Insulin itself is a prototypical TRA, its synthesis being virtually restricted to pancreatic β cells. Thymic insulin production is critical for establishing self-tolerance to β cells; simply abolishing insulin expression in the thymus leads to the rapid onset of autoimmune diabetes even in mice lacking a diabetogenic genetic background . In humans, thymic insulin expression is modulated by allelic variation and epigenetic effects at the insulin gene locus; this effect is largely mediated by a polymorphic variable nucleotide tandem repeat (VNTR) sequence. In the thymus, protective VNTR alleles are associated with 2.5- to 3-fold higher insulin gene transcription levels compared to predisposing alleles [27,28]; polymorphisms of the AIRE and MafA transcription factors also modulate insulin gene transcription in the thymus [29,30]. As with other self-molecules, thymic insulin levels impact negative selection of autoreactive T cells and/or positive selection of regulatory T cells; lower levels typically lead to less efficient selection processes, promoting the maturation of a T-cell repertoire that is enriched in insulin autoreactive T cells .
Immunologic tolerance to other islet autoantigens is likely shaped by thymic expression as well: splice variants of the IA-2 and IGRP islet autoantigens have expression patterns that differ in human pancreas and thymus, suggesting that self-tolerance to antigenic epitopes expressed only in the pancreas may not be fully achievable [32,33]. Thus, allelic variation, alternative splicing, and epigenetic regulation of gene expression can affect quantity and quality of TRA expression in the thymus, in turn impacting thymic selection processes and favoring the maturation of a T-cell repertoire that is prone to autoimmune responses to a variety of islet autoantigens.
Escape from negative selection may also be favored, albeit not in an antigen-specific fashion, by polymorphisms of the PTPN22 gene. This encodes for Lyp, an intracellular, lymphocyte-specific tyrosine phosphatase, which functions as a negative regulator of TCR signaling. PTPN22 predisposition may derive from more potent suppression of TCR signaling during thymic development, resulting in impaired negative selection . Furthermore, recent investigations of patients’ autoreactive CD4 T cells highlight abnormal features of the immunologic synapses that may also favor escape from negative selection and promote effector responses in the target organ, where larger amounts of antigen are present .
Overall, there is strong evidence that the earliest disease mechanisms and primary genetic predisposition effects are active in early life, through modulation of thymic selection processes, with the stronger effects likely being specific for selected islet cell molecules, primarily under the genetic control of HLA and insulin genes. Therefore, true primary prevention strategies should aim at improving the efficiency of thymic selection processes and should be attempted very early, perhaps in utero. For example, drugs are being sought to enhance thymic insulin expression , which could be given to genetically at-risk children if safe and effective.
Functional properties of HLA class II and class I molecules presenting diabetogenic peptides. As noted, HLA class I and class II molecules mediate TRA presentation to developing lymphocytes in the thymus. In patients with T1DM, these will commonly be the predisposing variants described earlier. Molecular studies reveal that certain amino acid residues impact the peptide binding properties of these molecules and in turn presentation of islet antigen peptides to autoreactive T cells; for example, DQB1*0302 carries a serine at position 57, where most non predisposing HLA-DQ molecules carry an aspartic acid residue . There are remarkable sequence (including position 57) and structural similarities between the human HLA-DQ8 molecule and its homologous I-Ag7 in the MHC (major histocompatibility complex) of the NOD (nonobese diabetic) mouse, a model of autoimmune diabetes. There are also similarities in the binding of insulin and other peptides to these two molecules . Recent landmark studies in the NOD mouse show that I-Ag7 binds a diabetes-associated insulin peptide (B9-23) in multiple registers: in the thymus, a register binding is used that results in poor negative selection of these autoreactive T cells , while presentation to autoreactive T cells in the periphery may use a different binding register that will trigger autoimmune responses . Of note, insulin is the only islet autoantigen shown to be critical for diabetes development in the NOD mouse ; therefore, these studies reveal that the molecular basis of a key disease mechanism is explained by the binding features of the predisposing MHC molecule. Likewise, in humans, the molecular interactions of a pre-proinsulin peptide with the HLA-A2 molecule are weak, resulting in low TCR-peptide-HLA binding affinities. These studies illustrate how the molecular features of predisposing HLA molecules can result in poor presentation of a pre-proinsulin peptide and in enhanced probability that autoreactive CD8 T cells may escape thymic selection and later become autoreactive . The discovery of target antigens and the definition of key molecular MHC-peptide interactions are critical for developing antigen-specific therapies to selectively delete or regulate autoreactive T cells .
