Abstract
Through animal models, particularly non-obesity diabetes model (NOD), pathological understandings of human autoimmune diabetes have been gained. However, features of those mouse models and the human disease are not sufficiently analogous; it is therefore not unexpected that interventions based on the mouse data fail at an alarming rate in clinical settings. An improvised model that maximally resembles the real pathological course is highly desirable. Here we devised a 'double-hit' strategy, pancreas was first hit by chemical damage (streptozotocin, STZ) to unleash auto-antigens, then hit second time by transient immune-inflammation (regulatory T cell depletion). Comparing to NOD model, this strategy not only induced classical diabetic symptoms, but also depicted the crucial pathogenic features absent in conventional models, such as CD8+ T cell dominant infiltrates, strong ketoacidosis and epitope-specific T cell responses. In addition, this model allowed synchronized control of disease onset, permitting more refined temporal analysis of disease progression. We believe that this model would yield research outcomes with clinically relevant prediction power unattainable previously.