Initiation of HeLa cell adhesion to collagen is dependent upon collagen receptor upregulation, segregation to the basal plasma membrane, clustering and binding to the cytoskeleton

Adhesion of cells to an extracellular matrix (ECM) is intimately involved in cell migration, metastasis and tissue development (e.g. see Burridge et al., 1988; Buck and Horwitz, 1987; Juliano, 1987; Thompson-Pletscher, 1986). An in vitro model of these normal and aberrant functions is the adhesion of cells to culture dishes coated with extracellular matrix components. The general events that are thought to occur during adhesion of cells in vitro are: first, an initial attachment phase; second, an intermediate phase where the cells spread or migrate by forming adhesion zones between the substratum and the cell; and third, a final phase where the cells either detach or the cell-substratum adhesion zones become more elaborate or are remodeled by incorporating existing or newly synthesized materials (cf. LeBaron et al., 1988; Singer et al., 1987, 1988; and reviews by Buck and Horwitz, 1987; Juliano, 1987; Burridge et al., 1988; Lark et al., 1985; Geiger, 1983; Geiger et al., 1984; Rollins et al., 1982; Culp et al., 1986; Rees et al., 1977).

Our aim has been to focus on the function of HeLa cell collagen receptors in the first two phases of cell adhesion, i.e. cell attachment and cell spreading. We have previously shown that HeLa cells have three Arg-Gly-Asp-dependent receptors for collagen that mediate cell-substratum adhesion but do not appear to be in the integrin family of ECM receptors (Lu et al., 1989; Beacham and Jacobson, 1990). The receptors are designated by their molecular mass and that of any prevalent proteolytic fragments (102/58 kDa, 87 kDa and 38/33 kDa). These cells are less complex than most other cells used to study cell adhesion in that they do not make an extracellular matrix, which is thought to be involved in the formation of the more complex adhesion zones referred to as focal contacts and fibronexuses (Izzard and Lochner, 1976; Singer, 1982; Singer et al., 1987; Couchman et al., 1983; Fairman and Jacobson, 1983; LeBaron et al., 1988). This is an important aspect since it is difficult to separate, in time, the formation of the complex adhesion zones from the attachment and spreading events. Therefore, since HeLa cells do not form complex adhesion zones, we are able to dissect the sequence of events that take place during attachment and spreading.

The objectives of the work reported here were to determine (1) whether the HeLa cell collagen receptors act cooperatively to mediate firm cell attachment; (2) whether the collagen receptors segregate to the basal plasma membrane (PM) by diffusion from the apical PM and/or are upregulated by exocytotic membrane flow from an internal pool of PM vesicles; (3) whether HeLa cell spreading on collagen is a cooperative process, and whether such cooperativity is due to clustering of the collagen receptors; and (4) whether clustering collagen receptors induces them to associate with the cytoskeleton to initiate cell spreading.

Suspension cultures of HeLa-S3 (ATCC) cells were grown to mid-log phase (3–5 ×10⁵ cells/ml) at 37°C in a humidified 5% CO₂ incubator in RPM 1640 medium (K.C. Biologicals, Lenexa, KS) supplemented with 5% calf serum (Gibco, Grand Island, NY), 0.3% NaHCCb, 100 μg/ml dihydrostreptomycin, 60 μg/ml penicillin and 0.002% butyl parahydroxybenzoate (Sigma, St. Louis, MO). If the cells were grown past 6–7 ×10⁵ cells/ml they began to clump and while they attached to gelatin-coated culture dishes the percent of cells that spread was much less than cells grown from 3–5 ×10⁵/ml.

