Further Analysis of Pemphigus Autoantibodies and Their Use in Studies on the Heterogeneity, Structure, and Function of Desmosomes J o n a t h a n C. R. Jones , Karen M. Yokoo, and Rober t D. G o l d m a n

Department of Cell Biology and Anatomy, Northwestern University Medical School, Chicago, Illinois 60611

Abstract. Pemphigus is an autoimmune disease that causes blistering of human epidermis. We have re- cently shown that autoantibodies in the serum of three pemphigus patients bind to desmosomes (Jones, J. C. R., J. Arnn, L. A. Staehelin, and R. D. Gold- man, 1984, Proc. Natl. Acad. Sci. USA., 81:2781- 2785), and we suggested that pemphigus blisters form, at least in part, from a specific antibody-induced dis- ruption of desmosomes in the epidermis. In this pa- per, experiments are described that extend our initial observations. 13 pemphigus serum samples, which in- clude four known pemphigus vulgaris (Pv) and four known pemphigus foliaceus (Pf) serum samples, have been analyzed by both immunofluorescence and by immunoblotting using cell-free desmosome prepara- tions. Tissue sections of mouse skin processed for double indirect immunofluorescence using each of the pemphigus serum samples and a rabbit antiserum di- rected against a component of the desmosomal plaque (desmoplakin) show similar punctate cell surface staining patterns. This suggests that all 13 pemphigus serum samples contain autoantibodies that recognize desmosomes. These autoantibodies appear specific for stratified squamous epithelial cell desmosomes and do not recognize desmosomes of other tissues (e.g., mouse heart and mouse intestine). Cultured mouse keratinocytes, which possess well-defined desmo- somes, were processed for indirect immunofluores-

cence using the pemphigus serum samples. Eight of the 13 sera (including the four known Pv samples but not the known Pf sera) stain desmosomes in these preparations. By double indirect immunofluorescence the desmoplakin antiserum stains a double fluorescent line along the contacting edges of cultured keratino- cytes, whereas the positive pemphigus serum samples stain a single fluorescent line along this same border. We believe that these pemphigus autoantibodies rec- ognize extracellular antigens located somewhere within the region between the two apposing mem- branes that ~omprise the desmosome. The pemphigus sera exhibit positive immunoblotting reactions with desmosome-enriched fractions obtained from bovine tongue epithelium. Three serum samples (including two of the four known Pf serum samples) react with 160- and 165-kD desmosome-associated polypeptides (Koulu, L., A. Kusimi, M. S. Steinberg, V. Klaus- Kovtun, and J. R. Stanley, 1984, J. Exp. Med., 160: I509-1518). Another eight serum samples (in- cluding the four known Pv sera) recognize a 140-kD desmosome-associated polypeptide. We propose that the antigens recognized by these human autoantibod- ies may play important roles in the adhesion of cells within the epidermis. Furthermore we discuss the het- erogeneity of desmosome structure in light of our im- munofluorescence and immunoblotting results.

p EMPHIGUS is a devastating autoimmune disease of the skin whose clinical features are characterized by the formation of numerous intraepidermal blisters that

cover varying portions of the body (4, 13, 24). There is variation in the disease, in that some patients have extensive blistering in most skin regions, while others, for example, exhibit only mouth blisters (4, 13, 24). Histological exami- nation of biopsies of blistering regions demonstrate that dys- adhesion of keratinocytes (acantholysis) produces blistering of the epidermis (4, 13, 24). According to most workers in the field, pemphigus has a number of variants, classified mainly by the clinical symptoms of patients and the histo-

pathology of skin lesions (4, 13, 24). The two main forms of the disease are termed pemphigus vulgaris (Pv) ~ and pemphi- gus foliaceus (Pf). In the former case, blisters result from dysadhesion of keratinocytes in the epidermis just above the basal cell layer (suprabasally), and in the latter case blisters form because of a loss of cell-cell contacts higher within the stratum spinosum (4, 13, 24). These variations of pemphigus appear to affect only stratified squamous epithelial tissues (4, 24).

i Abbreviations used in this paper." PBSa, a solution of 6 mM Na ÷, K + phosphate, 171 mM NaCI, 3 mM KC1, pH 7.4; Pf, pemphigus foliaceus; Pv, pemphigus vulgaris; TPBS, 0.05% Tween 20 in PBSa.

© The Rockefeller University Press, 0021-9525/86/03/1109/09 $1.00 The Journal of Cell Biology, Volume 102, March 1986 1109-1117 1109

The autoantibodies found in pemphigus patients appear to be directed against an intercellular cement substance (4, 13, 24). Pemphigus autoantibodies actually bind to the epidermis in blistering regions of the skin of diseased patients as deter- mined by microscopical analyses of biopsy material (4, 13, 24). Both immunofluorescence (3) and immunoelectron mi- croscopic (3 l) observations of these biopsies suggest that the autoantibodies bind to the surface coat or glycocalyx of indi- vidual epidermal cells. Indeed it is now considered that these autoantibodies play a pathogenic role in the disease (13, 24) since, for example, incubation of skin explants in the presence of pemphigus IgG induces acantholysis (26).

