Autoreactive T cells in the immune pathogenesis of pemphigus vulgaris
Kyle T. Amber,
Department of Dermatology and Cutaneous Surgery, University of Miami Miller School of Medicine, Miami, FL, USA
Correspondence: Kyle T. Amber, BS, Department of Dermatology and Cutaneous Surgery, University of Miami Miller School of Medicine, 10660 SW 75th Ave, Miami, FL 33156, USA, Tel.: 305-609-2110, Fax: 305-663-0844, e-mail: KAmber@med.miami.edu
Pemphigus vulgaris is a life-threatening autoimmune blistering disease caused by anti-desmoglein IgG autoantibodies that finally lead to acantholysis presenting clinically as progressive blistering. Whilst the production of pathogenic antibodies is key to the development of pemphigus vulgaris, many immunological steps are required prior to autoantibody induction. We review advances in the understanding of these immunologic processes with a focus on human leucocyte antigen polymorphisms and antigen recognition, epitope spreading, central and peripheral tolerance, T helper differentiation, induction of pro- and anti-inflammatory cytokines and T-cell regulation of B cells. Targeting autoaggressive T cells as regulators and stimulators of B-cell antibody production should allow for more specific therapeutic immune interventions, avoiding the global immunosuppression seen with many commonly used immunosuppressants in pemphigus vulgaris.
An introduction to the immune pathogenesis of pemphigus vulgaris
Pemphigus vulgaris (PV) is a life-threatening autoimmune blistering disease caused by anti-desmoglein (Dsg) IgG that leads to a loss of epidermal cell–cell adhesion, called acantholysis, leading to chronic, progressive blistering of the mucous membranes and skin. Whilst the production of pathogenic antibodies is key to the development of PV, many immunological steps are required prior to antibody production. This is evidenced by a series of hallmark animal experiments and clinical studies that have broadened our understanding of the pathogenesis of pemphigus .
Early studies of PV involved transferring anti-Dsg3 autoantibodies into newborn mice. The resultant PV phenotype identified IgG autoantibody involvement in the pathogenetically critical event of acantholysis [2, 3]. The murine model became more robust with the use of Dsg3−/− knockout mice, which also demonstrated the typical PV phenotype without the use of antibodies . These mice were then immunized with recombinant Dsg3, leading to the induction of Dsg3-specific autoreactive T cells and B cells. The Dsg3-primed splenocytes, consisting of T and B-lymphocytes, were then transferred into immunodeficient recombination activating gene 2 (Rag2)−/− mice, leading to the formation of clinically overt PV lesions of the mucous membranes . T cells and B cells were each isolated from Dsg3−/−, Dsg3+/− and Dsg3+/+ mice and injected into Rag2−/− mice in all possible combinations. Anti-Dsg3 IgG production and lesions were, however, only observed when both T cells and B cells from Dsg3−/− mice were transferred together . This indicates that both T cells and B cells with autoreactivity towards Dsg3 are necessary for the pathogenesis of PV.
Prevalence of human leucocyte antigens in PV
There is a genetic predisposition to PV [7-10], in which activation of autoreactive CD4+ T cells is first dependent on autoantigen presentation by specific human leucocyte antigen (HLA) class-II alleles . Numerous polymorphisms of HLA-II class alleles have been identified throughout several populations of patients with PV and are thought to occur due to changes in the charge of the active binding site on the HLA molecules for binding of autoantigenic peptides . A meta-analysis of 18 studies regarding HLA-DRB polymorphisms demonstrated that DRB1*04, DRB1*08 and DRB1*14 significantly increased susceptibility to PV, whilst DRB1*03, DRB1*07 and DRB1*15 significantly decreased the susceptibility to PV . HLA-B*50 and DQB1*02 have additionally been associated with a decreased susceptibility to PV in a large study of Turkish patients . Alterations of distinct HLA-DQ and HLA-DP alleles have been noted throughout many patient populations as well [7, 15]. Many HLA-II polymorphisms appear to be population dependent, however, and thus, their generalizability is limited [15, 16]. DRB1*04:02 and DQB1*05:03, two PV-associated HLA polymorphisms, were found to have similar avidity for binding of the PV autoantigen, Dsg3, but at different locations with DRB1*04:02 binding to peptides derived from the extracellular Dsg3 domain and DQB1*05:03 binding to COOH-terminal Dsg3 subdomains, suggesting multiple interactions of Dsg3 peptides with the restricting HLA class II alleles . The presence of the mentioned PV-associated HLA polymorphisms has also been linked to the presence of Dsg3-reactive T cells in healthy individuals, most of which were carriers of the aforementioned PV-associated HLA class II alleles [18, 19]. Dsg3-specific T cells in healthy individuals were, however, not directly associated with Dsg3-specific IgG autoantibodies. Only in a single study were low levels of serum anti-Dsg3 IgG autoantibodies detected in healthy relatives of patients with PV . This observation strongly suggests that additional, not yet fully understood, insults to immune regulation are critical for progression from subclinical autoimmunity against Dsg3 to clinically overt PV. These factors include drugs, tumors and infectious agents such as bacteria and viruses [21, 22].
