Graves disease is a common form of human autoimmune thyroiditis. It shares many pathological features and HLA associations with other, less easily studied, organspecific autoimmune conditions such as insulin-dependent diabetes mellitus, and hence it is also a useful model for understanding these other diseases. We have previously shown that thyroid-infiltrating T This contrasts with the limited heterogeneity recently reported in some animal models and has potentially important implications for both our understanding of the autoimmune process in humans and the design of immunotherapies to reverse it.The thyroid is the commonest target for organ-specific autoimmunity in humans. Graves disease is a form of thyroid autoimmunity characterized by lymphocytic infiltration of the thyroid and stimulation of the gland by anti-thyrotropin (TSH) receptor autoantibodies resulting in thyrotoxicosis (1). Frequently, high titers of autoantibodies to other thyroidspecific proteins, including thyroglobulin (Tg) and the thyroid microsomal antigen (TMA; ref. 2), are seen as well. The cDNAs for all three autoantigens have now been cloned (3-5), and the TMA has been shown to be identical with the tissue-specific enzyme thyroid peroxidase (TPO), which is required for the biosynthesis of thyroid hormones (6).Thyroidectomy is often performed as part of treatment for the overproduction of thyroid hormones in Graves disease. This provides large quantities of tissue for in vitro study. The lymphocytic infiltrate of the thyroid has been shown to contain a high proportion of T cells (7). Also, the autoantibodies to the TSH receptor, Tg, and TMA seen in patients' sera, the majority ofwhich are produced in the gland itself (8), are ofthe IgG isotype and often ofhigh titer. Their production is therefore likely to be dependent on T-cell help (9). Furthermore, the target epithelium in the disease has been shown to aberrantly express HLA class II antigens (10), as well as adhesion molecules (11) and cytokines (12) required for antigen presentation to T cells. As a result, studies of autoantigen recognition by T cells in the thyroid in this disease may be expected to contribute key elements to our understanding of the autoimmune process involved.We (13), and subsequently others (14, 15), have studied T cells from thyroidectomy specimens of multiple patients with autoimmune thyroiditis. A high proportion of the in vivo activated [interleukin 2 (IL-2) receptor-expressing] T cells present in these specimens show specific reactivity to autologous thyroid epithelial cells (TECs) in the absence of additional antigen-presenting cells (APCs) (13-15). In contrast, T cells derived from the peripheral blood of the same patients show no such response (16). This recognition of TECs is known to be HLA class II-restricted (13,16), but the antigens involved have not been defined.In the current study, we selected a single patient with "typical" HLA type (B8, DR3) and active disease at the time of thyroidectomy for more detailed analysis. A large number of...