Characterization of a 125K glycoprotein associated with bovine epithelial desmosomes

The spot desmosome is an intercellular junction that appears to be involved in epithelial cell adhesion (Staehelin, 1974). The desmosome is intimately associated with the cytoskeleton (Staehelin, 1974). In particular, in epithelial cells bundles of intermediate filaments (IF) appear to attach to the cytoplasmic plaques of desmosomes. The biochemical properties of epithelial desmosomes have been studied using desmosomes isolated with and without their associated IF (Skerrow & Matoltsy, 1974a, b; Drochmans et al. 1978; Jones & Goldman, 1985). A number of glycosylated and non-glycosylated components of bovine epithelial desmosomes isolated using the procedure described by Skerrow & Matoltsy (1974a) or modifications thereof, have been characterized (Skerrow & Matoltsy, 1974b; Gorbsky & Steinberg, 1981; Cohen et al. 1983; Mueller & Franke, 1983; Tsukita & Tsukita, 1985; Cowin et al. 1984a, 1986; Gorbsky et al. 1985; Jonesef al. 1986a, b).

All desmosomes possess a 250K (K = M_r× 10⁻³) polypeptide termed desmoplakin I (Mueller & Franke, 1983; Cowin et al. 1985). In stratified epithelia a related polypeptide termed desmoplakin 11 of 22OK has also been found (Mueller & Franke, 1983; Cowin et al. 1985). Immunoelectron microscopical localization of antibodies directed against the desmoplakins reveals that these components reside in the desmosomal plaque (Mueller & Franke, 1983; Jones & Goldman, 1985; Jones et al. 1986b). Other minor plaque components of 200K (DI antigen) and 240 K have also been studied (Franke et al. 1987; Tsukita & Tsukita, 1985). Plakoglobin of 83 K has recently been shown to reside in the desmosome plaque in addition to other symmetrical adherens junction types (Gorbsky et al. 1985; Cowin et al. 1986). Bovine epithelial desmosomal glycoproteins ranging in molecular weight from 100K to 165K have also been characterized immunologically (Gorbsky & Steinberg, 1981; Cohen et al. 1983; Cowin et al. 1984a, b;,Jones et al. 1986a,b). Some of these glycoproteins appear to be located along the intercellular region of the desmosome while others are transmembranous (Schmelz et al. 1986a,b; Jones et al. 1986b; Miller et al. 1987). Furthermore, a recently described 140K bovine tongue epithelial desmosomal glycoprotein is not restricted to the desmosome but is found along the entire surface of bovine epithelial cells (Jones et al. 1986b).

In this paper a 125K glycoprotein component of bovine epithelial desmosomes is identified and characterized. Its distribution in bovine tongue epithelia at the electron microscope level of resolution is compared with that of a number of other desmosomal glycoproteins.

Primary mouse epidermal cells were prepared and maintained in culture as reported (Hennings et al. 1980; Jones et al. 1982).

A 100-mm Petri dish of confluent mouse epidermal cells was first rinsed briefly in phosphate-buffered saline (PBS) and then 1 ml of standard Laemmli sample buffer containing 8 M-urea and 2% ß-mercaptoethanol (Laemmli, 1970) was added to the cells. The cells were scraped off the dish in the sample buffer. The solution was stirred at room temperature for 30 min and then stored at −80°C for later use. In the case of bovine tongue epithelium, the epithelium was minced finely with scissors in the above sample buffer. The sample buffer solution containing the minced tissue was stirred until the tissue was solubilized and then stored at −80°C for later use.

Desmosomes were prepared from bovine muzzle and tongue epithelium using the technique of Skerrow & Matoltsy (1974a) with the modifications described by Mueller & Franke (1983). The glycoprotein components of such isolates were prepared according to S. J. Jones et al. (1987). In brief, isolated whole desmosomes were treated with 9·5M-urea in 10mM-Tris HC1, pH9, for lh with constant agitation. Following this treatment, the sample was centrifuged at 180 000g for 45 min and the supernatant containing the ureasoluble desmosomal components and the urea-insoluble pellet were collected. These separate fractions and whole desmosomes were solubilized in the sample buffer mentioned above.

