The three most important antigens involved in thyroid autoimmunity are clearly defined. First to be recognized was thyroglobulin (TG), the 670 kD protein synthesized in thyroid cells and in which T 3and T 4are produced. Although considered to be a single unique protein, the proteins prepared from different thyroid glands, especially those with Graves' disease and thyroid malignancy, react differently with polyvalent rabbit anti-TG antisera (40), suggesting that the fine structure of TG differs from person to person. Four to six B cell epitopes of TG are known to be involved in the human autoimmune responses. Animal studies suggest that antigenicity of the molecule is related to iodine content, but studies on human antisera do not bear this out (40). A recent review has described these species differences in detail and discussed the role of measuring TG antibodies in thyroid disease (41). Antibodies appear to recognize conformation of large fragments of TG, whereas T cells recognize peptide segments and their primary structure. Mouse experiments suggest that, to induce autoimmunity to TG, initial tolerance to dominant epitopes must be overcome, and the immune response then spreads to cryptic epitopes that are the major inducers of thyroidal T cell infiltration (42). There is also renewed interest in the concept that TG itself may be present in orbital tissue in ophthalmopathy patients where it could act as a co-antigen in ophthalmopathy (43). The fact that it is present in fat but not muscle from the orbit may indicate its involvement in a subset of such patients, but the failure to find TG in normal orbital tissue is certainly very suggestive of a possible role.
The second antigen to be identified was the TSH receptor (TSH-R), a 764 aa glycoprotein. Antibodies to TSH-R mimic the function of TSH, and cause disease by binding to the TSH-R and stimulating (or inhibiting) thyroid cells, as described later. The human TSH-R has been cloned and sequenced in several laboratories, and is known to be a member of a family of cell surface hormone receptors which are characterized by an extra-membraneous portion, seven transmembrane loops, and an intracellular domain which binds the G Ssubunit of adenyl cyclase (44, 45). It undergoes complex post-translational processing, forming a Z subunit structure comprising the A subunit. Graves' patients' IgGs are reported to bind to specific sequences in the extracellular domain, and much effort is currently being directed to definition of the B and T cell epitopes (46). Human TSH-R epitopes are non-linear or "conformational", perhaps composed of several segments of the protein.
The recent description of two mouse and a hamster monoclonal antibody has significance for several reasons (47-49). Firstly, these antibodies confirm that a single antibody is sufficient to activate the receptor, rather than two or more simultaneously. Secondly, they will permit epitope mapping, already partially achieved. One antibody preferentially recognizes the free A subunit, not the holoreceptor, suggesting that free A subunit, shed from thyroid cells, may initiate or amplify the autoimmune response. The hamster monoclonal antibody, in contrast to TSH, does not enhance post-translational TSH-R cleavage, which may extend the receptor half-life and thus account for the prolonged thyroid stimulation seen following antibody binding (49). It has become clear that in the hamster model, high-affinity TBAb recognize at least 2 conformational epitopes, one of which was indistinguishable from the epitope for thyroid stimulating antibodies (50). The epitopes for the latter are more restricted but these antibodies need not have identical epitopes, although their binding does require interaction with the highly conformational N terminus of the A subunit (51). Finally, these pave the way for the development of human monoclonal antibodies which will allow a greatly improved understanding of the mechanisms involved in Graves’ disease.
A single human monoclonal TSH-R stimulating antibody has been produced (52). Both the intact IgG and its Fab bound to the TSH-R with high affinity and the monoclonal had similar features in all respects to known TSAb. This observation indicates that indeed only a single species of antibody is needed to stimulate the receptor and opens the way to a more detailed analysis of receptor-antibody interaction. More conventional approaches based on different methods of expressing the TSH-R have shown that TSAb preferentially recognize the free A subunit rather than the holoreceptor, either because of steric hindrance from the plasma membrane or membrane spanning region of the receptor or because of TSH-R dimerisation (53). The epitope for TBAb are largely in the C terminus and are able to recognize holoreceptor more efficiently. These observations have provided support from the hypothesis that shedding of free TSH-R A subunits may be critical in initiating or amplifying the autoimmune response in Graves’ disease. Further evidence comes from immunization of mice with adenoviruses expressing different structural forms of the TSH-R: goiter and hyperthyroidism occur more frequently when mice are given virus that expresses the free A subunit rather than a receptor with minimal clearage into subunits (54).
