ULTRASTRUCTURAL LOCALIZATION OF IMMUNOGLOBULINS IN BULLOUS PEMPHIGOID SKIN

Transcription

1 THE JOUIINAL OF INVJISTIGATJVE DBRMATOLOG~, 64: , by The Williams & Wilkins Co. Vol. 64, No.4 Printed in U.S.A. ULTRASTRUCTURAL LOCALIZATION OF IMMUNOGLOBULINS IN BULLOUS PEMPHIGOID SKIN Employment or a New Peroxidase-Antiperoxidase Multistep Method K. HoLUBAR, M.D., K. WoLFF, M.D., K. KoNRAD, M.D., AND E. H. BEUTNER, PH.D. Division of Experimental Dermatology, Department of Dermatology I., University of Vienna, Vienno, Austria, and Department of Microbiology, State University of New York at Buffalo, Buffalo, New York A multistep immunochemical method was developed for light and electron microscopy utilizing horseradish peroxidase as an immunologically bound marker. The method was applied to studies of in vivo bound IgG in bullous pemphigoid skin and of circulating antibasement membrane antibodies using rhesus esophagus as substrate. The immunocytochemical chain employed consisted of sequential binding of goat antihuman IgG, rabbit antigoat lgg, goat antihorseradish peroxidase antibody, and horseradish peroxidase. The sites of the latter were revealed cytochemically. At the light microscope level, results closely paralleled those obtained with regular immunofluorescence. With the electron microscope, IgG deposits were shown to be confined to the basal lamina and lamina Iucida including the basilar surface of basal cells. Immunoglobulin deposits in bullous pemphigoid appear to be a true basal lamina phenomenon. Immunofluorescence (IF) methods are widely used tools in immunodermatology. They are specific and sensitive provided the appropriate technical parameters are observed [1,2] but they also have disadvantages inherent in the nature of the marker radical, e.g., lack of permanence of the preparations, decay of fluorescence, and the fact that they cannot be employed in electron microscopical studies. A search for other labels has led to the utilization of heavy metals [3], ferritin [3], and enzymes [4] as markers, and horseradish peroxidase (HRP)-Iabeled antibodies have been repeatedly used in light [5-7] and electron microscopical [8--11] studies on in vivo bound immunoglobulins {Ig) in the skin. The procedures employed have several advantages that have been recently reviewed in detail [12], but since they are all based on the preparation of HRP-antibody conjugates they also have limitations [12]. These are mainly related to the labeling procedure and the determination of conjugate characteristics. We have therefore searched for an immunologically more specific and less cumbersome approach to the immuno- chemical detection of antigepic sites and have devised a multistep peroxidase-anti peroxidase procedure that circumvents chemical labeling of antibodies. The present communication details the application of this method to light microscopical and ultrastructural studies of immunoglobulins in bullous pemphigoid. MATERIALS AND METHODS Principles of the Procedure (Fig. l) Based on the general principles of enzyme-antienzyme procedures [4] the following immunologic chain was devised. Goat antihuman lgg antibody was to react. in the initial step of the procedure, with immunoglobulin (antibody) deposits in biopsy specimens of patients or in animal substrate tissues. In a second step, rabbit antigoat!gg was designed to react with the bound goat antihuman lgg of the first step and, owing to its bivalent nature, with a goat anti-hrp antibody to be added in a subsequent third step. In the last (fourth) step of the immunologic chain HRP was to react with its respective antibody and. after immunologic binding, to be visualized cytochemically by employing an established cytochemical procedure. Manuscript received August 26, 1974; in revised form November 18, 1974; accepted for publication November 20, Supported, in part, by Fonds zur F"orderung der wissenschaftlichen Forschung, Grant No. 1416, and a research grant from Schering AG, Berlin. Presented, in part, at the 4th Annual Meeting of the European Society for Dermatological Research, Amsterdam, April 23, 24, Reprint requests to: Dr. K. Holubar, Division of Experimental Dermatology, Department of Dermatology I., University of Vienna, A-1090 Vienna, Austria. 220 Antibodies Goat anti HRP. Two outbred goats were immunized according to Triftshauser et al [131 utilizing HRP (Sigma. grade VI, RZ 3) as an antigen. One-milliliter aliquots of complete Freund's adjuvant containing 1 mg ofhrp were injected intracutaneously into multiple sites of the shoulder and hip regions of the animals at weekly intervals and antibody titers were monitored biweekly in an Ouchterlony type system with HRP (1 mg/ml in phosphate-buffered saline [PBS]) serving as an antigen in the center well and doubling dilutions of goat serum in

