Characterization of 230-kD Bullous Pemphigoid Antigen Associated with the Detergent-Insoluble Fraction of Cultured Keratinocytes

Scott M. Thacher and Patricia L. Hefti Department of Medical Biochemistry and Genetics, Texas A & M College of Medicine, College Station, Texas, U.S.A.

The 230-kD protein identified by antibodies from patients with bullous pemphigoid (BP) has a dual location in cultured normal human epidermal keratinocytes: part in a high-speed supernatant of homogenized cells and part in a particulate fraction, where it is resistant to extraction by non-ionic de­tergent or mild base. Antibodies were affinity purified from the particulate 230-kD BP antigen, which can be extracted in the presence of urea. The affinity-purified antibodies bind not only the cytosolic 230-kD protein, showing that it is related, if not identical, to the particulate form, but also pro-

B ullous pemphigoid (BP) is an autoimmune blistering disease that leads to dermal-epidermal separation; the sera of BP patients contain IgG that reacts primarily with intracellular [1- 3) but also, to a lesser extent, extracellular (4 ,5] sites on the hemidesmosome. As im­

plied by these ultrastructural data, the protein targets ofBP antibod­ies are diverse. A 230-kD antigen has been identified in extracts of epidermis by immunoblotting [6] . This same antigen, which has a basic isoelectric point, can be immunoprecipitated with BP antibod­ies from extracts of cultured rodent and human keratinocytes [3,7,8). On immunoblots, BP antibodies also recognize severa l other proteins in the molecular weight range of77- 200 kD (9 , 10]. Although the majority of BP sera are reported to immunoprecipi­tate the 230-kD antigen (8) , other studies show that when immu­noblotting is chosen as the method of detection, the 230-kD anti­gen is recognized by only about 50% of BP sera [9,10] . Because multiple antigens are recognized by BP sera, care must be taken in assigning subcellular locations for them. For example, some BP sera can be positive by immunoblotting for one of the recognized

Manuscript received April 19, 1990; accepted for publication August 6, 1990.

Pam of this work were presented at the 1989 combined meeting of the American Societies of Cell Biology and Biochemistry & Molecular Biology in San Francisco, CA.

Supported by in part by American Cancer Society Grant CD-352 and grant #RROS814 from the National Institutes of Health.

Reprint requests to: Dr. S.M. Thacher, DepartmentofMedical Biochemis­try & Genetics, 440 Medical Sciences Building, Texas A & M College of Medicine, College Station, TX 77843.

Abbreviations: BP: bullous pemphigoid OTT: dithiothreitol EDTA: disodium (ethylenedinitrilo) tetraacetic acid NP-40: Nonidet P-40 PBS: phosphate-buffered sa line PMSF: phenylmethylsulfonyl fluoride SDS: sodi um dodecyl sulfate TBS: Tris-buffered saline

duce a discontinuous granular pattern by indirect immunoflu­orescence in the basement membrane zone of rabbit esopha­gus. In stratifying epidermal cultures, expression of the 230-kD BP antigen is limited to basal cells. These data are consistent with 230-kD BP antigen involvement in keratin­ocyte basal cell interaction with extracellular matrix and in­dicate that the cultured cell may provide a useful model for analysis of 230-kD BP antigen function. J Invest Dermatol 96:139-143, 1991

bands but negative by indirect immunofluorescence on sections of skin [10).

The staining pattern with BP sera suggests that BP antigens, such as the 230-kD antigen, function in keratinocyte interaction with extracellular matrix. The cultured human epidermal keratinocyte is a useful model for study of adhesion as well as other processes of epidermal homeostasis (11). This study characterizes the 230-kD BP antigen that is present in the non-cytosolic or particulate frac­tion of homogenized cultured keratinocytes. Its subcellular location in the cultured cell may be compatible with function in processes of cell adhesion to substratum.

MATERIALS AND METHODS

Cell Culture Newborn foreskin keratinocytes, serially cultured to confluence with a 3T3 feeder cell layer (12], were rinsed in PBS-EDT A, harvested in 10 nU\1 benzamidine, 5,ug/mlleupeptin, 50 KIU (kallikrein inhibitor units)/ml aprotinin, 10 mM Tris pH 7.4, and 2 mM EDTA (low salt buffer), and stored at -70oC with addition of 0.5 mM PMSF.

