In a rosette assay, 63 patients with recent-onset type I (insulin-dependent) diabetes mellitus had a higher (P < .001) number of lymphocytes adhering to rat insulinoma RINm5F cells (diabetic rosettes) than 153 healthy control (background rosettes) or 20 nondiabetic subjects with other organ-specific autoimmune diseases. Furthermore, lymphocytes from diabetic patients displayed a highly correlated (r = .97, P < .001) binding on two different xenogeneic β-cell lines (RIN and hamster insulinoma HIT cells). This phenomenon was not found on a panel of seven non β-cell lines (e.g., exocrine pancreatic cells, endocrine cells). By increasing lymphocyte-to-RIN ratios (0.25:1 to 30:1), the supernumerary RIN-adherent lymphocytes from diabetic patients, expressed as the percentage of lymphocytes involved in conjugates, were only detectable at lower ratios (0.25:1 to 4:1), and their binding efficiency was two times higher than that of control lymphocytes. This efficiency fell at higher ratios (>4:1) to the level of background rosettes that remained constant through the ratio scale. This specific RIN-rosette formation was abrogated when lymphocytes from diabetic patients were preabsorbed on β-cells (either HIT or RIN) but not on non-β-cells, whereas preabsorption of control lymphocytes did not modify the number of background rosettes. In addition, diabetic rosettes, but not background rosettes, were inhibited by competition with RIN membrane extracts but not by non-β-cell extracts. Moreover, diabetic rosettes were inhibited during blocking experiments with anti-CD3 monoclonal antibody (MoAb) but not with unrelated MoAbs. Finally, diabetic rosettes were abolished after depletion of CD3+ or CD4+ cells but remained after elimination of CD8+ or CD16+ cells. Enrichment experiments confirmed the implication of CD3+/CD4+ cells but not of B-lymphocytes. Thus, CD3+/CD4+ cells from diabetic patients specifically recognize xenogeneic β-cell membrane antigens. This reinforces our hypothesis that diabetic rosettes are markers of cellular immunity in type I diabetes.

This content is only available via PDF.