A rabbit antiserum directed against the 160/165K glycoprotein components of bovine tongue epithelial desmosomes (desmoglein 1 (Cohen et al. 1983)) was used in this study (J. C. R. Jones et al. 1987). A rabbit antiserum directed against the 100/115K glycoprotein components of bovine tongue epithelial desmosomes (the ‘desmocollins’ (Cowin et al. 1984b) was prepared as follows. An enriched preparation of bovine tongue epithelial desmosomes was subjected to SDS-polyacrylamide gel electrophoresis (SDS-PAGE) and following a brief stain in Coomassie Blue the 100K polypeptide was excised from the gel. The gel pieces were homogenized in PBS and the resulting solution used to immunize a rabbit. Serum from the immunized rabbit was collected as described previously (Jones et al. 1982) and analysed by immunofluorescence microscopy and Western immunoblotting to determine antibody specificity. This antibody preparation recognizes both the 100K and 115K desmosomal glycoproteins, which have already been shown to be related immunologically (Cohen el al. 1983; Cowin & Garrod, 1983; see Results, Fig. 1). A rabbit antiserum directed against E-cadherin was a gift from Dr M. Takeichi (Yoshida & Takeichi, 1982). In addition, Dr Takeichi provided a rat monoclonal antibody directed against mouse E-cadherin, ECCD-1, for this study (Yoshida-Noro et al. 1984).

Rabbit antibodies were affinity purified using the appropriate region of nitrocellulose blots of bovine tongue desmosome preparations (see below) as reported by Olmsted (1981).

Pieces of bovine tongue were frozen in Freon 22 cooled by liquid nitrogen. Sections of approximately 1 μm were prepared on a Reichert Ultracut E microtome fitted with an FC4D cryoattachment (Reichert Instruments, Buffalo, NY) and placed on slides. Sections were fixed for 5 min at −20°C in acetone and air dried. To minimize non-specific binding of secondary’ antibody the sections were first incubated for 30 min in normal goat serum diluted 1:20 in PBS in a moist chamber at 37°C. Following this step either the rabbit anti-100/1 15K, the anti-160/165K or the E-cadherin antibodies were overlaid on the sections and the slides incubated for a further 45 min at 37°C. The slides were next extensively washed in PBS and then the sections were incubated for 45 min at 37°C in fluorescein-conjugated goat anti-rabbit immunoglobulin (IgG) (Kirkegaard and Perry Labs. Inc., Gaithersburg, MD). After washing the sections were mounted with a glass coverslip in gelvatol (Monsanto, St Louis, MO).

Pieces of bovine tongue were fixed in 4% paraformaldehyde in 0·1 M-phosphate buffer, pH 7 ·2 and were then embedded in Lowicryl K4M as described by Roth et al. (1981). Thin sections were prepared and placed on nickel grids. To minimize non-specific binding sections were incubated in normal goat serum diluted 1:20 in 20mM-Tris-buffered saline containing 1% bovine serum albumin (BSA buffer) for 30 min at room temperature. Sections were then incubated overnight in the 100/115K, 160/165K or cadherin antibody preparations. The sections were washed in BSA buffer and then incubated in 5 nm gold-conjugated goat anti-rabbit IgG (Janssen Pharmaceutica, Beerse, Belgium). The sections were washed in BSA buffer, rinsed briefly in water and then stained for 15 min in 3% aqueous uranyl acetate and viewed in a Jeol 100CX electron microscope at 60 kV.

Sodium dodecyl sulphate-polyacrylamide gel electrophoresis (SDS-PAGE) using 7·5% acrylamide gels with 4·5% stacking gels was performed on enriched preparations of desmosomes, the urea-soluble and urea-insoluble fractions of such desmosome preparations, whole bovine tongue epithelial cell extracts and whole cell extracts of cultured epidermal cçlls. The separated polypeptides were then transferred to nitrocellulose according to Towbin et al. (1979).

Two-dimensional gel electrophoresis was performed on the enriched preparations of bovine tongue epithelial desmosomes as described by O’Farrell (1975). Following electrophoresis, the two-dimensional gels were either silver stained using a Biorad silver stain kit (Biorad, Richmond, CA) or separated polypeptides were transferred to nitrocellulose (Towbin et al. 1979).