2 Aprill975 the peripheral wells. Immunization was terminated at 11 weeks when the titers had reached 128 and 64, respectively. The animals were sacrificed and part of the raw goat serum was subjected to a Cohn fractionation and freeze-drying procedure as modified after Deutsch 114]. This was kindly performed by Grand Island Biological Co., Division of Mogul Co., Grand Island, N. Y. Titer controls of redissolved freeze-dried goat protein revealed a drop in titer to 1/5 the original value. Goat antihuman lgg (wh) antiserum. This antibody was produced in goats according to Triftshauser et al [13]. It was conjugated to fluorescein isothiocyanate (FITC) following established procedures. The conjugate had the following characteristics: FITC, 127 ~tg/ml: total protein mg/ml; specific antibody (as assayed by reverse immunodiffusion (RID)), 1200 ~tg/ml; fluorescein/ protein (F/P) ratio (molar) or 9.92 x w- (weight). Rabbit antigoat!gg antiserum. This antibody was also an FITC conjugate with the following characteristics: FITC, 33.9 ~tg/ml; total protein, mg/ml; specific antibody (as assayed against doubling dilutions of goat serum in an Ouchterlony type procedure). 2 units/ml; F/P, 1.3 (molar) or 3.16 X w- (weight). Specificity controls of the antisera were performed with a technique employing immunoglobulin particles of either human lgg or goat lgg [15,16]. Tissues Employed Multiple biopsy specimens were obtained under local anesthesia from the Jesional and juxtalesional skin of 2 patients with bullous pemphigoid. Monkey esophagus (Maccacus rhesus) kept in liquid nitrogen for not longer than 1 week served as test tissue for the indirect procedure. Immunocytochemical Procedures Cryostat and chopped sections (see below) were rinsed in PBS (ph 7.2. ionic strength 0.15) for 30 min at room temperature before and after each incubation step. Incubations were carried out at room temperature, and incubations were 30 min for each step. Direct procedure (demonstration of in vivo bound t ~----G. anti H. lgg ab 4. **~HRP 3. ~G. anti HRP ab 2. -R. anti G. lgg ab 1 ~ -H. lgg (au<o ab; 4' Antigen Tissue FIG. l. Antibody "chain" of individual antibodies linked to each other immunologically during the different incubation procedures. Step 1, goat antihuman lgg antibody (G. anti H. lgg a b). Step 2, rabbit antigoat lgg antibody (R. anti G. lgg ab). Step 3, goat anti-hrp antibody (G. anti HRP ab). Step 4, horseradish peroxidase (HRP) (See also text). IMMUNOGLOBULINS IN BULLOUS PEMPHIGOID 221!gG). Specimens were incubated with goat antihuman (whole) lgg, diluted with PBS-4% bovine serum albumin (BSA), the antibody (ab) concentrations being 37 ~tg/ml. After rinsing in PBS they were incubated with rabbit antigoat lgg antiserum, diluted to about 2 units of specific antibody/ml, eqt~al parts of PBS and normal human serum (NHS) serving as diluent. After another rinse in PBS they were incubated with goat anti-hrp ab (final concentration 1 unit/mli, equal parts of PBS and NHS serving as diluent. After rinsing in PBS the specimens were immersed in HRP-PBS solution (30 llg HRP/ml PBSl and subsequently rinsed in PBS. Indirect procedure. Sections of monkey esophagus were incubated with bullous pemphigoid sera diluted 1/10, 1/20, or higher with PBS-4% BSA and after rinsing in PBS, subjected to the immunochemical chain described above. Procedure for Light and Electron Microscopy Light microscopy. Five-micron cryostat sections were cut from snap-frozen biopsy specimens and monkey esophagus stored in liquid nitrogen. After thawing they were dried at room temperature and were subjected without fixation to the immunochemical procedure. After incubation in HRP they were rinsed in PBS and fixed in a 1: 1 dilution of