C-cam (cell-CAM 105) is an adhesive cell surface glycoprotein with homophilic binding properties

Short-range interactions between cells of multicellular organisms are of fundamental importance in embryonic development and for the integrity and proper function of mature tissues. Our knowledge about the structural basis at the molecular level of such interactions has increased profoundly during recent years. It is now clear that distinct cell types possess several different systems for recognition and binding to each other (Obrink, 1986a,b). These systems include morphologically identifiable cellular junctions, which to a large extent have been dissected into molecular constituents. Cell adhesion molecules (CAMs) make another group of components that are central in various types of adhesive cell-cell interactions. Recent progress in the field of CAMs indicates that two major CAM families exist (Edelman, 1988; Takeichi, 1988). One family contains cell surface proteins that participate in CAM binding that does not require calcium ions for proper function. Several members of the calcium-independent family have been shown to belong to the immunoglobulin superfamily. Prominent examples are N-CAM (Edelman, 1988), Ll/Ng-CAM (Moos et al. 1988), F3 (Gennarini et al. 1989), contactin (Ranscht, 1988), MAG (Lai et al. 1987), P_o (Lemke et al. 1988) and ICAM-1 (Staunton et al. 1988). The other family requires calcium ions for binding. The most well-characterized calciumdependent CAMs are the cadherins of which three distinct members, E-cadherin, N-cadherin and P-cadherin, have been cloned and sequenced (Takeichi, 1988). The cadherins do not share structural features with the immunoglobulin superfamily. Several of the CAMs of both the calciumdependent and the calcium-independent families have been demonstrated to participate in homophilic binding reactions, i.e. one individual of one CAM can bind to another individual of the same kind (Edelman, 1988; Hoffman and Edelman, 1983; Nagafuchi et al. 1987; Edelman et al. 1987).

Cell-CAM 105 (or C-CAM) is a cell surface molecule that originally was identified as being involved in intercellular adhesion between isolated mature rat hepatocytes in vitro (Ocklind and Obrink, 1982). Immunochemical and immunohistochemical analyses have demonstrated that C-CAM is not confined to liver but is also expressed in a number of different epithelia, vessel endothelia, platelets and polymorphonuclear granulocytes (Odin and Obrink, 1987; Odin et al. 1988). Some variations in the biochemical properties, primarily detected as differences in the apparent molecular weights, of C-CAM in the different tissues were discovered (Odin et al. 1988). The liver C-CAM is an integral membrane glycoprotein consisting of two structurally similar peptide chains of apparent molecular weights 105000 and 110000, respectively (Odin et al. 1986). Liver C-CAM can be phosphorylated on serine residues (Odin et al. 1986) and has been shown to bind calmodulin in a specific manner (öbrink et al. 1988; Blikstad et al. unpublished data).

The involvement of C-CAM in cell-cell adhesion of hepatocytes was demonstrated by the ability of both monospecific, monovalent antibodies (Fab) (Ocklind and öbrink, 1982) and purified C-CAM itself (Odin et al. 1986) to inhibit hepatocyte aggregation. The elucidation of the mechanism of the involvement of C-CAM in cell adhesion requires different approaches, preferably involving binding studies of the purified protein. In this report we present the results of such studies, which demonstrate that C-CAM can participate in a calcium-independent homophilic binding.

Hepatocytes were isolated from young Sprague-Dawley rats (obtained from the local breed at the animal care department of the Biomedical Center, Uppsala) by collagenase perfiision of the liver as previously described (Obrink, 1982).

Rabbit antisera (sera nos 8 and 9) against purified liver C-CAM were produced as previously described (Odin et al. 1986). The production of IgG, Fab fragments and affinity-purified immunoglobulins from these antisera has been described (Odin et al. 1986). The exclusive reaction of these antibodies with the A- and B-chains of liver C-CAM has been demonstrated by immunoprecipitation and immunoblotting analyses (Odin et al. 1986).

Protein was determined either according to Lowry et al. (1951), or by the procedure of Bradford (1976) adapted for microtiter plates and analysis in a Titertek ELISA-photometer. C-CAM was determined with a specific solid-phase radioimmunoassay (Odin and öbrink, 1987).

C-CAM was purified from rat liver according to Odin et al. (1986). The purity was checked by SDS—PAGE. This procedure yields C-CAM dissolved in Triton X-100. Before radiolabeling or incorporation of C-CAM into liposomes this detergent was exchanged for octyl glucoside (l-O-n-octyl-β-D-glucopyranoside) by adsorbing C-CAM to a column of wheatgerm agglutinin (WGA)-Sepharose, washing the column with phosphate-buffered saline (PBS) containing 1 % octyl glucoside, and eluting the protein with 0.3 M N-acetylglucosamine in PBS containing 2% octylglucoside. C-CAM was labeled with “‘I using lodo-Beads (Pierce Chemical Co.) as previously described (Odin et al. 1986).