Type I gelatin from swine skin (Sigma, St. Louis, MO) or rat tail collagen isolated by the method of Chandrakasan et al. (1976) was coupled to 35 mm polystyrene tissue culture dishes (Nunc, Kamstruck, Denmark) as previously described (Jacobson and Ryan, 1982) at a concentration such that the surface of the culture dishes was saturated with protein unless otherwise stated. The dishes were sulfonated with fresh, concentrated reagent grade sulfuric acid at 70°C. Rat tail collagen or 45°C denatured collagen at 1–2 mg/ml in 4 mM pyridine HC1, pH 4.5, was coupled to the sulfonate group with l-ethyl-3-(3-dimethylaminopropyl)carbodiimide (Sigma, St. Louis, MO). Unless otherwise stated the coupling was done at room temperature. The gelatinized dishes were then washed with 1 M NaCl to remove non-specifically absorbed collagen, followed by rinsing with distilled H₂O and were stored desiccated at room temperature. To determine the amount of gelatin bound to the culture dish, radioactive iodine ¹²⁵I-labeled gelatin was used. The ¹²⁵I-gelatin (5×10⁷ cts min ⁻¹/mg) was covalently bound to the dishes as above, and the amount of label bound was determined with a gamma counter. It should be noted that culture dishes (35 mm) from Falcon did not work as reliably as those from Nunc.

Cells used for attachment or spreading studies were harvested from suspension culture (3–5 ×10⁵ cells/ml) by centrifuging at 300 g and washed with 37°C pre-warmed RPMI-1640 buffered with 20 mM HEPES, pH 7.2. The cells were resuspended in the buffered RPMI-1640 to give a concentration of 2–5 ×10⁵ cells/ml, then plated on collagen-coated culture dishes preequilibrated at 37°C prior to scoring for per cent cell attachment or spreading. In some experiments, where indicated, the cells were labeled with [³H]thymidine and the number of attached cells determined by the amount of trichloroacetic acid-precipitable counts associated with the culture dish.

Polyclonal antibodies were raised in Balb/c mice as previously described (Lu et al., 1989). 100 μg of column affinity-purified collagen binding proteins or 25 μg of the individual collagen receptors were electroeluted from SDS-polyacrylamide gels and emulsified in 0.1 ml of 50% (v/v) Freund’s complete adjuvant (Sigma). Two booster shots were given to each mouse at one week intervals two weeks after the first immunization. Hyperimmune ascites fluid was collected from mice after injecting them with 5×10⁷ mouse sarcoma cells three days after the third booster shot.

Detergent extraction of HeLa cells to separate cytoskeletal-associated proteins from non-associated ones followed the procedure previously described by Patton et al. (1989) with some modifications. HeLa cells in suspension or on culture dishes were extracted with extraction buffer containing 1% Triton X-100 in PBS, with 1 mM CaCl₂, 1 mM MgCl₂, 2 mM phenylmethylsulfonyl floride, 10 μg/ml leupeptin, 10 μg/ml pepstatin. The extractions were carried out on ice for 3 min. When cells were in suspension, the cell extraction and detergent non-soluble fraction were separated by centrifugation. Cells on culture dishes at various stages of spreading were extracted in the same buffer, and the detergent-extractable fraction was collected directly from dishes. The detergent-insoluble fraction was solubilized in hot 1% SDS buffered with 25 mM Tris, pH 7.8, containing protease inhibitors and scraped off from the dish.

The procedures to separate the apical plasma membrane domain exposed to the culture medium from the basal plasma membrane domain adjacent to the culture dish were performed as described by Mason and Jacobson (1985). Briefly, cell mono-layers or cells that were allowed to attach to a gelatin substratum were washed two times with 37°C PBS and once with 20 mM MES, 135 mM NaCl, pH 6.0. The cells were coated with an ice cold 1% colloidal suspension of positively charged silica in the MES-buffered NaCl, washed and overcoated with 1 mg/ml polyacrylic acid (M_r 50,000 Polyscience, Warrington, PA) in the same buffer. After washing away excess polyacrylate, the monolayers were then incubated for 5 min at room temperature in a 50% (v/v) glycerol solution buffered with 2.5 mM imidazole at pH 7.25. All buffers contained 2 mM phenylmethylsulfonyl floride, 10 μg/ml leupeptin, 10 μg/ml pepstatin and 1 mM EDTA. The coated apical plasma membrane fragments and intracellular components were removed by squirting the monolayer with hypotonic buffer forced from a 5 ml syringe equipped with a blunt 18 gauge needle. The buffer was centrifuged for 10 min at 970 g yielding a pellet containing the silica-coated apical plasma membrane fragments and a supernatant containing the cytoplasmic fraction. The cytoplasmic fraction was then centrifuged with a microfuge at 14,500 g for 15 min to obtain a fraction containing large intracellular vesicles termed “the internal plasma membrane.” The basal plasma membrane attached to the culture dish was washed two times with cold PBS and scraped into 2% SDS containing the protease inhibitor cocktail. The “internal PM” fraction was separated from the cell surface PM the same way as that done with the apical PM and the internal membrane fraction, except the entire cell surface for the cells in suspension was coated with colloidal silica and the cells were disrupted in a parr pressure bomb at 600 psi. All fractions were solubilized in the same SDS-protease inhibitor cocktail. The samples were heated to 90°C for 10 min and sonicated with a Branson sonifier (Branson Ultrasonics Co., Danbury, CT) set at 35 W and sonicated 3 times for 5 s durations. The microbeads and other insoluble materials were then removed by microfuging the samples for 5 min.

