lipids and stain brown, as opposed to blue for the GM3.

required for 50% inhibition of bead binding is 0.08 &ml when the heparin is coincubated with beads and cells and 5 &ml when the beads are pre-incubated with heparin and then washed. The inhibition of Fn-bead-cell binding caused by coincubation with heparin is accompanied by considerable bead-bead aggregation; however, Fn beads pretreated with heparin and then washed do not aggregate appreciably, but nonetheless fail to bind to cells. It has been reported previously [ 11, 121that sialoglycolipids can block fibronectin-mediated adhesion. In like manner, such gangliosides can also block Fnbead-cell binding (fig. 3) with highly sialylated compounds, such as GTl and GDla being most effective. In contrast to the case of heparin, gangliosides did not cause obvious aggregation of Fn beads. Despite these observations it seems unlikely that sialoglycolipids are the endogenous ‘receptors’ for Fn. Thus one can extensively treat CHO cells with neuraminidase under conditions which destroy most of the total cell ganglioside (fig. 4) without any effect on Fn-bead-cell binding (table 3) nor on cell adhesion to Fn-coated substrata (data not shown). Treatment of cells with neuraminidase did not enhance non-specific binding of ‘H-BSA-coated beads to cells and thus a simple reduction in surface charge cannot account for the maintenance of bead binding in enzyme-treated cells. The Erp Cell Res 152 (1984)

Interaction offibronectin-coated

beads with CHO cells 309

Table 3. Effect of neuraminidase treatment upon ‘H Fn bead-binding to CHO cells Neuraminidase” w 100 50

10 1 0 load

cpm ‘H-Fn in cell pellet’ 8 553fl 643 8 973+403

10 933+630 10 4615813 10 671+2 325

Decrease in GM3’ Ganglioside 82% 88%

-

456+64

LI 2x 10’ CHO cells in 5 ml were incubated with V. cholera neuraminidase, as described by Markwell et al. [19]. ’ Fn-bead-binding assays to control and neuraminidase-treated cells were performed as described above. Data are means of triplicate determinations. ’ Analysis of gangliosides was performed as described in Methods. GM3 ganglioside content in control and neuraminidase-treated cells was determined by densitometric measurement of the plates. d Pre-incubation of 3H-Fn beads with anti-Fn antiserum.

only detectable ganglioside in CHO cells is GM3 [20] (fig. 4) and 80-90% of this material is cleaved by the neuraminidase treatment applied. Higher gangliosides, if present at all, must be very scarce; moreover, Markwell et al. [19] have shown that the neuraminidase treatment used here cleaves most of the higher gangliosides in cells, where they are present. Thus one can markedly reduce the amount of CHO cell sialoglycolipid without interfering with Fn bead-binding. It should be noted that neuraminidase also removes a substantial fraction of the sialic acid residues of CHO cell surface glycoproteins [22]; however, it is unlikely that these residues are involved in adhesion [ 141. Treatment of cells with proteolytic enzymes such as papain or thermolysin can ablate the ability of the cells to bind Fn beads (fig. 5). The cells remain viable after enzyme treatment, and pretreated cultures eventually (overnight) regain bead-binding capacity. The inhibition requires active enzymes, since treatment of the protease with an appropriate inhibitor (e.g., EGTA for thermolysin or iodoacetate for papain) can totally prevent the effect. Thus the cell surface-binding sites for Fn are protease-sensitive, indicating that glycoproteins or possibly proteoglycans are involved. Our results on protease sensitivity qualitatively agree with those of Grinnell [lo] in the BHK cell system; however, we find that the cell surface components involved in Fn bead-binding are relatively insensitive to trypsin as compared with thermolysin, papain or ‘pronase (pronase data not shown). Proteolytic treatment may either destroy cell surface-binding sites for Fn beads or, alternatively, release them into the medium. In preliminary experiments (table 4) we have been able to block Fn-bead-cell binding with cleavage products liberated from cell surfaces by thermolysin or papain. Thus protease treatment can liberate factors which may be fragments of cell surface glycoproExp Cell Res 152 (IYtWJ

310 Schwarz and Juliano Table 4. Inhibitory effects of cell surface cleavage fragments of binding on Fnbeads to CHO cells Inhibitor

% input beads bound vs control

None (control) Thermolysin digest of cells Inactivated thermolysin Papain digest of cells Inactivated papain