Immunoblot Detection of 230-kD BP Antigen Several BP sera were used during initial characterization of230-kD BP antigen distribution in cultured cells; however, only one serum (#897, a gift of Dr. J.R. Stanley, NIH, Bethesda, MD) was the source of affinity­purified antibodies . Proteins extracted from keratinocytes were sepa­rated on SDS-polyacryla1mde gels (6.5%) 7 em in length (Mighty Small, Hoefer Instruments, San Francisco, CA) according to the method of Laemmli [13), electrophoretically transferred to nitro­cellulose filters [14], and stained with the BP serum according to Stanley et al [ 6]. Pre-stained molecular weight standards migrated as follows, according to the manufacturer (BRL, Bethesda, MD): myosin (211 kD), phosphorylase b (107 kD), and BSA (69 kD). Serum was diluted 1:600 in 0.3% NP-40, 0.5% BSA, and TBS (20 mM Tris pH 7.4, 0.15 M NaCl), and the second antibody, affinity-purifi~d goat anti-human IgG coupled to peroxidase (TAGO, Burlmgame, CA) was used at 1: 2,500. Peroxidase activity was detected by reaction with 0.003% H20 2 and 0.5 mg/ml diami­nobenzidine in TBS for 40 min at room temperature.

0022-202X/91/S03.50 Copyright © 1991 by The Society for Investigative Dermatology, Inc.

139

' 140 THACHER AND HEFTI

Extraction of 230-kD BP Antigen The cultured cells were thawed and Dounce homogenized in low-salt buffer, and the partic­ulate material recovered (10 min, 10,000 X g) was reextracted with low-salt buffer. The pooled supernatants were further clarified (1 h, 100,000 X g) to produce a high-speed supernatant. The combined pellets were then treated sequentially with 0.5% NP-40 and 0.5 mM DTT in low-salt buffer (0.25 ml per 10-cm plate of cells harvested) and 2 mM NaHC03 , pH 9.0, and recovered by centrifu­gation (20 min, 35,000 X g) after each step. The final pellet was further homogenized in 2 M or 8 M urea and 1 mM DTT in low­salt buffer and cleared by centrifugation at 100,000 Xg for 1 h.

Partial Purification of 230-kD BP Antigen The 230-kD. BP antigen in 8 M urea was extracted in a polypropylene tube with sequential additions of methanol, chloroform, and water (4: 1.5: 2). The aqueous phase was removed and three parts methanol added to the precipitated protein at the solvent interface. The precipitate was recovered by centrifugation and further washed in 95% methanol. The residue was dried under nitrogen gas, resuspended in 10 mM Tris pH 7.4, with 2 mMEDTA, 20 mM DTT, and 5% SDS, heated to 95 o C for 2 min, and clarified by centrifugation at 100,000 X g for 50 min. This procedure rapidly concentrates the protein while simultaneously removing excess urea.

The extracted material contained primarily keratins. These were separated from BP antigen on a 2.6- X 64-cm Sephadex S-500 col­umn. The resin was sterilized by heating to 95°C for 2 min in 50 mM Tris and 0.5% SDS. Column buffers, likewise sterilized, contained 0.5% SDS, 1 mM EDT A, and 0.02% sodium azide in either 50 mM Tris, pH 7.4, or 50 mM Borate-succinate, pH 7.4. The 230-kD BP antigen positive peak identified by immunoblot­ting overlapped only slightly with cell keratins (Fig 2). The column effiu ent was continuously monitored at 280 nm with an ISCO Type 6 absorbance detector and UA-5 chart recorder. The elution volume of the column was identified with 3H 20 added to the sam­ple. Purity of peak samples was analyzed on 3% polyacrylamide gels containing 0.5% agarose.

Affinity Purification of Antibody Specific antibodies to the 230-kD BP antigen were obtained by a modification of an earlier method [15] . A narrow strip of nitrocellulose filter containing par­tially purified 230-kD BP antigen was cut out following transfer from a 6.5% acrylamide gel and incubated in a 1:200 dilution of patient serum. A vertical strip of the same nitrocellulose filter, from which the 230-kD BP antigen was excluded, was treated similarly as a control. Both strips were washed twice in 30 mM Tris, pH 7.4, with 0.5% NP-40 and 0.3 M NaCl, rinsed briefly in water, and treated for 2 h with a minimal volume of2 M KI at 37°C. The filter eluate was cleared of particulate material (80 krpm, 15' in a Beck­man TL-100 ultracentrifuge) and concentrated twice, with dilution in TBS and centrifugation for 15' at 5500 rpm against a 30,000 molecular weight cut-off Centricon filter (Amicon Corp). The concentrated antibody was stabilized in the presence of 2% fetal bovine serum in TBS with 0.02% NaN3 and used without further di lution. The material recovered from the control strip was referred to as "sham"-purified antibody.