Recently, we have shown that certain autoantibodies pres- ent in the serum of pemphigus patients appear to bind specif- ically to desmosomes (19), which are intercellular junctions considered to play a vital role in the adhesion of the cells within the epidermis (2). Indeed, we have proposed, based on observations of living epidermal cells, that pemphigus blisters form, at least in part, from a specific antibody-induced dis- ruption of desmosomes in the epidermis.

Our finding that certain autoantibodies in pemphigus rec- ognize desmosomes has been supported, at least in part, by a recent publication by Koulu et al. (22). These authors showed that autoantibodies from certain Pf patients recognize poly- peptides found in cell-free, so-called "core" (15) preparations of desmosomes. However, these same authors report that Pv autoantibodies do not react with the same desmosome frac- tions. Gorbsky et al. (16) have also suggested that Pv patients do not possess any circulating anti-desmosome antibodies. The results of these two publications are based primarily on either immunoprecipitation and immunoblotting analyses or ELISA using enriched fractions of bovine muzzle desmo- sprees as antigenic substrates. Furthermore, in both of these studies (16, 22), the authors either did not or were unable to show by means of immunofluorescence that any of their pemphigus serum samples contained anti--desmosome auto- antibodies. This has led Koulu et al. (22) to suggest that our previous finding, which demonstrated the presence of auto- antibodies directed against desmosomes, could be explained in light of their data, if all of our serum samples were derived from Pf patients (19, 22).

In this study, in contrast to the aforementioned reports of Koulu et al. (22) and Gorbsky et al. (16), we show that the serum samples of known Pv patients, in addition to certain Pf patients, recognize desmosomal components as determined by both immunofluorescence and immunoblotting criteria using cell-free preparations of desmosomes. We propose that the antigenic determinants of desmosomes recognized by these autoantibodies may play important roles in the adhesion of cells within the epidermis. Furthermore, data is presented that suggests that these autoantibodies reveal heterogeneity in desmosome components both within and amongst the cells that comprise different epithelial tissues. These results are discussed in light of the possibility that desmosomes may contain specific cell-cell adhesion molecules and not merely represent spot welds between neighboring cells.

Materials and Methods

Cell Culture

Primary mouse keratinocytes were prepared and maintained in culture as reported elsewhere ( 18, 2 l).

Antisera

Antisera from 13 patients with pemphigus were used in these studies. Five of these were obtained from Dr. Robert Marder, the Director of the Clinical Immunology Laboratory, Northwestern Memorial Hospital, Chicago, and Dr. Ruth Frienkel, Department of Dermatology, Northwestern University Medical School, Chicago. These Northwestern pemphigus serum samples were not subclassified into Pfand Pv. Drs. John Stanley and Leena Koulu (Department of Dermatology, Uniformed Services University of the Health Sciences, Be- thesda, MD) provided four serum samples from patients diagnosed as having Pf and four serum samples from patients diagnosed as having Pv. Human serum samples from normal individuals were used as controls. A rabbit anti- serum directed against the desmoplakin 1 (250-kD polypeptide) component of bovine muzzle desmosomes (a gift of Drs. James Arnn and Andrew Staehelin) (1, 19), was also used for immunotluorescence analyses.

Immunofluorescence Pieces of fresh neonatal mouse skin and bovine tongue were frozen in liquid nitrogen. Cryostat sections, 1-3-pm thick, were prepared on a Sorvall MT2 Ultramicrotome with an FTS cryo-attachment fiTS Systems Inc., Stone Ridge, NY). These sections were placed on glass slides coated with polylysine (Sigma Chemical Co., St. Louis, MO). After drying for -60 min at room temperature, the sections were fixed for 5 min in -20"C acetone and air dried. Cultured cells were grown on glass coverslips and, after removal of culture medium, were fixed for 2-3 min in -20"C acetone and then air dried. To minimize nonspecific binding of antibody, the fixed material was preabsorbed with normal goat serum diluted 1:20 in PBSa (6 mM Na ÷, K ÷ phosphate, 171 mM NaC1, 3 mM KCI, pH 7.4) for 20 min at 37"C. Such preparations were then processed for single and double indirect immunofluorescence as previously reported (19).

Isolation o f Desmosomes The epithelium of either the muzzle or the tongue of a recently slaughtered cow with its attached connective tissue was cut into small pieces ~ l-cm square. These pieces were soaked overnight at room temperature in 2 mM EDTA in PBSa that contained 1 mM phenylmethylsulfonyl fluoride. Then the epithelium was peeled from the connective tissue and was minced finely with scissors. Isolation of desmosomes was then done using the technique described by Skerrow and Matoltsy (27), with the modifications of Mueller and Franke (25).