Polymorphisms of HLA class I alleles also vary amongst patients with PV compared with healthy patients. HLA-A polymorphisms have been associated with PV in Jewish as well as in Chinese patients [13, 17]. Variations within HLA-A, HLA-B and HLA-C have additionally been noted amongst patients with PV, although these variations were similarly population dependent [23-25]. Polymorphisms of the transporter-associated antigen-processing genes (TAP1 and TAP2), transporters required for the loading of certain antigens to HLA class-1 molecules, were not found to be significant in patients with PV .
Non-classical HLA-I class molecules appear to be also involved in PV. HLA-G transcription, a non-classical HLA-I class molecule that does not demonstrate polymorphisms, appears to be altered in patients with PV, with a noted reduction in HLA-G2 and increase in HLA-G1 within epidermal cells of patients with PV [27, 28]. Likewise, HLA-E, another non-classical HLA that mediates natural killer cell (NK) as well as CD8+ cytotoxic T function, was found to vary in patients with PV, with a significant increase in HLA-E*01:03X in patients with PV .
The role of various HLA alleles for increased susceptibility to PV is mainly attributed to the electric charge of Dsg3-derived epitopes and their ability to bind to distinct HLA alleles upon recognition by autoreactive T cells. The best studied interaction of Dsg3-derived peptides has been described with HLA-DRB1*04:02 . A limited number of Dsg3 peptides with a positive charge at position four were found to avidly bind to DRB1*04:02 that contains an aspartic and glutamic acid residue at positions DRB70 and DRB71, respectively [12, 30]. Because HLA-DQB1*05:03 also has a positive charge at position 71, it may allow binding of similar or identical peptides of Dsg3. This contention is supported by previous findings that autoreactive T cells from HLA-DQB1*05:03+ PV patients recognized epitopes of Dsg3 which were also recognized by T cells from DRB1*04:02+ PV patients .
Epitope spreading in the pathogenesis of PV
Epitope spreading is defined as an autoimmune response that extends from the initial to additional epitopes within the primary target antigen or from the initial autoantigen to an unrelated secondary target antigen . In PV, the majority of patients show IgG reactivity to Dsg3 and develop predominantly mucosal blisters, as Dsg3 is the main desmosomal component of mucosal surfaces. Frequently, patients show additional IgG recognition of Dsg1 that is mainly expressed in the granular layer of stratified epithelia such as the skin, developing cutaneous and mucosal lesions . In paraneoplastic pemphigus, IgG autoantibodies do not only target Dsg3 and Dsg1 but also target components of the desmosomal plaque such as plakins and additional proteins that are not related to cell adhesion such as the protease inhibitor, alpha-2-macroglobulin-like 1 [22, 33].
Epitope spreading from recognition of Dsg to unrelated proteins has also been demonstrated such as recognition of desmocollins, which are also components of desmosomes, plakins and additional target autoantigens, which are not components of the desmosome. Specifically, cholinergic receptors such as pemphaxin and α9-AchR were identified as target antigens of IgG autoantibodies from pemphigus sera. However, the role of these receptors in PV has not been determined [34, 35].
