Portal de Periódicos da CAPES

IgG Binds to Desmoglein 3 in Desmosomes and Causes a Desmosomal Split Without Keratin Retraction in a Pemphigus Mouse Model

2004; Elsevier BV; Volume: 122; Issue: 5 Linguagem: Inglês

10.1111/j.0022-202x.2004.22426.x

1523-1747

Atsushi Shimizu, Akira Ishiko, Takayuki Ota, Kazuyuki Tsunoda, Masayuki Amagai, Takeji Nishikawa,

Urticaria and Related Conditions

Pemphigus vulgaris (PV) is an autoimmune blistering disease caused by IgG autoantibodies against desmoglein 3 (Dsg3). In this study, we characterized the ultrastructural localization of in vivo-bound IgG, Dsg3, and desmoplakin during the process of acantholysis in an active mouse PV model, using post-embedding immunoelectron microscopy. In non-acantholytic areas of keratinocyte contact, IgG labeling was restricted to the extracellular part of desmosomes, and was evenly distributed throughout the entire length of the desmosome. The distribution of in vivo IgG was similar to that of anti-Dsg3 labeling in the control mouse. Within the acantholytic areas, there were abundant split-desmosomes with keratin filaments inserted into the desmosomal attachment plaques. These split-desmosome extracellular regions were also decorated with anti-Dsg3 IgG and were associated with desmoplakin staining in their cytoplasmic attachment plaques. No apparent split-desmosomes, free of IgG-labeling were observed, suggesting that Dsg3 was not depleted from the desmosome before the start of acantholysis in vivo. Desmosome-like structures (without keratin insertion) were found only on the lateral surfaces of basal cells, but not on the apical surfaces at the site of acantholytic splits. These findings indicate that anti-Dsg3 IgG antibodies can directly access Dsg3 present in desmosomes in vivo and cause the subsequent desmosome separation that leads to blister formation in PV. Pemphigus vulgaris (PV) is an autoimmune blistering disease caused by IgG autoantibodies against desmoglein 3 (Dsg3). In this study, we characterized the ultrastructural localization of in vivo-bound IgG, Dsg3, and desmoplakin during the process of acantholysis in an active mouse PV model, using post-embedding immunoelectron microscopy. In non-acantholytic areas of keratinocyte contact, IgG labeling was restricted to the extracellular part of desmosomes, and was evenly distributed throughout the entire length of the desmosome. The distribution of in vivo IgG was similar to that of anti-Dsg3 labeling in the control mouse. Within the acantholytic areas, there were abundant split-desmosomes with keratin filaments inserted into the desmosomal attachment plaques. These split-desmosome extracellular regions were also decorated with anti-Dsg3 IgG and were associated with desmoplakin staining in their cytoplasmic attachment plaques. No apparent split-desmosomes, free of IgG-labeling were observed, suggesting that Dsg3 was not depleted from the desmosome before the start of acantholysis in vivo. Desmosome-like structures (without keratin insertion) were found only on the lateral surfaces of basal cells, but not on the apical surfaces at the site of acantholytic splits. These findings indicate that anti-Dsg3 IgG antibodies can directly access Dsg3 present in desmosomes in vivo and cause the subsequent desmosome separation that leads to blister formation in PV. desmoglein electron microscopy keratin filament pemphigus vulgaris Pemphigus vulgaris (PV) is a life-threatening autoimmune blistering disease of the skin and mucous membranes, which was clinically characterized by flaccid blisters and erosions and histopathologically, by cell–cell detachment between basal and suprabasal keratinocytes, resulting in suprabasal acantholysis. The PV target antigen is desmoglein 3 (Dsg3), a transmembrane desmosomal component that belongs to the cadherin supergene family (Amagai et al., 1991Amagai M. Klaus-Kovtun V. Stanley J.R. Autoantibodies against a novel epithelial cadherin in pemphigus vulgaris, a disease of cell adhesion.Cell. 1991; 67: 869-877Abstract Full Text PDF PubMed Scopus (870) Google Scholar,Amagai et al., 1992Amagai M. Karpati S. Prussick R. Klaus-Kovtun V. Stanley J.R. Autoantibodies against the amino-terminal cadherin-like binding domain of pemphigus vulgaris antigen are pathogenic.J Clin Invest. 