Immune dysregulation of pemphigus in humans and mice


Masayuki Amagai, M.D., Ph.D., Department of Dermatology, Keio University School of Medicine, 35 Shinanomachi, Shinjuku-ku, Tokyo 160-8582, Japan. Email:


Pemphigus vulgaris is an autoimmune blistering disease of the skin and mucous membranes that is caused by immunoglobulin (Ig)G autoantibodies against the cadherin-type adhesion molecule desmoglein (Dsg)3 expressed on stratified epithelial cells. Interaction between antigen-specific T and B cells, which is selectively achieved through T-cell receptor/major histocompatibility complex–peptide complex association and subsequently corroborated by co-stimulatory molecules such as CD40/CD154, is required for production of pathological anti-desmoglein 3 antibody. Some genetically and environmentally susceptible individuals harbor desmoglein reactive B and T cells, and anti-desmoglein antibodies were detected in their serum. Analysis of the anti-desmoglein antibody clones derived from pemphigus patients or pemphigus model mice revealed that pathogenic antibodies mostly react with conformational epitopes on mature form desmogleins, whereas non-pathogenic ones tend to react with non-conformational epitopes. Surprisingly, antibodies to the Dsg1 precursor pro-protein are also cloned from individuals without pemphigus. These observations suggest that active suppression by regulatory cell subsets is dominant in these susceptible individuals. In fact, Dsg-reactive T-inducible regulatory type 1 (Tr1) cells are readily detected in healthy carriers of pemphigus-related human leukocyte antigen haplotypes, but rarely in pemphigus patients. These Tr1 cells can be functionally converted to T-helper 2-like cells which secrete interleukin-2 by inactivation of Foxp3 through antisense oligonucleotides. Thus, delicate balance between self-reactive lymphocytes and regulatory T cells may be a key element in determining whether individuals produce pathogenic antibodies and develop pemphigus phenotypes or not.


The adoptive immune system is based on the random generation of immunoglobulin (Ig) and T-cell receptors (TCR) to create a repertoire that can react with a broad range of exogenous antigens. However, this strategy harbors potential risks to generate repertoire-recognizing self-antigen and to cause the “horror autotoxicus”. To prevent the generation of an autoreactive repertoire, developing lymphocytes undergo a series of checkpoints including clonal deletion, receptor editing, and anergy induction. It is induced at primary sites of lymphocyte development, namely, the bone marrow for B cells and the thymus for T cells.1,2 Unfortunately, these checkpoints cannot eliminate self-reactive lymphocytes. Such self-reactive lymphocytes are under active suppression by the regulatory T-cell subsets, as represented by CD4+ CD25+ naturally occurring regulatory T cells.3 Despite these elegant mechanisms, self-tolerance is breached leading to the initiation of autoimmunity.

Pemphigus is a chronic blistering disease of the skin and mucous membranes that is caused by IgG autoantibodies against the cadherin-type adhesion molecule desmoglein (Dsg)1 and Dsg3 expressed on stratified epithelial cells.4 There are two major forms of pemphigus: pemphigus vulgaris (PV) and pemphigus foliaceus (PF).5 In PV, essentially all patients have mucosal membrane erosions and more than half of them also have skin blisters and erosions. The blisters of PV develop in the deeper part of the epidermis, just above the basal cell layer. In PF, patients have only cutaneous involvement without mucosal lesions and the splits occur in the superficial part of the epidermis, mostly at the granular layer. PF and PV antigens are termed Dsg1 and Dsg3, respectively. Thus, pemphigus could be redefined to be an anti-Dsg IgG-mediated autoimmune disease. Pemphigus is one of the most clearly defined autoimmune diseases mediated by autoantibodies and serves as a suitable model for investigation of the mechanisms of breakdown of self-tolerance. In this review, we discuss the immunopathogenesis of pemphigus, mainly focusing on T- and B-cell tolerances, interaction of these two populations, active immune suppression by regulatory T cells, and recently emerging advances in the analysis of anti-Dsg antibodies.

