Beta cell antigens in type 1 diabetes: triggers in pathogenesis and therapeutic targets.

Recognition of pancreatic beta cell antigens by autoreactive T lymphocytes plays a central role in the pathogenesis of insulin-dependent type 1 diabetes. Recent results suggest that non-conventional antigenic epitope processing and presentation may contribute to triggering and maintaining autoreactive responses. Moreover, promising results raise hope that autoantigens may become safe and specific therapeutics for type 1 diabetes in the future.

Introduction and context

Ample evidence documents that both CD4+ helper and CD8+ killer T lymphocytes are critical in the pathogenesis of autoimmune type 1 diabetes (T1D) [1]. Although the initial events triggering autoreactive responses remain unclear, specific autoantigen presentation by strongly disease-associated major histocompatibility complex (MHC) class II molecules and, according to recent results, weakly disease-associated MHC class I molecules is thought to favor priming and expansion of pathogenic T cells [2]. Among the steadily expanding list of autoantigens, many are shared between patients and non-obese diabetic (NOD) mice, the murine model of the disease. In patients, autoantibodies recognizing four antigens - insulin, glutamic acid decarboxylase (GAD), insulinoma antigen-2, and the recently discovered ZnT8 - have strong disease-predictive value and are measured in clinical routine diagnostics [3]. Other autoantigens such as islet-specific glucose-6-phosphate catalytic subunit-related protein (IGRP) and the very recently identified chromogranin A [4] have been studied mainly or exclusively in mice. The critical issue of whether there is one autoantigen whose recognition by T cells is an obligatory triggering event at the onset of the autoimmune response has been much discussed. In the NOD model, recent evidence suggests that insulin is one such primary triggering antigen [5]. Mice tolerized to insulin [6] or mice expressing exclusively an insulin molecule lacking the immunodominant T-cell epitopes recognized by CD4+ and CD8+ T cells [7] develop neither diabetes nor insulitis, whereas similar tolerization to other autoantigens has no such effect [8,9]. It remains unclear whether insulin or any other antigen plays a similar role in humans.

The pivotal role of T cells in T1D has prompted sustained and ongoing efforts to identify autoantigenic epitopes, which can be used to develop T-cell assays potentially useful for disease prediction or monitoring of immunotherapeutic intervention (or both). We [10] and others [11] have identified a large number of mainly HLA-A2 restricted autoantigenic CD8+ T-cell epitopes. Using these epitopes, we could develop an enzyme-linked immunosorbent spot (ELISpot) T-cell assay that detected interferon-gamma-secreting T cells in the vast majority of patients at disease onset, whereas recognition seemed to vanish during longer-standing disease [12,13]. Recent studies have identified a number of epitopes derived from the proinsulin signal peptide, whose level of presentation may be proportional to the secretory activity of beta cells [14,15]. However, a recent blinded study evaluating CD4+ T-cell assays found that the sensitivity and specificity of ELISpot and proliferation assays using purified antigens or epitopes did not exceed 61% and 69%, respectively, a performance greatly inferior to autoantibody assays [16]. Moreover, despite much effort spent on T-cell assays, studies showing that T-cell assays are actually useful for predicting or monitoring intervention have not been published as of yet.

Major recent advances

What renders an islet cell protein autoantigenic and potentially pathogenic?

