1. ^*Department of Dermatology, University of North Carolina at Chapel Hill, Chapel Hill, North Carolina, USA
2. ^†Department of Microbiology and Immunology, University of North Carolina at Chapel Hill, Chapel Hill, North Carolina, USA

Pemphigus vulgaris (PV) is a life-threatening autoimmune blistering disease. Clinically, PV targets the skin and mucosal epithelia. Histologically, PV is characterized by detachment of suprabasal keratinocytes ("suprabasal acantholysis"), leaving a row of basal keratinocytes attached to the dermis producing a "tombstone-like" appearance (^{Lever, 1965}). PV patients are known to possess autoantibodies that recognize desmoglein-3 (Dsg3). These autoantibodies are detected bound to the keratinocyte cell surface of lesional epidermis and also in circulation. Dsg3 is a 130 kDa desmosomal glycoprotein that belongs to the desmoglein subfamily of the cadherin superfamily (^{Stanley et al, 1982;Amagai et al, 1991}), a class of molecules that play an important role in mediating cell–cell adhesion (^{Kowalczyk et al, 1999}). Anti-Dsg3 autoantibodies are pathogenic as determined by the passive transfer studies in which IgG isolated from sera of pemphigus patients was injected into neonatal mice and induced acantholysis with the classic immunohistological features seen in human patients (^{Anhalt et al, 1982;Amagai et al, 1992}). The pathogenic relevance of Dsg3 is further demonstrated by an active disease model of PV, in which splenocytes from Dsg3 knockout mice immunized with mouse Dsg3 were adoptively transferred to Rag2-/- mice, producing a PV disease phenotype (^{Amagai et al, 2000}).

Indirect immunofluorescence and immunoelectron microscopy (EM) studies showed that PV antibodies bind to the intercellular spaces of the epidermis where the ectodomains of Dsg3 interact to bring about cell–cell adhesion. Dsg3-specific autoantibodies predominantly stain the lower layers of epidermis, precisely corresponding to the site where blisters occur in PV. But the precise molecular mechanisms of anti-Dsg3 IgG-induced suprabasal acantholysis are not fully understood. The acantholytic process has been studied at the ultrastructural level in lesional epidermis or mucosa of PV patients. These changes have also been reported in the organ culture and epidermal cell culture systems where PV IgG was introduced in the growth media. These studies were repeated in skin specimens obtained sequentially from neonatal mice injected with PV IgG. Some of these studies showed PV IgG bound exclusively to desmosomal cores, whereas others demonstrated binding of PV IgG that was not restricted to desmosomes.

In this issue,^{Shimizu et al (2004)} employ post-embedding immunogold EM to study the in vivo binding of anti-Dsg3 IgG and the fate of IgG-bound desmosomes during acantholysis in the active PV mouse model. They found that in non-acantholytic areas, anti-Dsg3 antibodies bind to the extracellular portion of the desmosome but not the non-desmosomal plasma membrane. At the acantholytic sites, desmosomes were separated into two halves and there was no internalization of spilt-desmosomes. PV IgG was detected in the split desmosomes, indicating that Dsg3 was not depleted from the split desmosomes during acantholysis. One interesting finding in this study is the observation that the ultrastructural changes in desmosomes during acantholysis are different between apical and lateral surfaces of basal keratinocytes. At the apical surface of basal keratinocytes, desmosomes were intact before acantholysis and then split in half after acantholysis. These split desmosomes are connected to the keratin filament network with proper co-localization of desmoplakin, suggesting that there was no keratin retraction. In contrast, at the lateral surface of the basal keratinocytes, desmosomes at both lesional and peri-lesional areas showed keratin retraction as evidenced by absence of cytoplasmic desmosomal attachment plaque and keratin filament association with the attachment plaque.