Immunoblotting was carried out as described previously (Zackroff et al. 1984). Glycoprotein components of desmo-somes were detected on nitrocellulose paper following the procedure of Shiozawa et al. (1987). In brief the procedure was as follows. Following transfer of proteins to nitrocellulose paper, the paper was incubated for 1 h at room temperature in 50mM-Tris HC1, 0·lM-NaCl, lmM-MnCl₂, 1 mM-CaCl₂, pH 7 containing 1% BSA (ConA buffer). The nitrocellulose was then incubated for 30 min in 50 μgml⁻¹ of concanavalinA (ConA) in ConA buffer. Following extensive washing in ConA buffer, the nitrocellulose was incubated for a further 30min in ConA buffer containing 13μgml⁻¹ horse radish peroxidase (250 units mg⁻¹ protein). The nitrocellulose was then washed in ConA buffer lacking BSA. ConA-binding proteins were detected colorimetrically following incubation in ConA buffer (lacking BSA) containing 0·01% hydrogen peroxide and 6 mg ml⁻¹ 4-chloronaphthol.

In enriched preparations of bovine tongue epithelial desmosomes approximately five polypeptides of molecular weight 100K, 115K, 125K,140K, 160/165K bind ConA (Fig. 1). The 125K ConA-binding polypeptide is particularly evident in the two-dimensional blot (Fig. 1).

Two-dimensional electrophoretic analysis reveals that the 100K, 115K, 125K and 140K ConA-binding polypeptides appear as individual spots whereas the 160/165K polypeptide(s) appear as a streak (Fig. 1). The 100K, 115K, 125K and 140K polypeptides possess similar isoelectric points of around 4·6–4·8 (Fig. 1). A number of these ConA-reactive polypeptides have already been characterized by other workers. They include the desmocollins (M_r 100/115K) (Cowin et al. 1984b), a 140K glycoprotein (Jones et al. 1986a,b) and desmoglein 1 (M_r 160/165K.) (Gorbsky & Steinberg, 1981; Cohen et al. 1983). Western immunoblotting analysis of the bovine tongue desmosomes using polyclonal antibody preparations directed against these known desmosomal glycoprotein components reveals that the 125K glycoprotein is immunologically distinct from the desmocollins, desmoglein 1, and the 140K glycoprotein mentioned above (Fig. 2).

The 125K glycoprotein is of similar molecular weight to certain cell adhesion molecules (CAMs), e.g. E-cadherin (the same molecule as uvomorulin) and CAM 120/80 Yoshida & Takeichi, 1982; Yoshida-Noro et al. 1984; Hyafil et al. 1980; Vestweber & Kemler, 1984; Damsky et al. 1984). Thus an antibody preparation directed against E-cadherin was used in immunoblotting analyses of whole cell extracts of bovine tongue epithelia and whole isolated desmosomes (Fig. 2). The cadherin antibodies recognize a 125K polypeptide in these protein preparations. Furthermore, the cadherin antibodies recognize a 125K polypeptide in other bovine epithelial desmosome preparations, e.g. bovine muzzle desmosomes (Fig. 3).

Isolated bovine tongue desmosomes were separated into their urea-soluble and urea-insoluble (glycoprotein enriched) desmosomal fractions according to S. J. Jones et al. (1987). Immunoblotting analysis was then undertaken using the cadherin antibody preparation (Fig. 4). The cadherin antibodies recognize a 125K glycoprotein, which is enriched in the urea-insoluble compared to the urea-soluble desmosomal fractions (Fig. 4).