Western Blotting Procedure SDS PAGE using 7.5% acrylamide slab gels with 4.5% stacking gels was

done on the desmosome-enriched fractions after solubilization in 8 M urea, 0.19 M Tris-HC1, pH 6.8. 1% SDS, 1% B-mercaptoethanol, according to the procedure of Laemmli (23). The separated proteins were transferred to sheets of nitrocellulose using a Hoefer Transphor apparatus (Hoefer Scientific Instru- ments, San Francisco, CA) (30). Transferred proteins were visualized as discrete bands on nitrocellulose after staining for 15 min in 0.1% amido black in 5% methanol, 10% acetic acid, followed by a 10-min destain in 5% methanol, 10% acetic acid. Unstained nitrocellulose sheets that contained the desmosomal proteins were placed in blocking buffer (10% fetal calf serum and 2.5% bovine serum albumin [BSA] in PBSa) for 1 h, rinsed briefly in PBSa, and incubated for 1 h at room temperature in pemphigus serum samples (diluted 1:250 in 0.05% Tween 20 [Sigma Chemical Co.] in PBS [TPBS]). The sheets were then washed with four changes of TPBS over a 30-rain period and then incubated for 30 min in biotinylated goat anti-human IgG (Vectastain ABC Kit, Vector Laboratories, Inc., Burlingame, CA). Sheets were then washed for a further 30 min in four changes of TPBS and then incubated for an additional 30 min in an avidin/biotinylated horseradish peroxidase complex (Vector Laboratories. Inc.). After washing the sheets in four changes of PBSa over 30 min, bound IgG was visualized using 0.05 % 4-chloro- l-naphthol, 0.01% hydrogen peroxide. 15% methanol in PBSa.

Results Immunofluorescence Observations

The Visualization of Desmosomes in Tissues. We previously reported that cryostat sections of neonatal mouse skin proc- essed for single label indirect immunofluorescence using either the pemphigus serum samples or the desmoplakin antiserum show similar staining patterns (19). To compare more directly these fluorescent patterns, double indirect im- munofluorescence has been done on cryostat sections of

The Journal of CeU Biology, Volume 102, 1986 1110

neonatal mouse skin (Fig. 1), using the pemphigus serum samples and the desmoplakin ant iserum. The patterns ob- served with all the pemphigus serum samples are similar to

those generated by the desmoplakin ant iserum (i.e., in every case a coincident punctate staining at the surface of epidermal cells is revealed [Fig. l, a-d]).

Figure 1. Cryostat sections of mouse skin were processed for double indirect immunofluorescence using pemphigus serum samples (a and c) and the desmoplakin antiserum (b and d). a shows the punctate intercellular immunofluorescence pattern generated by a Pv serum sample in mouse skin (see Fig. 3, lane G for the immunoreactivity of this serum sample with an enriched fraction of desmosomes). Note the co- localization of this staining pattern with that generated by the desmoplakin antiserum in the same section (b). c shows the punctate intercellular staining pattern produced by a Pf serum sample (see Fig. 3, lane B for the immunoreactivity of this serum sample with a desmosome fraction), This staining pattern co-localizes with that produced by the desmoplakin antiserum in the same section (d). Bar, 10 urn.

1 1 1 1 Jones et al. Pemphigus A utoantibodies

Our results to date strongly suggest that pemphigus sera contain autoantibodies that are directed against components of desmosomes. Since desmosomes have been considered to be ubiquitous organelles of epithelial tissues (5), we wondered, therefore, whether the sera we analyzed recognize desmoso- mal antigens in tissues other than stratified squamous epithe- lia. Thus cryostat sections of several other desmosome-con- taining tissues were prepared for indirect immunofluorescence using the pemphigus serum samples and the desmoplakin antiserum as a control. The intercalated discs of mouse heart, which possess desmosome-like structures, and the desmo- somes of mouse intestine (5) are readily visualized in tissue sections using the desmoplakin antiserum, whereas these same structures are not stained by any of the pemphigus serum samples (results not shown).

The Visualization of Desmosomes in Cultured Cells. We selected a mouse keratinocyte culture system to determine the presence or absence of desmosome autoantibodies in our 13 pemphigus serum samples. In this culture system kerati- nocytes grown in low Ca2+-containing medium (~0.1 mM) possess no desmosomes (18). Desmosome formation is in- duced by increasing the Ca 2÷ to normal levels (-1.2 mM Ca -'+) (18).

Desmosome-possessing keratinocytes were therefore proc- essed for double indirect immunofluorescence using the pem- phigus serum samples and the desmoplakin antiserum. The four Pv serum samples and four of the five unclassified pemphigus serum samples yield a pattern of intermittent single lines along the borders of the contacting keratinocytes (19, Fig. 2b). This pattern was similar in overall distribution to that seen in similar preparations using desmoplakin anti- serum. However, this latter pattern usually revealed double lines instead of single lines at the rites of desmosomes (Fig. 2). In double label observations, the nonfluorescent line that separates a double band observed by immunofluorescence

using the desmoplakin antiserum appears to be coincident with the single fluorescent line stained by the pemphigus autoantibodies (Fig. 2, a and b).

The known Pf serum samples (see Materials and Methods), one of the unclassified serum samples, and normal human serum did not stain the desmosomes present at the contacting surfaces of kerafinocytes maintained in normal levels of Ca 2+ (results not shown).

Western Immunoblotting Analyses of Pemphigus Autoantibodies To determine the target antigen(s) of the pemphigus auto- antibodies, we used desmosome-enriched fractions from bo- vine tongue for analysis by Western immunoblotting (see Materials and Methods).

