Major T1D autoantigens include (pro)insulin, the 65-kDa isoform of glutamic acid decarboxylase (GAD65), the islet tyrosine phosphatase IA-2, and the islet-specific glucose-6-phosphatase catalytic subunit-related protein (IGRP). These pancreatic β-target autoantigens have been identified in both humans and nonobese diabetic (NOD) mice [1]. The generation of autoreactivity to islet β cells is the consequence of multiple various genetic defects that have an impact on fundamental immunological processes such as central and peripheral tolerance. In general, potentially autoreactive lymphocytes must encounter their cognate autoantigen during differentiation in order for the host to develop tolerance to self-antigens. This encounter may occur within the thymus or upon release into the periphery [2, 3]. Central self-tolerance occurs during T-cell differentiation in the thymus and involves deletion (negative selection) of self-reactive T cells during T-cell development in the thymus [4]. This process occurs to eliminate T cells that recognize ubiquitous or blood-borne self-antigens. T cells which recognize tissue-specific antigens (TSAs), however, were long-thought to be tolerized in the periphery [3]. Until recently, it was widely assumed that such TSAs are excluded from the thymus, precluding central self-tolerance. To explain tolerance to TSAs, numerous peripheral tolerance mechanisms have been suggested, including ignorance and immunoregulation. At the same time it has also been argued that TSAs may reach the thymus in significant quantities and thereby induce negative selection in situ, especially of high-affinity T cells. This idea became more plausible when others and we found that many TSAs are expressed by rare medullary thymic epithelial cells (mTEC) [5–7]. Findings such as these challenged historical assumptions about tolerance to TSAs and intense investigations, in both mice and humans, have elucidated the identity of TSA-expressing cells in the thymus, as well as their role in the process of central self-tolerance. Others and we showed that genetically determined low levels of TSAs (i.e., insulin) expression in the thymus is associated with the generation of high numbers of autoreactive T cells in the periphery and predisposition to autoimmune disease [5–7]. The ectopic expression of many tissue-restricted antigens in the thymus is controlled by the autoimmune regulator (AIRE). AIRE is a 54.5-kDa protein with a nuclear localization and several potential DNA-binding and protein interaction domains [8]. AIRE ^−/− mice exhibit a defined profile of autoimmune diseases including T1D. It is notable that all these diseases arise due to a lack of AIRE expression in stromal cells of the thymus. Furthermore, AIRE-deficient mTEC showed a specific reduction in promiscuous transcription of genes encoding peripheral antigens, demonstrating the importance of thymus-dependent tolerance in controlling autoimmunity [9].

The thymus is the central lymphoid organ responsible for the maturation and differentiation of bone-marrow-derived thymocytes. The random generation of the T-cell repertoire, including autoreactive T cells, is regulated in the thymus by mechanisms of central self-tolerance. Anatomically, the thymus is divided into subcapsular, cortical, and medullary compartments. The stromal cells include a variety of bone-marrow-derived professional antigen-presenting cells (dendritic cells [DC], macrophages, and B cells) and endoderm-derived cortical (c) TEC and mTEC [28]. A striking morphologic feature of the medulla is the presence of Hassall's corpuscles, which consist of concentric whorls of stratified keratinizing epithelium and share antigenic properties with ectodermic epithelium [29]. The thymus constitutes the central arm of immunological self-tolerance by two essential mechanisms that are intimately associated and paradoxically mediated by the same thymic self-antigens: (a) negative selection of self-reactive T cells issued from the random recombination of TCR genes (“recessive” self-tolerance) and (b) generation of self-antigen-specific nTreg that are able to inactivate in periphery self-reactive T cells having escaped intrathymic negative selection (“dominant” self-tolerance) [30, 31].

Several groups and ourselves have demonstrated that TEC from different species constitute a site for the promiscuous transcription of a great number of genes encoding tissue-restricted antigens or belonging to neuroendocrine families, such as the neurohypophysial family, tachykinins, neurotensins, somatostatins, atrial natriuretic peptides, and the insulin family. This demonstration has radically changed our common understanding of the pathogenesis of organ-specific autoimmune endocrine diseases such as T1D. From the investigation of intrathymic expression of neuroendocrine-related self-peptide precursor genes, a series of properties can be derived that define the nature of the “neuroendocrine self”. First, thymic neuroendocrine self-antigens usually correspond to peptide sequences that have been highly conserved throughout the evolution of their related family. Second, a hierarchy characterizes their expression pattern in the thymus. In the neurohypophysial family, oxytocin (OT) is the dominant peptide synthesized by TEC from different species. The binding of OT to its cognate receptor expressed by pre-T cells induces a very rapid phosphorylation of focal adhesion-related kinases. This event could play a major role in promoting establishment of synapses between immature T lymphocytes and TEC, as well as with macrophages and DC. All the genes of the insulin family are expressed in the thymus according to a precise hierarchy during fetal life: IGF2 (cTEC and mTEC) >IGF1>INS (a few subsets of mTEC). This hierarchical pattern is meaningful because the strength of self-tolerance to a protein is proportional to its intrathymic concentration [32]. Third, neuroendocrine precursors are not processed according to the classic model of neurosecretion, but they undergo an antigenic processing for presentation by—or in association with—MHC proteins [33]. Fourth, most of neuroendocrine self-antigens are transcribed in the thymic epithelium under the control of the autoimmune regulator gene AIRE (see below). Fifth, intrathymic OT transcription precedes OT and vasopressin (VP) expression in hypothalamic magnocellular neurons. Finally, epigenetic regulation of intrathymic gene expression is strongly suggested by the loss of IGF2 parental imprinting in human mTEC [34, 35].

