The human autoimmune thyroid diseases (AITDs) are characterized by profuse infiltrates of both CD4+ and CD8+ T cells. The intrathyroidal T-cell-receptor repertoire in Graves' disease, more than in Hashimoto's disease, has been shown to be biased as evidenced by phenotypic analysis and by the use of a restricted T-cell-receptor variable (V) gene repertoire seen in both TCR alpha and beta chains. Evidence for a bias in the T-cell repertoire has also been observed in animal models of induced and spontaneous autoimmune thyroiditis. We found a similar phenomenon of autoimmune thyroid-related T-cell bias in thyroid-humanized scid mice. In these studies we transplanted lymphocyte-depleted thyrocytes and autologous peripheral lymphocytes from AITD patients with a basement membrane preparation which allowed the formation of an artificial thyroid which we have called an "organoid". T-cell clonal expansion was present in these artificial mixed-cell organoids which appeared to mimic the in vivo process. Such clonal expansion was suggestive of an antigen-driven immune response and could also be identified in thyroid tissue from patients with Graves' disease. Our data on scid mice grafted with human mixed-cell thyroid organoids, therefore, suggested that the major antigens driving T-cell selection in patients with AITD were most likely to be thyroid specific. These antigens include thyroglobulin, thyroid peroxidase, and the receptor for thyroid stimulating hormone (TSHR) on the surface of thyroid epithelial cells and we found significant T-cell proliferation to synthetic TSHR peptides in patients with AITD as compared with normals. Our search for a TCR recognition motif for the autoantigen TPO did not reveal any specific sequence motifs. Instead, analysis of the physico-chemical characteristics i.e. hydrophobicity of the amino acids in the CDR3 (N) region of the TCR alpha chain, revealed a strong negative linear correlation between strength of stimulation and the average hydrophobicity of N-region amino acids. This led us to hypothesize that lower affinity T-cell clones were commonly more hydrophobic in their CDR3 alpha region amino acids in keeping with potential crossreactivity of such T cells as a consequence of promiscuous, hydrophobic CDR3 regions. This phenomenon would be analogous to polyreactive, natural autoantibodies which tend to be crossreactive and 'sticky'. Thus, the physico-chemical characteristics of the TCR alpha CDR3 region supported the interaction with antigen/MHC by potentially cross-reactive T cells of low affinity. It would seem likely that such low-affinity autoreactive T-cell populations serve as a pool of potentially pathogenetic cells. These cells would be able to respond to an insult which, via a number of possible mechanisms such as molecular mimicry, would initiate a thyroid lymphocytic infiltration in an antigen-driven fashion with intrathyroidal T-cell expansion and a marked bias in the utilization of T-cell-receptor V genes.