The tyrosine phosphatase like protein IA-2 is an important autoantigen in insulin-dependent diabetes mellitus (type 1 diabetes). Autoantibodies to IA-2 (IA-2A) are present in the serum of patients with type 1 diabetes even before the onset of the disease. Previously, we reported on a radioimmune assay to detect IA-2A, using E. coli-derived 125I-labelled IA-2 as antigen. Although this assay could be shown to be equivalent to the common reference method for IA-2A detection (radioligand assays using in vitro synthesised 35S-methionine labelled antigen), the disadvantages of both assays with respect to synthesis and handling of the radioactive antigen limit their use in routine laboratories. In this study, we have evaluated a non-radioactive enzyme-linked immunosorbent assay (ELISA) for the simple detection of IA-2A. We report on an ELISA where the biotinylated intracytoplasmic part of IA-2 (IA-2ic) is captured on streptavidin-coated plates. The sensitivity of the ELISA was similar to the validated radioligand assay, as it detected 47 of 69 (69%) patients with type 1 diabetes as compared to 46 of 68 (67 %) with the reference method for IA-2A detection (radioligand assays using in vitro synthesised 35S-methionine labelled antigen). Only 2 of 50 (4%) patients with autoimmune thyroid disease and 1 of 114 (1 %) healthy controls were detected in the ELISA, confirming specificity. There was a significant correlation between the ELISA and the radioligand assay (r = 0.64, p<0.001). We conclude that this ELISA is suitable to detect IA-2A in the serum of patients with type 1 diabetes with a similar sensitivity and specificity to the radioligand assay. This ELISA will allow rapid and simple measurement of IA-2A where the radioligand assay is inconvenient or not available.

Type 1 diabetes is thought to be an autoimmune disease mediated by T lymphocytes recognizing critical islet cell antigens. Recently, the tyrosine phosphatase like protein IA-2 was suggested as a putative autoantigen in type 1 diabetes since autoantibodies are detected in sera of diabetic patients and prediabetic subjects. Similarly, T cell responses of peripheral blood lymphocytes of type 1 diabetic patients to this protein have been described. Only very few data is available about immunodominant epitopes of IA-2 recognized by T cells. We have studied T cell responses in type 1 diabetic patients and age and partly HLA matched controls to IA-2 peptides designed to bind HLA risk alleles of IDDM as DR*0401 and DQ*0302. Both diabetic patients and controls responded to IA-2ic and some of the peptides. Three peptides of the C-terminal region of IA-2 were recognised by T cells of a fraction of diabetic patients but at least two of these peptides triggered also T cell responses in DR*0401/DQ*0302-matched controls. Most peptides bound to different HLA alleles ("promiscous binders"). The identification of autoantigenic epitopes may offer clues to related sequences e.g. of viral origin what relates to models of diabetes pathogenesis ("molecular mimicry"). Secondly, the design of antigen- or even epitope-specific immune intervention strategies aiming at tolerization of disease specific T cells in type 1 diabetes may profit from the knowledge of immunodominant T cell epitopes of a putative autoantigen.