T- and B-Cell Interaction in the pathogenesis of pemphigus

To produce antigen (Ag)-specific IgG, cognate interaction between Ag-specific T and B cells, which is selectively achieved through TCR/major histocompatibility complex (MHC)–peptide complex association, is required.6 This is called “linked recognition”. Subsequently, activated CD4+ T-helper cells (Th) express CD40L (CD154), which binds to CD40 expressed on the B cell. CD40–CD154 interaction induces the expression of activation-induced cytidine deaminase (AID) in B cells, which is essential for Ig-isotype class switch recombination and somatic hypermutation.7 Activated autoreactive B cells also act as potent antigen-presenting cells to T cells recognizing the cognate Ag.8 Thus, a positive feedback loop between T and B cells is crucial in initiation and amplification of autoimmune responses.9

Adaptive transfer of lymphocytes from Dsg3−/− mice, which have not acquired tolerance against Dsg3, into Rag2−/− immunodeficient mice which express Dsg3 results in stable production of anti-Dg3 antibody (Ab) in the recipient mice.10,11 Anti-Dsg3 IgG bound to keratinocyte cell surfaces in vivo, and caused suprabasilar acantholysis in the oral mucosa and esophagus. As a result, the mice showed weight loss due to the difficulty of food intake. The recipient mice also developed patchy hair loss or telogen hair loss.12 Using this PV model mouse system, a number of fundamental issues in PV pathogenesis have been investigated. When purified T and B cells from Dsg3−/−, Dsg3+/− or Dsg3+/+ mice were mixed with various combinations and transferred to Rag2−/− mice, apparent pathogenic anti-Dsg3 IgG production was observed only with a combination of Dsg3−/− T and Dsg3−/− B cells but not with other combinations.13 These results clearly indicated that loss of self-tolerance against Dsg3 in both T and B lymphocytes is required for efficient production of pathogenic anti-Dsg3 IgG antibodies.

The importance of Th cells in the pathogenesis of PV has been further suggested by recent studies. It has been shown that anti-Dsg3 IgG secreted by autoreactive B cells were detected by enzyme-linked immunosorbent spot (ELISPOT) upon in vitro stimulation of peripheral lymphocytes from PV patients with Dsg3. In contrast, the activation of autoreactive B cells was virtually absent when the peripheral lymphocytes were depleted of CD4+ T cells.14 B cells from Dsg3−/− mice could produce pathogenic anti-Dsg3 IgG and cause PV phenotypes when transferred into Rag2−/− mice with anti-Dsg3-reactive Th clones. However, Dsg3−/− B cells could not produce anti-Dsg3 IgG in the recipient mice when transferred without Th clones.15 These results in both human and mouse collectively show that autoreactive T cells are indispensable for the induction and regulation of anti-Dsg3 autoantibody production. The interaction between antigen-specific T cells and B cells through their co-stimulatory molecules is required for efficient Ab production. Indeed in PV model mice, anti-Dsg3 Ab production was almost completely blocked by administration of anti-CD154 monoclonal Ab (mAb) which blocks CD40/CD154 interaction.16 Interestingly, the tolerance to Dsg3 was transferable because co-transfer of splenocytes from anti-CD154 mAb-treated mice and naïve Dsg3−/− splenocytes significantly suppressed anti-Dsg3 IgG production in the recipient mice. This result indicated that antigen-specific immunoregulatory cell was induced by anti-CD154 mAb treatment. Recent studies revealed that CD4+ CD25+ regulatory T cells are induced in allograft transplant recipients given anti-CD154 mAb.17,18 Although further investigation should be performed to clarify the characteristics of induced immunoregulatory cells in PV model mice by co-stimulatory blockade, CD4+ CD25+ regulatory T cells are supposedly involved in the tolerance induction.