HeLa cells in suspension or after different times of spreading on gelatin-coated culture dishes were fixed with 3% paraformaldehyde in PBS, pH 7.4, for 30 min on ice. Cells were gently washed three times in PBS then incubated on ice for 5 min in 0.1 M glycine in PBS, pH 7.4, to block residual formaldehyde groups. After one wash with PBS, the cells were incubated on ice for 30 min in PBS containing 1% BSA and mouse primary antibodies to the collagen receptors. Cells were then washed with PBS containing 1% BSA and decorated with FITC-conjugated rabbit anti-mouse secondary antibodies for 30 min on ice, after which the cells were washed extensively with PBS/BSA. The specimen samples were covered by one drop of 3% (w/v) n-propylgallate in 90% (v/v) glycerol.

SDS-polyacrylamide gel electrophoresis (SDS-PAGE) was performed using a 4% stacking gel and an 8% resolving gel in the presence of βmercaptomethanol. Proteins were electro-phoretically transferred from 8% slab gels to nitrocellulose sheets as described by Towbin et al. (1979). For staining with antibodies, the nitrocellulose blots were blocked with 150 ml NaCl, 50 mM Tris, pH 7.5, containing 5% non-fat dry milk (Carnation Co., Los Angeles, CA). The decoration of primary antibodies with alkaline phosphatase-conjugated secondary antibody was completed in the same buffer. The unbound antibody was removed by washing extensively in wash buffer (150 mM NaCl, 50 mM Tris, pH 7.5, containing 1% Triton X-100, 0.5% sodium deoxycholate and 0.1% SDS) and the color was developed by incubating in 5-bromo-4-chloro-3-indoyl phosphate p-toluidin salt (Sigma) as chromogenic enzyme marker (Leary et al., 1983).

HeLa cell adhesion to collagen-coated culture dishes follows the same characteristic spreading morphology (Fig. 1) as that seen with the adhesion of most cells in vitro. Round cells attach to the substratum and spread by continually sending out new lamellipodia and filopodia at the edges of the cell as the spaces between the old lamellipodia and filopodia fill in with protoplasm by a process called webbing (Rajaraman et al.,1974; Knox, 1981). HeLa cells which attached to rat tail type I collagen or gelatin were partially spread by 15 min (Fig, 1, B and E) and fully spread by 60 min (Fig. 1, C and F). The collagen or gelatin is covalently bound to the culture dish since cell attachment is reduced and attached cells do not spread if the collagen or gelatin is only adsorbed to the dishes. Experiments showed that adsorbed ¹²⁵I-labeled gelatin is released from the culture dish during cell adhesion (data not shown). Desorption or removal of fibronectin from focal adhesion sites during fibroblast spreading has also been demonstrated (Avnur and Geiger, 1981; Grinnell, 1986). In addition, adsorbed fibronectin is even removed by antibodies to the surfaces of cells used in spreading assays to determine whether a particular antigen is involved in cell adhesion (Chen et al., 1985). For these reasons all experiments on cell attachment and spreading reported here were done with covalently bound substrata.

