A. P. Weetman, Clinical Sciences Centre, University of Sheffield S5 7SU, UK. Tel: +44 0114 2714160; Fax: +44 0114 2618775; E-mail: email@example.com
Recent research in autoimmune thyroid disease (AITD) has largely focused on delineation of the autoantigens and their epitopes, but there is now renewed interest in the immunoregulatory properties of T cells, an understanding of which may explain the emergence of AITD in experimental settings. T cell recognition of autoantigens has shown considerable intra- and interindividual heterogeneity, and a mixed pattern of cytokine production indicates that both the Th1 and Th2 limbs of the helper T cell response are involved in all types of AITD. It is now clear that secretion of chemokines and cytokines within the thyroid accounts for the accumulation and expansion of the intrathyroidal lymphocyte pool, and that the thyroid cells themselves contribute to this secretion. The thyroid cells also produce a number of proinflammatory molecules which will tend to exacerbate the autoimmune process. Thyroid cell destruction in autoimmune hypothyroidism is dependent on T cell-mediated cytotoxicity with the likely additional effect of death receptor-mediated apoptosis.
If you can't find a tool you're looking for, please click the link at the top of the page to "Go to old article view". Alternatively, view our Knowledge Base articles for additional help. Your feedback is important to us, so please let us know if you have comments or ideas for improvement.
It is almost 50 years since the discovery of autoimmune thyroid disease (AITD), with description of thyroglobulin (TG) autoantibodies in suitably immunized rabbits and in patients with Hashimoto's thyroiditis, and in the same year, the description of a long-acting thyroid stimulator in Graves’ patients which subsequently was identified as the thyroid-stimulating antibody directed against the TSH-receptor (TSH-R). Although these three types of AITD shared histological features of thyroid lymphocytic infiltration, it was another two decades before the probability of a similar underlying pathogenesis became established, and while we have much still to learn about what makes these diseases similar in some respects and different in others, it has become clear that a complex interaction between genetic susceptibility and environmental factors initiates the process, and that failure of immunological tolerance at multiple levels explains how this interaction operates (Parijs & Abbas, 1998; Goodnow, 2001; Tait & Gough, 2003).
In the last two decades attention has focused on the autoantigens in AITD, as in other autoimmune diseases, because delineation of the T and B cell epitopes of an antigen is critical to our understanding of an immune response, as well as offering the chance to modify such responses therapeutically. Particular emphasis has been placed on sequencing the thyroid autoantigens and defining B cell epitopes on the TSH-R, which will illuminate how the receptor is triggered by the rather unique antibodies that cause Graves’ disease. The recent cloning of human TSH-R antibodies (Sanders et al., 2003) will undoubtedly accelerate this process. However, it is also now clear that the tempo of the autoimmune response in AITD is usually slow, leading to spreading and diversification of the autoimmune response, in turn making therapy using modified autoantigens difficult to achieve. Reintroduction of tolerance or modifying the intrathyroidal autoimmune process offer more chance of success currently; this review will highlight recent developments in our understanding of these cellular immune responses in AITD. Earlier studies are reviewed extensively elsewhere (Weetman & McGregor, 1984, 1994). For simplicity, the term AITD will be used when similar findings have been made in both Graves’ disease and autoimmune hypothyroidism, otherwise the individual disorders will be highlighted.
Which T cells are involved?
Considerable energy has been expended in determining the phenotype of circulating and, more revealingly, intrathyroidal T cells, indicating that, in untreated AITD, a higher proportion of T cells express the activation marker HLA-DR. The CD4 : CD8 ratio is increased in the peripheral blood and both types of T cell are present within the thyroid. However, correlation between these changes and any functional consequences remains elusive as only a small fraction of each T cell population defined by monoclonal antibodies and flow cytometry is autoantigen-specific. For instance, postpartum thyroid dysfunction is more common in thyroid antibody-positive women with the highest levels of HLA-DR+ T cells, but there is considerable overlap with women who have antibodies but no thyroid dysfunction (Kuijpens et al., 1998).