T-Cell Tolerance in pemphigus

T-cell development occurs within the thymus. The development of T cells includes a series of selection processes that strictly eliminate autoreactive T cells. In thymus, a specialized subset of thymic stromal cells known as medullary thymic epithelial cells (mTEC) “promiscuously” express a wide array of tissue-restricted antigens (TRA) derived from nearly all organs in the body and regulates negative selection of autoreactive T cells against TRA.19,20 This mechanism to delete self-reactive T cells is termed “central tolerance”.2 However, autoreactive lymphocytes that escape clonal deletion in the thymus are kept from activation and expansion in the periphery by regulatory T-cell subsets such as CD4+ CD25+ regulatory T cells (“peripheral tolerance”).3 CD4+ CD25+ regulatory T cells (nTreg) develop in the thymus and specifically express forkhead box P3 (Foxp3), a transcription factor that plays a critical role in their development and function.21 The ectopic expression of Foxp3 converts the phenotype of naive T cells to a phenotype similar to that of nTreg.21,22 On the contrary, it has been shown that acute in vivo ablation of Foxp3+ Treg lead to activation of self-reactive T cells and fatal autoimmunity.23,24 Recent study revealed that a substantial percentage of Treg lost Foxp3 expression and became an activated-memory T-cell phenotype and produced inflammatory cytokines.25 Thus, the balance between Treg and effector T cells is critical for the proper control of the immune responses and existence of Foxp3+ Treg seems to be indispensable throughout life to keep immunological self-tolerance. Besides thymus-derived Foxp3+ nTreg, there are other types of regulatory T cells induced from naive T lymphocytes in the periphery. Such “adoptive Treg (also called iTreg)” include Foxp3+ iTreg,26 interleukin (IL)-10 secreting T-inducible regulatory type 1 (Tr1) cells,27 and transforming growth factor (TGF)-β secreting Th3 cells.28

In pemphigus patients, several epidemiological studies showed that human leukocyte antigen (HLA)-DRB1*0402 is associated with PV in Jewish and HLA-DQB1*0503 in non-Jewish populations and appear to recognize several epitopes on the extracellular domain of Dsg3.29–31 It is noteworthy that autoreactive Th cells recognizing identical epitopes of the Dsg3 ectodomain were also identified in healthy individuals that express the PV-associated HLA class II alleles.32,33 Active immune suppression may thus be operative in healthy individuals possessing Dsg3-responsive T cells. It has been shown that Dsg3-reactive IL-10-secreting Tr1 cells were identified in five of six healthy carriers of the PV-associated HLA class II alleles, DRB1*0402 and DQB1*0503, but were only rarely detected in PV patients.34 These Tr1 cells exert their Dsg3-dependent inhibitory function on the proliferative response of Dsg3-responsive, autoreactive Th clones by the secretion of IL-10 and TGF-β and in the cell contact independent manner. These cells are further divided into two subpopulations based on the cell size and granularity. The smaller subpopulation expresses Foxp3 and secretes IL-10, TGF-β and IL-5 upon stimulation with Dsg3. On the contrary, the larger subpopulations showed a Th-like phenotype, lacking Foxp3 expression and secrete IL-2.35 Downregulation of Foxp3 by antisense oligonucleotides converted these Tr1 cells into Th2-like cells which secreted IL-2 and lost an anergic state to Dsg3 antigen stimulation.35,36 These studies imply that the active control of Dsg3 reactive lymphocyte by Treg is critical in preventing the development of PV. There may be a physiological balance between these two populations required for maintenance of peripheral tolerance.36 It is unclear, however, whether most individuals have such Dsg3-reactive T cells or only a fraction of population have such autoreactive T cells and develop PV. It has also been reported that peripheral blood cells from PV patients contain a decreased number of CD4+ CD25high regulatory T cells compared with healthy controls.37 Expression of Foxp3 was also reduced in CD4+ CD25+ T cells from PV patients. However, functional analysis of Treg was not performed in PV patients because of the limited number of the Treg population.