1992; 90: 919-926Crossref PubMed Scopus (310) Google Scholar). Previously, the ultrastructural localization of the binding site of PV patients' IgG in normal human skin sections was localized to the extracellular portion of the desmosome using immunogold electron microscopy (EM) (Shimizu et al., 1995Shimizu H. Masunaga T. Ishiko A. Kikuchi A. Hashimoto T. Nishikawa T. Pemphigus vulgaris and pemphigus foliaceus sera show an inversely graded binding pattern to extracellular regions of desmosomes in different layers of human epidermis.J Invest Dermatol. 1995; 105: 153-159https://doi.org/10.1111/1523-1747.ep12316695Crossref PubMed Scopus (89) Google Scholar). But the ultrastructural localization of in vivo-bound IgG in PV patients' skin is still controversial. Some reports have demonstrated IgG localization only in the desmosomal plaque (Akiyama et al., 1991Akiyama M. Hashimoto T. Sugiura M. Nishikawa T. Ultrastructural localization of pemphigus vulgaris and pemphigus foliaceus antigens in cultured human squamous carcinoma cells.Br J Dermatol. 1991; 125: 233-237Crossref PubMed Scopus (22) Google Scholar;Karpati et al., 1993Karpati S. Amagai M. Prussick R. Cehrs K. Stanley J.R. Pemphigus vulgaris antigen, a desmoglein type of cadherin, is localized within keratinocyte desmosomes.J Cell Biol. 1993; 122: 409-415https://doi.org/10.1083/jcb.122.2.409Crossref PubMed Scopus (111) Google Scholar;Shimizu et al., 1995Shimizu H. Masunaga T. Ishiko A. Kikuchi A. Hashimoto T. Nishikawa T. Pemphigus vulgaris and pemphigus foliaceus sera show an inversely graded binding pattern to extracellular regions of desmosomes in different layers of human epidermis.J Invest Dermatol. 1995; 105: 153-159https://doi.org/10.1111/1523-1747.ep12316695Crossref PubMed Scopus (89) Google Scholar), whereas others showed it along the entire plasma membrane including parts of the non-desmosomal plasma membrane (Wolff and Schreiner, 1971Wolff K. Schreiner E. Ultrastructural localization of pemphigus auto-antibodies within the epidermis.Nature. 1971; 229: 59-60Crossref PubMed Scopus (88) Google Scholar;Patel et al., 1984Patel H.P. Diaz L.A. Anhalt G.J. Labib R.S. Takahashi Y. Demonstration of pemphigus antibodies on the cell surface of murine epidermal cell monolayers and their internalization.J Invest Dermatol. 1984; 83: 409-415https://doi.org/10.1111/1523-1747.ep12273480Crossref PubMed Scopus (59) Google Scholar;Bedane et al., 1996Bedane C. Prost C. Thomine E. et al.Binding of autoantibodies is not restricted to desmosomes in pemphigus vulgaris: Comparison of 14 cases of pemphigus vulgaris and 10 cases of pemphigus foliaceus studied by western immunoblot and immunoelectron microscopy.Arch Dermatol Res. 1996; 288: 343-352https://doi.org/10.1007/s004030050061Crossref PubMed Scopus (30) Google Scholar). There have been no in vivo studies using post-embedding immunogold as a marker due to technical difficulties in detecting IgG in resin-embedded samples, and this discrepancy is in part, due to the use of peroxidase as a probe, which forms a signal over a relatively large area that may obscure precise observation of the underlying ultrastructure. Recently, we have generated a novel experimental murine PV model by transferring splenocytes from a Dsg3 -/- mouse, which was immunized with recombinant mouse Dsg3, to Rag2 -/- immunodeficient mice that normally express Dsg3 (Amagai et al., 2000Amagai M. Tsunoda K. Suzuki H. Nishifuji K. Koyasu S. Nishikawa T. Use of autoantigen-knockout mice in developing an active autoimmune disease model for pemphigus.J Clin Invest. 2000; 105: 625-631Crossref PubMed Scopus (225) Google Scholar). These models stably produced anti-Dsg3 IgG and developed typical features of PV including oral erosions with suprabasal acantholysis and eosinophilic spongiosis (Ohyama et al., 2002Ohyama M. Amagai M. Tsunoda K. et al.Immunologic and histophathologic characterization of active disease mouse model for pemphigus vulgaris.J Invest Dermatol. 