The pemphigus serum samples react with several polypep- tides in the desmosome-enriched preparations, which are not recognized by normal human serum (Fig. 3). There are certain polypeptides that are recognized by subgroups within the I 1 pemphigus serum samples shown in Fig. 3. Indeed, in some cases these subgroups can be correlated with known Pv and Pf samples. For example, two of the four known Pf serum samples, in addition to one of the unclassified samples (which fails to stain the cell surface of cultured keratinocytes [see above]) recognize 160 and 165-kD polypeptides (Fig. 3, lanes B, C, and D) (22). Furthermore, the four known Pv serum samples and the remaining four unclassified serum samples all recognize a 140-kD polypeptide in the tongue desmosome preparations (Fig. 3, lanes G-L) (only the immunoblots of two of the known Pv serum samples [Fig. 3, lanes G and H] are shown).

To determine whether autoantibodies directed against the 140-kD desmosome-associated polypeptide contribute to the desmosome staining pattern we observe in cultured kerati- nocytes (19, see Fig. 2b), absorption of a Pv serum sample

Figure 2. Keratinocytes, 4 h after the Ca 2+ switch, were processed for double indirect immunofluorescence using the desmoplakin antiserum (a) and a pemphigus serum (b) (see Fig. 3, lane G for the immunoreactivity of this particular Pv serum sample with an enriched fraction of desmosomes). The desmoplakin antiserum stains a double line along an area of cell-cell contact (a, arrow). The pemphigus serum stains a single line along the same border (b, arrow). This single line appears to fill in the nonfluorescent area observed between the double line stained by the desmoplakin antiserum. Bar, 5 urn.

The Journal of Cell Biology, Volume 102, 1986 1112

Figure 3. An enriched preparation ofdesmosomes isolated from bovine tongue was subjected to SDS PAGE and electrophoretically transferred to nitrocellulose. Lane A shows an Amido black stain of transferred proteins. Note desmoplakins l and 2 (DI, D2) (arrows indicate, from top to bottom, 200-, 120-, 92-, 80-, and 50-kD). Lanes B-L are immunoblots using pemphigus serum samples, whereas lane M is an immunoblot using a normal human serum sample. Lanes B, C, E, and F are immunoblots of Pf serum samples. Lanes G and H are immunoblots of Pv serum samples. Note the 160- and 165-kD polypeptides (arrows) which are recognized by serum samples in lanes B, C, and D and the 140-kD polypeptide (arrow) recognized by serum samples in lanes G-L. Polypeptides recognized by both the normal human serum sample, in addition to the pemphigus serum samples, are arrowed in lane M.

was done using protein immobilized on nitrocellulose follow- ing a procedure described by Cox et al. (12). SDS PAGE was done on a desmosome-enriched fraction, and separated poly- peptides were transferred to nitrocellulose. A small vertical strip of this nitrocellulose sheet was stained for protein with Amido black (see Materials and Methods) for orientation. A horizontal piece of nitrocellulose that contained the 140-kD polypeptide was cut from the remaining nitrocellulose and incubated for 6 h at 4"C with the Pv serum sample diluted 1:10 in PBSa. Cultured desmosome-possessing keratinocytes were then processed for indirect immunofluorescence with the absorbed serum sample. After absorption no desmosome staining could be detected (results not shown). Immunoblot- ting analyses confirmed that autoantibodies to the 140-kD polypeptide were greatly reduced in the absorbed serum sam- ple (Fig. 4). Control experiments in which the Pv serum sample was absorbed against nitrocellulose pieces that con- tained desmosomal components other than the 140-kD poly- peptide were also done. No difference in staining intensity of desmosomes of cultured keratinocytes was detected after ab- sorption of the serum sample with such polypeptides.

Koulu et al. (22) were unable to show reaction of Pv serum samples with any desmosome proteins, either by immunopre-

cipitation or immunoblotting. Thus their results concerning Pv patients are different from ours. One possible explanation for these differences may lie in the use ofdesmosome fractions obtained from different tissues. We used enriched fractions of bovine tongue desmosomes, whereas these other workers used bovine muzzle desmosome fractions. We wondered, therefore, whether there were differences in cross-reactivity of pemphi- gus autoantibodies with bovine tongue and muzzle desmo- some preparations. To test this possibility, we isolated and compared desmosomes from bovine muzzle and tongue. These two fractions of desmosomes look identical by ultra- structural analyses (results not shown). However, the SDS PAGE profiles of such preparations differ quite considerably (Fig. 5). Immunoblotting analyses using two of each of the known immunoreactive Pf and Pv serum samples on both bovine tongue and muzzle desmosome preparations are shown in Fig. 6 (see also Table I). The Pf serum samples react with polypeptides in the 160-165-kD range in the tongue desmosome preparation (confirming the results we described above, see also 22) and also in the muzzle desmosome prep- aration (Fig. 6, lanes B, C, G, and H). In contrast, although the two Pv serum samples react strongly with a 140-kD polypeptide in the bovine tongue desmosome preparation

1 1 13 Jones et al. Pemphigus Autoantibodies

Figure 4. A bovine tongue des- mosome-enriched preparation was subjected to SDS PAGE and transferred to a nitrocellulose sheet. A thin vertical strip was cut from the sheet and stained with Amido black (lane 14). As described in the text, a 140-kD polypeptide (arrow in lane A) was used to absorb out autoantibod- ies from a Pv serum sample. Lane B shows the immunoreactivity of this Pv serum sample with the desmosome preparation before absorption (see also Fig. 3, lane G). Lane C shows the immuno- reactivity of the same serum sam- ple after absorption.