Control glycoproteins were isolated by the following procedure. Triton X-100-solubilized plasma membranes, that were used for isolation of C-CAM, were subjected to immunoaffinity chromatography on immobilized antibodies against C-CAM (Odin et al. 1986). The components not binding to the immunoaffinity column were then applied to a column of WGA-Sepharose. After washing the column the adsorbed glycoproteins were recovered by elution with 0.3 M N-acetylglucosamine in PBS containing 1 % octyl glucoside. The resulting glycoproteins were then subjected twice to immunoaffinity chromatography on anti-C-CAM-Sepharose and the non-binding fractions were recovered each time. By radioimmunoassay analysis the resulting glycoprotein fraction was found to contain no detectable C-CAM. The other major cell adhesion molecule in adult hepatocytes is E-cadherin (uvomorulin). However, previous results indicated that E-cadherin is not present in the type of membrane preparation used in the present work (Vestweber et al. 1985). Thus, the C-CAM-depleted glycoproteins should not contain E-cadherin.

The hydrophobic nature of ^12SI-labeled C-CAM was assessed by its relative partitioning into the detergent and aqueous phases of a 1 % solution of Triton X-114 according to Bordier (1981). Samples from the different fractions were analyzed by SDS-PAGE, the gels were dried and exposed to Fuji RX X-ray films with an intensifying screen.

Liposomes were prepared from dioleylphosphatidylcholine (Sigma Chemical Co.). The phospholipid (4 mg) dissolved in chloroform (1 ml) was dried to a thin film in a glass tube under a stream of nitrogen. The lipid was redissolved in diethylether and dried again under nitrogen to remove trace amounts of chloroform. The lipid films were then dissolved in 1ml of PBS, 2% octylglucoside, 0.3 M Macetylglucosamine with or without C-CAM or control glycoproteins, respectively. The lipids were mixed thoroughly with the respective solutions until clear solutions were obtained. The solutions were then dialyzed against PBS (500 ml), which was replaced five times with fresh PBS every 12th hour. Radiolabeled liposomes were prepared by adding [¹⁴C]phosphatidylcholine (The Radiochemical Centre, Amersham, Bucks, UK) to the phosphatidylcholine solution (5000 Bq mg^-1 phosphatidylcholine). In order to prepare fluorescent liposome vesicles 1 mM carboxyfluorescein (Eastman Kodak Company) was included in the solutions added to the lipid films and in the first three 500-ml portions of the dialysis solutions. The carboxyfluorescein-containing liposomes were then dialyzed against five 500-ml portions of PBS without carboxyfluorescein.

In order to test liposome integrity and to separate liposomes from unincorporated protein the liposomes were subjected to flotation in a discontinuous sucrose gradient by centrifugation for 16 h at 40 000 revs min^-1 in a Beckman SW 50.1 rotor. The sucrose gradients were prepared by mixing the liposome preparation (100/d) with 60% (w/w) sucrose (157/d) to give a final sucrose concentration of 40 % (w/w). This solution was overlaid with 30 % (w/w) sucrose (3 ml) and PBS (0.5 ml). In some experiments 0.1 % Triton X-100 was included in the sucrose gradients. After centrifugation the tubes were fractionated from the top of the gradients and samples of the fractions were analyzed for C-CAM, protein and radioactive lipid, respectively.

The aggregation of carboxyfluorescein-containing liposomes was determined qualitatively by fluorescence microscopy. Portions of the liposome samples were placed on glass microscope slides, covered with coverslips and viewed in a Leitz Ortoplan microscope equipped with epi-illumination.

Mixtures of liposomes (25 μl) and hepatocytes (1 × 10⁶ cells) were prepared in a final volume of 500 /d of Medium A (137 mM NaCl, 4.7mM KC1, 0.6mM MgSO₄, 1.2mM CaCl_a, 10mM Hepes, pH7.4) containing glucose (20mgml^-1) and DNase (lmgml^-r). The liposome-hepatocyte mixtures were incubated, in the presence or absence of Fab fragments, in bovine serum albumin (BSA)-coated wells (diameter 16 mm) of Linbro tissue culture plates. The plates were rotated at 130 revs min^-1 on a gyratory shaker for lh at 37 °C. In order to separate unbound liposomes from the cells the incubation mixtures were layered on top of a solution of Percoll (10 ml; 1.033 gml, Pharmacia Fine Chemicals) in a centrifuge tube and were centrifuged at 2000 revs minfor 10 min in a swinging bucket rotor. The cells and the cell-bound liposomes ended up in the pellet, whereas unbound liposomes remained on the top of the Percoll solution. The pellets were analyzed by fluorescence microscopy or liquid scintillation counting.

