Evidence for heterogeneity in the 160/165×10³m_r glycoprotein components of desmosomes

Desmosomes are intercellular junctions which are considered to play an important role in epithelial cell adhesion (Staehelin, 1974). Ultrastructurally, they possess electron-dense subplasma membrane plaques to which intermediate filaments (IF) appear to attach (Kelly, 1966). The intercellular space which separates the two closely apposed membranes of the desmosome is considered to be filled with ‘adhesive’ material and often contains an electron-dense midline (Staehelin, 1974).

Over the last few years the biochemical composition of desmosomes isolated from both bovine muzzle and tongue epithelial tissues using a technique described by Skerrow & Matoltsy (1974) has been studied. A number of cytoplasmic components of desmosomes have been characterized. These include the desmoplakins, which are high molecular weight polypeptides located in the region of the desmosomal plaque with which IF appear to associate (Mueller & Franke, 1983; Jones & Goldman, 1985). In addition, several laboratories have studied the glycoprotein constituents of desmosomes isolated from bovine epithelia (Cowin & Garrod, 1983; Cohen et al. 1983 ; Jones et al. 1986a,b; Schmelz et al. 1986a,b). For example, in bovine muzzle epidermis the major glycosylated polypeptides of desmosomes appear to be between 160 and 165×10³M_r (desmoglein 1; band 3) and polypeptides of approximately 115× 10³M_r and 130×10³M_r (desmoglein 2; band 4a and 4b; desmocollins) (Mueller & Franke, 1983; Cohen et al. 1983; Cowin et al. 1984). A role for the latter proteins in desmosome assembly appears likely as Fab′ fragments of antibody preparations directed against these desmosomal components inhibit desmosome formation in cultured cells (Cowin et al. 1984). Conversely, Fab fragments of antibody preparations directed against the 160/165×10³M_r desmosomal glycoproteins do not appear to inhibit desmosome formation in cultured epithelial cells (Cowin et al. 1984).

Our interest in the glycoproteins of desmosomes is based upon our finding that autoantibodies in the serum samples of certain patients afflicted with the blistering skin disease, pemphigus, recognize desmo-some-associated glycoproteins (Jones et al. 1986a,b). Indeed, our studies have revealed that pemphigus patients with the vulgaris variant of the disease possess circulating autoantibodies which react with a 140 × 10³M_r bovine tongue epithelial desmosome-associated glycoprotein. We recently showed that the 140× 10³M_r glycoprotein possesses the characteristics of a cell adhesion molecule (Joneset al. 1986a,b). The pemphigus vulgaris autoantibodies and rabbit 140×10³M_r antibodies that we have described appear to recognize only desmosomes in stratified squamous epithelia (Jones et al. 19866). Furthermore, the 140×10³M_r glycoprotein appears to be missing or in greatly reduced quantity in desmosomes prepared from bovine muzzle epidermis (Jones et al. 1986a,b). These findings and other reports in the literature appear to suggest that desmosomes vary from tissue to tissue with regard to their glycoprotein composition (Giudice et al. 1984; Jones et al. 1986a,b;Suhrbier & Garrod, 1986).

We have also shown that serum from other pemphigus patients (approximately 50% of patients with the foliaceus variety of pemphigus) contain autoantibodies directed against the 160/165×10³M_r desmosome-associated glycoproteins of both bovine muzzle and tongue epithelial desmosomes, confirming the work of Koulu et al. (1984)(Jones et al. 1986a). As in the case of pemphigus vulgaris, pemphigus foliaceus autoantibodies recognize only desmosomes in stratified squamous epithelia (Jones et al. 1986a). Furthermore, one of our pemphigus foliaceus serum samples which contains autoantibodies directed against the 160/165 × 10³M_r, desmosome-associated glycoproteins appears to recognize only those desmosomes in the upper layers of bovine tongue epithelium, as determined by indirect immunofluorescence microscopy (Jones et al. 1986a). Because of these findings we proposed that there are not only variations in the glycoprotein composition of desmosomes in different epithelial tissue types but there may be compositional differences in desmosomes in different layers of a stratified epithelial tissue (Jones et al. 1986a).