In animal models of autoimmune disease, the T cell subset which has gained most attention recently is CD4+, CD25+. These T cells prevent autoimmune disease in healthy animals: their depletion by neonatal thymectomy in mice and rats results in a variety of organ-specific autoimmune disorders including AITD (reviewed by Sakaguchi et al., 2001). Transfer of these CD4+, CD25+ T regulatory (TR) cells can inhibit the development of autoimmune disease in such animal models and this population is naturally anergic, i.e. the cells do not proliferate on stimulation with the antigen which is required for their activation. Once stimulated by specific antigen, the suppression mediated by TR cells is directed against CD4+ and CD8+ cells, and is antigen-nonspecific, with the capacity to inhibit the proliferation of nearby T cells which are not responding to the same antigen. It is now known that the glucocorticoid-induced tumour necrosis factor receptor family-related (GITR) gene and Foxp3 are key regulatory genes for the development of TR, which may be amenable to future manipulation in the treatment of autoimmune disease (Shimizu et al., 2002; Hori et al., 2003). There is an important interaction between activation state of antigen-presenting cells and T cells which determines the type of immune response generated (Pasare & Medzhitov, 2003) and this depends, in part, on the generation of TR (Fig. 1).
So far little work on TR has been done in human AITD, although experimental autoimmune thyroiditis induced by thymectomy and sublethal irradiation was the first model of autoimmunity defined as resulting from depletion of what are now known as TR (Penhale et al., 1973). Part of the difficulty lies in defining the TR population, which lies within the highest-expressing CD25+ T cells, so that measurement of all CD25+ T cells will not be adequate, while defining the cut-off for high expression of CD25+ is difficult without functional assays. A recent study showing an excess of CD4+, CD25+ in Hashimoto's thyroiditis also found that 30 ± 7% of healthy CD4+ cells expressed CD25, indicating that a mixed population of activated T cells as well as TR had been examined (Watanabe et al., 2002). We have found no change in CD4+, CD25high number in the circulation of patients with untreated Graves’ disease (unpublished observations). Much more work can be expected in this area with the likelihood of not only gaining fresh insights into pathogenesis but also attempting to enhance TR function therapeutically.
T cell recognition of thyroid autoantigens
A different approach to immunotherapy for autoimmunity was suggested over a decade ago by the observation that encephalitogenic T cells in the most used animal model of multiple sclerosis were sometimes so clonally restricted that their depletion, using antibodies against the relevant T cell receptor (TCR), could successfully treat diseases while leaving the rest of the immune repertoire intact (Acha-Orbea et al., 1988). Other therapeutic approaches have been directed at the trimolecular complex of TCR antigenic epitope and class II module including modifying the epitope to become tolerogenic or antagonistic (Gaur & Fathman, 1994). These developments explain the widespread interest in defining both the TCR V gene usage and the autoantigenic epitopes in AITD.
Initial reports of restricted TCR Vα gene expression in AITD, with only a handful of the 18–20 gene families tested being represented in the intrathyroidal lymphocyte population, suggested a very restricted T cell autoimmune response, in turn raising hopes that such T cells could be targeted and even eliminated without affecting the entire repertoire (Davies et al., 1991). However, further analysis has not been able to reproduce such restriction, even in the CD25+ population which contains the activated T cells most obviously involved in the ongoing autoimmune response (McIntosh et al., 1993). There must be restriction of the autoreactive T cells in the earliest phase of AITD, simply on a reductionist basis, but by the time disease is manifest the immune response has spread to involve many different T cells whose polyclonality is reflected in unrestricted TCR usage without evidence of a dominant clone.
Similar ‘spreading’ of the autoimmune response (Lehmann et al., 1993) is clearly observed in the pattern of epitope recognition by T cells in AITD. Both thyroid peroxidase (TPO) and TSH-R have been examined in detail and some weakly dominant epitopes are possible to discern for both autoantigens, such as the sequences aa 158–176, 237–263 and 343–362 for the TSH-R, but the picture overall is one of considerable heterogeneity within and between patients (Tandon et al., 1991; Soliman et al., 1995). An interesting, related observation is the high-affinity binding between HLA-DR3 molecules and certain TSH-R peptides, which may explain the genetic susceptibility conferred by this HLA specificity for AITD (Sawai & De Groot, 2000).
At a broader level, many studies have been undertaken, with a variety of assays, to show that autoantigen-specific T cells are indeed present in the circulation and in the intrathyroidal T cells of patients with AITD, most typically using uptake of 3H-thymidine as a marker of proliferation in response to whole antigen such as TG (Weetman & McGregor, 1984, 1994). Ingenious modifications of such assays, particularly that measuring migration inhibition factor production, have been introduced to assess suppressor T cell function; based on such results the dominant view in the 1980s held that AITD (and other autoimmune endocrinopathies) arose because of an antigen-specific defect in suppressor cells (Volpé, 1988). However, these assays have no proven physiological relevance, often employing allogeneic systems, and would not be capable of detecting any abnormality in TR function. It remains to be seen whether, or more likely how, TR abnormalities are involved in AITD.
