International workshop on lessons from animal models for human type 1 diabetes - Identification of insulin but not glutamic acid decarboxylase or IA-2 as specific autoantigens of humoral autoimmunity in nonobese diabetic mice

Several self-antigens have been reported as targets of the autoimmune respo nse in nonobese diabetic (NOD) mice. The aim of this workshop was to identi fy autoantibody assays that could provide useful markers of autoimmunity in this animal model for type I diabetes. More than 400 serum samples from NO D (4, 8, and 12 weeks of age and at diabetes onset), BALB/c, and B6 mice we re collected from six separate animal facilities, coded, and distributed to five laboratories for autoantibody measurement. Insulin autoantibodies (IA A) were measured by radiobinding assay (RBA) by four laboratories and by en zyme-linked immunosorbent assay (ELISA) in one laboratory. Using the 99th p ercentile of BALB/c and B6 control mice as the threshold definition of posi tivity, IAA by RBA were detected in NOD mice at frequencies ranging from 10 to 30% at age 4 weeks, from 26 to 56% at 8 weeks, from 42 to 56% at 12 wee ks, and from 15 to 75% at diabetes onset. With ELISA, IAA signals differed significantly between control mouse strains and increased with age in both control and NOD mice, with frequencies in NOD animals being 0% at 4 weeks, 14% at 8 weeks, 19% at 12 weeks, and 42% at diabetes onset. For IAA, the EL ISA results were relatively discordant with those of RBA. GAD autoantibody (GADA) and IA-2 autoantibody (IA-2A) signals obtained by RBA were low (maxi mum 2.5% of total) but were increased in NOD mice compared with control mic e at diabetes onset (GADA 29-50%; IA-2A 36-47%). ELISA also detected GADA ( 42%) and IA-2A (50%) at diabetes onset, with results concordant with those of RBA. Remarkably, GADA and IA-2A frequencies varied significantly with re spect to the source colony of NOD mice. Furthermore, whereas neither GADA n or IA-2A correlated with IAA, there was strong concordance between GADA and IA-2A in individual mice. Sera with increased binding to GAD and IA-2 also had increased binding to the unrelated antigen myelin oligodendrocyte glyc oprotein, and binding to GAD could not be inhibited with excess unlabeled a ntigen, suggesting nonspecific interactions. In sum, this workshop demonstr ated that IAA measured by sensitive RBA are a marker of autoimmunity in NOD mice and draw into question the true nature of GADA and IA-2A in this anim al model.