Among the vast number of proteins expressed by beta cells, only a few are prominent targets of the cellular autoimmune response, but the reasons for this selectivity have remained mysterious. One fairly constant but unexplained feature of prominent autoantigens is their association with secretory granules. Recent findings provide intriguing new leads with respect to this issue. The BDC2.5 CD4+ T-cell clone and T-cell receptor-transgenic mouse line, used in numerous studies, recognize the most recently described autoantigen, chromogranin A, another secretory granule protein [4]. Interestingly, the peptide recognized by these cells interacts with only the C-terminal half of the I-Ag7 peptide-binding site together with a flanking sequence. This suggests a non-conventional interaction with the T cell in a manner reminiscent of immunodominant CD4+ T cells involved in a mouse model of multiple sclerosis [17]. Non-conventional peptide-MHC class II complexes were also shown to play an important role in the recognition of the critical insulin epitope B9-23 by pathogenic T cells. Mohan and colleagues [18] found that many islet-infiltrating T cells recognizing this epitope are of ‘type B’; that is, these cells recognize complexes formed by incubation of antigen-presenting cells with the cognate peptide but not those formed by incubation with the source protein insulin, the latter being recognized only by ‘type A’ T cells. The molecular mechanism for formation of type B complexes is unclear, but Mohan et al. [18] suggest that extremely high antigen concentration, a hallmark of insulin in islets, together with the absence of type B complexes during thymic selection, plays a crucial role in the emergence and stimulation of type B insulin-specific T cells. A specific antigen-processing event could account for the fact that type B complexes may be produced only in islets. The notion that a specific potentially novel antigen-processing pathway may play a role in conferring dominant autoantigen status to insulin is supported by another study. Brosi et al. [19] showed that the priming of diabetogenic insulin-specific CD8+ T cells upon DNA vaccination in mice that express the co-stimulatory molecule CD80 in islets depends on translocation of the vaccine-encoded antigen into the endoplasmic reticulum. Although the underlying mechanism is entirely unclear, this together with the other cited studies raises the possibility that specific antigen-processing and presentation pathways and events play a role in the autoantigenicity of insulin and possibly other autoantigens.

Autoantigens as therapeutics

Although immunomodulation through CD3-specific antibodies remains the most successful biological therapeutic for T1D [20,21], the biological risk necessarily associated with similar strategies has prompted efforts to develop therapeutic strategies based on autoantigens, which are expected to be more specific and safer. In the NOD model, it is well established that treatment of prediabetic mice with various autoantigens prevents disease [22]. After a number of unsuccessful trials using insulin in humans, a recent study demonstrated for the first time a beneficial effect of autoantigen administration to patients at disease onset [23]. Injection of 20 µg of GAD in alum resulted in a slower decline of glucose-stimulated C-peptide levels in patients treated no later than 6 months after diagnosis. In these patients (unlike in anti-CD3-treated patients [21]), no reduction in insulin needs was observed, but this result encourages further development of autoantigen-based approaches in humans.

A promising novel strategy for the reversal of autoimmunity was very recently described in an elegant study by Tsai and colleagues [24]. These authors found that NOD mice harbor antigen-experienced CD8+ T cells recognizing the highly immunodominant autoantigen IGRP, which can be expanded by immunization with nanoparticles coated with specific peptide-MHC complexes. The expanded cells act as suppressors that prevent disease in young mice and, remarkably, reverse disease in overtly diabetic mice, a feat previously unmatched for antigen-based therapies. Nanoparticle stimulation of relatively minor T-cell populations also induced the generation of protective suppressor cells, suggesting that the application of the strategy does not require prior identification of immunodominant T-cell populations in every patient treated.

Future directions

As so often is the case, these intriguing studies leave more questions open than they answer. With respect to the parameters determining autoantigenicity, the general validity of the observed phenomena needs to be studied. Is partial filling of the MHC class II peptide-binding site a common feature of autoantigenic CD4+ T-cell epitopes? Can ‘type B’ T cells with specificity for autoantigens other than insulin be found? Given that transferred pre-processed epitopes are considered highly unstable and prone to degradation (at least in the context of MHC class I cross-presentation), how does insulin peptide B9-23 produced in beta cells ‘survive’ to be presented by MHC class II molecules of professional antigen-presenting cells? Why does proinsulin targeting to the endoplasmic reticulum increase the priming of pathogenic insulin-specific CD8+ T cells?

With regard to therapeutic approaches, the efficacy must be ameliorated greatly if true therapeutic benefits are to be obtained in patients. Combination with low-dose immunomodulatory treatments and/or earlier treatment of patients with relatively high C-peptide levels could be roads to follow. T-cell assays with better performance or novel T-cell assays likely will be required before clinical use can be envisaged. Finally, T-cell assays should be used to monitor patients at T1D risk and/or patients undergoing immunotherapy to prove the clinical interest of these assays.