Karnovsky's formaldehyde-glutaraldehyde fixative [17] for 30 min at room temperature. They were then incubated in Graham and Karnovsky's cytochemical medium [17] for 10 min at room temperature, to reveal peroxidase activity, rinsed in 0.05 M Tris-HCl buffer, ph 7.6, and coverslipped with Difco mounting medium, sealed with wax. and examined with a regular laboratory microscope. The slides have been stored at room temperature for 8 months without apparent loss of staining. Electron microscopy. Immediately after biopsy, the specimens were cut into thin strips measuring 1 x 1 x 5 mm and rinsed for 30 min in PBS at room temperature. They were then embedded in agar and cut into 50-1-! sections using a Smith and Farquhar tissue chopper. Fixation of any kind was omitted since previous investigations using HRP conjugates [5] and pilot experiments performed for this study had revealed that antigenicity of in vivo bound immunoglobulins was lost during fixation employing either formalin or glutaraldehyde. The sections were then subjected to the immunologic chain and, after incubation in HRP and rinsing in PBS, they were fixed for 30 min at room temperature in 1: 1 dilution of Karnovsky's formaldehyde-glutaraldehyde fixative [17] and rinsed overnight at 4 oc in 0.1 M sodium cacodylate buffer, ph 7.2. They were then incubated in the Graham and Karnovsky medium [18] to reveal the sites of peroxidase. rinsed, and subjected to the postosmification and embedding procedure routinely used in our laboratory [19]. Ultrathin sections were cut with LKB ultratome III and Reichert OM U2 ultramicrotomes and contrasted with lead citrate. A Zeiss EM 9S electron microscope was used to examine the sections. Controls The following control procedures were performed: ( 1) Omission of each individual step of the incubation sequence including incubation with HRP and HRP cytochemistry. (2) In vitro blocking of HRP by an excess of anti-hrp antibody. The blocking procedure was performed by reacting the HRP solution with 20-fold excess of HRP antibody for 60 min at 37 C. The solution was used as such or was centrifuged; in the latter case the supernatant was used as final step of the incubation

3 222 HOLt BAR ET AL chain. t31 Reference control b) IF. Due to the fact that the goat antihuman IgG antiserum happened to be an F'ITC-Iabeled conju~ate. reference control was performed utilizing the very same antiserum. Actually. even the tissue spectmens designated to be processed along the.. chain" incubation for light microscop) "ere occas10n allv checked with the UV microscope after the first incubation step and then processed further in ihe an tiperoxidase peroxidase procedure. (-II Utilization of normal human skin a~ substrate instead of bullous pem phigoid skin. 15) ln the indirect procedure. utilization of the known positive and negative reference 'era routinely employed ur laboratory. Li{fhl Micro.~copy RE ULT!' Direct procedure. The sites ot IgG deposits were marked by an intensely brown reaction product. ect ions of lt>sional skin from patients with bullous pemphigoid l->howed the linear basement mem brane staining pattern tn>ical for the disease (Fig. 2A). Connective tissue and cytoplasm of epidermal cells exhibited a uniform brownish tinge and thi.'i was considered to represent nonspecific back ground; nuclear areas of keratinocytes were free of precipitate. All controls were devoid ot specific stainmg. Control sections of pemphigoid skin incubated without goat antihuman antiserum hut with all other antbera exhibited only a slight nonspecific background. Similarly. absem e of specific staining wa~ noted in sections that had heen l'arrif'd through the entire incubation sequence but were finally incubated with HRP that had been reacted with, and thus blocked b:-.. an excess of anti-hhp antibody. Fig. 28 represents an IF control that was stained with the same antihuman antiserum as the specimen of Fig. ZA, employtng the same dilution and same incubation time, and, iewed bv immunofluorescence after step #1. Applicatio~ of the multistep procedure to normal human skin exhib ited only a \ ellowish background staining but no reaction product in the basement membrane zone. Vol. 64, No. 4 Indirect procedure. Rhesus monkey esophagus sections onto which