No real advances have been made in the last decade in employing tests of T cell sensitization to understand better how AITD develops, partly due to the shift in attention to epitope recognition and partly due to wider awareness that existing methods are cumbersome and often give equivocal results because so few sensitized T cells exist in the circulation or even in the thyroid-infiltrating population. New assays, using small volumes of whole blood and interleukin-2 (IL-2) supplementation, give superior results following TG stimulation of T cells to conventional methods and may permit population-based studies (Butscher et al., 2001), but so far no results have been produced that allow the quantification of responding T cell number, especially in conjunction with measures of function. Such methods would be useful, for instance in studying the progression of subclinical thyroid dysfunction and the genetic susceptibility to AITD. Multiparameter flow cytometric methods, combining tetramer-binding, cytoplasmic cytokine content and proliferation (Bercovici et al., 2003) may prove useful in this regard, especially with the current pace of development of major histocompatibility (MHC) class II, as well as class I, tetramers.
Cytokine production and patterns in AITD
A seminal development in immunology has been the recognition that T helper cells can be categorized as Th1 or Th2, identified by different patterns of cytokine production and effector function (reviewed by Mosmann & Coffman, 1989; Table 1). Originally defined in the mouse, a similar dichotomy has been observed in man, albeit with less precision. Moreover, some CD4+ T cells produce both Th1 and Th2 cytokines, and have been termed Th0, while a Th3 subset, producing TGF-β1, has been identified which is important in ameliorating autoimmunity (Prud’homme & Piccirillo, 2000). Determination of which Th limb is activated during an immune response depends on the concentration and type of antigen, the nature of the initial antigen-presenting cell and, most likely, on ill-defined genetic and environmental factors. A final crucial facet of the Th1/Th2 paradigm is that stimulation of one limb leads to reciprocal inhibition of the other; this ‘deviation’ of the immune responses may have important immunomodulatory consequences, and indeed represents an additional pathway to TR cells in the suppression of some types of autoimmune response (reviewed in Liblau et al., 1995).
Table 1. Summary of main characteristics of Th1 and Th2 cells
Key chemokine receptors
CCR3, 4, 8
Cell – mediated immune responses
Antibody production (especially IgE and non cytotoxic IgG isotypes), eosinophil and mast cell infiltrates
For obvious reasons, therefore, analysis of cytokine production in AITD has been a dominant exercise in the field over the last decade. In interpreting the data, the limitations of the available technology have not always been appreciated, nor have more prosaic concerns such as the sample size, the stage of disease and the production of cytokines by cells other than the CD4+ lineage. Analysis of the peripheral blood T cell population, particularly after mitogen-stimulation, will clearly be biased by the inclusion of the majority of lymphocytes that do not have specificity for thyroid autoantigens and even the intrathyroidal population will not be free from such biases. In vitro culture after cell fractionation, with measurement of cytokine release into culture supernatant, has clear advantages over RT–PCR methods in terms of directly quantifying cytokine-as-protein but requires large numbers of cells for purification of populations such as the CD4+ cells, and may not be free from the possible artefacts of any in vitro system. So what can we deduce from the results so far?
Cytokine mRNA patterns in both Graves’ disease and autoimmune hypothyroidism reveal both a Th1 and Th2 response (reviewed in Ajjan et al., 1996), although some have found a pattern more consistent with the prediction that Graves’ patients with their antibody-dependent disease should have a predominant Th2 profile, while Th1 predominates in Hashimoto's thyroiditis (Heuer et al., 1996). In the BB rat model of AITD, IL-12 mRNA is a prominent cytokine in the initial phase of disease; this product of dendritic cells and macrophages is crucial in deviating an immune response towards a Th1 pattern (Zipris et al., 1996).