Indirect Immunofluorescence Cryostat sections (6 J.l.m thick) of rabbit esophagus or cultured keratinocytes, released as an intact sheet by the neutral protease, Dispase [16] , were air dried, fixed in acetone for 2 min, washed in TBS (2 X 5' each) and incubated in TBS containing 4% Dextran T-70 (Pharmacia) and 0.3% DMSO (37"C for 30 min) in preparation for incubation with the BP serum (1: 100 in TBS, 1% fetal bovine serum) or the affinity-purified antibody for 2 hat 37"C, according to methods described elsewhere (17]. After washing in TBS (3 X 5'), rabbit anti-human IgG coupled to Texas Red (Accurate, Westbury, NY) was applied in TBS with 0.5% BSA for one hour at 37 oC. After washing in TBS, the slides were mounted in a 80-mM Na-carbonate buffer, pH 9.5, with 20% glycerol and 0.1 mM a-phenylenediamine. Lumicon 2415 film, which is a gas-hypersensitized Kodak Type 2415 Technical Pan (Lumicon, Co., Livermore, CA) or Kodak T m.x 400 was used for

THE JOURNAL OF INVESTIG ATIVE DERMATOLOGY

• •

1 2 3 4 5 Figure 1. Identification of 230-kD BP antigen in cultured keratinocyte extracts by immunoblotting. Lane 1, particulate fraction after two extrac­tions in low salt buffer; lm1e 2, 0.5% NP-40 extract of cell particulate frac­tion; latle 3, 2M urea, low-salt buffer extract. Lanes 4 at1d 5, comparison of urea-extracted particulate (!a ti e 4) and cytosolic (lane 5) fractions of cultured keratinocytes. In proportion to number of cells extracted, the amount of cytosol run on the gel was approximately 40% less than the urea extract. The molecular weight marker shown is 211 kD.

photography on a Zeiss Fluorescence photomicroscope III equipped with a 63 X oil objective.

RESULTS

Cultured human keratinocytes have been shown to contain a pool of 230-kD BP antigen that can be extracted by homogenization in detergent-free buffer, as previously reported in studies of rat, mouse, and _human_ cultured keratinocytes [3,7,8). In this study, there was s1gm_ficant msoluble 230-kD BP antigen in the rernaining cellular de_bns recovered by centrifugation, as shown. by immunoblotting w1tl~ BP serum_(Fig 1, lane 1). The 230-kD BP antigen of this partlculate fracuon was clearly different from the cytosolic high­speed supernatant, because further extracts with 11-on-ionic deter­gent (Fig 1, lane 2) or mild alkali (2 mM NaHC0

3, pH 9.0 (not

shown]) were unable to release the antigen. However, the particu­lat_e 230-kD BP antigen could be partially extracted in 2 M urea, (F1g ~· lane 3)_. The urea-extracted 230-kD BP antigen appeared to co-~mgrate w1th t~1e cytosolic 230-kD BP antigen (Fig 1, latus 4, 5). Esumated by the mtensity of staining on several immunoblots, the partlculate fraction contained more than 50% of total cellular 230-kD BP antigen.

For furthe~ purification, the antigen was completely extracted from the part_lculate fraction in 8 M urea and chrornatographed by gel filtratlonm the presence of SDS (Fig 2). The 230-kD BP anti­gen (F_ig 2, imet) was largely separate from the keratins, which compnsed the majority of proteins in the second m:;~.jor protein peak from the column. The 230-kD BP antigen was not pure, as shown by the CoomaSSle Blue-stained gel of pool~d peak fractions (Fig 3), but nonetheless could be blotted onto nitrocellulose in a concen­trated form and used as a ligand to affinity purify antibodies from BP serum.

