Type 1 diabetes (T1D) is an autoimmune disease caused by several genetic and environmental factors leading to targeted destruction of insulin-producing cells. However, T1D cannot be explained by inherited susceptibility and environmental factors alone. Concordance in monozygotic twins is much <100% despite shared environment (pediatric T1D onset). Similarly, only <50% of males in the inbred NOD (non-obese diabetic) mouse, get the disease despite genetic identity and standardized environment. These observations point to stochastic determinants, one of which could consist of somatic mutations in the expanding antigen-specific autoreactive T-cell lineages.

To test this hypothesis, we examined memory CD4+ cells from pancreatic lymph nodes of NOD mice and T-cells from blood of newly diagnosed T1D patients, sorted for in vitro proliferation to proinsulin. We concentrated on copy-number alterations (CNA) because they are known to happen in blood cells, are easier to detect than point mutations and their functional consequences are easier to infer. These were detected by high-resolution comparative genomic hybridization using germline DNA as reference and confirmed by MLPA. In both the mouse and human samples we found a number of CNAs with highly non-random independent recurrence of the same gene(s) across samples, including SIRPB, at a known germline T1D locus, and two genes (MAP3K3 and SLC12A7) expressed exclusively in immune cells. We found no CNAs in lymphocytes expanded as part of normal host defense, (regional lymph nodes of mice with Leishmania major infection).

Our data strongly support a causal role for post-zygotic copy-number mutations in autoreactive T-cells in T1D.