The rate of HeLa cell attachment (Fig. 2) and spreading (Fig. 3) on gelatin or collagen is rapid and the percentage of attached cells that spread is high. Generally, 85-95% of the attached cells spread (Fig. 3; cf. Mason and Jacobson, 1985). While attachment is rapid on a positively charged substratum such as polyethylenimine (PEI), cell spreading is significantly slowed (Figs 2 and 3). Low temperature (20°C) also significantly reduced cell attachment (Fig. 2) and inhibited cell spreading (data not shown). Interestingly, while cells attached and spread at 37°C on culture dishes covalently coated with gelatin at room temperature, cells attached but did not spread at 37°C on culture dishes covalently coated with gelatin at 45°C (Fig. 3). It is possible that the gelatin bound to the culture dishes at 45°C further denatures its cell binding conformation as would be expected of gelatin coated on dishes at room temperature. Gelatin molecules above their critical melting temperature of 41°C are in an extended conformation and do not have a tertiary structure as does gelatin below its critical melting temperature. Below 41°C even gelatin has alpha helices, although the helix is not the same as the quaternary, triple-helical structure of collagen (cf. Traub and Piez, 1971).

A convenient way to ascertain if a receptor-ligand interaction might be a cooperative reaction is to determine whether it has a threshold or critical concentration below which the reaction does not take place (Weigel et al., 1978, 1979; Aplin and Hughes, 1981; Oka and Weigel, 1986). We followed this kinetic approach with HeLa cell attachment and spreading on gelatin. Gelatin was used instead of collagen since (1) HeLa cells attach and spread equally well on collagen or gelatin (Figs 1, 2 and 3); (2) HeLa cell collagen receptors bind to both type I collagen and gelatin, and type IV collagen; (3) antibodies to the collagen receptors inhibit cell spreading on gelatin or conversely, substitute for gelatin and facilitate cell spreading (Lu et al., 1989); and (4) it is easier to control the amount and distribution of gelatin bound to the sulfonated polystyrene culture dishes. Collagen aggregates in solutions used to covalently couple it to the culture dishes resulting in an uneven surface distribution.

Cell attachment was followed by metabolically labeling the cells with [³H]thymidine. The percentage of cells attached was determined from the percent of trichloroacetic acid-precipitable counts associated with the culture dish. The percentage of cells attached to dishes with different surface concentrations of gelatin differs depending upon the time of incubation before the unbound or weakly attached cells are washed away. If the percentage of cells attached is determined after 2 or 4 min incubation and plotted against the gelatin concentration on the dish, a sigmoidal relationship is observed (Fig. 4). After 2 min incubation, the concentration of gelatin at which cells begin to attach is between 18 and 20 μg/cm², clearly indicating a threshold effect. The sigmoidal relationship disappears with longer times of incubation (Fig. 4, 10 and 20 min). Additionally, gelatin in solution at 1 mg/ml blocks the attachment, which is consistent with the reasoning that specific cell surface receptors are involved in the attachment (Lu et al., 1989).

The percentage of attached HeLa cells that spread in 60 min was also found to have a threshold concentration of gelatin below which the cells do not spread. The concentration is 19 to 20 μg/cm² of culture dish surface (Fig. 5). Interestingly, when percent cell spreading was measured after 15 min, when the cells were only partially spread, the threshold concentration of gelatin was still 19 to 20 μg/cm² (data not shown). In other words, the cells below the threshold concentration do not even partially spread.

The distribution of the receptors in HeLa cells in suspension and after spreading was determined by fluorescence microscopy using mouse monospecific collagen receptor antibodies and FITC-conjugated goat anti-mouse antibodies. Antibodies to the 102 kDa collagen receptor and its 58 kDa proteolytic fragment (102/58 kDa CR; Lu et al., 1989) show a diffuse pattern of fluorescence when the cells are in suspension (Fig. 6A), and a punctate fluorescence in the basal PM when the cells are fixed and stained after they have spread (Fig. 6B). Antibodies to two other collagen receptors, an 87 kDa protein and a 38 kDa protein and its 33 kDa fragment (38/33 kDa CR; Lu et al., 1989) show similar punctate patterns of fluorescence in spread cells except that the fluorescence is less intense (data not shown). On the other hand, antibodies to a 45 kDa HeLa cell surface collagen binding protein, which does not appear to be involved in cell spreading (Lu et al., 1989), show a non-punctate, diffuse fluorescence staining pattern in the basal PM (Fig. 6C).