Specificity of the affinity-purified antibodies was demonstrated b_r imm_unoblott_ing of the initial cytoplasmic and particulate frac­twns (F1g 4). A smgle band at 230 kD was stained in each case (lafl es 1, 2) . As a control for non-specific adsorption of i:mmunoglobulin to nitrocellulose, sham-purified antibodies were obtained by elution of bound antibodies from vertical strips of nitrocellulose lying below the 230-kD band region. These antibodies did not react to the 230-kD band (lanes 3, 4). It is notable that there: was no staining in the molecular weight range between 77 kD and 230 kD, in which other BP antigens have been found [9,10], in either the experimental or control i1mnunoblots. At lower molecular weights, the staining was non-specific and was due to cross-reaction to epi­dermal keratinocyte proteins by the second antibody. These results

VOL. 96, NO. 1 JANUARY 1991

2.0

E c:

0 co ~

.__,-

> 44 46 48 50 52 54 -·c;; 1.0 c:

C1)

c c;; 0

. ·.;::; a. 0

0 20 30 40 50 60 70 80

Fraction (4.8ml)

Figure 2. Partial purification of 230-kD BP antigen by gel filtration. The optical density at 280 nm (-- )shows two peaks: one at the void volume and a second peak containing primarily keratins, as determined by SDS-gel electrophoresis. I11set, immunoblot of 230-kD BP antigen in column frac­tions. The peak of 230-kD BP antigen lies in the region between the rwo main protein peaks. Fractions 45-53 were pooled for further study.

show that the same subset of IgG in BP sera recognizes both the particulate and the cytoplasmic forms of the 230-kD BP antigen.

T he affinity-purified BP antibodies also reacted with the base­ment membrane zone of rabbit esophagus. In comparison to control staining with sham-purified antibodies (Fig Sa), the basement membrane zone was clearly stained (Fig Sb) by indirect immunoflu­orescence. In oblique sections, at higher magnification, BP sera revealed a variegated pattern of fluorescence (Fig 6a). An earlier study of neonatal mouse epidermis also reported a similar macular staining [18] and suggested that it was due to the discrete nature of hemidesmosomes. The affinity-purified antibodies produced a com­parable pattern (Fig 6b), but with reduced intensity at the individual spots. The diffusely staining regions shown in Fig 6a,b could be

1 2

Figure 3. Anal ysis of 230-kD BP antigen pooled peak fractions following gel chromatography. Samples were separated on a 3% acrylamide, 0.5% agarose ge l. La11e 1, stained with Coomassie Blue. Arrofl;lzead indicates possi­ble 230-kD BP antigen band. La11e 2, detection of 230-kD BP antigen by immunoperoxidase following transfer to nitrocellulose. Molecular weight standard, 211 kD.

230-kD BULLOUS PEMPHIGOID ANTIGEN 141

• • •

1 2 3 4 Figure 4. :Sp,·.: ,tic,ty ot .!thiuty-purified antibody to the 230-kD BP anti­gen. Cytoplasmic extract, la11es 1 a11d 3. Total cell particulate extracted in 8 M urea, la11es 2 and 4. The affinity purified antibodies were used to stain lan es 1 and 2; the sham-purified antibody, lan es 3 and 4. Molecular weight markers are 211 kD, 107 kD , and 69 kD.

resolved into the same discontinuous pattern by adjustment of the focal plane of the objective lens .

To further evaluate 230-kD BP antigen expression, we stained stratified normal human keratinocytes in culture. These cells were lifted from the culture dish without loss of cell orientation by treat­ment with the neutral protease, Dispase [16] . Indirect immunofluo­rescence of frozen sections with affinity-purified antibodies showed that the 230-kD BP antigen was expressed in basal cells of the stratifying cultured epidermal keratinocytes and was concentrated towards the basal portion of the individual cells (Fig 7).

Figure 5. Detection of230-kD BP antigen at the basement membrane zone of rabbit esophagus by indirect immunofluorescence staining with: a, sham-purified antibodies; b, affin ity-purified antibodies (bar, 10 pm). •

142 THACHER AND HEFTI

Figure 6. Indirect immunofluorescence along oblique sections of the rabbit esophageal basement membrane zone. Comparison of unfra~twnated BP sera and affi nity punfied antibodies: a, BP serum; b, affimty-punfied antibod­ies to the 230-kD BP antigen (bar, 5 f1m).

DISCUSSION

Human keratinocytes cultured in the presence of 1.6_ mM Ca++ _and a 3T3 feeder layer express the 230~kD ~p antigen 111 two d1stmct subcellular fractions. One of these IS a h1gh-speed supernatant, ob­tained following Dounce-homogenization of keratmocytes 111 buf­fered aqueous solutions lacking de~er~ents or den~turants. The sec­ond pool of 230-kD BP antigen Is 111 the result111g pellet ?f cell fragments. The 230-kD protein in this pool resists extractiOn by non-ionic detergent but can be solubilized in 2M and 8 M ur~a.