2002; 118: 199-204https://doi.org/10.1046/j.0022-202x.2001.01643.xCrossref PubMed Scopus (40) Google Scholar). Furthermore, these model mice showed an ultrastructural PV phenotype that included rows of basal keratinocytes with a "tombstone-like" appearance and formation of split-desmosomes (Shimizu et al., 2002Shimizu A. Ishiko A. Ota T. Tsunoda K. Koyasu S. Amagai M. Nishikawa T. Ultrastructural changes in mice actively producing antibodies to desmoglein 3 parallel those in patients with pemphigus vulgaris.Arch Dermatol Res. 2002; 294: 318-323PubMed Google Scholar). These results suggest that the PV model mice faithfully represent the disease phenotype from both an immunological and ultrastructural point of view, and that this mouse in vivo model can help to elucidate the immunomolecular mechanisms of blister formation in PV. The purpose of this study is to clarify the ultrastructural localization of in vivo-bound IgG using post-embedding immunogold EM and to elucidate the molecular pathological mechanisms of blister formation by demonstrating the fate of IgG-bound desmosomes in PV model mice. Here we demonstrate that, at non-acantholytic sites, the in vivo-bound IgG is exclusively localized to the extracellular portion of the desmosome and this distribution is identical to that of Dsg3 which was detected in normal control mice using immunogold EM. Our findings also indicate that split-desmosomes, which are a dramatic morphological feature of the acantholytic process, were also decorated with anti-IgG gold particles and that these split-desmosomes maintained their connections with the keratin filaments (KF) via desmoplakin. The ultrastructure of the post-embedding immuno-EM in control mice was well preserved. Low magnification views of the sections showed that the majority of gold probes, which labeled Dsg3 using the serum from the PV model mouse, were localized at the desmosome and almost no other labeling was observed in the interdesmosomal cell membrane or intracellular region of the keratinocytes Figure 1a. High magnification views of the sections demonstrated that most of the polyclonal anti-Dsg3 antibodies bound between the two attachment plaques of a desmosome and almost no other labeling was observed Figure 1b, c. This result is consistent with previous binding pattern of PV patients' sera to the desmosomes of normal human epidermis (Shimizu et al., 1995Shimizu H. Masunaga T. Ishiko A. Kikuchi A. Hashimoto T. Nishikawa T. Pemphigus vulgaris and pemphigus foliaceus sera show an inversely graded binding pattern to extracellular regions of desmosomes in different layers of human epidermis.J Invest Dermatol. 1995; 105: 153-159https://doi.org/10.1111/1523-1747.ep12316695Crossref PubMed Scopus (89) Google Scholar). No desmosomal labeling was seen on the sections of control mice using normal mouse serum (data not shown). The ultrastructure of the PV model mice was also well preserved. Low magnification views demonstrated that the gold probes, which labeled in vivo-bound IgG in the PV model mice, were mainly localized at desmosomes in the non-acantholytic areas and very little labeling was observed in the intracellular portion of the cells Figure 2a, b. In addition, there was both linear and discontinuous labeling on the lateral surfaces of the basal keratinocytes (arrowheads in Figure 2b), but these findings were not observed in the control mice. High magnification views demonstrated that the majority of desmosomes were decorated with colloidal gold and that the IgG autoantibodies were specifically deposited between the two cells attachment plaques of the desmosome, i.e., within the extracellular region of the desmosome Figure 3a, b. No significant labeling was seen over the control mouse sections in mucous membrane samples (not shown). Double-sided immunogold labeling of the sections demonstrated extensive desmoplakin (5 nm gold) labeling localized to between the plasma membrane of the desmosomes and KFs over the attachment plaque, whereas mouse IgG was deposited over the extracellular portion of the desmosome Figure 3c. Furthermore, double-sided labeling of mouse IgG (15 nm gold) and β-catenin (5 nm gold) showed distinct binding sites within extracellular portion of the cell over the desmosome and within the intracellular region of adherens junctions in the interdesmosomal areas, respectively Figure 3d.Figure 3In vivo-bound anti-Dsg3 IgG in desmosomes of PV model mice was localized to the extracellular region. High magnification views of the PV model mice incubated with rabbit anti-mouse IgG followed by 5 nm gold anti-rabbit IgG and enhanced by silver staining for 3 min are shown (a–d). The gold labeling was bound to all desmosomes and no labeling was observed between the desmosomal areas (a, b). The majority of Dsg3 IgG gold probes were observed over the extracellular portion of the desmosome (b). Using double immunolabeling, mouse IgG (15 nm gold) and desmoplakin (5 nm