Figure 5. SDS PAGE profiles of a bovine tongue desmosome-en- riched fraction (A) and a bovine muzzle desmosome-enriched fraction (B) are shown. Note that although both these preparations are enriched in desmoplakin 1 and 2 (D1 and D2), there are differences between the two pro- files.

(Fig. 6, lanes D and E), one of these Pv serum samples shows an extremely weak, barely detectable reactivity with a 140- kD polypeptide in the bovine muzzle desmosome preparation (Fig. 6, lane I), whereas the other Pv serum sample does not show a reaction with any polypeptide in the 140-kD molecular weight range in the same preparation (Fig. 6, lane J). Fur- thermore, neither the two other Pv serum samples nor the unclassified serum samples show immunoreactivity with poly- peptides in the 140-kD range in the muzzle desmosome preparation (see Table I).

Our results would thus appear to suggest that desmosomal antigens recognized by the Pf and Pv autoantibodies are different as determined by their different immunoblotting reactions with the bovine tongue desmosome preparation (Fig. 3). Moreover, the 140-kD polypeptide recognized by the Pv autoantibodies is either absent or in greatly reduced amounts in bovine muzzle desmosomes as compared to bo- vine tongue desmosomes (Fig. 5). These data, in addition to the histology of Pv and Pf blisters (Pv blisters form suprabas- ally, whereas Pf blisters form higher up within the stratum spinosum) led us to determine whether we could detect dif- ferences in the location of Pf and Pv autoantibody binding in cryostat sections of epidermis. Currently all of the Pv and Pf serum samples that we have tested stain most areas of cell contact in neonatal mouse epidermis (Fig. 1). However, one Pf serum sample generates an intercellular staining pattern only in the upper layers of the epidermis of bovine tongue (Fig. 7). This particular serum recognizes the 160/165-kD polypeptides of the desmosome fractions (Fig. 3, lane B). We have found this differential staining pattern only in bovine tongue epidermis and not in neonatal mouse epidermis using this serum sample. These results are probably partly due to the large number of epidermal cell layers in the tongue epithelium, which may afford a better opportunity to resolve such differential staining characteristics.

Table I summarizes both our immunofluorescence and immunoblotting results from the 13 pemphigus serum sam- ples.

Discussion

We previously reported that the sera of three pemphigus patients contained autoantibodies that recognized the des- mosomes of keratinocytes maintained in tissue culture (19). We have now extended these observations and have analyzed 13 pemphigus serum samples by immunofluoreseence, using both cryostat sections of a variety of tissues and cultured keratinocytes, and by immunoblotting, using enriched prep- arations of bovine tongue and muzzle desmosomes. The 13 pemphigus serum samples display immunofluoreseence pat- terns on cryostat sections of mouse epidermis that are similar to those observed using the desmoplakin antiserum. Indeed double indirect immunofluorescence using the pemphigus serum samples and a desmoplakin antiserum reveal that the pemphigus autoantibodies localize in the same region as the desmoplakin antibodies, which suggests that they are directed against desmosome components. Eight of the pemphigus serum samples (including the known Pv samples and the three samples we used in our previous study [19]) stain a single fluorescent line at the site of desmosomes of cultured kerati- nocytes as determined by indirect immunofluorescence. This line appears to co-localize with the material that comprises

The Journal of Cell Biology, Volume 102. 1986 1114

Table I. lmmunofluorescence and Immunoblotting Reactivity* of Pemphigus Autoantibodies

Tongue desmosome Muzzle desmosome fraction fraction Desmosome staining pattern

in cryostat sections of epider- Desmosome staining patterns Serum sample Source Classification mis in cultured keratinocytes 160/165 kD 140 kD 160/165 kD 140 kD

1 A P + - + - + -

2 B Pf + - + - + - 3 B Pf + - + - + - 4 B Pf + . . . . . 5 B Pf + . . . . . 6 B Pv + + - + - WR 7 B Pv + + - + - - 8 B Pv + + - + - - 9 B Pv + + - + - -

10 A P + + - + - - 11 A P + + - + - - 12 A P + + - + - - 13 A P + + - + - - 14t A Normal . . . . . .

A, obtained from Northwestern University. B, obtained from Dr. J. Stanley, National Institutes of Health (see Materials and Methods). P, pemphigus. WR, weak reactivity. * Only reactivity that appears common to groups of patients is tabulated. * Four normal serum samples were tested.