Initially, only IgG autoantibodies that target NH2-terminal epitopes of the Dsg3 ectodomain [Dsg3[1-87]] were thought to be pathogenic in PV . This is consistent with the observation that IgG reactive with the NH2-terminal subdomains, Dsg3 EC1 and Dsg3 EC2 induced suprabasilar acantholysis, whilst IgG reactive against the COOH-terminal EC3-5 regions did not . A recent study has shown that IgG autoantibodies from PV sera bind not only to trans but also to cis interaction sites of adjacent Dsg3 molecules in desmosomes . Following studies, however, strongly suggest that IgG against COOH-terminal epitopes may be also pathogenic (Y. Exner, V. Spindler, D. Rafei, L. Dittmar, N. Kurrle, R. Tikkanen, M. Hertl, J. Waschke, R. Eming in preparation). Still, intra-molecular epitope spreading from IgG recognition of COOH-terminal epitopes of Dsg3, that is, Dsg3(87–566), to NH2-terminal epitopes, such as Dsg3[1-88], is presumably a critical step in the pemphigus pathogenesis.
Although the precise mechanism of epitope spreading in PV is unknown, it does appear to be contingent on T-cell activity. A single Dsg3-reactive T cell has been shown to induce polyclonal anti-Dsg3 IgG production in vivo . In turn, a small group of T-cell receptors can give rise to antibodies with a vast array of epitopes. Furthermore, the ability of B cells to act as antigen presenting cells suggests that a positive feedback loop exists between B cells and T cells, which can lead to rapid acceleration of epitope spreading with disease advancement [37, 38].
Central and peripheral T-cell tolerance in PV
Dsg3-specific T cells can be detected in the peripheral blood of patients with PV and healthy carriers of PV-associated HLA class II alleles . Moreover, healthy carriers of DRβ1*04:02 and DQβ1*05:03 exhibit CD4+ T-cell responses against the same epitopes of the Dsg3 ectodomain as patients with PV [18, 19]. Thus, the lack of a clinical disease phenotype in healthy carriers of PV-associated HLA class-II alleles clearly shows that loss of T-cell tolerance alone does not lead to PV.
T-cell tolerance is regulated centrally and peripherally. Central tolerance occurs during T-cell development in the thymus. Medullary thymic epithelial cells are present in the thymic stroma and make use of a transcription factor called autoimmune regulator (AIRE) that allows for the expression of an enormous collection of antigens representing almost all of the organs in the body. AIRE has been found to promote the expression of Dsg3 within the thymus . Generally, negative selection results in the deletion of potentially autoaggressive T cells with high affinity to self-antigens expressed on thymic epithelium . Accordingly, alteration of central tolerance in thymoma may lead to a loss of self-tolerance .
Peripheral tolerance involves the inactivation of autoreactive T cells that escape from the thymus. This is generally handled by CD4+CD25+ regulatory T cells (nTreg). These regulatory cells also develop in the thymus; however, they are unique in their expression of the transcription factor forkhead box P3 (FoxP3). It was shown that a naïve T cell that begins to express FoxP3 obtains attributes comparable to a regulatory T cell . Additionally, it was found that in vivo removal of FoxP3+ Treg cells results in the proliferation of self-reactive T cells leading to clinically overt autoimmunity . This suggests that autoreactive T cells and nTregs have an inverse relationship and that their functional differences appear to be based on the expression of FoxP3. In general, nTreg cells make up 10% of all peripheral CD4+ T cells and operate in a cell contact-dependent and non-antigen-specific manner as they are presumably already activated in vivo .
Forkhead box P3 was found to be constitutively expressed in Dsg3-specific type 1 T-regulatory (Tr1) cells that were initially thought to inhibit the activation of T effector cells by the secretion of IL-10 . These Dsg3-specific Tr1 cells were found at higher frequencies in healthy carriers of PV-associated HLA class II alleles than in patients with PV . Furthermore, antisense-driven inactivation of FoxP3 mRNA in these Tr1 cells led to loss of their immunosuppressive abilities, gain of proliferative response to Dsg3 and secretion of a cytokine profile similar to T-helper 2 (Th2) cells . Likewise, the suppression of FoxP3 in Dsg3-specific Tr1 cells was found to induce their production of interleukin 2 (IL-2), whilst IL-5, IL-10 and TGF-β secretion remained unchanged . Because the secretion of IL-10 and TGF-β by the Tr1 cells was not altered by Foxp3 antisense treatment, the suppressor function of the Dsg3-reactive Tr1 cells is presumably due to the consumption of exogenous IL-2 secreted by bystander T effector cells, decreasing the available pool of stimulatory IL-2 for the activation of T effector cells. These findings are in line with an independent study which showed that down-regulation of Foxp3 in Treg cells leads to Th2 differentiation .