To investigate further whether collagen receptors are redistributed among the apical PM adjacent to the culture medium, the basal PM contacting the culture dish, and an internal or intracellular pool of PM vesicles, we took advantage of a recently developed procedure to isolate the three PM domains (Mason and Jacobson, 1985). The goal was to determine whether the relative abundance of the collagen receptors in these three PM fractions changes during cell adhesion. The relative abundance of three collagen receptors designated 102/58 kDa CR, 87 kDa CR and 38/33 kDa CR (Lu et al., 1989) was determined by probing immunoblots of the PM fractions resolved by SDS-PAGE with antibodies to the receptors. For comparison, we also probed the blots with antibodies to a 45 kDa collagen-binding protein (45 kDa CBP) not thought to be involved in HeLa cell adhesion to gelatin (Lu et al., 1989). The relative abundance of the receptors in the immunoblots was quantified with a video densitometer and the results are expressed as the percent of total for each receptor in the PM fractions. In the case of cells in suspension, the percent of total for a receptor is given only for the internal PM fraction with the balance being in the external or cell surface fraction. For cells spreading, the percent of total is given for all three PM fractions, i.e. the internal PM domain, and the apical and basal PM domains that make up the cell surface. The most obvious finding is a decrease in relative abundance of all the receptors in the internal PM fraction during cell adhesion (Fig. 7, bars designated INTERNAL). For example, 45% of the total 102/58 kDa collagen receptor is found in the internal PM fraction when cells are in suspension (open bar) while only 27% for cells spread for 15 min (diagonally hatched bar) and 22% for cells spread for 60 min (horizontally hatched bar). All the collagen receptors increased in the basal PM fraction during cell spreading (Fig. 7, bars designated BASAL) and tended to decrease or remain the same in the apical PM fraction (Fig. 7, bars designated APICAL). In all cases the relative abundance of the collagen receptors was greatest in the basal PM, which accounts for 27% of the total PM in cells spread for 60 min (cf. Mason and Jacobson, 1985; Mason et al., 1987).

It has been hypothesized that cell attachment and spreading might be mediated by the segregation of gelatin receptors to the basal plasma membrane domain where they become clustered and bind to the cytoskeleton (cf. Rees et al., 1977; Grinnell, 1978, 1980; Rubin et al., 1984; Geiger, 1983; Cody and Wicha, 1986; Rapraeger et al., 1986; Jacobson, 1988). To determine whether clustered HeLa cell collagen receptors bind to the cytoskeleton, cytoskeletal preparations with associated PM proteins were made by extracting the cells with a nondenaturing detergent (e.g. see Flanagan and Koch, 1978; Prives et al., 1982; Woods et al., 1986; Cody and Wicha, 1986; Rapraeger et al., 1986; Patton et al., 1989). The intent of the experiments reported below was to determine the baseline level of collagen receptor binding to a detergent-resistant cytoskeletal fraction, and to determine whether larger amounts of receptors co-isolate with the cytoskeletal fraction during cell adhesion.

Nondenaturing detergent-resistant cytoskeletons with associated collagen receptors were prepared from HeLa cells: (1) treated in suspension with polyclonal antibodies raised against collagen receptors; or (2) treated in suspension with 1 mg/ml gelatin; or (3) after the cells were allowed to partially spread (15 min); or (4) after the cells were allowed to fully spread (60 min) on gelatin-coated culture dishes. The cells were then extracted with 1% Triton X-100 in PBS on ice for 20 min. After the supernatants were removed, the cells were extracted once in PBS without the detergent. The supernatants and the insoluble material were solubilized in SDS and resolved by electrophoresis. The relative amounts of collagen receptors in either the supernatant or the cytoskeletal fractions were determined by immunoblotting using mouse anti-collagen receptor antibodies. The relative abundance of the receptors in each fraction was determined by videodensitometry and the percentage in the cytoskeletal fraction is presented in Table 1. In all cases, incubating the cells with polyclonal anti-receptor antibodies or 1 mg/ml gelatin to cluster the receptors, or allowing the cells to partially spread for 15 min or fully spread for 60 min, increases the amount of receptors associated with the cytoskeletal fraction (Table 1). The correlation between the increased receptor-cytoskeletal association due to antibody or gelatin presentation and cell spreading is consistent with the hypothesis that spreading is initiated by receptor-cytoskeletal binding that comes about as a consequence of receptor clustering in the basal PM.