Proteins (1–5 μl) were spotted on a nitrocellulose filter (Schleicher and Schüll) (0.22 μm). The spots were dried using a hairdryer. Excess binding sites on the filter were blocked by incubating for 30 min at 37 °C in a solution containing 5 % BSA in 0.15 M NaCl, 10 mw Tris-HCl, pH 7.5 (TBS). The filter was then incubated with ¹²⁵I-labeled C-CAM (l×10⁶ctemin^-1ml^-1) in TBS containing 5% BSA at room temperature for 2h. Thereafter, the filter was washed twice with TBS, twice with TBS containing 0.05 % Triton X-100, and twice with TBS. The nitrocellulose filter was exposed to Amersham Hyperfilm MP X-ray film for 18h with an. intensifying screen.

Purified C-CAM (1.3 μg) was electrophoresed in the presence of SDS on 10% polyacrylamide gels (Laemmli, 1970). The proteins were electrophoretically transferred to nitrocellulose sheets as described by Burnette (1981). The nitrocellulose sheets were incubated with 5 % BSA in TBS for Ih at 37 °C and thereafter with ¹²⁵I-labeled C-CAM (10® to 2×10⁶ctsmin^-1) in TBS containing 5% BSA for 5h. The filter was then washed twice with TBS, once with TBS containing 0.01% octylglucoside and twice with TBS. To visualize bound C-CAM the filter was subjected to autoradiography as described above for the dot-blot analysis.

In order to answer the question of whether C-CAM can bind to itself we set up simple binding assays in which radiolabeled C-CAM was challenged to C-CAM and other proteins bound to nitrocellulose filters. In the simplest assay different amounts of proteins were spotted onto the filters, which were then incubated with [^I2BI]C-CAM. As demonstrated in Fig. 1 a significant binding to C-CAM occurred, but no detectable binding was observed to BSA or to ovalbumin (data not shown). In order to ensure that C-CAM bound to C-CAM and not to any other component present in the unlabeled C-CAM preparation that was spotted onto the nitrocellulose, the binding experiment was repeated with C-CAM and control proteins that first had been separated by SDS-PAGE and then transferred to nitrocellulose filters. When such filters were incubated with [¹²⁶I]C-CAM it was found that the binding indeed occurred to C-CAM (Fig. 2). Only one labeled band with an apparent molecular weight of 105000 appeared. Both reduced and non-reduced C-CAM could bind [¹²⁵I]C-CAM. No binding was detected to any of the control proteins, comprising myosin, β-galactosidase, phosphorylase b, BSA, ovalbumin and carbonic anhydrase (Fig. 2, lane 1). The binding to C-CAM did not require calcium ions.

For further characterization of the binding properties of C-CAM it was desirable to expose the protein to viable hepatocytes. However, previous studies (Ocklind and Cbrink, 1982) have shown that C-CAM requires detergent to be kept in solution without aggregating. In order to learn more about the hydrophobicity and solubility properties of C-CAM it was subjected to partitioning between the water and detergent phases of Triton X-114. As demonstrated in Fig. 3 intact ¹²⁶I-labeled C-CAM ended up exclusively in the detergent phase of Triton X-114. This behaviour strongly indicates that C-CAM is an integral, amphiphilic membrane protein. It would thus be possible to use C-CAM incorporated into liposome membranes for binding studies with viable cells.

C-CAM and phosphatidylcholine were reconstituted into liposomes by dialysis of octylglucoside-containing solutions. To monitor the reconstitution the liposomes were subjected to centrifugation in sucrose gradients. In pilot experiments employing ¹²⁵I-labeled C-CAM we found that all of the protein moved to the top of the gradient together with the lipid vesicles. Non-labeled C-CAM reconstituted into liposomes behaved in the same way, and after centrifugation of intact liposomes all the lipid and C-CAM were found in the top fraction of the gradient (Fig. 4). When Triton X-100 was added to the liposomes before centrifugation all the protein and most of the lipid remained in the bottom of the centrifuge tubes (Fig. 4). These results thus demonstrate that C-CAM became efficiently incorporated into non-leaky liposomes.