Yokoyama et al.  demonstrated that Treg cells created in a Dsg3−/− mouse were capable of suppressing anti-Dsg3 IgG secretion when transferred into mice which actively produced autoantibodies. This again highlights the inverse relationship between Th2 and Treg subsets. Modulating of FoxP3 expression could thus potentially allow restoration of tolerance in PV.
In contrast to nTreg whose antigenic stimulus is not known, type 1 regulatory T (Tr1) cells are activated upon exposure to their nominal antigen leading to the secretion of the immunosuppressive cytokine IL-10 [45, 46]. Dsg3-specific Tr1 cells were detected in the majority of healthy carriers of the DRβ1*04:02 and DQβ1*05:03 alleles, yet they represented less than 20% of IL-10+, Dsg3-specific T cells in patients with PV . Indeed, a high ratio of Th2/Tr1 cell was found in patients with PV, whilst a low Th2/Tr1 cell ratio was found in healthy carriers of PV-associated HLA class-II alleles . Thus, deficiencies in Tr1 cells may be directly related to the loss of tolerance against Dsg3 in PV consistent with the detection of low peripheral Treg cells in patients with active PV . The loss of tolerance against Dsg3 in PV is presumably not only dependent on a deficiency of nTreg or Tr1 cells, as a subset of CD8+CD28− T cells also appear to have regulatory function in patients with newly remittent, but not inactive PV . CD8+CD28− cells in a small sample of newly remittent patients secreted an increased level of IFN-γ in response to Dsg3, leading to a Th1 and repressive response.
Pathogenic Th-cell responses in PV
Several T-cell subsets appear to be involved in the pathogenesis of PV. Elevated serum levels of IL-4 and IL-10 were detected in PV, suggesting Th2 involvement . IL-6, another stimulator of Th2 differentiation, has also been implicated in PV. IL-6 levels are elevated in patients with PV during active disease as well as in clinical remission [52, 53]. Following treatment, however, a significant reduction in serum IL-6 concentrations was noted . Similarly, elevated levels of IL-6 were also found in blister fluids  and correlated with disease severity [55, 56]. Serum levels of IL-6 were also elevated in glucocorticoid-resistant PV . Serum concentrations of IL-5, another stimulator of Th2 differentiation, were also elevated in active PV .
A decreased activation of Th1 cells may be the consequence of an altered function of NK cells that regulate Th1 proliferation. In PV, NK cells were found to drive differentiation to Th2 by impairing IL-12 signalling . As IL-12 leads to a decreased production of IL-10 which inhibits Th2 cells, impaired IL-12 signalling causes an increase in Th2 cells which are the dominant Th subset in PV.
Serum levels of the pro-inflammatory cytokine, TNF-α, are also elevated in PV and correlate with disease severity . Likewise, TNF-α serum levels are also elevated in glucocorticoid-resistant patients with PV , and distinct polymorphisms of TNF-α were found to be associated with PV . It is, however, unclear how these TNF-α polymorphisms contribute to the pathogenesis of PV. Altered TNF binding to target cells such as NK cells that regulate Th1 proliferation may contribute to this association. Macrophage-derived cytokines that are elevated in PV include IL-1 and IL-23 [60, 61]. Following treatment with intravenous immunoglobulins (IVIG), serum IL-1 levels decreased, whilst IL-1Ra, an endogenous antagonist to the IL-1 receptor, was elevated in the patients' sera. Macrophage migration inhibitory factor (MIF), a pro-inflammatory chemokine as well as inhibitor of glucocorticoid-related anti-inflammatory effects, was also found to be elevated in patients with PV . MIF can additionally drive antibody production, and thus, an MIF inhibitor has been suggested as a steroid sparing therapeutic alternative.
Both Dsg3-reactive Th2 and Th1 cells appear necessary for PV to occur . Dsg3-reactive Th2 and Th1 cells as well as their corresponding cytokines have been identified in patients with PV and are thought to require autoantigen presentation in association with PV-associated HLA class-II alleles [30, 64, 65]. These Th cells were capable of recognizing multiple portions of the extracellular domain of Dsg3 in context with PV-associated HLA class-II alleles. Dsg3-reactive Th2 lymphocytes were present at different disease stages , that is, acute (i.e. clinically active), chronic (residual activity on immunosuppressive therapy) and remittent (clinically healed) PV . Moreover, Th2 cytokine levels were elevated in perilesional skin .