The functional relation of these two pools of 230-kD BP ant1ge_n is uncl ear; however, the 230-kD BP antigen in the cytopla_smic fraction of cell extracts is antigenically related to that asso~l~t.ed

with the urea-soluble matrix (Fig 4) , leaving open the poSSibility that there is a precursor-product relationship between .the two. The cytosolic 230-kD BP antigen has. not b~en reported 111 .extracts of human epidermis; instead, SDS IS required f?r ex~ract1on [6). In sequential extractions of monkey (Macaca Jascr~ulans) epi?er~ms or bovine tongue epithelium, the 230-kD BP antigen likewise IS r~n­dered soluble only in the presence of SDS or urea and reducmg agent.* Thus, the cytosolic 230-kD BP antigen of cultured cells has

• Tang H-Y , Thacher SM (unpublished observations).

THE JOURNAL OF INVESTIGATIVE DERMATOLOGY

Figure 7. Stratifying normal human keratinocytes in cu lture released from the culture dish by treatment with Dispase. Indirect immunofluorescence staining with a, affinity-purified antibodies; b, sham-purified an tibodies (bar, 10 f1m).

no obvious parallel in vivo. An earlier report stated that the 230-kD BP antigen could be immunoprecipitated from the soluble extracts of cultured mouse keratinocytes grown in low (0.07 mM) ca++, but not from urea extracts of cell particulates [7]. Given the pres~me.d location of the 230-kD BP antigen in the hemidesmosome, wh1ch IS

a highly insoluble structure [2, 19], this observation suggests that cultured keratinocytes might be of questionable relevance to under­standing 230-kD BP antigen function. The discrepancy betwe~n this earlier study and the present work may be due either to species differences or to the method of detection of the 230-kD BP antigen. Keratinocytes grown in th~ low (0 .15 mM) Ca++ medium of Boyce & H am [20] also contain a particulate pool of 230-kD BP antigen recognized by immunoblotting, * so that the level of Ca++ in cell culture appears unlikely to be a critical factor. The urea extract­ability of 230-kD BP antigen from the cell particulate fraction is similar to cytoskeletal proteins such as keratins, suggesting the possi­bility that the 230-kD BP antigen is involved in some aspects of cytoskeletal interactions as well.

The staining by affinity-purified antibodies to the 230-kD BP antigen on the rabbit esophagus validates, as other studies have [8,21], the assumption that the 230-kD BP antigen is in fact pre.sent at the basement membrane zone. The variegated pattern in oblique sections (Fig 6) is consistent with binding to the hemidesmosome; monospecific antibodies to 230~kD BP anti g~n have also been shown to bind the cytoplasmic side of the hemtdesmosome m bo­vine tongue and human skin [21,22] . Th~ di~tri~ution ~f 230-kp BP antigen to the basal pole of basa l cells 111 vivo IS recapttulated li1

stratifying normal keratinocytes in culture (F1g 7), suggest111g that

VOL. 96, NO. 1 JANUARY 1991

the processes that localize 230-kD BP antigen in vivo are still opera­tive in the absence of dermis. Structures closely resembling the hemidesmosome cytoplasmic dense plaque and sub-basal dense plate have also been described in human epidermal cells cultured by the same 3T3 feeder cell method used here (23,24]. The cultured keratin.ocyte may be a valuable system for testing the hypothesis that the 230-kD BP antigen is functionally important in maintenance of cell polarity or cell-matrix adhesion.

We thank Dr.J.R. StatJley for gifts ofBP patiwt sera; Dr. R.C. BtJrglwrdt, Texas A & M University, for advice and assistance concern ing ittttlttlttojluorescwce photogra­phy; attd Dr. S. Zhao for help with some of these experiments.

REFERENCES

1. Yamasaki Y, Nishikawa T: Ultrastructural localization of in vitro binding sites of circulating anti-basement membrane zone antibod­ies in bullous pemphigoid. Acta Derm Venereol63:501 - 506, 1983

2. Mutasim DF, Takahashi Y, Labib RS, Anhalt GJ, Patel HP, Diaz LA: A pool of bullous pemphigoid antigen(s) is intracellular and associated with the basal cell cytoskeleton-hemidesmosome complex. J Invest Dermatol84:47-53, 1985

3. Westgate GE, Weaver AC, CouchmanJR: Bullous pemphigoid anti­gen localization suggests an intracellular association with hemides­mosomes.J Invest Dermatol84:218-224, 1985

4. Mutasim DF, Morrison LH, Takahashi Y, Labib RS, Skouge J, Diaz LA, Anhalt GJ: Definition of bullous pemphigoid antibody binding to intracellular and extracellular antigen associated with hemides­mosomes. J Invest Dermatol 92:225-230, 1989