Figure 6. A bovine tongue desmo- some-enriched preparation (lanes A- E) and a bovine muzzle desmosome enriched preparation (lanes F - J ) were subjected to SDS PAGE and electrophoretically transferred to ni- trocellulose. Lanes A and F show Amido black stains of transferred proteins. Lanes B and G are immu- noblots using one Pf serum sample. An immunoblot of this same serum sample is shown in Fig. 3, lane B. Lanes C and H are immunoblots using a second Pf serum sample (see also Fig. 3, lane C). Note that in both tongue and muzzle desmosome frac- tions these two serum samples rec- ognize 160- and 165-kD polypep- tides (arrowed). Lanes D and I are immunoblots using one Pv serum sample (see also Fig. 3, lane G). Lanes E and J are immunoblots us- ing a second Pv serum sample (see also Fig. 3, lane H). Note that these sera recognize a 140-kD polypeptide in the tongue desmosome prepara- tion (lanes D and E). However, these same sera either do not react or react extremely weakly with polypeptides in the 140-kD molecular weight range in the muzzle desmosome preparation (lanes I and J).

t h e d a r k l ine s epa ra t i ng t h e t w o ha lves o f a d e s m o s o m e as d e l i n e a t e d by use o f t h e d e s m o p l a k i n a n t i b o d y p r e p a r a t i o n s (see Fig. 2 a n d r e f e r ence 14). T h e s e o b s e r v a t i o n s suggest t ha t t h e a n t i g e n i c d e t e r m i n a n t s r e c o g n i z e d by t h e s e p e m p h i g u s a u t o a n t i b o d i e s lie s o m e w h e r e w i t h i n t he r eg ion b e t w e e n t h e t w o s u b p l a s m a l e m m a l d e s m o s o m a l p l a q u e s r e c o g n i z e d by the d e s m o p l a k i n a n t i b o d i e s (14). F u r t h e r m o r e , we feel t h e p e m -

ph igus a u t o a n t i b o d i e s p r o b a b l y b i n d to t h e o u t e r sur face o f l iving ke r a t i nocy t e s as i n d i c a t e d by o u r ear l ie r s tudy t h a t d e m o n s t r a t e d tha t a loss o f ce l l -ce l l c o h e s i o n o c c u r r e d af ter e x p o s u r e o f l iv ing ke r a t i nocy t e s to p e m p h i g u s s e r u m s a m p l e s o n l y a t cell su r face si tes o c c u p i e d by d e s m o s o m e s (19). Based o n these da ta , we fu r t he r p r o p o s e tha t p e m p h i g u s an t i gens are m o s t l ikely p o s i t i o n e d in t h e ex t race l lu la r r eg ion b e t w e e n

1115 Jones et al. Pemphigus Autoantibodies

Figure 7. Cryostat section of bovine tongue processed for indirect immunofluorescence using a Pf serum sample (the immunoblot of this serum on a desmosome preparation is shown in Fig. 3, lane B). Note that the serum stains mainly in the upper layers of the epidermis, while cells in the lower layers of the epidermis (lower right) are unstained. Bar, 20 ~m.

the two apposing membranes that comprise the desmosome. The pemphigus serum samples exhibit a variety of reactions

as determined by immunoblotting using enriched prepara- tions of bovine tongue desmosomes. However, even though we have tested a relatively small number of known Pv and Pf serum samples, the immunoblotting data are very provoca- tive. These data suggest that there may be a correlation between the 160- and 165-kD desmosome-associated poly- peptides and the autoantibodies found in some but not all Pf patients (see also reference 22) and a 140-kD desmosome- associated protein and the autoantibodies found in the serum of Pv patients. This latter polypeptide may be that described by Stanley et al. (28) as a cell surface glycoprotein recognized by Pv autoantibodies. Furthermore, it is also possible to divide the unclassified pemph~gus serum samples into those that react with the 160- and 165-kD desmosome-associated poly- peptides (one serum sample) and those that cross-react with the 140-kD desmosome-associated polypeptide (four serum samples). Whether this type of immunoblotting analysis will allow us to differentially diagnose patients with different forms of pemphigus awaits the analysis of a significantly larger number of known Pf and Pv serum samples. However, our results tempt us to suggest that the four unclassified serum samples that recognize the 140-kD polypeptides are from Pv patients, whereas the remaining serum sample is from a Pf patient. Indeed, additional evidence for this supposition is

provided by our finding that the four unclassified serum samples stain the desmosomes of cultured keratinocytes sim- ilarly to that observed using the known Pv serum samples, whereas the one remaining unclassified serum sample in our collection, like the known Pf serum samples, does not.

Our immunoblotting results are, in part, in conflict with two reports in the literature, both of which have suggested that Pv autoantibodies do not react with desmosomal poly- peptides by immunoblotting or immunoprecipitation anal- yses (16, 22). Our results and those from other laboratories appear to differ because of the use of different tissues as sources of desmosomes. The Pv samples show immunoreac- tivity with desmosome-associated polypeptides obtained from enriched fractions of tongue desmosomes, which appear either to be missing or are available in much lower amounts in similar preparations isolated from muzzle.

These data suggest that desmosomes in different types of stratified squamous epithelia are biochemically distinct. In addition, pemphigus autoantibodies cross-react only with des- mosomes of stratified squamous epithelial tissues as deter- mined by immunofluorescence. This raises the distinct pos- sibility that the composition of desmosome components var- ies both amongst different cells in different epithelial tissues and from cell to cell within a single type of tissue (e.g., the stratified squamous epithelia of tongue and muzzle). Thus it is our beliefthat desmosomes are more biochemicalIy diverse than previously thought (10, 11). Indeed, this possibility has been suggested previously (6). Moreover, Guidice et al. (17), using monoclonal antibodies directed against desmosome components, have shown diversity of desmosome composi- tion in different tissues.