100 21.5 95.1 19.3 91.7

CHO cells (2x lO*/ml) were incubated with thermolysin or papain at 100 @ml enzyme in PBS at 37°C for 30 min. After an initial centrifugation at 2 000 g to remove cells the enzymes were blocked by the addition of inhibitors and the hydrolysate supematant was further centrifuged at 40000 g for 30 min. The enzyme-inactivated cell hydrolysates were then tested for their ability to block Fn-bead-cell binding; equivalent amounts of inactivated enzyme was used as a control (thermolysin was inactivated with 5 mM EDTA, papain with 1 mg/ml iodoacetic acid). To test bead-binding, 1x 10’ CHO cells were incubated with 5 x 10’ Fn beads plus 1x IO* cell equivalents of hydrolysate or a comparable amount of inactivated enzyme for 30 min at 4°C. Bead-cell binding was determined as described in Methods. Results represent the means of triplicate experiments differing by less than 10%.

teins (or perhaps proteoglycans) that retain the ability to interact with Fn. The fractionation and biochemical characterization of these factors awaits future investigations. DISCUSSION In this communication we have studied the initial interaction of CHO cells with fibronectin (Fn). To discriminate CHO cell-Fn binding from the more complex secondary events in the adhesion process, we have examined Fn-bead-binding under conditions which have been shown to either block or modulate adhesions to Fir-coated culture dishes [13, 141. As summarized in table 1, Fn-bead-binding was not dependent upon temperature or the presence of divalent cations. Incubation of CHO cells with inhibitors or modulators of cytoskeletal activity (which alter Fn-mediated cell adhesion), and even mild fixation, had no effect on beadbinding. Considering these results, the binding of Fn beads to CHO cells seems to represent an early step in the adhesion process which is independent of subsequent cytoskeletal or metabolic activities. The events we describe may resemble the early, divalent cation independent, cell-cell aggregation events reported in a number of other systems [24, 251. The contention that glycoiipids, specifically gangliosides, maybe the ‘cell surface receptors’ for tibronectin has been a topic of controversy. Kleinman et al. 11I] and Yamada et al. 1121were able to effectively block adhesion of fibroblastic cells to fibronectin substrata which had been preincubated with gangliosides. However, recently it has also been shown that adhesion to highly charged surfaces (e.g., polylysine-coated culture dishes) was similarly blocked by ganglioEXP Cell Res 152 11984)

Interaction of fibronectin-coated

beads with CHO cells

311

side pretreatment via an unknown mechanism [2]. The fact that neuraminidase treatment of cells did not effect Fn-bead-to-cell binding causes us to question further the role of gangliosides in fibronectin-mediated adhesion. The amount of GM3, the predominant CHO cell surface ganglioside, is greatly reduced following neuraminidase treatment (fig. 4) without affecting Fn bead-cell binding; thus although we cannot totally discount the involvement of a minor neuraminidaseresistant glycolipid in the binding of Fn beads, our evidence suggests that gangliosides are not the cell surface-binding site for Fn, although exogenous gangliosides may indirectly inhibit Fn-cell association. The associations of fibronectin with glycosaminoglycans (GAGS) and proteoglycans have been well documented [7, 231. We have demonstrated that Fn-beadbinding can be blocked by heparin, heparan sulfate, and dermatan sulfate, but not by hyaluronate, chondroitin-4-sulfate or chondroitin-6-sulfate. Our results are consistent with the findings of Laterra et al. [7], who observed the ability of similar classes of GAGS to bind to fibronectin aftinity matrices. Structural analysis of cleavage fragments prepared by limited proteolysis of fibronectin has revealed two distinct heparin-binding regions [26], one of which is in close proximity to the domain purported to mediate cell attachment. Therefore, the inhibition of Fn-bead-binding by heparin which we have observed may result from (1) steric blocking of the cell attachment domain by tibronectin-bound heparin; or (2) a heparin-induced conformational change in one domain which could modulate the cell attachment activity of another. Alternatively, a third possibility might be that Fn beads actually bind to cell surface glycosaminoglycan chains and the action of hepatin we observe may be a simple competitive inhibition for this binding. It should be noted that Fn bead-cell-binding is far more sensitive to inhibition with heparin than is Fn-mediated cell adhesion [27] (M. Schwarz, unpublished observations) but the reasons for this are unclear. Perhaps the conformation of Fn absorbed to latex beads is different from that assumed when the molecule is bound to tissue culture plastic and the cell binding domain and adjacent heparin binding domain may be in closer proximity when Fn is adsorbed on a latex surface. We have examined the relative sensitivity of the CHO cell-Fn interaction to proteolytic digestion. The inability of low doses of trypsin (10 ug/ml) to block Fnbead-binding correlates well with the trypsin insensitivity of CHO cell adhesion to Fn-coated culture dishes observed by Harper & Juliano [18, 281. In a similar vein Tarrone et al. [9] have recently suggested the role of trypsin-insensitive cell surface proteins in tibronectin-mediated adhesion of baby hamster kidney (BHK) cells. The inhibition of Fn-bead-binding following incubation of CHO cells with thermolysin, pronase, and papain which we have observed (fig. 5) is in close agreement with the findings of Tarrone et al. [9]. Since proteolytic digestion causes loss of fibronectin-binding components from the cell surface, one might suspect that these components are released into the medium; thus one might be able to block bead-binding by preincubation of Fn beads with proteolytic lysates. Exp Cell Res 152 (1984)