bullous pemphigoid antibody (BPAI had been fixed in an initial incubation step and subsequently carried through the peroxidase antiperoxidase multistep procedure showed a crisp delineation of the basement membrane zone due to brown reaction product (Fig. 3A). The staining pattern obtained was bandlike as is typical for BPA!Fig. 38). Preliminary determinations of titers in the sera tested disclosed titers that were m the range of. or one doubling dilution lower than. the com entional lf procedure. Controls utilizinl( normal human serum instead of BPA exhibited only a yellowish nonspecific background but no delineation of the basement Ftc. 2. Peribullous skin of a patient with bullous pemphigoid. A: HRP multilayer technique. Linear staining of the basement membrane (arrou:sl.!original mag : Conventional immunofluorescence staining of adjoining sect ion. Pattern of fluorescence is identical with staining pattern shown in A. Goat antihuman lgg conjugate, F/P: 4.08 (M), diluted to 40 ~tg ab/ml (Original mag. A 31). Ftr.. :J. Cirt ulatmg antibasement membrane ab of hulluu~ pemphigoid type: rhesus monkey esophagus served as a substrate tissue in the indirect procedure. A. Patient's serum diluted 1/20 with PBS- 4% BSA; basement membrane of ewphagus ir outlined bv brm'n reaction product (arrow.~). epithelial layers and connective tissue are stained nonspecifically!original mag. " : Bandlike stai111ng pattern of basement membrane zone. Specimen treated as described for Fig. 3A (Original mag. 400).

4 Apri/1975 IMMUNOGLOBULINS IN BULLOUS PEMPHIGOID 223 membrane zone by reaction product. The same was true for controls in which one of the incubation steps had been omitted and for specimens that, in the final step, had been reacted with HRP blocked by anti-hrp antibody. Background. The nonspecific yellowish staining of all sections, save in the specimens incubated with in vitro blocked HRP that were free of reaction product, increased with time of incubation in the histochemical medium. Since one molecule of HRP can catalyze a series of substrate molecules, an increasing amount of reaction product is created with increasing incubation time. Such reaction product may lloat off into the incubation medium and may possibly be bound, in a nonspecific manner, to the tissue; therefore, it had to be determined at which point the optical contrast was optimal between "specific" and "nonspecific" staining_ This was found to be between 10 and 15 min of incubation time for light microscopy, at which time staining of specific sites was marked and crisp whereas the background showed only a faint yellowish discoloration. Electron Microscopy Morphology. As compared to the standards of the aldehyde-osmium fixation routinely used in our laboratory [19] the general preservation of the tissue was poor. However. it was still surprisingly good taking into account that the tissue had been subjected to the various immunochemical treatments before being subjected to fixation. The cells appeared adequately preserved in their general structural appearance and so were the cell organelles (Figs. 4-6). The cytoplasm, however, was edematous and although the cytomembranes were generally well preserved they exhibited occasional breaks and discontinuities. This was particularly evident in basal cells of bullous lesions (Fig. II. Structural detail of desmosomes, half-desmosomes, basal lamina, and the fibrous components of the dermis was adequately preserved (Figs. 4-6). Immunochemistry. The reaction product indicating peroxidase activity and thus the site of IgG was granular in appearance and, owing to its intrinsic electron density. was easily recognized. In lesional (ervthematous. nonbullousl skin it was distributed.in a bandlike pattern along the basal lamina which was thus heavily accentuated against the moderately electron-dense background of epidermis and dermis (Figs. 4A, 51. Higher magnifications disclosed that it was present both in the lamina Iucida between the basal lamina and the cytomembranes of keratinocytes and in the basal lamina proper (Fig. 6). However, the amount of reaction product present in the lamina Iucida exceeded by flu that within the basal lamina and extended to the very surface of the keratinocyte membranes. In areas, it