These studies collectively indicate that breakdown of immune suppression by Treg on the autoreactive Th cells is operative in the PV pathogenesis. However, it is not clear whether Treg directly suppress autoreactive B cells. Recently, Ota et al.38 have shown that in AK7-transgenic mice expressing non-pathogenic AK7 IgM against Dsg3, autoreactive transgenic B cells were neither deleted nor inactivated under normal conditions but eliminated from the periphery when a pathogenic anti-Dsg3 IgG1 mAb (AK23) capable of inducing blisters was transferred. T-cell transfer studies into AK7-Rag2−/− mice revealed that autoreactive B-cell elimination required CD4+ T cells from wild-type mice and that CD4+ T cells from gld (FasL mutant) mice were unable to eliminate autoreactive B cells.38 Furthermore, AK23 mAb injection increases the frequency of CD4+ CD25+ Treg in the recipient mice. These studies raised the possibility that Treg directly act on autoreactive B cells when activated by the target antigen. These data also suggest that dysregulation of regulatory T-cell function leads to the activation and expansion of autoreactive B cells and the onset of PV.39

Autoreactive T cells are thought to be involved in the pathogenesis of PV, but evidence for their direct pathogenicity is almost lacking. Recently, CD4+ T-cell lines reactive against Dsg3 are established.15 Dsg3-reactive CD4+ T-cell lines generated in vitro were adoptively transferred into Rag2−/− immunodeficient mice with primed B cells derived from Dsg3-immunized Dsg3−/− mice. Seven of 20 T-cell lines induced IgG anti-Dsg3 Ab production and acantholytic blister, a typical disease phenotype, in recipient mice. Comparison of the characteristics between pathogenic and non-pathogenic Dsg3-reactive T cell lines led to the identification of IL-4 and IL-10 as potential factors associated with pathogenicity. Further in vitro analysis showed that IL-4, but not IL-10, promoted IgG anti-Dsg3 Ab production by primed B cells. Additionally, adenoviral expression of soluble IL-4Rαin vivo suppressed IgG anti-Dsg3 Ab production and the PV phenotype, indicating a pathogenic role of IL-4.

As reviewed in this section, Dsg3-reactive T cells play cardinal roles in PV pathogenesis. Regulatory T cells actively suppress these autoreactive T cells in healthy individuals. Thus, manipulation of regulatory T cells may be an ideal therapeutic strategy for PV patients. Cytokines produced by autoreactive T cells are also expected to be a good target for the treatment of pemphigus.

B-Cell Tolerance in pemphigus

Rearrangement of Ig genes during B-cell development potentially leads to the generation of B-cell repertoire containing self-reactive B cells. Such self-reactive B cells are either eliminated or inactivated during lymphocyte development, resulting in self-tolerance.40,41 B-cell receptor avidity defines the fate of the developing B cells. B cells with a high-avidity receptor undergoes editing–deletion pathway, whereas those with weaker avidity permit anergic cell development. B cells expressing receptors that are weakly cross-linked by self will emerge from the bone marrow and appear in the periphery as normal non-self-reactive B cells. This has been termed “clonal ignorance”.1 These insights in the regulation of B-cell tolerance have been acquired by the studies using transgenic animals with B-cell receptor reactive with various endogenous and genetically expressed neo-self-antigens.42