Dsg3-reactive Th1 cells were primarily found in chronic active PV and in healthy individuals with PV-associated HLA class-II alleles. Healthy patients, however, lack Dsg3-reactive Th2 lymphocytes. As during acute-phase PV, both autoreactive Th1 and Th2 cells appear to circulate with similar frequencies, both Th subtypes may share responsibility at different phases of the disease process .
CD8+ T cells that secrete IL-2 and IFN-γ upon in vitro stimulation with Dsg3 have also been identified . Likewise, Dsg-3-sensitive CD4+ Th1-cells were found to directly induce interface dermatitis as well as PV in a mouse model, suggesting a direct role of T-cell involvement in PV . Although there continue to be new advances regarding T cells in PV, much of the focus of T-cell involvement in PV has focused on its regulation of B-lymphocytes.
T-cell regulation of B-cell function
As several studies have demonstrated that depletion of CD4+T-cells reduces the amount of anti-Dsg3 antibodies produced in patients with PV, further insight into the interaction of B lymphocytes with T lymphocytes may help identify novel therapeutic targets [70, 71]. Following B-cell depletion upon treatment with rituximab, there is marked and prolonged reduction in serum IgG autoantibodies against Dsg3 and Dsg1 [72-74]. Noteworthy, serum levels of IgG antibodies against antigens such as tetanus toxoid and Epstein–Barr virus were not reduced, suggesting the presence of long-lived plasma cells specific for these antigens which were not affected by rituximab . The potential B-cell regulation of Dsg3-specific, autoreactive Th cells was supported by the finding that the frequency of Dsg3-specific autoreactive T cells was markedly reduced with the depletion of peripheral B cells, which may also act as antigen presenting cells . In contrast, the frequencies of Th cells specific for recall antigens, such as tetanus toxoid, were not reduced by rituximab-induced B-cell depletion .
Interaction of activated T cells and B cells requires binding of CD40L (CD154) on activated T cells with CD40 on B cells. This interaction promotes the expression of activation-induced cytidine deaminase in B cells, which promotes somatic hypermutation and Ig-isotype class switching . It was thus found that an anti-CD154 monoclonal antibody (mAb) was able to block the CD40/CD154 interaction resulting in a suppression of anti-Dsg3 IgG production . Whilst this appears to have therapeutic value, an equally interesting result was obtained when splenocytes from anti-CD154 mAb-treated mice and splenocytes from Dsg3−/− mice were co-transferred in to syngeneic Dsg3+/+ animals. The recipient mice had dramatically depressed anti-Dsg3 IgG production . This is surprising given that sudden injection of Dsg3−/− splenocytes into a Dsg3 environment would be expected to cause a large immune response. Recent work has shown that anti-CD154 treatment of allograft transplant patients actually up-regulates Treg cells [77, 78]. The splenocytes from the anti-CD154 mAb-treated mice therefore contained high enough levels of Treg to suppress the immune response against Dsg3 in the recipient mice. These findings strongly suggest that anti-CD154 mAb treatment may restore tolerance against autoantigens via the induction of Treg cells.
There is good evidence that Dsg3-specific autoreactive T cells activate B cells by the secretion of IL-4 . Twenty Dsg3-reactive T-cell lines generated by immunization of Rag2−/− deficient mice were co-cultured with primed B cells derived from Dsg3-immunized Dsg3−/− mice. Seven of the 20 T-cell lines induced anti-Dsg3 IgG antibody production, leading to mucocutaneous erosions in the Dsg3+/+ recipient mice. These pathogenic T-cell lines expressed IL-4 and IL-10 at higher levels than the non-pathogenic T-cell lines. Additional analysis showed that IL-4, but not IL-10, promoted IgG antibody production in vitro and that adenoviral expression of soluble IL-4Rα, an antagonist of the IL-4 receptor, suppressed IgG antibody production. These results suggest that IL-4 could be a viable therapeutic target . Support for a critical role of autoreactive Th2 cells in the pathogenesis of PV is provided by the recent identification of Dsg3-reactive IgE autoantibodies in active disease stages, but not in clinical remission . An illustration of the potential interaction of autoaggressive T cells and B cells in PV is given in Fig. 1, as well as an overview of possible therapeutic targets.