gold) were seen at distinct sites over the extracellular portion of the desmosome and at the junction between the desmosomal attachment plaque and keratin filaments, respectively in the intact epidermis (c). Furthermore, by double immunolabeling with anti-mouse IgG (15 nm gold) and β-catenin (5 nm gold, adherens junctions), the β-catenin labeling was clearly seen over the intracellular region of interdesmosomal adherens junctions (arrow) whereas IgG was localized at desmosomes (open arrows) (d). Dotted line indicates basement membrane zone. n; nucleus. Scale bars (a–d)=200 nm.View Large Image Figure ViewerDownload (PPT) By careful measurement of desmosome thickness, 369 gold particles that labeled the PV antigen in control mice and 383 that labeled the deposited IgG in the PV model mice were converted into their respective percentage frequencies, because of the thickness of each desmosomes was subtly different. In control mice, 88.9±3.21% (95% confidence interval) of the Dsg3 labeling recognized on section using the PV model mouse serum were located within the extracellular portion of the desmosome Figure 4a. In PV model mice, 93.0±2.56% (95% confidence interval) of the IgG deposited in vivo was also located within the extracellular portion of the desmosome Figure 4b. Both distributions showed a major peak at the midpoint between the extracellular region of the desmosome and the frequency of labeling gradually decreased approaching the desmosomal plasma membrane. In order to measure the labeling distribution along the desmosome length, 433 gold particles that labeled Dsg3 in control mice and 541 gold particles that labeled IgG in the PV model mice were used. Dsg3 in the control mice and the IgG deposition in the PV model mice were both regularly and evenly distributed throughout the entire length of desmosome Figure 5. Observing the apical surface of acantholytic basal cells in the PV model mice showed that separated keratinocytes maintained numerous split-desmosomes (arrows in Figure 6a, conventional EM). High magnification views of immunolabeled areas using cryofixed samples revealed that the extracellular portion of the split-desmosome was decorated with IgG Figure 6b. Almost all split-desmosomes were decorated with IgG. There was no labeling on the plasma membrane in the absence of split-desmosomes and no internalization of split-desmosomes was observed. Desmoplakin was detected over regions between KFs and the split-desmosome attachment plaques Figure 6c similar to that observed in desmosomes in the non-acantholytic areas. At the lateral surface of the basal cells in the PV model mice there were clusters of IgG over areas where two adjacent cell plasma membranes were in close apposition (arrowheads in Figure 2b and Figure 7a). This unique feature without any electron dense cytoplasmic desmosomal attachment plaque and KF association with the attachment plaque was seen only at the lateral surface of the basal keratinocytes in both acantholytic and non-acantholytic epithelia. The distance between the opposing plasma membranes was almost identical to that of the desmosome (approximately 30 nm) and there was a faint, dense mid-line within the desmoglea, mid-way between the adjacent plasma membranes (arrowhead in Figure 7a). Some of these structures were observed continuous with and adjacent to small desmosomes (open arrows in Figure 7d, e). Double-labeling with mouse IgG (15 nm gold) and desmoplakin (5 nm gold) demonstrated that the desmoplakin was only sparsely detected at these unique desmosome-like structures at the lateral surface of the basal keratinocytes Figure 7b. Instead, desmoplakin was seen at the ends of KFs in the cytoplasm around the desmosome-like structures Figure 7e. Using double-sided labeling of mouse IgG (15 nm gold) and β-catenin (5 nm gold), the desmosome-like structures were shown to be positive for clusters of IgG but not β-catenin, indicating that these structures were not adherens junctions Figure 7c. In this study, we have demonstrated that the majority of Dsg3 epitopes recognized by PV model mouse sera were expressed within the extracellular region of desmosomes in the control mouse, and that the in vivo-bound PV IgG in the model mice was also localized to the extracellular portion of the desmosomes. The intra-desmosomal distribution of the target epitope in the control mouse epithelia and IgG deposition in the