Patients with autoimmune thyroid disease may have both stimulating and blocking antibodies in their sera, the clinical picture being the result of the relative potency of each species. The major T cell epitopes are heterogeneous and T cell reactivity against certain TSH-R epitopes has been demonstrated in high frequency in normal subjects (55). Identification of the critical epitopes has proved elusive although peptides 132-150 do appear to constitute one key epitope; there is poor correlation between binding affinity and T cell immungenicity in experiments to attempt such localization (56). Apparent hTSH-R mRNA transcripts and protein have been identified in retrobulbar ocular tissue, particularly the preadipocyte fibroblast, suggesting that TSH-R expression in the orbit could well be involved in the development of autoimmunity and ophthalmopathy (57) and this is partially supported by experiments showing that activation of the TSH-R stimulates early differentiation of preadipocytes, but terminal differentiation is not induced (57a). It should be noted that an alternative pathway for fibroblast involvement in ophthalmopathy may depend on the production of insulin-like growth factor antibodies in these patients but it is difficult to reconcile these findings with the orbital specificity of the autoimmune process in thyroid eye disease (57b)
The third thyroid antigen was described as "microsomal antigen" was identified as thyroid peroxidase (TPO) in 1985 (58) (Fig. 7-8). For more than three decades, since antisera from humans with thyroid autoimmunity reacted with an easily denatured protein present on the surface of thyroid cells and in cell cytoplasm. DeGroot’s laboratory demonstrated that human antisera reacting to "microsomal antigen" precipitated human thyroid peroxidase (TPO) prepared from Graves' disease thyroid tissue (59) (Fig. 7-8) and at the same time Czarnocka et al. purified human TPO and confirmed identity with the microsomal antigen (60). The cDNA was cloned and sequenced in several laboratories (61 - 63). Alternative splicing of the mRNA probably provides the explanation for the 101 and 107kD forms of the protein (64). As yet a functional difference for the two forms has not been described. The interaction of human anti-TPO antisera and monoclonal antibodies also indicate the presence of several B cell epitopes which map to two main domains, A and B (reviewed in 65). The three-dimensional structure of TPO has been modeled and the location of the B determinant has been defined (66). Recently, further experiments with monoclonal antoibodies have defined individual amino acid residues that are critical for the the two immunodominant regions (67). The epitopes recognized by antibodies are stable within a patient and may be genetically determined (68). Investigation of linear epitopes of TPO recognized by T cells from patients with AITD has produced conflicting results to date but certain sequences are beginning to emerge which are shared between reports on various patients (69, 70).
Figure 8. Precipitation of peroxidase activity by sera from a patient with autoimmune thyroid disease and positive "microsomal" antibodies, and from a control subject without circulating antibodies. TPO was precipitated by primary incubation with human sera, and removal of TPO-Ig complexes was achieved by addition of Protein H-Sepharose CL-4B. Residual hTPO activity in the supernatant was assayed in a guaiacol assay.
The microsomal antigen/TPO is expressed on the thyroid cell surface as well as in the cytoplasm, and may represent the cell-surface antigen involved in complement-mediated cytotoxicity (71). Intracytoplasmic binding of antibodies to TPO indicates that there is access to this compartment, but the consequences in vivo are unclear. Expression of the antigen is increased by incubation of thyroid cells with TSH and lectins, and this response is augmented by interferon-g (IFN-g) (72). IFN-g alone does not stimulate production of the antigen. Surprisingly, TPO and TG are reported by some workers to share common epitopes (73). There are small areas of amino acid sequence homology, but it is uncertain that these are important B cell epitopes. The most recent study in this area, using monoclonal antibodies, has concluded that TG-TPO autoantibodies are polyreactive rather than bispecific (74).
It is of considerable interest that the three major antigens involved in AITD are involved in production of thyroid hormone. On the other hand, perhaps this is circuitous reasoning, since clearly it is sensible that autoimmunity to unique thyroid antigens would produce "thyroid disease." If autoimmunity cross-reacted with a variety of other organs, the disease produced would present as a different syndrome. These antigens are unique to the thyroid gland although fat and possibly other tissues may express TSH-R (57), and cross-reactivity of antibodies with proteins in other organs is extremely limited.
Antibodies against the sodium/iodide symporter (NIS) were first shown functionally in cultured dog thyroid cells (75). With the cloning of NIS, antibodies were also demonstrated in the majority of Graves’ disease sera by immunoblotting (76), albeit using the rat sequence. We have recently found that a third of Graves’ disease sera contain antibodies capable of blocking NIS-mediated iodide uptake in cells transfected with the human NIS but the relevance of this for thyroid function is unclear (77). The same antibodies have also been detected using an immunoprecipitation assay (78). Others have found no such blocking activity using assays with cell lines displaying much higher 131I uptake, in turn suggesting that any NIS blocking activity only occurs at limiting conditions (79). This implies that NIS autoantibodies probably have no effect in vivo. NIS expression on TECs is upregulated by TSH and downregulated by cytokines and the latter could impair thyroid function in the setting of AITD when such cytokines are synthesised in the thyroid (80).
Antibodies to a variety of other thyroid cell components are also occasionally present in AITD, including antibodies that react with thyroxine or triiodothyronine (81), antibodies reacting with tubulin and megalin, calmodulin and antibodies reacting to DNA or DNA associated proteins (82 - 84).