In order to determine more precisely the distribution of the 160/165× 10³M_r glycoprotein components in epithelial tissues, we have now produced both monoclonal and polyclonal antibody preparations directed against the 160/165 × 10³M_r desmosome associated polypeptides of bovine tongue epithelium.

Enriched preparations of desmosomes were prepared from bovine tongue epithelium according to the procedure of Skerrow & Matoltsy (1974) as modified by Mueller & Franke (1983). /Cvtoskeletal’ preparations (i.e. a high salt, Triton X-100-insoluble residue, the constituents of which include intermediate filaments and associated components e.g. desmosomes (Jones & Goldman, 1985)) of mouse skin, bladder, trachea and heart were prepared according to Zackroff & Goldman (1979) with the following modifications. Tissue was first minced finely and then homogenized in phosphate buffered saline (PBS) containing 0·6M-KC1, 1% Triton X-100, 10 mM-MgCl, 1 mM-phenylmethyl sulphonyl fluoride (PMSF) and 1 mM-tosyl arginine methvl ester (TAME). DNase was added to the homogenate to a concentration of 1 mg ml⁻¹ and after 5 min the resulting solution was centrifuged at 2000g in a Beckman TJ-6, centrifuge (Beckman Instruments). A pellet of ‘cytoskeletal’ material was collected. The bovine tongue epithelial desmosomes and the ‘cytoskeletal’ preparations were solubilized in standard Laemmli sample buffer containing 8 M-urea and 2% mercaptoethanol. Whole cell extracts of tissues were prepared as follows. Tissues were minced finelv and then homogenized in the above sample buffer.

A monoclonal antibody preparation directed against the 160/165×10³M_r glycoprotein components of bovine tongue desmosomes was raised as previously described (Jones et al. 1986c). A rabbit antiserum directed against these same glycoproteins and a rabbit antiserum directed against desmo-plakin were prepared as follows. An enriched preparation of bovine tongue epithelial desmosomes (see above) was subjected to SDS-PAGE and following a brief staining in Coomassie Blue the 160/165× 10³M_r and the desmoplakin polypeptides of 250 and 220× 10³M_r were excised from the gel. The gel pieces were homogenized in PBS and the resulting solutions were used to immunize rabbits. Serum from the immunized rabbits was collected as described previously (Jones et al. 1982) and analysed both by immunofluorescence and by Western immunoblotting. Antibodies directed against the 160/165× 10³M_r glycoproteins were affinity purified by using the appropriate region of DPT paper blots of bovine tongue epithelial desmosome preparations as described by Olmsted (1981).

A human serum sample from a patient suffering from the blistering skin disease, bullous pemphigoid (BP) was obtained from Robert Marder, Chief of the Clinical Immunology Laboratory, Northwestern Memorial Hospital, Chicago, and Ruth Frcinkei, Department of Dermatology, Northwestern University Medical School. This serum sample contained autoantibodies directed against hemides-mosomes as determined by immunoelectron microscopy (Jones et al. 1986c).

Tissues were prepared for double label indirect immunofluorescence as described elsewhere (Joneset al. 1984). In brief, tissues were frozen in Freon 22, cooled to liquid nitrogen temperatures. Sections 1 μm thick were prepared on a Reichert Ultracut E with an FC4D cryosystem attachment (Reichert Instruments, Buffalo, New York) and placed on Coverslips. Sections were fixed for 5 min in —20°C acetone and air dried. Desmoplakin antiserum or bullous pemphigoid serum was mixed with monoclonal antibody medium to a final dilution of 1:60. This mixture was overlayed on the tissue sections, which were then incubated for 1 h at 37°C in a moist chamber. Following extensive washing, sections were incubated in a mixture of fluorescein-conjugated goat antimouse and either rhodamine-conjugated goat anti-rabbit IgG (for the desmoplakin antiserum) or rhodamine-conjugated goat anti-human IgG (for the human autoantibodies) (Kirke-gaard and Perry Labs., Inc.) for 1 h at 37°C. Sections were washed in PBS and mounted on slides in Gelvatol (Monsanto, St Louis, Missouri). Bullous pemphigoid serum was also mixed with the affinity purified 160/165 × 10³M_r. rabbit polyclonal antibodies to a final dilution of 1:60. Sections were incubated in this mixture as above and then, after washing, were treated with a mixture of fluorescein-conjugated goat anti-rabbit IgG and rhodamine-conjugated goat anti-human IgG for 1 h at 37°C. Sections were then washed and mounted in gelvatol.