5. Regnier M, Vaigot P, MichelS, Prunieras M: Localization of bullous pemphigoid antigen (BPA) in isolated human keratinocytes. J Invest Dermatol85:187-190, 1985

6. Stanley JR, Woodley DT, Katz SI: Identification and partial character­ization of pemphigoid antigen extracted from normal human skin. J Invest Dermatol 82:108-111, 1984

7. Stanley JR, Hawley-Nelson P, Yuspa SH, Shevach EM, Katz SI: C har­acterization of bullous pemphigoid antigen: a unique basement membrane protein of stratified squamous epithelia. Cell 24:897-903, 1981

8. Mueller S, Klaus-Kotvun V, Stanley JR: A 230-kD basic protein is the major bullous pemphigoid antigen. J Invest Dermatol 92:33-38, 1989

9 . Labib RS, Anhalt GJ, Patel HP, Mutasim DF, Diaz LA: Molecular heterogeneity of the bullous pemphigoid antigens· as detected by immunoblotting. J Immunol 136:1231-1235, 1986

10. Bernard P, Didierjean L, Denis F, Saurat J-H , Bonnetblanc J-M: Het-

230-kD BULLOUS PEMPHIGOID ANTIGEN 143

erogeneous bullous pemphigoid antibodies: detection and character­ization by immunoblotting when absent by indirect immunofluores­cence. J Invest Dermatol 92:171 - 174, 1989

11. Toda K-1, Tuan T -L, Brown PJ, Grinnell F: Fibronectin receptors of human keratinocytes and their expression during cell culture. J Cell Biol105:3097-3104, 1987

12. Thacher SM, Rice RH: Keratinocyte-specific transglutaminase of cul­tured epidermal cells: relation to cross-linked envelope formation and terminal differentiation. Cell 40:685-695, 1985

13. Laemmli UK: Cleavage of structural proteins during the assembly of the head of bacteriophage T 4 • Nature 227:680 - 685, 1970

14. Tow bin H, Staehlin T , Gordon J: Electrophoretic transfer of proteins from polyacrylamide gels to nitrocellulose sheets: procedure and some applications. Proc Nat! Acad Sci USA 76:4350-4354, 1979

15. O lmsted JB: Affinity purification of antibodies from diazotized paper blots of heterogeneous protein samples. J Bioi Chem 256:11955-11957, 1981

16. Watt FM: Selective migration of terminally differentiating cells from the basal layer of cultured human epidermis. J Cell Bioi 98:16-21, 1984

17. Thacher SM: Purification of keratinocyte transglutaminase and its expression during squamous differentiation. J Invest Dermatol 92:578-584, 1989

18. Mutasim DF, Anhalt GJ, Diaz LA, Patel HP: Linear immunofluores­cence staining of the cutaneous basement membrane zone produced by pemphigoid antibodies: the result of hemidesmosome staining. JAm Acad Dermatol16:75-82, 1987

19. Skerrow CJ, H unter!, Skerrow D: Dissection of the bovine epidermal desmosome into cytoplasmic protein and membrane glycoprotein domains. J Cell Sci 87:411-421, 1987

20. Boyce SA, Ham RG: Calcium-regulated differentiation of normal human epidermal keratinocytes in chemically defined clonal culture and serum-free serial culture. J Invest Dermatol 81:33S-40S, 1983

21. Klatte DH, Kurpakus MA, Grelling KA, Jones JCR: Immunochemical characterization of three components of the hemidesmosome and their expression in cultured epithelial cell s. J Cell Bioi 109:3377-3390, 1989

22. Tanaka T, Korman NJ, Shimizu H,EadyRAJ, Klaus-Kotvun V, Cehrs K, Stanley JR: Production of rabbit antibodies against carboxy­terminal epitopes encoded by bullous pemphigoid eDNA. J Invest Dermatol 94:617 - 623 , 1990

23. Chapman SJ, Eady RAJ: Blistering in keratinocyte cultures: a regular phenomenon associated with differentiation. Eur J Cell Bioi 39:352-359, 1985

24. Compton CC, Gi llJM, Bradford DA, Regauer S, Gallico GG, O'Con­nor NE: Skin regenerated from cultured epithelial autografts on full-thickness burn wounds from 6 days to 5 years after grafting. A light, electron microscopic and immunohistochemical study. Lab Invest 60:600-612, 1989