PV mouse epithelia were statistically analyzed in two dimensions or axes (distance from the plasma membrane and distribution along the length of the desmosome) and the two patterns were almost identical. These results indicate that the anti-Dsg3 IgG antibodies can access the entire desmosomal Dsg3 protein complex and that the binding of autoantibodies to this antigen does not immediately cause any changes in the antigen localization in the non-lesional epithelia. In the cultured keratinocytes, it is known that Dsg3 exists unbound on the plasma membrane before being recruited into desmosomes (Aoyama and Kitajima, 1999Aoyama Y. Kitajima Y. Pemphigus vulgaris-IgG causes a rapid depletion of desmoglein 3 (Dsg3) from the Triton X-100 soluble pools, leading to the formation of Dsg3-depleted desmosomes in a human squamous carcinoma cell line, DJM-1 cells.J Invest Dermatol. 1999; 112: 67-71https://doi.org/10.1046/j.1523-1747.1999.00463.xCrossref PubMed Scopus (131) Google Scholar). In the presence of PV autoantibodies in the culture medium, Dsg3 was shown to be depleted from the plasma membrane that resulted in the formation of abnormal desmosomes, lacking Dsg3. But in our mouse model experiments in vivo, the antibody-bound Dsg3 showed the same distribution as the Dsg3 distribution in control mice and Dsg3-deficient desmosomes were not found. Even in the split-desmosomes, IgG binding was regularly observed on the extracellular surface of the plasma membrane adjacent to the attachment plaques suggesting the presence of Dsg3. Furthermore, we could not detect free Dsg3 labeling along the non-desmosomal plasma membrane in normal epithelia. Thus, depletion of Dsg3 from desmosomes is unlikely to be either a major or early step in the pathogenesis of blister formation in this mouse PV model although such depletion may occur under certain in vitro conditions. We have identified a unique membrane structure with IgG deposition at the lateral surface, but not at the apical surface, of the basal keratinocyte in the PV mouse model. This was not observed in any of the control mice but was regularly seen between the basal cells in the PV model mice regardless of the presence of acantholysis or not in the epidermis. These unique structures comprised flat plasma membranes in close apposition, with a distance similar to that between the desmosomal plasma membranes, and these new structures also possessed a dense midline. Some of them were continuous with small desmosomes, strongly suggesting a close relation to desmosomes, although they lacked tonofilament insertion and attachment plaques. These structures were not adherens junctions because the distance between opposing plasma membranes was significantly larger and β-catenin labeling was absent from these structures. Using double labeling, desmoplakin was localized to the ends of KFs in the cytoplasm around these unique structures. From these observations, there are two possibilities regarding the formation of these structures; they might represent degraded desmosomes formed by the retraction of KFs and the loss of desmoplakin from the desmosomal attachment plaque or they might represent the early stages of desmosome formation before other plaque components were recruited. The latter, however, is less likely for the following reasons. (1) The link between keratin filaments and desmoplakins would make assembly into the desmosome a slow and highly complex process. (2) There is no ultrastructural evidence for any molecular transfer such as the presence of microtubules between the desmosome-like structure and the desmoplakins accompanied by keratin. Furthermore, keratin filaments that usually fill the cytoplasm are lacking in the space between them. (3) These structures are seen regardless of the presence of acantholysis in the PV model mouse. Previous ultrastructural analysis of a desmoplakin gene knockout in mouse keratinocytes (Gallicano et al., 1998Gallicano G.I. Kouklis P. Bauer C. Yin M. Vasioukhin V. Degenstein L. Fuchs E. Desmoplakin is required early in development for assembly of desmosomes and cytoskeletal linkage.J Cell Biol. 1998; 143: 2009-2022https://doi.org/10.1083/jcb.143.7.2009Crossref PubMed Scopus (264) Google Scholar;Vasioukhin et al., 2001Vasioukhin V. Bowers E. Bauer C. Degenstein L. Fuchs E. Desmoplakin is essential in epidermal sheet formation.Nat Cell Biol. 2001; 12: 