Whilst the immunosuppressive effects of Treg cells on Th1 and Th2 cells are well documented, it was unclear how Treg cells suppress B cells in PV. Ota et al.  demonstrated that pathogenic self-antigens induce Treg-cell proliferation which then acts directly on B cells through an apoptotic mechanism. Further studies demonstrated the ability of Dsg3-reactive Th2 cells to cause anti-Dsg3 IgG1 secretion by B lymphocytes in an unexposed mouse . Interestingly, IgG4 is the most common IgG autoantibody subtype in pemphigus, whilst IgG1 autoantibodies are more commonly seen in chronic PV [84, 85].
Th17 is a recently identified helper T-cell subset which is generated upon IL-23 stimulation of naïve T cells and is characterized by the secretion of IL-17, a pro-inflammatory cytokine. Th17 cells were found in skin lesions of patients with PV, but their presence correlated neither with IgG autoantibody serum levels nor with the clinical severity of PV . Likewise, Th17 does not appear to regulate B-cell function such as Th2 or Treg cells, making a pathogenic role in PV unlikely. This is in contrast to bullous pemphigoid, which demonstrates Th17 involvement .
Therapeutic targets in the PV pathogenesis
Global immunosuppressants such as corticosteroids, cyclophosphamide, azathioprine and methotrexate, whilst often effective, have significant side effects and lead to generalized immunosuppression . Thus, the development of therapeutic strategies that target critical events in the PV pathogenesis is highly desirable.
Rituximab, an anti-CD20 mAb, has been used with great success in patients with PV [71, 89-91]. Rituximab depletes peripheral CD20+ pre-B-cells and mature B cells and is currently mainly used as a second-line treatment for PV. Long-lasting remission following rituximab therapy was associated with the disappearance of autoreactive B cells . Noteworthy, a study demonstrated a significant decrease in T-cell function in patients with PV who received rituximab, highlighting the importance of the proper interplay of autoreactive T cells with B cells which may present a very specific therapeutic target in PV .
TNF-α inhibitors have shown variable success in the treatment for PV based on small treatment groups. Etanercept and infliximab were successful in the treatment of a handful of PV patients [93-95]. On the contrary, a case report noted the spontaneous development of PV in a psoriatic patient upon treatment with infliximab . A therapeutic study in a larger cohort of PV patients with sulfasalazine and pentoxifylline, non-biologic TNF-α inhibitors provided evidence for clinical improvement upon TNF-α blockade which was associated with significantly decreased serum levels of TNF-α . As the serum concentrations of TNF-α correlate with the clinical severity of PV, further research into the therapeutic potential of TNF-α inhibitors in PV may be warranted.
Removal of circulating IgG autoantibodies against Dsg3, such as through the use of plasmapheresis, IVIG, and recently, by the use of immunoadsorption, has major potential in rapidly reducing the clinical activity of PV [98-100]. Agents that directly target T-helper differentiation currently have been tested for the treatment for PV and other autoimmune conditions. Daclizumab, an anti-CD25 antibody, was reported to successfully treat a patient with PV refractory to many other conditions . Daclizumab functions at the level of IL-2 binding, affecting T-cell function in a similar manner as the calcineurin inhibitors, cyclosporine, tacrolimus and sirolimus. On the contrary, a case was reported of a patient who had a PV relapse when receiving IL-2 therapy . The ultimate goal certainly remains to develop therapeutic strategies that specifically target key players in the pathogenesis of PV such as auto-aggressive T cells, B cells and plasma cells.
KT Amber designed the concept of the article, wrote the article and approved of the final article. P Staropoli wrote the article and approved of the final article. MI Shiman and GW Elgart contributed to critical revision and approval of the final article. M Hertl wrote the article and critically revised and approved of the final article.
Conflict of interests
Part of the work of M.H. was funded by grants from the German Research Council, Bonn, Germany (DFG; He 1602/7-2). M.H. has received honoraria from Fresenius Medical Care, Germany; Biotest, Germany; and TEVA, Germany. The authors have declared no conflicting interests.