A striking difference between Pf and Pv serum samples is that the Pv samples stain the desmosomes of cultured kerati- nocytes, whereas the Pf samples do not. Since all our serum samples appear to contain autoantibodies that co-localize with the desmoplakin antiserum in cryostat sections of mouse skin, we wondered why these same serum samples did not all stain the desmosomes of cultured keratinocytes. We are tempted to speculate that the autoantibodies in Pv serum samples recognize desmosome-associated cell surface antigens respon- sible for cell adhesion in basal cells, i.e., the types of kerati- nocytes that grow in cell cultures. On the other hand it may be that the antigens responsible for cell adhesion recognized by Pf autoantibodies (we propose these to be desmosome- associated based on the immunoblotting data and the tissue staining observations) are accessible for antibody binding only in the more differentiated cells present in the upper layers of the epidermis. Indeed, our finding that one Pf serum sample stains primarily the borders of epidermal cells in the superfi- cial layers supports this contention and also confirms reports in the literature (7, 8). In addition, this observation suggests that desmosomal components undergo some form of modi- fication during differentiation of keratinocytes. In support of this notion, it has already been shown that the keratin subunits that comprise intermediate filaments, which are associated with desmosomes (2), undergo modification during kerati- nocyte differentiation in vivo (29).

Our results indicate that autoantibodies in many pemphigus patients recognize the desmosomes found in stratified squa- mous epithelia. The desmosomal antigens recognized by cer- tain Pf and Pv autoantibodies are distinct and may be differ-

The Journal of Cell Biology, Volume 102, 1986 1116

entially localized within the cells associated with the different layers of the epidermis. Furthermore, some desmosome-as- sociated antigens may be found in varying amounts in differ- ent areas of the body within the same tissue type (e.g., strati- fied squamous epithelia). If these desmosome-specific auto- antibodies are pathogenic in the disease, these results may begin to explain why only stratified squamous epithelial tis- sues are affected by the autoantibodies in pemphigus, and why there is regional blister formation in many pemphigus patients (e.g., in Pv, the oral mucosa may be the only region of the body affected by the disease).

In summary, the autoantibodies in pemphigus may prove to be valuable tools in the determination of the composition of desmosomes of stratified squamous epithelia. These auto- antibodies can be used as probes to investigate the structural heterogeneity of desmosomes at both the morphological and biochemical levels. To this end we are currently attempting to further characterize desmosome-specific autoantibodies from pemphigus serum samples by affinity purification of monospecific antibody fractions using desmosome-derived polypeptides. We are also in the process of making chicken and rabbit antibodies to those desmosomal proteins recog- nized by autoantibodies in several pemphigus sera (e.g., the 165-, 160-, and 140-kD desmosome-associated polypeptides mentioned above). Ultimately we hope to analyze the ability of such antibodies to mimic pemphigus autoantibodies by immunofluorescence, immunoblotting, and in in vivo model systems (e.g., skin explants [26] and cultured keratinocytes [ 19]). We are optimistic that these studies will lead ultimately to the analysis of specific cell-cell adhesion molecules local- ized within desmosomes found in cells that comprise different types of epithelial tissues. In addition, our studies should provide new insights into the cellular and molecular mecha- nisms involved in the blistering which characterizes pemphi- gus, and ultimately the data obtained may be useful in design- ing more specific diagnostic protocols and treatments for the disease.

We wish to thank Drs. James Arnn and Dr. Andrew Staehelin for the gift of the desmoplakin antiserum and also Ms. Laura Davis for her considerable patience during the preparation of this manuscript.

Research support has been provided by grants to Dr. Robert D. Goldman from the National Institutes of Health.

Received for publication 17 April 1985, and in revised form 6 November 1985.

References

1. Arnn, J. 1983. Ph.D dissertation. University of Colorado, Boulder. 2. Arnn, J., and L. A. Staehelin. 1981. The structure and function of spot

desmosomes. Dermatology. 20:330-339. 3. Beutner, E.H.,andR. E. Jordan. 1964. Demonstration ofskin antibodies

in sera of pemphigus vulgaris patients by indirect immunofluoreseent staining. Proc. Soc. Exp. Biol. Med. 117:505-510.

4. Beutner, E. H., T. P. Chorzelski, and R. E. Jordan. 1970. Autosensiti- zation. In Pemphigus and Bullous Pemphigoid. Charles C Thomas, Springfield,

IL. 5-116. 5. Bloom, W., and D. W. Fawcett. 1986. A Textbook of Histology. W. B.

Saunders Co., Philadelphia. 65, 303, 646. 6. Borysenko, J. Z., and J. R. Revel. 1973. Experimental manipulation of

desmosome structure. Am. J. Anat. 137:403-422. 7. Bystryn, J.-C., and J. Rodriquez. 1978. Absence of intercellular antigens

in the deep layers of the epidermis in pemphigus foliaceus. J. Clin. Invest. 61:339-348.