1076-1085https://doi.org/10.1038/ncb1201-1076Crossref Scopus (248) Google Scholar) demonstrated desmosomes with hypoplastic attachment plaques without KFs insertion indicating that the loss of desmoplakin is critical not only for keratin insertion but also for attachment plaque formation. The lack of desmoplakin in the desmosome-like structure is thus critical in plaque formation. If they are really degrading desmosomes, the keratin retraction in the PV model mice might also be expected to occur on the lateral sides of the basal keratinocytes. This therefore may by the first ultrastructural report demonstrating keratin retraction in vivo, which has not previously been recognized without the aid of double immunolabeling. Some previous reports have demonstrated in vivo-bound IgG along the plasma membrane including the non-desmosomal plasma membrane (Wolff and Schreiner, 1971Wolff K. Schreiner E. Ultrastructural localization of pemphigus auto-antibodies within the epidermis.Nature. 1971; 229: 59-60Crossref PubMed Scopus (88) Google Scholar;Takahashi et al., 1985Takahashi Y. Patel H.P. Labib R.S. Diaz L.A. Anhalt G.J. Experimentally induced pemphigus vulgaris in neonatal BALB/c mice: A time-course study of clinical, immunologic, ultrastructural, and cytochemical changes.J Invest Dermatol. 1985; 84: 41-46https://doi.org/10.1111/1523-1747.ep12274679Crossref PubMed Scopus (85) Google Scholar;Bedane et al., 1996Bedane C. Prost C. Thomine E. et al.Binding of autoantibodies is not restricted to desmosomes in pemphigus vulgaris: Comparison of 14 cases of pemphigus vulgaris and 10 cases of pemphigus foliaceus studied by western immunoblot and immunoelectron microscopy.Arch Dermatol Res. 1996; 288: 343-352https://doi.org/10.1007/s004030050061Crossref PubMed Scopus (30) Google Scholar). Non-desmosomal IgG on the cell surface might result from the desmosome-like structures that lack attachment plaques. We could not detect IgG on the plasma membrane that was not associated with either normal or aberrant desmosomes in our study. We have shown in this study that the ultrastructural changes in desmosomes during acantholysis are different between apical surface and lateral surface of basal cells Figure 8. In non-acantholytic areas, or in the epithelium before acantholysis, desmosomes on the apical side of basal cells and those in the suprabasal layer were intact and the majority of PV-IgG was deposited at the extracellular portion of the desmosomes as described above. On the lateral sides of the basal keratinocytes, both keratin-retracted desmosomes and intact desmosomes with IgG deposition were seen. In the acantholytic area, many split-desmosomes were found at the blister roof (on the basal surface of suprabasal cells) as well as the blister floor (i.e., the apical surface of the basal cells) as previously described (Shimizu et al., 2002Shimizu A. Ishiko A. Ota T. Tsunoda K. Koyasu S. Amagai M. Nishikawa T. Ultrastructural changes in mice actively producing antibodies to desmoglein 3 parallel those in patients with pemphigus vulgaris.Arch Dermatol Res. 2002; 294: 318-323PubMed Google Scholar). These split-desmosomes were decorated with IgG along the extracellular surface and connected to the KF network. Moreover, keratin-desmosome association in the presence of desmoplakin was detected in these split-desmosomes, i.e., the keratin retraction did not take place on the apical surface of basal cells. From our results, the fate or local conditions of apical and lateral desmosomes might be different. We suggest the former might be split in half without keratin retraction, whereas the later are degraded by keratin retraction. The cleavage of desmosomes at the apical surface therefore occurs without retraction of KF. The reason for these different desmosome fates remains unknown. The pathophysiological mechanisms of blister formation in PV are still uncertain. Currently, two possible mechanisms for blister formation in PV include; (1) interference of desmosomal cadherin-bound antibody with intracellular events, and/or (2) direct inhibition of the extracellular Dsg3 adhesive ability by autoantibody binding to Dsg3. Recently,Caldelari et al., 2001Caldelari R. de Bruin A. Baumann D. Suter M.M. Bierkamp C. Balmer V. Muller E. A central role for the armadillo protein plakoglobin in the autoimmune disease pemphigus vulgaris.J Cell Biol. 2001; 153: 823-834Crossref PubMed Scopus (162) Google