To confirm the immunological relationship between the 125K desmosome associated glycoprotein and cadherin the following experiments were performed. A preparation of bovine tongue epithelial desmosomes was subjected to SDS-PAGE and subsequently transferred to nitrocellulose. A small vertical strip of this nitrocellulose sheet was stained for protein with Amido Black for orientation. A horizontal piece of nitrocellulose containing the 125K desmosome-associated glycoprotein was then sliced from the remaining nitrocellulose and incubated for 12 h in the cadherin antibody preparation. The absorbed antibody preparation was then collected and the antibodies adsorbed onto the nitrocellulose were eluted from the strip following the procedure of Olmsted (1981). These two antibody preparations and the whole cadherin antibody preparation were analysed by immunoblotting on a bovine tongue desmosome preparation and extracts of mouse epidermal cells (Fig. 5). Whereas the whole cadherin antibody preparation and the eluted antibodies recognize 125 K polypeptides present in both preparations, the absorbed antibody preparation shows extremely weak reactivity with these polypeptides (Fig. 5). Furthermore, the 125K polypeptide in the mouse epidermal cell extracts appears to be cadherin since it is recognized by a monoclonal cadherin antibody, ECCD-1 (Fig. 5). This monoclonal antibody shows limited cross-species reactivity and does not react with bovine tissues as determined by both immunoblotting (Fig. 5) and immunofluorescence analysis (result not shown).

The immunoblotting analyses suggest that the cadherin antibody preparation recognizes a desmosome component. However, it is possible that the co-purification of cadherin-reactive polypeptide and desmosomes is merely fortuitous and that the 125K glycoprotein is not a bona fide component of bovine epithelial desmosomes. Thus the distribution of the 125K glycoprotein was studied in bovine tongue epithelium by immunofluorescence. Parallel experiments were also undertaken using antibody preparations directed against known components of desmosomes (i.e. the 100/115K and 160/165K glycoproteins). The cadherin antibody preparation generates an intercellular staining pattern in cryostat sections of bovine tongue epithelium as determined by indirect immunofluorescence (Fig. 6A). This pattern is similar to that produced in the same tissue by antibody preparations directed against the 100/115K and 160/165K desmosomal glycoproteins (Fig. 6B,C).

Because of the similarity in the immunofluorescence patterns generated by the cadherin antibodies and desmosomal glycoprotein antibody preparations it is tempting to speculate that cadherin antibodies recognize desmosomes. To confirm this an immunoelectron microscopic analysis was undertaken. Thin sections of Lowicryl K4M embedded bovine tongue epithelium were processed for indirect immunogold localization using the cadherin, 100/115Kand 160/165K antibody preparations (Fig. 7). The cadherin antibodies localize along the entire surface of an epithelial cell, i.e. gold particles are found both along the intercellular region of the desmosome and along non-desmosomal cell surfaces (Fig. 7A,B). This staining pattern contrasts with those produced by the 100/115K and 160/165K antibody preparations. In the case of the former antibody preparation, gold particles are mainly found over the intercellular region of the desmosome and do not occur over non-desmosomal cell surfaces (Fig. 7C). The 160/165K antibodies, like the 100/115K antibodies, also appear to bind specifically to desmosomes (Fig. 7C,D). However, the 160/165K antibodies not only appear to localize to the intercellular region of the desmosome but are also found over the desmosomal plaque (Fig. 7D) (Schmelzer al. 1986b). The latter immunolocalization results using the 160/165K and the 100/115K antibody preparations confirm a recent report (Miller et al. 1987).

Bovine epithelial desmosomes contain a variety of glycoproteins. The major desmosomal glycoproteins have been studied by a number of laboratories (see, e.g., Cohen et al. 1983; Cowin & Garrod, 1983; Jones et al. 1986a,b; Schmelz et al. 1986a,b). In this paper the immunochemical characterization of a 125K glycoprotein constituent of enriched preparations of bovine tongue and muzzle epithelial desmosomes is described. This protein shows considerable similarity to the cell adhesion molecule E-cadherin (uvomorulin), i.e. it is of the same molecular weight, it binds ConA, and it is immunologically related to cadherin (Yoshida & Takeichi, 1982; Yoshida-Noro et al. 1984; Vestweber & Kemler, 1984).