In vivo deposition of PV IgG both in the desmosomal plaque and the non-desmosomal plasma membrane has been reported (^{Bedane et al, 1996}). In cultured keratinocyte system, PV autoantibodies bind to the plasma membrane and are internalized, resulting in the formation of desmosomes without Dsg3 (^{Aoyama and Kitajima, 1999}). The current study demonstrated that anti-Dsg3 IgG is predominantly deposited at the extracellular region of desmosomes at both peri-lesional and lesional sites. These data show that in this PV animal model, (1) pathogenic PV IgG antibodies get full access to the desmosomal Dsg3 and (2) Dsg3 is still present in the split desmosomes, suggesting that depletion of Dsg3 from desmosomes does not occur and hence is not the cause of suprabasal acantholysis. Whether this is the case for human PV should be determined by revisiting skin biopsies from human PV patients.

This study also demonstrated that IgG binding to Dsg3 causes desmosomal splitting without keratin retraction. In the IgG passive transfer PV mouse model, however,^{Takahashi et al (1985)} showed that the binding of PV autoantibodies to the epidermal cells is followed by widening of the intercellular space between desmosomal junctions and then by splitting of the desmosomes. Coincident with cell detachment, intracellular tonofilaments retracted from the cell periphery and clustered in a perinuclear position (^{Takahashi et al, 1985}). Whether a higher level of circulating anti-Dsg3 IgG in an acute disease setting in the IgG passive transfer model causes this discrepancy needs to be further investigated.

What causes this unique desmosome structural changes at the lateral surface of the basal keratinocyte? It has been shown that desmoplakin is crucial for assembly of functional desmosomes and cytoskeletal linkage (^{Vasioukhin et al, 2001}). It is likely that anti-Dsg3 IgG binding to Dsg3 in the lateral side desmosomes causes the retraction of keratin filaments and the loss of desmoplakin from the desmosomal attachment plaque, resulting in desmosomes lacking the inner plaque and the attachment to the keratin cytoskeleton.

The most intriguing finding in this study is that anti-Dsg3 IgG binding to its target causes structural/morphological changes of desmosomes differently at the apical and lateral surfaces of the same basal keratinocytes. The question is why IgG binding causes the apical side desmosomes to split without keratin retraction and the lateral side desmosomes to have keratin retraction without splitting? Is it due to the differences in the distribution of desmosomes and/or adherens junctions or is it due to the differences in desmosome composition in these two sides? More detailed investigations are required to explore these issues.

Several mechanisms for acantholysis have been proposed including complement activation, proteinase activation, steric hindrance blocking the adhesive site of Dsg3 and activation of transmembrane signaling that downregulates cell–cell adhesion. The finding that monovalent Fab' fragments of PV IgG were able to induce blisters in complement-sufficient and complement-deficient mice demonstrated that neither complement activation nor antigen cross-linking was required for PV antibodies to induce blisters (^{Mascaro et al, 1997}). Passive transfer experiments using mice deficient in plasminogen activator system (tPA and uPA), neutrophil elastase and gelatinase B demonstrated that these proteolytic enzymes were also unnecessary for pemphigus blister formation (^{Liu et al, 1999;Mahoney et al, 1999}). In cell culture systems, PV IgG has been shown to activate intracellular inositol 1,4,5-trisphosphate (IP₃) and Ca²⁺ flux, increase protein kinase C activity, and trigger Dsg3 phosphorylation (^{Kitajima, 2002}).^{Caldelari et al (2001)} recently showed that plakoglobin, a desmosomal plaque protein, played an important role in PV IgG-induced acantholysis. Plakoglobin+/+ keratinocytes, but not plakoglobin-/- keratinocytes, respond to PV IgG with keratin retraction and loss of cell adhesion, suggesting that steric hindrance alone might not be sufficient for suprabasal blistering. Taken together, these results suggest that binding pathogenic antibodies to Dsg1 or Dsg3 may impair the adhesive function of these molecules either directly or indirectly by triggering signaling events, leading to cell–cell detachment by mechanisms that need to be elucidated. The ultrastructural observation by Shimizu et al that IgG binding to Dsg3 causes the desmosome to split in half without keratin retraction is compelling because it suggests that PV suprabasal blistering could be caused by IgG binding that directly blocks the adhesive functions of Dsg3. A definite proof for this "direct hit" theory requires further investigations.