A number of abnormalities of thymic morphology and cytoarchitecture have been described in T1D. Central tolerance and apoptosis of self-reactive T cells are defective in the thymus of NOD mouse [39, 40]. Transcription of insulin-related genes (Ins, Igf1, and Igf2) has been analyzed in the thymus of diabetes-resistant (BBDR) and diabetes-prone (BBDP) rats, another model of T1D. Ins and Igf1 transcripts were detected in all thymi from BBDP and BBDR rats. Igf2 transcripts were also present in the thymus from all BBDR rats, but were not detected in the thymus from more than 80% of BBDP rats, in close concordance with the incidence (86%) of autoimmune diabetes in those rats. This defect in Igf2 transcription in BBDP thymus could also explain both their lymphopenia (including CD8+ T cells and RT6+ Treg) and the absence of central self-tolerance to insulin-secreting islet β cells [41, 42]. As already mentioned, we have shown that susceptibility to autoimmune diabetes is correlated with the level of Ins2 transcription in the mouse thymus [6]. Breeding of Ins2 ^−/− mice onto the NOD background markedly accelerated insulitis and onset of diabetes [43]. In contrast, insulitis and diabetes were considerably reduced in Ins1 ^−/− congenic NOD mice [44]. These observations are explained by the dominance of Ins2 encoding proinsulin in the murine thymus, while Ins1 encodes proinsulin in islet β cells. In the human species, INS transcripts were measured at lower levels in the fetal thymus with short class I VNTR (variable number of tandem repeats) alleles, a genetic trait of T1D susceptibility as discussed above [45, 46]. The contribution of thymic insulin in mediating central self-tolerance to islet β cells was definitively demonstrated by the rapid onset of autoimmune diabetes following thymus-specific deletion of Ins1 and Ins2 through an elegant transgenic construction in mice [47].

The identification of AIRE led to further demonstration that a thymus dysfunction plays a crucial role in the pathogenesis of organ-specific autoimmune diseases [48, 49]. Loss-of-function AIRE single mutations are responsible for a very rare autosomal recessive disease named autoimmune polyendocrinopathy, candidiasis, and ectodermal dystrophy (APECED), or autoimmune polyendocrine syndrome type 1 (APS-1). This syndrome develops in early childhood and is characterized by multiorgan autoimmunity and insufficiency of several endocrine glands such as parathyroids, adrenal cortex, and gonads. AIRE expression is maximal in the thymus, mainly in mTEC, but is absent in TEC of NOD mice [50]. Depending on their genetic background, AIRE ^−/− mice exhibit several signs of peripheral autoimmunity, which are associated with a significant decrease in thymic transcription of neuroendocrine genes (including Ot, Npy, Igf2, and Ins2), as well as other TSAs [9, 51, 52]. However, as shown for GAD67, AIRE does not control the intrathymic expression of all TSAs. By different aspects, AIRE-regulated transcription in the thymus differs from expression of these antigens in eutopic tissues. For example, loss of Igf2 imprinting with biallelic transcription has been observed in mTEC [53]. The same study has shown that many TSAs in mTEC are clustered in their chromosomal location, including AIRE-dependent and AIRE-independent gene targets [53]. Extrathymic AIRE expression has been evidenced in secondary lymphoid organs where AIRE also controls the expression of TSA genes that are different from those regulated by AIRE in mTEC [54]. The precise molecular mechanisms by which AIRE controls transcription of TSA are not completely elucidated [55, 56]. However, it more and more appears that AIRE uses one of its two zinc finger plant homeodomains for binding to nonmethylated histone H3K4 and activating gene expression, which establishes a relationship between chromatin regulation and tissue-specific central tolerance [57, 58]. Finally, there is mounting evidence that AIRE is closely implicated in mTEC differentiation (reviewed in [59]).