8. Bystryn, J.-C., E. Abel, and C. DeFeo. 1974. Pemphigus foliaceus: subcorneal intercellular antibodies of unique specificity. Arch. Dermatol. 110:857-861.

9. Cohen, S. W., G. Gorbsky, and M. S. Steinberg. 1983. Immunochemical characterization of related families of glycoproteins in desmosomes. J. Biol. Chem. 258:2621-2627.

10. Cowin, P., and D. S. Garrod. 1983. Antibodies to epithelial desmosomes show wide tissue and species cross-reactivity. Nature (Lond.). 302:148-150.

11. Cowin, P., D. Mattey, and D. S. Garrod. 1984. Distribution of desmo- somal components in the tissues of vertebrates, studied by fluorescent antibody staining. Z Cell Sci. 66:119-132.

12. Cox, J. V., E. A. Schenk, and J. B. Olmsted. 1983. Human anticentro- some antibodies: distribution, characterization of antigens and effect on micro- tubule organization. Cell. 35:331-339.

13. Dahl, M. 1981. Clinical Immunodermatology. Year Book Medical Pub- lishers Inc., Chicago. 144-147.

14. Franke, W. W., R. Moll, H. Mueller, E. Schmid, C. Kuhn, R. Krepler, U. Artlieb, and H. Denk. 1983. Immunocytochemical identification of epithe- lium-derived human tumors using antibodies to desmosomal plaque proteins. Proc. Natl. Acad Sci. USA. 80:543-547.

15. Gorbsky, G., and M. Steinberg. 1981. Isolation of the intercellular glycoproteins of desmosomes. Z Cell Biol. 90:243-248.

16. Gorbsky, G., S. Cohen, and M. S. Steinberg. 1983. Desmosomal antigens are not recognized by the majority of pemphigus autoimmune sera. J. Invest. Dermatol. 80:475-480.

17. Guidice, G. J., S. M. Cohen, N. H. Patel, and M. S. Steinberg. 1984. Immunological comparison of desmosomal components from several bovine tissues. J. Cell. Biochem. 26:35-45.

18. Hennings, H., D. Michael, C. Cheng, P. Steinert, K. Holbrook, and S. H. Yuspa. 1980. Calcium regulation of growth and differentiation of mouse epidermal cells in culture. Cell. 19:245-254.

19. Jones, J. C. R., J. Arnn, L. A. Staehelin, and R. D. Goldman. 1984. Human autoantibodies against desmosomes: possible causative factors in pem- phigus. Proc. Natl. Acad. Sci. USA. 81:2781-2785.

20. Jones, J. C. R., A. E. Goldman, H.-Y. Yang, and R. D. Goldman. 1985. The organizational fate of intermediate filament networks in two epithelial cell types during mitosis. Z Cell Biol. 100:93-102.

21. Jones, J. C. R., A. E. Goldman, P. M. Steinert, S. Yuspa, and R. D. Goldman. 1982. Dynamic aspects of the supramolecular organization of inter- mediate filament networks in cultured epidermal cells. Cell Motil. 2:197-213.

22. Koulu, L., A. Kusimi, M. S. Steinberg, V. Klaus-Kovtun, and J. R. Stanley. 1984. Human autoantibodies against a desmosomal core protein in pemphigus foliaceus. J. Exp. Med. 160:1509-1518.

23. Laemmli, U. K. 1970. Cleaverage of structural proteins during the assembly of the head of the bacterophage T-4. Nature (Lond). 227:680-685.

24. Lever, W. F., and G. Schaumberg-Lever. 1983. Histopathology of the Skin. 6th Ed. J. B. Lippincott Co., Scranton, PA. 106-114.

25. Mueller, H., and W. W. Franke. 1983. Biochemical and immunological characterization of desmoplakins I and 11, the major polypeptides of the desmosomal plaque. J. Mot. Biol. 163:647-671.

26. Schlitz, J. R., and B. Michel. 1976. Production of epidermal acantholysis in normal human skin in vitro by the lgG fraction from pemphigus serum. J. Invest. Dermatol. 67:254-260.

27. Skerrow, C. J., and A. G. Matoltsy. 1974. Isolation of epidermal des- mosomes. J. Cell Biol. 63:515-523.

28. Stanley, J. R., M. Yaar, P. Hawley-Nelson, and S. 1. Katz. 1982. Pemphigus antibodies identify a cell surface glycoprotein synthesized by human and mouse keratinocytes. J. Clin. Invest. 70:281-288.

29. Steinert, P. M., and W. W. Idler. 1979. Post synthetic modifications of mammalian epidermal a-keratin. Biochemistry. 18:5664-5669.

30. Towbin, H., T. Staehelin, and J. Gordon. 1979. Electrophoretic transfer of proteins from polyacrylamide gels to nitrocellulose sheets: Procedure and some applications. Proc. Natl. Acad Sci. USA. 76:4350-4354.

31. Wolff, K., and E. Schreiner. 1971. Ultrastructural localization of pem- phigus autoantibodies within the epidermis. Nature (Lond). 229:59-61.

1 1 17 Jones et al. Pemphigus A utoantibodies