Recently, work in cell attachment has focused on the interaction of specific cell surface receptors with specific ECM components (cf. reviews by Ruoslahti et al., 1985; Buck and Horwitz, 1987; Hynes, 1987) but there is a large body of information indicating that attachment can also be mediated by nonspecific physicochemical forces (e.g. see Maroudas, 1975, 1977; Curtis and McMurray, 1986). HeLa cell attachment is initiated by both nonspecific substrata such as BSA and the specific substrata, gelatin or collagen (Figs 1 and 2). However, the cells do not spread on the nonspecific substrata although they will spread slowly on the positively charged polymer PEI (Fig. 3; Fairman and Jacobson, 1983). Other cells have been shown both to attach and spread on nonspecific substrata, but unlike HeLa cells they are capable of making their own ECM (cf. Damsky et al., 1984; Lark et al., 1985).

Experiments to measure attachment kinetics (Fig. 4) and experiments designed to follow the segregation of receptors to different PM domains during cell adhesion (Fig. 7) indicate that HeLa cell attachment to gelatin involves the upregulation of collagen receptors by exocytosis to the basal PM. The kinetic measurements indicate that HeLa cell attachment is a threshold effect with regard to the surface concentration of gelatin (Fig. 4) and is consistent with previous suggestions (Oka and Weigel, 1986; Weigel et al., 1978) that a minimum number of substratum-receptor interactions must take place before the cells can attach. However, there appears to be more to attachment than just a minimum number of initial receptor-substratum interactions, since with increased times of incubation (10 or 20 min), the threshold effect of attachment is dampened and the cells become more adherent at lower surface concentrations of gelatin (Fig. 4). These results indicate that the initial contacts of the cell with gelatin become more stable over time. Consistent with this interpretation is the observation that, as the cells are allowed to attach to the culture dish for longer times, gelatin in solution becomes less effective at inhibiting cell attachment (Fig. 4). The increased strength of attachment does not appear to be simply a result of the onset of spreading because the decrease in inhibition of cell attachment by gelatin occurs with dishes coated with surface concentrations of gelatin either above or below that which induces cell spreading (Fig. 5). The more firm attachment with longer incubation times at surface concentrations of gelatin below the threshold for cell spreading is consistent with a time-dependent increase in the number of collagen receptors in the basal PM.

Previous work indicates that cell attachment is made more strong by a nonspecific substratum attachment-induced exocytosis of collagen receptors from an intracellular pool of PM. We have shown that HeLa cells in suspension have 55% of their total PM protein in intracellular vesicles as indicated by cell surface iodination. Upon cell attachment to gelatin, which induces cell spreading, or to BSA, which does not induce spreading, there is a stimulation of membrane efflux from an intracellular pool of PM, presumably by exocytosis. During the first 10 min of spreading the intracellular pool decreases to 35% of the total PM. This is followed by a re-endocytosis over the next 20 min, so that when the cells are fully spread on gelatin the internal pool contains 46% of the total PM. 55% of the PM is found in the internal membrane pool of unspread cells attached to BSA (Mason et al., 1987). This information taken together with the results on the redistribution of collagen receptors during cell adhesion (Fig. 7) indicates that the major movement of collagen receptors to the basal PM domain probably occurs by a nonspecific substratum attachment-stimulated exocytosis. It cannot be concluded that the exocytosis of receptors from the intracellular pool of PM vesicles occurred only in the direction of the basal PM and not the apical PM. HeLa cells are like many other cells in that they constantly recycle their plasma membrane between an internal domain and the cell surface (Mellman et al., 1980; Fishman and Cook, 1982; Widnell et al., 1982; Mason et al., 1987). It is possible that the attachment-stimulated upregulation of collagen receptors during HeLa cell adhesion occurred by random exocytosis from the internal PM domain to both the apical PM and the basal PM.