Scholar speculated that steric hindrance alone might not be sufficient to induce PV lesions without keratin retraction using plakoglobin deficient embryonic keratinocytes. Our observations of the lateral surface of basal cells are supportive of keratin retraction under in vivo conditions. The splitting of desmosomes, however, cannot be explained by the keratin retraction alone. Mechanical stress may be the final trigger initiating the separation of desmosomes, and these desmosomes are susceptible to stress for reasons other than the loss of keratin retraction itself because split-desmosomes still retain keratin filaments and Dsg3 labeling. Some as yet, unknown mechanism, which requires plakoglobin, for cell–cell detachment may be an important event after autoantibody binding. Previously, we have reported that the PV model mice showed ultrastructure strikingly similar to that of Dsg3 knockout mice (Shimizu et al., 2002Shimizu A. Ishiko A. Ota T. Tsunoda K. Koyasu S. Amagai M. Nishikawa T. Ultrastructural changes in mice actively producing antibodies to desmoglein 3 parallel those in patients with pemphigus vulgaris.Arch Dermatol Res. 2002; 294: 318-323PubMed Google Scholar). Here we have proven the presence of Dsg3 in the desmosome during acantholysis. A loss of Dsg3 function may be one mechanism causing acantholysis in our PV model mice. Recently,Tsunoda et al., 2003Tsunoda K. Ota T. Aoki M. et al.Induction of pemphigus phenotype by a mouse monoclonal antibody against the amino-terminal adhesive interface of desmoglein 3.J Immunol. 2003; 170: 2170-2178Crossref PubMed Scopus (261) Google Scholar identified a monoclonal antibody isolated from our PV model mouse that has pathogenic activity and this antibody is able to induce acantholysis with the epitope recognizing the amino-terminal adhesive interface of Dsg3. Together with the previous findings, our results may indicate that autoantibody binding to Dsg3 in the desmosome causes desmosomal splitting by direct inhibition of the adhesive function(s) of Dsg3 at least in some cellular regions of this in vivo PV mouse model. PV model mice were produced as described previously (Amagai et al., 2000Amagai M. Tsunoda K. Suzuki H. Nishifuji K. Koyasu S. Nishikawa T. Use of autoantigen-knockout mice in developing an active autoimmune disease model for pemphigus.J Clin Invest. 2000; 105: 625-631Crossref PubMed Scopus (225) Google Scholar). In brief, Dsg3 -/- mice, which did not have any immunological tolerance to Dsg3, were immunized using recombinant mouse-Dsg3 and their splenocytes were transferred to 7-week-old Rag2 -/- immunodeficient mice (Schulz et al., 1996Schulz R.J. Parkes A. Mizoguchi E. Bhan A.K. Koyasu S. Development of CD4−CD8− abTCR+NK1.1+ T lymphocytes: Thymic selection by self antigen.J Immunol. 1996; 157: 4379-4389PubMed Google Scholar), which were obtained from Taconic Farms (Germantown, New York). The Rag2 -/- mice stably produced anti-Dsg3 autoantibodies and showed erosions in oral mucous membranes and patchy hair loss 15–25 d after the cell transfer. Rag2 -/- mice without this transfer of Dsg3 -/- mouse splenocytes were used as the control mice in this study. Samples taken from the oral mucous membrane of the PV model mouse were fixed with 2% glutaraldehyde and 1% osmium tetroxide, and processed for EM using conventional methods. Samples were taken from the oral mucous membrane of three PV model mice and two Rag2 -/- mice for immuno-EM. Post-embedding immunogold EM with cryofixation and freeze substitution without using chemical fixatives was performed, as described previously (Shimizu et al., 1994Shimizu H. Masunaga T. Ishiko A. Hashimoto T. Garrod D.R. Shida H. Nishikawa T. Demonstration of desmosomal antigens by electron microscopy using cryofixed and cryosubstituted skin with silver-enhanced gold probe.J Histochem Cytochem. 