appes red preferentially associated with half-desmosomes (Figs. 5, 6). It was absent from the intercellular spaces between the basal cells and this was even true when the intercellular spaces were considerably widened (Fig. 51. There were no precipitates in the upper layers of the subjacent superficial dermis save on occasional anchoring fibrils whose periodicity was thus accentuated (Fig. 51. Very sparse amounts of granular precipitate were found dispersed between the fibrous components of the dermis (Figs. 4A, 5, 6). They were considered to be nonspecific and to correspond to the background staining observed at the light microscope level. Control specimens were free of reaction product (Figs. 4B, 7A). Lesional skin from the periphery of bullae exhib ited incipient or manifest detachment of the epidermis from the underlying dermis (Fig.7A, B). Again, the reaction product was confined to the basal lamina and the lamina Iucida but, depending on whether the basal lamina was attached to the dermis (Fig. 7 B) or had lifted off together with the epidermis, the distribution of the reaction product followed two patterns. In areas where the basal lamina had separated from the dermis, all reaction product was found attached to the undersurface of the epidermis forming the roof of the blister; most frequently, however, the basal lamina was found resting on the dermis which was thus outlined by a moderately electron-dense band of granular material. In such instances, the detached epidermis was also lined by a similarly granular but heavier precipitate which corresponded to the precipitates found in the lamina Iucida of non bullous lesional skin (Fig. 7Bl. DISCI'SSIO!\ Previous studies on the ultrastructural localization of lg in skin disease have dealt with in vivo bound lg deposits in pemphigus and LE using HRP conjugates [8-ll]. Although the first attempts in this field were fraught with technical ditliculties and thus yielded morphologic results that left much to be desired [8 ], further improvements of the technique resulted in a satisfactory resolution of the ultrastructural localization of lg and thus actually permitted th(> mapping of these deposits at the ultrastructural level [10, 12]. However, since these studies were performed with HRP-labeled antibodies. they bore the limitations of HRP-conjugate procedures in general [12]. The approach presented in this paper appears superior to the HRP-conjugate-dependent methods f{lr the following reasons: 1. The quality of HRP conjugates depends on the conjugation rate, as is the case with all labeled antibody, and has to be checked by spectrophotometry [12]. Since our present method does not require conjugation, a monitoring of conjugation rates becomes unnecessary. 2. During the conjugation procedure. tracer molecules are covalently bound to antibody but HRP dimers and polymers are also formed [4,20 ]. These complexes cannot be reliably removed from the conjugate mixture [4,20] and since the latter also contains free, nonbound enzyme as well as anti-

5 224 HOLUBIIR ET IlL Vol. 64, No.4 FtG. 4.A: Immuno-eleetron microscopy. erythematous, non bullous, lesional skin. Electron-dense reaction product (arrow,), indicating peroxidase activity and thus the site of lgg deposits, is distributed in a band like pattern along the dermal-epidermal interface marking the sites of the basal lamina and lamina Iucida. The general structural appearance of the keratinocyte (KCJ appears adequately preserved ( x ). B: Control. The initial step of the procedure (goat antihuman!gg antibody! was omitted. The subsequent incubation steps reveal no electron-dense reaction product at the dermal-epidermal junction. The arrows mark tnc basal lamina. C: collagen\/ !. body, there are several potential sources of error both in the spectrophotometric measurements and in the binding of eonjugated or nonconjugated antibody to the antigenic site 112]. No such difficulty need be expected with <Jur technique as HRP is hound immunologically and not chemically. :l. The HRP-anti-HRP method provides greater versatility than the conjugate-dependent procedures: different monospecific antisera may he interchangeably employed in the first step and may then readily be combined with the rest of the antibody chain.