Kinetics of HeLa cell spreading revealed that there is a threshold concentration of gelatin on the culture dish below which the cells will not spread (Fig. 5). This can be interpreted in three ways with regard to the initiation of cell spreading (cf. Weigel et al., 1978, 1979): (1) there must be a minimum number of receptorgelatin interactions required for spreading to occur; (2) gelatin on the culture dish must be present at a sufficiently high concentration where it assumes a unique conformation which only then allows the HeLa cells to spread; or (3) the collagen receptors must be clustered before cell spreading can take place. It is not likely that the number of receptors in the basal PM is the limiting factor in cell spreading. HeLa cells left in contact with dishes coated with low levels of gelatin still do not spread (Fig. 4) even though they increase their strength of attachment, presumably by delivery of receptors to the basal PM (see above). It is also unlikely that a conformational change in the substratum-bound gelatin takes place at the higher surface concentrations of gelatin. Transmission electron microscopy demonstrated that gelatin covalently bound to the surface of polystyrene cell culture microcarriers at surface-saturating levels maintains a random appearance of overlapping fibrous molecules and does not form any detectable supercoiled structures or regions of higher density (Fairman and Jacobson, 1983). Thus, the cooperativity in cell spreading is consistent with the need to cluster receptors.

To further understand whether cooperativity in cell spreading is indicative of receptor clustering, we employed immunofluorescence microscopy using anti bodies to the receptors to determine if visible clusters are formed during cell adhesion. Antibodies to collagen receptors that inhibit cell spreading on gelatin-coated culture dishes and conversely, substitute for the gelatin on the dish and facilitate cell spreading (Lu et al., 1989), exhibit a diffuse fluorescence staining pattern when the cells are in suspension and a punctate pattern in the basal PM of spread cells (Fig. 6). On the other hand, antibodies to a cell surface collagen-binding protein that does not inhibit spreading or substitute for gelatin as the extracellular matrix (Lu et al., 1989) give a diffuse staining pattern of fluorescence in both suspension and spread cells (Fig. 6). Punctate patterns of fluorescence for extracellular matrix receptors that mediate cell-substratum adhesion have been repeatedly shown with fibronectin receptors (c.f. Chen et al., 1985; Damsky et al., 1984; Giancotti et al., 1986), collagen receptors (Mollenhauer et al., 1984) and cell surface proteoglycans (Rapraeger et al., 1986). Furthermore, extensive morphological studies indicate that there is a co-localization of the punctate fluorescence of the receptors and cytoskeletal elements (e.g. Rogalski and Singer, 1985; Chen et al., 1985).

To further explore whether clustering of collagen receptors is essential for HeLa cell spreading, we followed the adhesion of the cells to a positively charged substrate, polyethyleneimine (PEI). The rate of spreading was significantly slowed relative to cell spreading on collagen or gelatin (see Figs 2 and 3). The rationale was that if receptor clustering is a precondition to cell spreading, then cell spreading will be impeded on a substrate that has a high positive charge, which slows the rate of lateral diffusion and therefore, receptor clustering. Positive substrates have been shown to bind so tightly to the plasma membrane that the lateral diffusion of Con A receptors is restrained (Patton et al., 1990). It is likely the PEI substrate nonspecifically binds to the collagen receptors and not necessarily at the “active” collagen binding site. While it cannot be ruled out that the receptors might not be in the proper conformation to initiate cell spreading, it should be noted that monoclonal antibodies that bind to EGF or insulin receptors, at sites other than the hormone binding sites, still induce receptor clustering and endocytosis (cf. Schlessinger et al., 1983; Maron et al., 1984; Forsayeth et al., 1987). Furthermore, endocytosis of membrane proteins is induced by binding to highly cationized ferritin (Simionescu et al., 1981). The above, taken together with the observation that collagen receptors exhibit a punctate pattern of fluorescence in the basal PM of spread cells, and that cell spreading is a threshold function of the gelatin concentration on the culture dish, strongly supports the hypothesis that receptor clustering is a requirement for cell spreading.