In the pemphigus animal model, anti-Dsg3-IgM transgenic mice were developed using variable regions of both the H and L chains of AK7, one of the Dsg3-reactive but not pathogenic mAb.43 In these mice, Dsg3-reactive B cells were readily detected in the bone marrow and the spleen. These B cells were not inactivated nor anergized, and produced a substantial amount of anti-Dsg3 IgM.44 This study clearly showed the presence of self-reactive B cells in the periphery, implying that Dsg3-reactive B cells are also present in normal individuals. Actually, a low level of anti-Dsg3 Ab was detected in healthy relatives of PV patients.45–47 Furthermore, anti-Dsg1 IgM was highly detected in the serum of patients with fogo selvagem (FS), an endemic form of pemphigus foliaceus among endemic regions of Brazil.48 Interestingly, anti-Dsg1 IgM was also detected in the serum of healthy individuals residing in rural settings in or near an endemic area.48 These results collectively indicate that mere presence of anti-Dsg antibodies is not enough to develop pemphigus. A key determinant is the epitopes recognized by anti-Dsg Ab. Certain additional factors may trigger the pathogenic Ab production and development of pemphigus phenotypes in the genetically susceptible individuals.

Characterization of pv autoantibodies

Characterization of the autoantibodies in pemphigus patients was performed using different approaches.49–51 One of the most recent advances is phage display, which has been used to obtain single-chain V-region fragments (scFv) and several human mAb against Dsg3 and Dsg1 from a PV patient49 and mAb against Dsg1 from a PF patient.51 In the phage display system, an Ab phage display library is constructed from the peripheral blood lymphocytes isolated from a pemphigus patient. Dsg-reactive clones are then selected and expanded by panning from this library. These scFv are pathogenic against keratinocytes in both human cell culture and neonatal mice. Epitope mapping of the two pathogenic scFv derived from a mucocutaneous PV patient on the domain swapped mouse desmogleins shows that these pathogenic scFv bind the amino terminal domains of Dsg, which is consistent with previous data.43,52,53 Interestingly, scFv clones reactive with both Dsg3 and Dsg1 are isolated from mucocutaneous PV patients, indicating that a single mAb can bind and pathologically inactivate both Dsg.49 Genetic analysis of cloned scFv reveals that the usage of the VH gene is limited.

From a PF patient, two unique clones of anti-Dsg1 scFv that were pathogenic in mouse and human skin were isolated.51 Both pathogenic scFv bound conformational epitopes in the N-terminus of Dsg1. Genetic analysis of pathogenic and non-pathogenic scFv revealed that their heavy-chain usage was limited to only five genes, that their immunological properties correlated with Ab heavy-chain gene usage, and that, although the light-chain gene usage was more promiscuous, only certain light chains enable the binding with Dsg1. Interestingly, some anti-Dsg1 scFv isolated from a PF patient share the heavy-chain gene usage with scFv isolated from a PV patient. Rather restricting usage of the heavy-chain gene enables more targeted therapy of pemphigus. Rabbit anti-idiotype antisera against recombinant scFv clones from a PV patient specifically inhibit the pathogenicity of PV mAb.54 It may be also possible to design “pathogenic antibody enzyme-linked immunosorbent assay” by blocking non-pathogenic Dsg epitopes with non-pathogenic scFv.51 These therapeutic and diagnostic implications in clinical settings deserve further investigation.

Autoantibody against premature form desmogleins

Interestingly, in immunoprecipitation experiments using recombinant human Dsg1 produced by a baculovirus expression system, some antibodies immunoprecipitated a Dsg1 polypeptide with a slightly greater apparent molecular weight on sodium dodecylsulfate polyacrylamide gel electrophoresis than other clones. At the same time, these Ab showed cytoplasmic staining in the superficial layers of epidermis by indirect immunofluorescence on human skin. These results collectively indicate that they bind to the pro-protein but not to the mature protein produced by the baculovirus system.51 Dsg1 is initially synthesized in the endoplasmic reticulum as an inactive precursor protein (pro-protein) with an N-terminal pro-peptide, and transit through the cis-and trans-Golgi.55 Subsequently, they are processed by pro-protein convertase such as furin to become biologically active mature proteins that is assembled into desmosomes on the cell surface.56,57 Detailed characterization of anti-Dsg mAb revealed that pathogenic anti-Dsg antibodies bind only the mature form Dsg, whereas non-pathogenic ones bind only the precursor form or bind both mature and precursor forms.58,59 A longitudinal study of pemphigus patients showed that immunoreactivity against the mature form Dsg1 reflects the disease activity well.59 These findings suggest that pemphigus patients might first develop antibodies against the intracellular precursor form Dsg to which they would not be expected to have tolerance, and in some susceptible patients the Ab response might extend to the mature molecule.59 Recently, mAb against premature form Dsg1 are cloned from a healthy person and two patients with thrombotic thrombocytopenic purpura.60 These non-pathogenic clones have distinct VH gene usage from pathogenic anti-mature Dsg1 antibodies cloned from a PF patient, suggesting that pathogenic anti-Dsg1 Ab producing B cells in PF patients are not derived through somatic mutation from the B cells expressing anti-pre-Dsg1 Ab.60