Single-cell analysis of intrathyroidal lymphocytes and in vitro experiments including TSH-R-specific T cells have also produced conflicting results on the predominance of a Th2 response in Graves’ disease (Fisfalen et al., 1997; Roura-Mir et al., 1997; Okumura et al., 1999). Given that TG and TPO antibody production within the thyroid is a pronounced feature of all types of AITD, and that such antibodies reach concentrations at least 100-fold higher in the circulation than those against the TSH-R, it seems hardly surprising that a clear cytokine distinction between diseases within the AITD spectrum is difficult to pin down. However, by comparison with type 1 diabetes mellitus, which is believed to be primarily a Th1-mediated disease, whole blood cultures from Graves’ patients showed a bias in cytokine production away from Th1 towards Th2 (Kallmann et al., 1997). Radio-iodine treatment of Graves’ disease may bias the immune response to Th1, which in turn may contribute to hypothyroidism in some patients, but only small numbers of individuals have been studied (Jones et al., 1999). Further support for the importance of the Th2 pathway in Graves’ disease comes from work showing that recurrence after antithyroid drugs is more likely after attacks of allergic rhinitis, and elevated IgE levels and IL-13 levels in the circulation, which are markers for aTh2 response, are also adverse factors (Sato et al., 1999; Komiya et al., 2001). At some odds with these findings is the observation that CD4+, CD30+ T cells, believed to reflect the Th2 population, are elevated in patients with remission rather than relapse after antithyroid drugs (Watanabe et al., 2003).
Animal models of Graves’ disease remain incomplete in mirroring the full range of immunological features found in the human disorder. This most likely accounts for the equally confusing picture which has emerged recently of the relative importance of the two cytokine networks in experimental Graves’ disease. Using a model based on injection of mice with adenovirus expressing TSH-R, immune deviation towards Th2 was accompanied by a decrease rather than the expected increase in thyroid-stimulating antibody production, while Th1 responses to TSH-R were critical for disease induction (Nagayama et al., 2003). Using another murine model based on injection of syngeneic B cells expressing TSH-R, skewing of the response initially to Th1 or Th2 had no effect, but the complete absence of IL-4 prevented disease (Dogan et al., 2003). In the only animal model so far which resembles Graves’ ophthalmopathy, the cytokine milieu is believed to play a critical role but formal proof that the Th2 pathway is essential for eye disease initiation will require additional models (Many et al., 1999).
Therapeutic doses of γ-interferon (γ-IFN), surprisingly, are not associated with the development of AITD despite the key role of this cytokine in the effects of Th1 cells (De Metz et al., 2000). IL-2 can induce AITD (Krouse et al., 1995), but α-IFN is the cytokine most clearly associated with AITD, especially hypothyroidism, despite being neither Th1- nor Th2-dependent (Carella et al., 2001). β-IFN, used for treatment of multiple sclerosis, does not cause AITD (Durelli et al., 2001). This last result makes the finding of Graves’ disease in a third of multiple sclerosis patients treated with the Campath-1 lymphocyte-depleting antibody very striking (Coles et al., 1999). It is possible that this experiment of nature is the result of lymphocyte recovery in the aftermath of treatment being biased towards a Th2 response, but it is also possible that it mirrors the emergence of specific autoimmune disease in rodents after thymectomy (see above), in which case the treatment is interfering with TR function. A parallel but less clearly defined example is the emergence of Graves’ disease after immune restoration in AIDS patients treated with highly active antiretroviral therapy (Gilquin et al., 1998). Prospective studies should allow dissection of these important phenomena.
Why do lymphocytes accumulate in the thyroid?
It seems intuitive that the thyroid, as the only source of TG, TPO and (largely) TSH-R, should be infiltrated by the autoreactive lymphocytes causing AITD, but it is now clear that a complex set of steps is necessary for lymphocytes to leave the circulation (Fig. 2), with the endothelial cell having a pivotal role in controlling their entry into the thyroid (Marazuela, 1999). Once within the thyroid, the infiltrate becomes increasingly organized into germinal centres, microanatomically virtually identical to the secondary lymphoid follicles of lymph nodes, including the development of well-formed high endothelial venules and a distinct pattern of adhesion molecule and chemokine expression (Armengol et al., 2001).