Genetic predisposition also impairs peripheral tolerance to islet autoantigens. An immunologically relevant expression of insulin and other peripheral antigens is not limited to the thymus, but it is now well described in peripheral lymphoid tissues as well. Human DCs express insulin and several other TRAs [23,42]; studies in mice demonstrate that insulin expression by
AIRE-expressing DCs maintains peripheral tolerance to this critical diabetes autoantigen . Several other cell types have been linked to TRA expression in the periphery [24,44,45]; for example, lymph node stromal cells express many TRAs under AIRE transcriptional control and mediate deletion of autoreactive T cells specific for a model autoantigen in transgenic mice; however, these cells do not express insulin . Thus, the immune system has mechanisms ensuring redundancy and complementarity of peripheral TRA expression, to help maintain peripheral tolerance after the involution of the thymus. It is plausible that the same predisposing genes (HLA, insulin, PTPN22), alternative splicing, and epigenetic mechanisms that impact thymic selection may also impart similar influences on TRA expression in the periphery. In the NOD mouse, insulin gene expression in the pancreatic lymph node (PLN) is lost in relation to proinflammatory changes developing as mice age and progress towards diabetes . These studies link reduced insulin expression to another transcription factor, Deaf1, which is alternatively spliced into a less active form in the PLN of NOD mice; similar findings have been replicated in the pancreatic lymph nodes of patients with T1DM . These observations link inflammation to impaired peripheral tolerance, via effects on TRA expression of a critical islet autoantigen. Then, perhaps, anti-inflammatory therapies may have a role in disease prevention and or treatment if they can help preserve peripheral tolerance mechanisms mediated by the presentation of self-antigens.
Multiple risk loci impair peripheral immune regulation. Non-antigen-specific abnormalities of peripheral tolerance are also involved in the pathogenesis of T1DM . Multiple defects impair the function of patients’ regulatory T cells and patients’ effector T cells are resistant to suppression by regulatory T cells . There is also an imbalance between Th17 immunity and regulatory T cells both in peripheral blood and the PLN of T1DM patients . Several susceptibility loci promote enhanced immune reactivity and less effective control over T-cell selection, activation, and perhaps differentiation into memory and regulatory phenotypes . The effects of these loci are not antigen- or disease-specific; indeed, these loci confer increased risk for several autoimmune disorders. The effects of a selected few, PTPN22, CTLA-4, and IL2RA are summarized here and illustrated in Figure 29.1.
Through its effects on TCR signaling, the PTPN22 locus affects T-cell function in the periphery, and reports indicate that it also impacts the function of regulatory T cells and B lymphocytes [51,52]. The CTLA-4 (Cytotoxic T lymphocytes- antigen-4; CD152) gene encodes for a protein, expressed by activated CD4 and CD8 T cells, which downregulates T-cell responses. CTLA-4 polymorphisms linked to T1DM are associated with reduced levels of a soluble form of CTLA-4 with a negative impact on regulatory T-cell function . Of note, a recent trial showed that CTLA-4-Ig therapy mitigated loss of insulin secretion in new onset patients, albeit the effect was limited in time . Polymorphic variants of the IL2RA gene (coding for the α chain of the Interleukin-2 receptor, IL-2Rα, or CD25) modulate T1DM risk through effects on several functions, including reduced levels of soluble IL-2 receptor , reduced STAT5a signaling in antigen-experienced CD4 T cells, and impaired regulatory T-cell function with decreased expression of the FOXP3 transcription factor . Studies from both NOD mice  and other autoimmune diseases indicate that impaired IL-2 signaling compromises T regulatory function. Hence, there is interest in using low dose IL-2 to ameliorate autoimmunity by selectively stimulating regulatory T cells . Overall, imperfect regulation of peripheral tolerance mechanisms, largely but not exclusively through effects on regulatory T cells, plays an important role in T1DM pathogenesis.