To rule out the possibility that the 125K glycoprotein is a co-purifying contaminant of bovine epithelial desmosomes, immunoelectron microscopy was undertaken and revealed that the 125K glycoprotein is a bona fide desmosome component, even though it also occurs along non-desmosomal cell surfaces. Indeed the distribution of the 125K glycoprotein along the surface of bovine epithelial cells appears identical to that recently described for a 140K glycoprotein component of stratified squamous epithelial cells, in that both occur not only along the intercellular space of the desmosome but also over non-desmosomal cell surfaces (Jones et al. 1986b). Moreover, Jones et al. (1986b) raised the possibility that the 140K glycoprotein was related to E-cadherin (uvomorulin). However, this does not appear to be the case since the 125K and 140K glycoprotein components are immunologically distinct.

If the 125K desmosome component is indeed cadherin, as the results presented here suggest, then this would appear to contradict a report in which it has been shown by immunoelectron microscopy that in intestinal epithelium uvomorulin (E-cadherin) is a component of adherens junctions but not of desmosomes (Boiler et al. 1985). However, this apparent contradiction may be explained by differences in the glycoprotein composition of desmosomes in different tissues. Such variability in desmosomal glycoproteins has already been suggested (Giudice et al. 1984; Suhrbier & Garrod, 1986). In support of this possibility, it has recently been reported that a 140K glycoprotein component of bovine tongue epithelial desmosomes is missing or in reduced quantity in bovine muzzle desmosomes (Jones et al. 1986a,b; see Fig. 4), and that there is heterogeneity in the 160/165K desmosomal glycoproteins in different epithelial tissues (J. C. R. Jones et al. 1987). Indeed, variability in desmosomal glycoproteins may relate to a possible role for desmosomes in determining specific epithelial cell-cell interactions.

Because of the relationship of the 125K desmosomal glycoprotein to E-cadherin it is tempting to speculate that the 125K glycoprotein also possesses the adhesive properties of cadherin (Yoshida & Takeichi, 1982; Yoshida-Noro et al. 1984). If this is the case then the desmosomes of bovine tongue epithelium contain multiple molecules playing a role in cell adhesion, i.e. a cadherin-like molecule, the desmocollins and the 140K component (Cowin et al. 1984; Jones et al. 1986b). It is not clear why the desmosome possesses several adhesion molecules but certain of them may confer on the desmosome its tissue specificity as discussed above.

Cadherin is a Ca²⁺-dependent cell adhesion molecule and therefore, by analogy, it is possible that the 125K desmosome-associated glycoprotein described here may also somehow be involved in Ca²⁺-dependcnt adhesion, i.e. as in the case of cadherin the conformation of the 125K glycoprotein may be modified by reaction with Ca²⁺ (Yoshida-Noro et al. 1984). In this regard it is interesting to note that Steinberg et al. (1987) recently described an uncharacterized 130K bovine muzzle desmosome component that binds Ca²⁺. The close similarity in molecular weight between the latter protein and the I25K glycoprotein characterized in this study suggests that these proteins may be one and the same. Whether or not the I25K desmo-some-associated glycoprotein plays a role in the Ca²⁺-dependent assembly of desmosomes in stratified squamous epithelial cells as reported by Hennings et al. (1980) and Jones et al. (1982) remains to be determined.

Although we now have an extensive knowledge of the molecular components of desmosomes isolated from two bovine epithelial tissues we still have little information concerning the function of these molecules. For example, a number of workers have suggested that the desmoplakins and/or desmocalmin may be involved in the association of intermediate filaments with the desmosomal plaque but experimental support of this has not been forthcoming (Jones & Goldman, 1985; Tsukita & Tsukita, 1985; Kapprell et al. 1985). Some advances have been achieved in the analysis of the desmosomal glycoproteins, particularly with regard to the identification of putative cell adhesion molecules (Cowin et al. 1984b; Jones et al. 1986b). However, the role that desmosomal glycoproteins play in desmosome assembly, i.e. whether they initiate desmosome formation, is not yet clear. Furthermore, little analysis of desmosomal CAMs in a variety of disease states in which it is considered that there are perturbations of cell adhesion, e.g. epithelial carcinomas, has been attempted (Parrish et al. 1986). We are currently undertaking such analyses and hope that such studies will allow a better understanding of the role that desmosomal glycoproteins play in the assembly and functions of desmosomes.

The author thanks Mary Beth Bosshart for technical assistance. This work was supported by a grant from NIH (2SO7RR05370).