In order to locate C-CAM in the intact liposomes, the liposomes were subjected to immunofluorescence rni-croscopy employing antibodies against C-CAM. The intact liposomes became strongly fluorescent (data not shown), indicating that C-CAM was exposed on the external surface of the liposome membrane and not just trapped inside the vesicles.

Binding of C-CAM-containing liposomes was investigated by incubating liposomes with hepatocytes in suspension in a balanced salt solution at 37 °C for various times. In order to obtain good binding it was essential to achieve an optimal colliding frequency between the cells and the liposomes without damaging the cells. This was determined by a systematic variation of the rotation rate. It was found that maximal binding was obtained at 130 revs min^-1 and an incubation time of 60 min. Under these conditions we observed an increasing binding of C-CAM-containing liposomes to the hepatocytes with increasing ratios of C-CAM to phospholipid in the reconstituted liposomes (Fig. 5). At the maximal protein-to-lipid ratio that was tested (0.125, w/w) the maximal binding at 60 min incubation time amounted to 30–40 % of the added lipid radioactivity.

In order to obtain further evidence that the liposome binding was mediated by C-CAM, the binding experiments were done in the presence of antibodies against C-CAM. As demonstrated in Fig. 5, monovalent antibodies (Fab fragments) effectively inhibited the liposome-cell binding, whereas Fab fragments from pre-immune serum had no effect. In another set of experiments the hepatocytes were preincubated with intact anti-C-CAM IgG and washed free of non-bound antibodies prior to incubation with liposomes. Also under these conditions a significant and specific inhibition of the C-CAM-containing liposomes was observed (Fig. 6).

The binding of C-CAM-containing liposomes to hepatocytes could also be observed in the microscope when liposomes containing carboxyfluorescein were used. Fluorescent liposomes were primarily seen between hepatocytes that occurred in large aggregates, giving the impression that they acted as bridges between the cells (Fig. 7A). When anti-C-CAM Fab fragments were included in the binding experiments no hepatocyte-associated liposomes were observed (Fig. 7B). Instead the hepatocytes formed aggregates by direct cell-cell binding (Fig. 7B), as expected when the Fab fragments were neutralized by the C-CAM-containing liposomes.

Control experiments were done to investigate the possibility that any kind of membrane glycoprotein reconstituted in liposomes would bind to hepatocytes. Thus binding experiments were performed with liposomes containing various amounts of liver membrane glycoproteins that were depleted of C-CAM. The results of these experiments, which are presented in Fig. 5, demonstrate that no binding above background level occurred, due to incorporation of these glycoproteins in liposomes. Furthermore, neither pre-immune Fab fragments, nor anti-C-CAM Fab fragments affected the background binding of these glycoprotein-containing liposomes.

During the preparation of C-CAM-containing liposomes we observed a marked flocculation of liposomes having high protein-to-lipid ratios. For further examination of this phenomenon we prepared fluorescent liposomes by including carboxyfluorescein during the reconstitution process. The vesicles were then viewed in a fluorescence microscope. Liposomes having a protein-to-lipid ratio of 0.06–0.125 (w/w) formed large aggregates (Fig. 8A) compared to liposomes without C-CAM (Fig. 8C). These aggregates were effectively dissociated by anti-C-CAM Fab fragments (Fig. 9B) but not by pre-immune Fab fragments (Fig. 9A). The liposome aggregates could also be dissociated by increasing the pH to 11.5 (Fig. 10B). After neutralization a rapid reaggregation occurred (Fig. 10 C-E). Anti-C-CAM Fab fragments prevented this reaggregation (data not shown). The effect of salt concentration on the dissociation of C-CAM-containing liposomes was also analyzed. Increasing the salt concentration to 1 M NaCl at neutral pH did not dissociate the liposome aggregates. We could also conclude that calcium ions were not required for C-CAM-mediated liposome aggregation, since aggregation to the same extent occurred when the liposome reconstitution was performed without calcium ions in the presence of the specific calcium chelator EGTA (10 mw).

Control liposomes reconstituted with C-CAM-depleted liver membrane glycoproteins were also analyzed for aggregation by fluorescence microscopy. No tendency to liposome aggregation was observed at any protein/lipid ratio (data not shown).