Epitope spreading in pemphigus

Epitope spreading is a phenomenon in which autoimmunity extend to include new epitopes, within the same or a different molecule and is thought to be involved in the disease onset of autoimmune disorders.61,62 The involvement of epitope spreading in pemphigus pathogenesis was first demonstrated in fogo selvagem, endemic pemphigus foliaceous.53 FS patients in preclinical stage have Ab recognizing epitopes on the C-terminal EC5 domain of Dsg1, and disease onset was associated with the emergence of Ab specific for epitopes on the N-terminal EC1 and/or EC2 domains. In several cases of pemphigus, it has been described that intermolecular epitope shift between Dsg1 and Dsg3 occurs along with the transition of pemphigus phenotype as rare events.63–65 Although the exact immunological mechanisms of the epitope spreading in human PV are largely unknown, analysis of Dsg3-reactive T-cell clones established from PV model mouse revealed that a single Dsg3-reactive T-cell clone was able to help polyclonal naive B cells produce anti-Dsg3 IgG independent of T-cell epitopes on Dsg3.66 B cells not only produce an Ab but also function as very efficient antigen-presenting cells. Because a B cell can present any antigen that it takes up as part of a macromolecular complex, it can in principle activate and get help from a broad range of T cells.8,67 Thus, positive feedback loop between T and B cells may boost the autoimmune response and induce epitope spreading.9

Future direction

Pemphigus is one of the most clearly defined autoimmune diseases mediated by autoantibodies and serves as a suitable model for investigation of the mechanisms of breakdown of self-tolerance. Recent studies revealed that genetically susceptible individuals harbor Dsg3-reactive B and T cells which escaped central tolerance mechanisms. These autoreactive lymphocytes are actively suppressed by regulatory T cell subsets such as CD4+ CD25+ Treg and IL-10 producing Tr1 cells. Once this delicate balance between self-reactive lymphocytes and regulatory T cells collapses, a positive feedback loop between activated self-reactive T and B cells augments the rapid diversification of the humoral immune response resulting in generation of pathogenic antibodies (Fig. 1). Further investigation will provide new insights in PV pathogenesis, and this will bring us a more targeted remedy of this potentially fatal autoimmune disorder.

Figure 1.

 Central tolerance mechanism does not necessarily eliminate desmoglein (Dsg)3 reactive B and T cells. These autoreactive lymphocytes are actively suppressed by regulatory T-cell subsets such as CD4+ CD25+ regulatory T cells (Treg) and interleukin (IL)-10 producing T-inducible regulatory type 1 (Tr1) cells. CD4+ CD25+ Treg may directly act on autoreactive B cells through the Fas–FasL pathway when activated by the target antigen. Dsg3-reactive Tr1 cells are able to be functionally converted to T-helper (Th)2-like cells by inactivation of Foxp3 through antisense oligonucleotides. Once this delicate balance between self-reactive lymphocytes and regulatory T cells collapses, a positive feedback loop between activated self-reactive T and B cells augments the rapid diversification of the humoral immune response resulting in generation of pathogenic antibodies (Ab). Teff, effector T cells.