1994; 42: 687-692Crossref PubMed Scopus (23) Google Scholar). In brief, small pieces (<1 mm2) of mucous membrane samples were cut into small pieces and cryofixed by rapidly plunging them into cooled liquid propane (-190°C) and cryosubstituted in 100% methanol for 48 h at -80°C followed by embedding in Lowicryl K11M (Chemische Werke Lowi Waldkraiburg, Germany) at -60°C. Specimens were polymerized with UV radiation at 60°C for 48 h and at room temperature for another 48 h. The ultrathin transverse sections were cut and collected on nickel grids covered with a polyvinyl folmvar support film and processed for immunogold labeling. The ultrathin sections of the mucous membrane of the PV model mice, which were mounted on nickel grids, were placed on a drop of blocking buffer, 5% normal goat serum (NGS), 0.8% bovine serum albumin (BSA), 0.1% gelatin and 2 mM NaN3 in phosphate-buffered saline (PBS), pH 7.4, for 30 min. The grids were incubated in affinity purified rabbit anti-mouse IgG (H+L) (American Qualex Antibodies; San Clemente, California) at 1:320 or rabbit anti-desmoplakin (Research Diagnostics, Flanders, New Jersey) at 1:40 diluted in incubation buffer (PBS pH7.4 with 1% NGS, 0.8% BSA, 0.1% gelatin and 2 mM NaN3) overnight at 4°C and washed in the washing buffer (PBS pH7.4 with, 0.8% BSA, 0.1% gelatin and 2 mM NaN3). Then, they were incubated in 5 nm or 15 nm colloidal gold-conjugated goat anti-rabbit IgG (Amersham Life Science, Olen, Belgium) at 1:40 diluted with the incubation buffer for 2 h at room temperature and washed by washing buffer and distilled water. To obtain suitable labeling sizes for observation, the 5 nm gold probes were amplified by incubating the grids with the IntenSE silver enhancement kit (Amersham). Then, the sections were stained with saturated uranyl acetate for 6 min and lead citrate for 1 min and examined by EM (Model 1200EX, JEOL, Tokyo, Japan). For the control study, Rag2 -/- mice mucous membranes were used as a substrate instead of the samples from the PV model mice. The ultrathin sections of the mucous membrane of control mice were immunolabeled by the same method described above using the PV model mouse serum diluted 1:40 and 5 nm colloidal gold-conjugated goat anti-mouse IgG (Amersham) at 1:40 dilution. For the controls, normal mouse serum was used as primary antibody instead of PV model mouse serum. Double immunohistochemical labeling was performed by methods of Bendayan (Bendayan, 1982Bendayan M. Double immunocytochemical labeling applying the protein A-gold technique.J Histochem Cytochem. 1982; 30: 81-85Crossref PubMed Scopus (291) Google Scholar) with a small modification. In brief, the ultrathin sections of the mucous membrane of PV model mice were collected on nickel grids, which were not covered with support films. First, face A of the tissue section, corresponding to the face where sections were completely exposed, was immunostained with rabbit anti-β-catenin (1:40, H-102, Santa Cruz Biotechnology, Delaware Avenue, California) or with rabbit anti-desmoplakin (1:40, Research Diagnostics) by floating the grids on a drop of antibody with face A down. After washing and drying, the grids were turned face-over, and face B, the opposite face of the sections/grids were immunolabeled with rabbit anti-mouse IgG using a different size of gold particle as the marker by floating the grids with face B down. After being washed and stained with uranyl acetate and lead citrate, carbon coating was performed and the sections were observed under the EM. For the control study, Rag2 -/- mice mucous membranes were used instead of the sample from the PV model mice. Non-oblique immunoelectron micrographs of the PV model mice in which IgG was labeled, and control mice in which Dsg3 was labeled, were taken at the same magnification (×30, 000) and digitally scanned at a high resolution for image analysis using the Image-Pro Plus (Media Cybernetics; Media Cybernetics; Silver Spring, Maryland). The distance or thickness (in nm) between the two desmosomal plasma membranes (L1 in Figure 9a) and the distance of the gold particles from the mid point of L1 (L2 in Figure 9a) were measured in pairs for more than 400 gold particles. The horizontal length of the desmosomal attachment plaque (L3 in Figure 9b) and the distance between a gold particle and the mid point of L3 (L4 in Figure 9b) were also measured. Both vertical and horizontal length data were converted into relative distances using the formula: L2/L1×100 and L3/L4×100, respectively. The relative distance and relative frequencies were plotted as histograms. This work was supported by Health Science Research Grants for Research on Specific Diseases from the Ministry of Health and Welfare and Grant-in Aids for Scientific Disease from the Ministry of Education, Science and Culture of Japan. We would like to thank Toshihiko Nagai and Hiroshi Oka for their technical assistance and James R. McMillan for proofreading this manuscript.