6 Apri/1975 IMMUNOGLOBULINS IN BULLOUS PEMPHIGOID 225 FIG. 5. Erythematous skid in the vicinity of a bulla. The reaction product marks the sites oflgg in the lamina Iucida and in the basal lamina (arrows). It extends to the very surface of the keratinocyte membrane and appears preferentially associated with half-desmosomes (HD). The intercellular spaces UCS) between keratinocytes are free of reaction product. There are no precipitates in the subjacent superficial dermis (D) save on occasional anchoring fibrils whose periodicity is thus accentuated (circles)( x 22,200). Inset: Anchoring fibrils at higher magnification ( x 83,600). FIG. 6.1esional. nonhullous skin. Heavy precipitates of the granular reaction product (asterisks) are present in the lamina Iucida and on the surface of the cvtomembrane of the basal cell (BC). Finely granular reaction product is also seen in the basal lamina (BL) whereas the subjacent dermis is almost free of reaction product. TF. tonofilaments; HD, half desmosomes ( x 44,000). 4. Immunologic sensitivity should be higher in any more "indirect'" method, as the one described in this paper. than in a "direct" approach utilizing conjugated antibody. To our knowledge there have been no previous reports on the ultrastructural localization of IgG in pemphigoid skin. In studies on lupus erythematosus (LE) lesions, dense deposits of Ig have been observed in the entire junctional zone extending from the superficial layers of the dermis to the cytomembrane of the basal cells [10]. In other words. the basal lamina was not the only site where Ig deposits were traced. The zone of the lamina Iucida was sho-wn to be involved and occupied by Ig deposits in places. whereas it appeared free in others. The deposits tended to form aggregates which were most pronounced in chronic lesions and oft en occupied entire micropapillae of the dermis \10.12]. ln contrast. our results indicate that, in bullous pemphigoid. there is a more selective involvement of the basal lamina and the lamina Iucida. These findings may be considered well in accord with present concepts on the pathogenic mechanisms operative in this disease: in bullous pemphigoid. autoantibodies are believed to react with antigens in the basement membrane zone [21 ], whereas, in LE, preformed antigen-antibody complexes are thought to be deposited along the basement membranes (of skinl in an immunologically rather passive manner [22]. It is interesting that in the pemphigoid skin studied in this investigation we found IgG to be deposited on half-desmosomes, but at present it would be speculative to ascribe to them a pathogenic role in the dermalepidermal separation that occurs in this disease. A similar localization of lgg on half-desmosomes has also been found in LE 110 J whi<:h appears to reduce the significance of this finding.

7 226 HOLUBAR ET AL Vol. 64, No.4 FIG. 7. Lesional skin from periphery of a bulla with detachment of the keratinocytes fkci from the dermis. The basal lamina (BL) rests on the dermis. A: Control (omission of the first step in the immunochemical chainl. There is no reaction product (" 22,200). B: Specimen run through the entire immunochemical chain. Electron-dense precipitate (arrows) is localized on the cytomembrane of the detached keratinocytes. Small amounts of the granular reaction product are present in the basal lamina (BL) resting on the dermis. ( Y 22,200).lnset: High-power view of the reaction product (arrolrs) on the membrane of a keratinocyte which shows detachment from the underlying basal lamina (BL) and subjacent dermis ( x 83,6(\0). Though a satisfactory correlation of IgG deposits and submicroscopic structure was obtained in the present study. it can, of course. give no information on the nature of the antigen involved. It does. however. confirm the light microscopical evidence that in bullous pemphigoid the lgg deposits represent a basal lamina phenomenon. in contrast tole where the entire basement membrane zone lineluding the