Type 1 diabetes is characterized by T-cell–medi-ated destruction of insulin-producing-cells. The strong association between autoimmune diabetes and certain susceptible major histocompatibility complex (MHC) class II alleles suggests that T-cell activation by self-peptides presented via these MHC class II alleles plays a critical role in the disorder’s pathogenesis. A diverse repertoire of T-cells is generated in the thymus, first through positive selection on MHC and self-peptide within the thymic cortex. This process requires adequate peptide presentation through interactions with MHC and sufficient T-cell receptor (TCR) signaling through the TCR/MHC/self-peptide complex (Fig. 1). As such, all T-cells in normal physiology are intrinsically self-reactive. However, subsequent negative selection of self-reactive T-cells in the thymic medulla should lead to clonal deletion for TCRs that recognize self-peptide/MHC with high affinity. Although some self-reactive, high-avidity T-cells do escape into peripheral circulation, suboptimal recognition of MHC/self-peptide by the TCRs may be required for T-cells to escape tolerance mechanisms. This idea is supported by experimental observations that the affinity of TCRs for MHC/self-peptide is generally lower than that for MHC/foreign peptide and that the interactions of autoreactive TCRs to MHC/self-peptide appear to be less extensive than to foreign peptide (1, 2). In light of the opposing mechanisms of positive and negative selection, fundamental questions remain regarding the affinity of TCR/MHC/self-peptide interactions that give rise to autoreactive T-cell responses. There is increasing evidence that insulin may be the primary autoantigen in the nonobese diabetic (NOD) mouse model (3). Several studies have emphasized the insulin B9–23 epitope. This peptide binds to IA g7, the MHC class II molecule in NOD mice, with moderate affinity (4). Wegmann et al.(5) reported that the majority of CD4 T-cells isolated from pancreatic islets of NOD mice recognized this epitope. Furthermore, an amino acid substitution within this peptide has also been shown to abrogate diabetes development in a transgenic mouse line expressing this modified insulin transgene, implying an important role for this epitope in disease development (6). In humans, HLA-DR0401-DQ0302 is a major disease susceptibility haplotype. Considerable effort has been devoted to identifying the relevant T-cell epitopes in human type 1 diabetes because this knowledge is essential for the study of autoreactive T-cells. In addition to insulin, multiple autoantigens have been reported, and many class II epitopes within each antigen have been identified (7, 8). These antigenic peptides have generally demonstrated moderate-to-high affinities when tested for their MHC binding. Thus, there has been no compelling evidence that autoreactive T-cells recognize unusually low-affinity peptides. However, there is also no conclusive evidence that T-cells recognizing these moderate-and high-affinity peptides are the primary pathogenic cells in humans with type 1 diabetes.

In this issue of Diabetes, Levisetti et al.(9) report their study of the proinsulin 1 47–64 (PI-147–64) epitope in NOD mice. This peptide is located in the C-peptide region and has been shown to be naturally processed and presented (10). In contrast to insulin B9–23, PI47–64 binds to IA g7 with a very weak affinity (30-fold lower). In spite of this, PI-147–64–specific T-cell lines could be isolated from NOD mice and expanded in vitro. More notably, these cells induced diabetes when transferred into NOD. scid recipient mice with an incidence rate of 50%. These results demonstrate that high-affinity peptide binding to …