312 Schwarz and Juliano In preliminary experiments we have been able to block Fn-bead-binding with cleavage fragments prepared by thermolysin and pronase treatments (table 4). We can summarize our findings as follows. (1) Binding of Fn beads represents an ‘early recognition’ step in cell adhesion which is not dependent upon temperature, divalent cations, or active cytoskeletal elements; this provides an attractive model to pursue the biochemistry of the cell surface-binding sites for Fn. (2) Although gangliosides and glycosaminoglycans interact with Fn and block the binding of Fn beads to cells, they are unlikely to be the endogenous ‘receptors’ for fibronectin. (3) Rather, CHO cell-Fn interactions appear to be mediated by protease-sensitive surface components, which are most likely glycoproteins. This

work

was supported by grant NIH-GM 26165 and NSF PCM 7905752 to R. L. J.

REFERENCES 1. Aplin, J D, Hughes, R C, J&e, C L & Sharon, N, Exp cell res 134 (1981) 488. 2. Carter, W G, J biol them 253 (1982) 3249. 3. Damsky, C H, Knudson, K A & Buck, C A, J cell biochem 18 (1982) 1. 4. Pena, S D J & Hughes, R C, Nature 276 (1978) 80. 5. Yamada, K & Olden, K, Nature 275 (1979) 179. 6. Gulp, L A & Domen, C, Arch biochem biophys 213 (1982) 726. 7. Laterra, J, Ansbacher, R & Gulp, L A, Proc natl acad sci US 77 (1980) 6662. 8. Perkins, M E, Ji, T H 8z Hynes, R 0, Cell 16 (1979) 941. 9. Tarrone, G, Galetto, G, Prat, M & Comoglio, P M, J cell biol 94 (1982) 179. 10. Grinnell, F, J cell biol 86 (1980) 104. 11. Kleinman, H K, Martin, G R & Fishman, P H, Proc natl acad sci US 76 (1979) 3367. 12. Yamada, K M, Kennedy, D W, Grotendorst, G R & Momoi, T, J cell physiol 109 (1981) 343. 13. Juliano, R L 8~ Gagalang, E, J cell physio192 (1977) 209. 14. Juliano, R L, J cell biol 76 (1978) 43. 15. McDonald, J A L Kelley, D G, J biol them 255 (1980) 8848. 16. Vuento, M & Vaheri, A, J biochem 183 (1979) 331. 17. Rice, R H & Means, G E, J biol them 246 (1971) 831. 18. Harper, PA & Juliano, R L, Nature 290 (1981) 136. 19. Markwell, M A K, Svennerholm, L & Paulson, J C, Proc natl acad sci US 78 (1981) 5406. 20. Stanley, P, Sudo, T & Carver, J P, J cell biol 85 (1980) 60. 21. Laterra, J & Gulp, L A, J biol them 257 (1982) 719. 22. Juliano, R L & Behar-Bannelier, M, Biochemistry 114 (1975) 3816. 23. Gulp, L A, Current top memb transport 11 (1978) 327. 24. Brackenbury R, Rutishauser, U 62 Edelman, G M, Proc natl acad sci US 78 (1981) 387. 25. Thomas, WA & Steinberg, M S, Dev biol 81 (1981) 106. 26. Ruoslahti, E, Haymon, E G, Engvall, E, Cothran, W C & Butler, W J, J biol them 256 (1981) 7277. 27. Klebe, R J & Mock, J, J cell physiol 112 (1982) 5. 28. Harper, P A & Juliano, R L, J cell biol 87 (1980) 755. Received September 19, 1983 Revised version received December 28, 1983

Exp Cell Res 152 (1984)

Printed

in Sweden