Multiple reports have implicated Treg in T1D prevention. Treg depletion or interference with B7/CD28 pathway in NOD mice has been shown to accelerate T1D onset [70, 71]. In aged NOD mice, autoimmune diabetes resistance has been correlated with the expansion of CD25+ CD4+ T cells with regulatory activity within inflamed pancreatic lymph nodes [72]. Recently, Chen et al. reported that Foxp3-deficient NOD mice, which are deficient in Treg, display an increased incidence and earlier onset of T1D compared with normal NOD mice, strongly implying a role for Treg in the control of T1D pathogenesis [73]. Obviously, alterations in the frequency and/or function of Treg are associated with various autoimmune diseases, including T1D [74, 75]. The quantification of naturally occurring Treg is hampered by a lack of unique surface markers. Perhaps not surprisingly, there is some controversy in the literature regarding the frequency of Treg in NOD mice, as determined by the identification of CD4+CD25 high cells in the spleen and lymphoid organs. More importantly, however, it appears that NOD mice harbor Treg that prevent disease development at an early age, but lose their functional capacity later on in life, thereby allowing pathogenic effector T cells to attack pancreatic islets [76, 77]. Similarly, the numbers of Foxp3+ Treg in the peripheral blood of T1D patients do not appear to be significantly different from healthy control subjects but there is evidence for functional defects in their suppressive capacity [78–80]. Therefore, there has been much interest in the potential use of adoptive transfer of in vitro expanded islet-specific Treg as a way of suppressing autoimmunity in T1D patients, perhaps preserving islet-cell function in newly diagnosed T1D.

Other genetic linkage and association studies have identified a second locus for T1D susceptibility that corresponds to a high polymorphic minisatellite constituted by a variable number of tandem repeats (VNTR) [87, 88]. This VNTR is embedded on chromosome 11p15 and controls the transcription of the insulin (INS) and insulin-like growth factor 2 (IGF2) genes downstream. Short VNTR class I alleles contain 20–63 repeats of 14-15 base pairs, while intermediate class II and long class III alleles include 64–139 and 140–210 repeats, respectively. VNTR class I alleles are associated with TID susceptibility, whereas class III alleles confer protection.

The CTLA4 gene region on chromosome 2q33 is also associated with susceptibility to T1D [89]. The signaling between B7, expressed by professional antigen-presenting cells such as DC and cytotoxic T-lymphocyte-associated protein 4 (CTLA-4), expressed by T cells, plays a pivotal role in peripheral T-cell tolerance. CTLA-4 is expressed neither by thymocytes nor resting T cells, but it is detectable after antigen-mediated T-cell activation and downregulates responses of activated T cells. Ctla4 deletion in mice results in an extremely severe lymphoproliferative and an autoimmune phenotype with lethal multiorgan tissue destruction [90].

Another mutation in a non-HLA gene conferring significant susceptibility to T1D is a variant of the lymphoid tyrosine phosphatase Lyp gene (PTPN22), a suppressor of T-cell activation [91]. Lyp normally interacts with a C-terminal Src kinase (Csk) complex to dephosphorylate positive regulatory tyrosines and downregulate signaling from the TCR pathway. The minor allele derived from the single nucleotide polymorphism (SNP) differs in a single but crucial amino acid residue (R620W) involved in interaction of Lyp with Csk. Interestingly, the same variant R620W also increases risk to other common autoimmune diseases, such as rheumatoid arthritis, Graves' disease, and systemic lupus erythematosus [92]. However, the variant PTPN22 620W is a gain-of-function mutant, since it is associated with a higher catalytic activity of the encoded Lyp, a marked decrease of T-cell response to antigen stimulation, CD25 expression and IL-10 secretion from TCR stimulation, and an increase in peripheral memory CD4+ T cells [93, 94]. The role of this mutation in the pathogenesis of T1D and other autoimmune diseases remains to be further elucidated.

Different studies, including a genome-wide association analysis, have identified association of T1D with noncoding SNPs on the chromosome 10p15 region containing CD25, which encodes the high-affinity α chain of the IL2R complex [95]. Further mapping of the association between the IL2RA locus and T1D supported a role of IL2Rα in the pathogenesis of the disease, most possibly through modulation of Treg activity [96].

An association has also been found between T1D and a polymorphism of the IGF2 receptor gene (IGF2R), which seems to be subject to parental imprinting since only maternal alleles at this polymorphism are associated with the disease [97]. Human T1D differs from other common autoimmune disorders, which preferentially affect females (like autoimmune diabetes in the NOD mouse). Evidence was also recently provided for an association between T1D and polymorphisms in CYP27B1, which encodes 1α-hydroxylase, the enzyme that transforms 25(OH) vitamin D into bioactive 1,25(OH)₂ vitamin D3 [98].