Sodium dodecyl sulphate/polyacrylamide gel electrophoresis (SDS-PAGE) using 7·5% acrylamide gels with 4·5% stacking gels was performed on enriched preparations of bovine tongue desmosomes, whole tissue extracts and the cytoskeletal’ preparations (Laemmli, 1970). Following SDS-PAGE, separated polypeptides were transferred to sheets of nitrocellulose (Towbin et al. 1979). Immunoblotting was carried out using the monoclonal and affinity-purified rabbit antibody preparations directed against the 160/165×10³M_r glycoproteins, in addition to the desmoplakin antiserum, as described previously (Zackroff et al. 1984).

The specificity of the monoclonal and rabbit antibody preparations directed against the 160/165× 10³M_r bovine tongue epithelial desmosome components was determined by immunoblotting using an enriched preparation of bovine tongue epithelial desmosomes and whole cell extracts of the same tissue (Fig. 1). The monoclonal and rabbit 160/16S× 10³M_r antibody preparations recognize 160/165 × 10³M_r polypeptide(s) present in the high salt, Triton-insoluble residue of mouse skin (Fig. 2). In the bladder, trachea and heart ‘cytoskeletal’ preparations the rabbit antibodies also recognize 160/165 × 10³d/_r polypeptide(s) whereas the monoclonal antibodies do not (Fig. 2). Note that in the immunoblots of the cytoskeleton preparations the 160/165 × 10³M_r glycoproteins sometimes appear as a smear (Fig. 2).

In confirmation of these immunoblotting data, immunofluorescence microscopy reveals that both the monoclonal and polyclonal 160/165× 10³M_r antibody preparations generate desmosome-like staining patterns in mouse skin (Figs 3, 5) while only the polyclonal antibody preparation generates staining at the sites of desmosomes in non-stratified squamous epithelial tissues (e.g. pseudostratified epithelium (trachea), transitional epithelium (bladder), stratified cuboidal epithelium (the epithelium of. esophageal glands)) and heart (data not shown). These results, in part, appear to confirm a recent report which showed that 160/165 × 10³M_r desmosomal proteins occur in a number of different tissue types (Schmelz et al. 1986a,b). However, our immunoblotting and immunofluorescence results also suggest that there are modifications in the 160/165 × 10³M_r polypeptides in desmosomes of heart and non-stratified squamous epithelial tissues as compared to desmosomes in epidermal cells, at least as far as the epitope recognized by our monoclonal antibody preparation is concerned.

When sections of mouse skin were processed for double label indirect immunofluorescence using the monoclonal antibody and the desmoplakin antiserum, we observed that the monoclonal antibodies fail to stain certain desmosomes in basal cells which are, none the less, recognized by the desmoplakin antibodies (Fig. 3).

In order to determine more precisely which desmosomes in basal cells do not stain with the monoclonal antibody preparation we undertook double label indirect immunofluorescence analysis of mouse skin using the monoclonal antibody in combination with BP autoantibodies used as markers for the basal surfaces of the basal epithelial cells (Jones et al. 1986c) (Fig. 4). No staining along the lateral and basal surfaces of the basal cells is seen with the monoclonal antibodies (Fig. 4A). However, punctate staining is observed along the apical surfaces of these same basal cells (Fig.4A).

The above result suggests either that desmosomes located along the lateral surfaces of basal cells lack the 160/165 × 10³M_r glycoproteins or that these components are in some way altered in laterally displaced desmosomes. However, the former possibility appears to be ruled out because of the finding that the polyclonal 160/165× 10³M_r antibodies recognize desmosomes located along the lateral surfaces of basal epithelial cells (Fig. 5).