Several cell adhesion molecules that are involved in various recognition and binding reactions between avian and mammalian cells have been described during the last decade. The nature of the binding mechanisms is, however, known only for a few of these molecules. N-CAM and the cadherins (including L-CAM) participate in homophilic binding (Hoffman and Edelman, 1983; Nagafuchi et al. 1987; Edelman et al. 1987), i.e. each respective molecule can bind to other molecules of the same kind. N-CAM binds to N-CAM (Hoffman and Edelman, 1983), E-cadherin binds to E-cadherin (Nagafuchi et al. 1987) and so on. Two different approaches have been used to demonstrate homophilic binding of CAMs. The most direct approach involves physical binding studies of purified CAMs incorporated into liposomes (Hoffman and Edelman, 1983). In a more indirect approach non-expressing cells have been transfected with expression vectors coding for various CAMs (Nagafuchi et al. 1987; Edelman et al. 1987). Binding of CAM-containing liposomes has been demonstrated only for N-CAM (Hoffman and Edelman, 1983). Cadherins require interaction with cytoplasmic actin filaments to establish stable binding (Nagafuchi and Takeichi, 1988). Thus, homophilic binding of cadherins has only been demonstrated indirectly by transfection (Nagafuchi et al. 1987; Edelman et al. 1987).

In this study we investigated the binding properties of C-CAM by physical binding experiments employing both reconstituted liposomes and solid-phase methods. By both techniques we found that C-CAM has specific adhesive properties. Furthermore, the liposome experiments showed that C-CAM, in the same manner as N-CAM, retains its binding ability when incorporated in liposome membranes.

The following three independent observations demonstrated that C-CAM has homophilic binding properties: (1) C-CAM in detergent solution bound specifically to C-CAM immobilized on nitrocellulose filters; (2) C-CAM-containing liposomes self-aggregated; (3) blocking of C-CAM on hepatocytes by antibodies prevented binding of C-CAM-containing liposomes. An important question concerns the specificity of this binding, i.e. is the binding of C-CAM to C-CAM unique, or is C-CAM generally sticky and able to bind to a variety of structures? Another question related to the binding specificity is whether liposomes reconstituted with any membrane protein become sticky. The inability of C-CAM to bind to several different proteins immobilized on nitrocellulose filters demonstrated that the binding is specific. The uniqueness of C-CAM as a ligand for C-CAM was also evident from the observation that selective blocking of C-CAM on hepatocytes inhibited binding of C-CAM-containing liposomes. We demonstrated that liposomes containing any membrane protein do not become sticky, by showing the inertness of liposomes reconstituted with C-CAM-depleted membrane glycoproteins. We can thus conclude that the homophilic binding of C-CAM is specific and selective. Furthermore, this binding is independent of calcium ions.

The explicit demonstration of the adhesive properties of C-CAM suggests that C-CAM can mediate physical binding between adjacent cells via direct molecular contacts. This raises a number of questions. Does C-CAM cause physical cell aggregation only, or does the binding generate signals that affect the behavior and function of the participating cells? Can the C-CAM-mediated cell-cell binding be regulated? How might such signals and/or regulations be transmitted? The answers to these questions can only be obtained by further experimental work, e.g. transfection with mutagenized C-CAM. However, available knowledge about the biochemical properties of C-CAM indicate that C-CAM-mediated cell binding involves more than mere physical cell-clumping. Thus, the detergent partitioning and the liposome incorporation confirmed earlier data indicating that C-CAM is an amphipathic integral membrane protein that is anchored by hydrophobic interactions in the lipid bilayer of the plasma membrane (Odin et al. 1986). Previous analyses demonstrated that C-CAM becomes phosphorylated on serine residues (Odin et al. 1986), and recent results have shown that C-CAM is a calmodulin-binding protein (Obrink et al. 1988; Blikstad et al. 1989). These observations suggest that C-CAM is a transmembrane protein in which both the phosphorylation and the calmodulin binding presumably take place in an area that is exposed on the cytoplasmic face of the membrane. Both the phosphorylation and the calmodulin binding might be parts of regulatory events in which C-CAM participates.

Future analyses of the function of C-CAM and other cell adhesion molecules should provide us with answers to the very important and intriguing questions about the general function of cell adhesion molecules and the selective roles of different cell adhesion molecules. This is a central theme in tissue development and maintenance and should increase our knowledge about tissue integrity, tissue repair and tissue regeneration.

This work was supported by grants from the Swedish Medical Research Council (project no. 05200), the Swedish Cancer Foundation (project no. 1389), Konung Gustaf V:s 80-àrs fond, Torsten and Ragnar Soderbergs Stiftelser, OE och Edla Johanssons Stiftelse and The Swedish Natural Science Research Council. The excellent technical assistance of Ms Tarja Wikstrom is gratefully acknowledged.