The observation that clustering HeLa cell collagen receptors causes them to bind to the cytoskeleton as determined by nondenaturing detergent extraction (Table 1), and experiments indicating that the gelatin substratum clusters collagen receptors (Figs 5 and 6) are consistent with previous work with other cells indicating that receptor-cytoskeletal binding must occur before cell spreading can take place (e.g. see Rees et al., 1977; Grinnell, 1978; Geiger, 1983). However, the mechanism by which the HeLa cell collagen receptors interact with the cytoskeleton is not known. Clustering laminin receptors (Cody and Wicha, 1986) or a transmembrane proteoglycan that is an extracellular matrix receptor (Rapraeger et al., 1986) induce them to bind to a detergent-resistant cytoskeletal fraction. Both the laminin receptor (Brown et al., 1983) and the proteoglycan extracellular matrix receptor (Rapraeger and Bemfield, 1982) bind directly to F-actin in vitro. Fibronectin receptors have been shown to bind to talin in vitro (Horwitz et al., 1986).

It should be emphasized that the measurements of receptor-cytoskeletal binding determined by nondenaturing detergent extraction procedures do not indicate whether the receptors that were associated with the cytoskeleton in the absence of an extracellular ligand or substratum were either unclustered or clustered. The data only indicate that clustered receptors, or making larger clusters from smaller ones, are correlated with an enhancement of receptor association with the cytoskeletal preparations (cf. Brandts and Jacobson, 1983). If the receptors were in the form of dimers, trimers, etc., and exhibited a weak or even a strong affinity for the cytoskeleton, clustering them into larger oligomers would markedly increase the affinity (Brandts and Jacobson, 1983; Shiozawa et al., 1989). For example, one estimate of the enhanced affinity reveals that, depending upon the mole fraction of the receptors in the plane of the membrane, a tetrameric cluster would bind to the cytoskeleton 10¹² times more effectively than four unclustered receptors and an octomer 10²⁷ times (Jacobson, 1988). It should be noted, that these increases in cytoskeletal binding upon receptor clustering are consistent with the cooperativity seen in cell spreading versus the concentration of gelatin on the culture dish (Fig. 5).

Is the clustering and binding of the collagen receptors to the cytoskeleton sufficient to induce cell spreading or are other messages involved? It is tempting to speculate that clustering might induce a second messenger which signals the cells to spread. It has been shown that mutants of Chinese hamster ovary cells that do not adhere to fibronectin have an altered type I protein kinase. Adhesion of these mutants can be induced by the addition of cyclic AMP (Cheung and Juliano, 1985; Cheung et al., 1987). It is also known that one of the subunits of chick fibronectin receptor has a protein kinase C substrate-binding site in its cytoplasmic domain (cf. Burridge et al., 1988). Interestingly, Danilov and Juliano (1989) recently reported that phorbol ester, a protein kinase C activator, does not alter the phosphorylation state of fibronectin receptor or talin, or the number or affinity of cell surface fibronectin receptors, and concluded that the increase in cell adhesion induced by the phorbol ester is not due to a direct effect on the receptors. Work is currently in progress to determine whether a second messenger is involved in signaling HeLa cell spreading. Preliminary experiments indicate that HeLa cells have a dramatic rise in cyclic AMP and a spike of intracellular free calcium during cell spreading; however, both the rise and the spike occur after spreading begins indicating that neither cyclic AMP nor Ca²* are the second messengers initiating cell spreading (Lu, Chun and Jacobson, unpublished observations).

We are deeply indebted to Dr Peter Mason and Deidra Gramas for the micrographs of spreading HeLa cells, and to Mr. Jang-Soo Chun for providing us with his initial observations on second messengers during cell adhesion. This work was supported in part by a grant from the National Institutes of General Medical Sciences, GM 29127.