Giudice et al. (1984) have suggested that desmosomal glycoproteins differ biochemically from tissue to tissue. The evidence for this was based on comparative Western immunoblotting analysis of desmosomes prepared from three stratified squamous epithelial tissues using a number of monoclonal and polyclonal antibody preparations. Indeed they were able to show that the molecular weights of immunologically related glycoproteins varied from tissue to tissue. In support of this Suhrbier & Garrod (1986) have also shown that desmosomal glycoproteins vary in both molecular weight and antigenicity in a number of tissues in a number of species.

However, in contrast to the aforementioned reports Schmelz et al. (1986a,b) have provided evidence that a 165×10³M_r bovine muzzle desmosomal glycoprotein (we presume that this polypeptide is identical, or at least related, to the one described here) is a general constituent of desmosomes which does not display cell type diversity. These authors base their evidence on immunoblotting and immunolocalization data obtained using two monoclonal antibody preparations directed against the 165×10³M_r glycoprotein components of bovine muzzle epidermal desmosomes. In our study we have been able to confirm, in part, the evidence of Schmelz et al. (1986a,b) by showing that a polyclonal antibody preparation directed against the 160/165 × 10³M_r glycoprotein components of bovine tongue epithelial desmosomes recognizes 165×10³M_r polypeptides in a number of different mouse tissue types. However, the monoclonal antibody preparation that we have produced is restricted in its tissue reactivity, i.c. our monoclonal antibody preparation only appears to recognize stratified squamous epithelial desmosomes. Thus it would appear likely that even though there are conserved portions of the 160/165 × 10³M_r desmosomal glycoproteins, as evidenced by the cross-reactivity of our 160/165 × 10³M_r rabbit polyclonal antibody preparation, there are also variable portions of the same glycoprotein.

The basal cells of the epithelial tissue types we have studied appear to exhibit a distinct polarity with regard to the 160/165 × 10³A7_r desmosomal glycoproteins. Using the 160/165 ×10³M_r monoclonal antibody preparations we have been able to show that there appear to be differences in the 160/165× 10³M_r components of desmosomes located along the lateral surfaces of basal cells compared to those desmosomes located along the apical surfaces of the same cells in two examples of stratified squamous epithelial tissues. Furthermore, the 160/165 × 10³M_r glycoproteins appear absent from hemidesmosomes located along the basal portion of these basal cells (Jones et al. 1986c). Indeed, our results and those recently reported by Parrish et al. (1986) who showed that mouse antisera directed against certain other desmosomal glycoproteins of 130 and 115×10³M_r (the ‘desmocollins’) distinguish between basal and suprabasal cells of bovine and human epidermis, suggest that the glycoproteins of desmosomes may be modified depending upon where they are located in an epithelial tissue.

These are intriguing findings since these are not the first components of epithelia which have been shown to vary within an epithelial tissue. For example, the expression of certain keratins (Woodcock-Mitchell et al. 1982), involucran (Rice & Green, 1979; Banks-Schlegel & Green, 1981), filaggrin (Harding & Scott, 1983) and L-CAM, a cell adhesion molecule, (Thiery et al. 1984) appears to depend on the state of differentiation of an epithelial cell. It is thus possible that there may be coordinate changes in expression of these components as epithelial cells undergo differentiation.

It has been suggested that variability in desmosomal glycoproteins may be the result of differences in their carbohydrate moieties (Suhrbier & Garrod, 1986). If this is the case then one would predict that our monoclonal antibody preparation recognizes an epitope of the 160/165×10³M_r glycoproteins that involves carbohydrate. We are currently attempting to determine whether this is the case. Our preliminary results indicate that the epitope of the 160/165 ×10³M_r glycoprotein that is recognized by our monoclonal antibody is not periodic acid-sensitive. This would argue against the possibility that our antibody preparation is directed against carbohydrate. We are now in the process of enzymically deglycosylating the 160/165 × 10³M_r glycoproteins to determine if this treatment prevents binding of our monoclonal antibody preparation. We hope in this fashion to be in a position to determine whether the tissue-specific epitope of the 160/165 × 10³M_r glycoproteins that we have described above lies within the protein or carbohydrate portion of the molecule.

The authors thank Mary Beth Bosshart for technical assistance. The work was supported by grants from Nl 11 to RDG (ROI GM36806–01) and to JCRJ (biomedical research support grant, 2S07RR05370).