superficial dermis! is im olved. Thus. ultrastructural distribution patterns of lg in pemphigoid and LE permit a distinction between the two diseases. The technical assistance of Mrs. Lotte Polasek. Mr>. Susan C'segezi. and Mrs. Hilde Martinjak is gratefully acknowledged. REFERE~CES I. Beutner EH. Chorzelski TP.. Jordon RE: Autosensitt zation in pemphigus and bullous pemphigoid. Springfield, Ill. Thomas, Beutner EH, Hale WL. :'iisengard R-J, Chorzelski TP. Holubar K: Defmed immunofluorescence in clinical immunopathology, Immunopathology of the Skin. Labeled Antibody Studies. Edited by EH Beutner. TP Chorzelski. SF Bean. RE.Jordon. Stroudsburg, Pa, Dowden, Hutchinson and Ross. 1973, pp Sternber!(er LA: Electron microscopic immunocytochemistry: a review.,j Histochem Cvtochem 15: Avrameas S: lmmunoenzyme techniques: enzymes as markers for the localization of antigens and antibodies. lnt Rev Cvtol 27: Wolff K, Schreiner E: -lmmunohistochemische Untersuchungen mittels eines Enzym-Immunoglobulin-Konjugates. Nachweis in vivo gebundener Immunglohuline hei Lupus erythematosus, Pemphigus vulgaris. Pemphigus erythematosus und Pemphigoid. Arch Klin Exp Dermatol 238: , Thivolet J, Leung-Tack J, Page Y. Schmitt D: Utilisation d'anticorps marquees a Ia peroxidase pour Ia recherche et la diagnostic immunologiques. Presse Med 79: , Fukuyama K, Douglas SD. Tuffanelli DL. Epstein WL: Immunohistochemical method for localization of antibodies in cutaneous disease. Am J Clin Pathol 54: , Schreiner E. \Volff K: Systemic lupus erythematosus. Electron microscopic localization of in vivo bound globulins at the dermal epidermal junction.,j Invest Dermatol 55: , 1970

8 April Wolff K, Schreiner E: Ultrastructural localization of pemphigus autoantibodies within the epidermis. Nature (Lond) 229: Wolff-Schreiner E, Wolff K: Immunoglobulins at the dermal-epidermal junction in lupus erythematosus. Ultrastructural investigations. Arch Dermatol Forsch 246: , 1973 J 1. Ueki H, WolffHH. Braun-Falco 0: Cutaneous localization of human gamma-globulins in lupus erythematosus. Arch Dermatol Forsch 248: , Wolff K, Fukuyama K: Peroxidase conjugates in immunopathologic studies of skin, Immunopathology of the Skin. Labeled Antibody Studies. Edited bv EH Beutner, TP Chorzelski, SF Bean. RE Jt>rdon. Stroudsburg, Pa. Dowden. Hutchin"m and Ross, 1973, pp :J. Triftshauser C, Hayden DW. Beutner EH: Procedures for the immunization of goats with human immunoglobulins and complement. lnt Arch AI lergy 38: , Deutsch HF: Separation of antibody-acti, e proteins from various animal sera bv ethanol fractionation techniques. Methods Med Res 5: , Bergquist NR Holubar K. Diaz GA. Beutner EH: Manufacture of protein microspheres by suspen sion polymerization.!nt Arch Allergy 43: , Holubar K. Bergquist NR. Lesser GR. Beutner EH: JMMUNOGLOBULINS IN BULLOUS PEMPHIGOID 227 Comparative microfluorometrlc studies of activity and size of suspension polymerized anti-igg antibodies at different ph levels. Int Arch Allergy 44: Karnovsky MJ: A formaldehyde-glutaraldehyde fixative of high osmolality for use in electron microscopy (ahstrl. J Cell Bioi 27:137A-138A Graham RC, Karnovsky M,J: The early stages of absorption of injected horseradish peroxidase in the proximal tubules of mouse kidney: ultrastructural cytochemistry by a new technique. J Histochem Cvtochem 14: , Honigsma-nn H, Viiolff K: Continuity of intercellular space and endoplasmic reticulum of keratinocytes. Exp Cell Res 80: Kraehenbiihl JP, de Grandi PB. Campiche MA: Ultrastructural localization of intracellular antigen using enzyme labelled antibody fragments. J Cell Bioi 50: , Sams M: Pemphigus and pemphigoid, Immunologic Aspects of Skin Diseases. Edited by L Fry. PP Seah. St. Leonardgate, Lancaster, England, MTP Medical and Technical Pub! Co, 1974, pp Jablonska S, Chorzelski TP: Lupus ery1hematosus: Immunological Aspects of Skin Disease. Edited by L Fry, PP Seah, St. Leonardgate, Lancaster, Eng land, MTP Medical and Technical Pub! Co. 1974, pp