Chemokines are a cytokine family with the ability to direct leucocyte migration (Table 2) and overexpression of several subfamilies has been observed in the thyroid in AITD, including macrophage inflammatory protein (MIP) − 1α (CCL3) and β, (CCL4), γ-IFN-inducible protein – 10 (IP-10; CXCL-10) monokine induced by γ-IFN (Mig; CXCL19) and RANTES (Yaqoub et al., 1999; Garcia-Lopez et al., 2001; Kemp et al., 2003). The infiltrating T cells express the IP-10/Mig receptor, CXCR3, and the RANTES receptor, CCR5, indicating that these chemokines are likely to be particularly important for the local accumulation of lymphocytes in AITD (Garcia-Lopez et al., 2001). Recently, chemokines specific for the process of lymphoid follicle organization, namely CXCL12, CXCL13 and CCL22, have been found to be overexpressed in AITD patients with thyroid germinal centres (Armengol et al., 2003) and, as might therefore be predicted, expression of these chemokines correlates with thyroid antibody level. This study also reported a reduction in CXCR4+ T cells and CCR7R+ T and B cells in the circulation of patients with intrathyroidal lymphoid follicles when compared with the intrathyroidal population, suggestive a novel effect of the formation of extralymphoid germinal centres on the circulating T cell pool. Another noteworthy development is the description of coordinate expression of IL-16 (a CD4-specific T cell chemoattractant) and RANTES by Graves’ (but not control) fibroblasts in response to EGF-1-receptor autoantibodies present in Graves’ sera, which may explain the infiltration of lymphocytes into the orbit and skin as well as thyroid in Graves’ disease, although the wider anatomical distribution of the EGF-1 receptor is difficult to fit into such a model (Pritchard et al., 2003).
Table 2. Summary of main chemokines described in the thyroid in AITD. There are four chemokine families classified on the basis of the position of two cysteine (C) residues near the N-terminus of the molecule: CXC chemokines have an amino acid residue (X) between the cysteine molecules, CC chemokines do not. (L = ligand, R = receptor); C and CX3C chemokines have not yet been found in the thyroid
CXCR1, 2 are involved in neutrophil migration; CXCR3, 5 in migration of effector T cells, CCR1, 2, 5 in migration of dendritic cells and monocytes, CCR2, 5 in migration of Th1 cells, CCR3, 4 in migration of Th2 cells and eosinophils.
1. CXC (α) chemokines
2. CC (β) chemokines
CCR1, CCR3, CCR5
An additional critical element in the accumulation of lymphocytes within the thyroid gland in AITD is the expression of selectins and integrins on the endothelium (Fig. 2) but this has not been studied to quite the same extent as the expression of chemokines or cytokines (the latter being responsible for expansion rather than migration of lymphocytes). Cutaneous lymphocyte-associated antigen, a selectin found on high endothelial venules, occurs on the endothelium in AITD thyroid tissue (Armengol et al., 2001). β1-integrins mediate cell attachment to extracellular matrix proteins and one of these, VLA-α2 (very late antigen-α2), is upregulated on endothelial cells in AITD, as are vascular cell adhesion molecule-1 (VCAM-1), intercellular adhesion molecule-1 (ICAM-1) and E- and P-selectins (Marazuela et al., 1995, 1997; Jungheim et al., 2001). Interestingly, the increased intrathyroidal vasculature which typifies Graves’ disease is associated with an increase in circulating vascular endothelial growth factor, which appears to be derived from the thyroid cells in response to TSH-R stimulation by TSH-R stimulating antibodies (Iitaka et al., 1998). This last response is an example of what is now recognized to be a key feature of AITD, namely the involvement of the thyroid follicular cells (TFC) themselves in the autoimmune process, as summarized below.
The role of TFC in AITD
Since the ground-breaking description of MHC class II molecule expression by thyroid cells (Pujol-Borrell et al., 1983), many more pathways of potential communication between TFC and the infiltrating lymphocytes have been uncovered (Table 3). Far from being innocent victims of the thuggery of a dysregulated immune system, it increasingly appears that the strength of some of these interactions between TFC and lymphocytes is crucial in determining the progression of AITD. Thus, MHC class II expression by TFC can induce anergy in naïve T cells, and may therefore be protective against AITD in the face of a viral thyroiditis, yet can stimulate memory T cells, leading to expansion of the infiltrate once AITD becomes established (Fig. 3; Marelli-Berg et al., 1997). This pathway may explain some of the heritability of AITD, as the magnitude of class II expression by TFC is genetically determined (Sospedra et al., 1995).
Table 3. Possible pathways of participation of thyroid follicular cells in the autoimmune process
Enhanced MHC class I
De novo MHC class II
Anergy of naïve autoreactive T cells; exacerbation of ongoing autoimmune response involving memory T cells
Increased lymphocyte homing and cytotoxicity.
Interaction with CD40 ligand on T cells; increased IL-6 production
Cytokines (e.g. IL-1, IL-6, IL-12, IL-13, IL-18)
Various, especially enhancing local autoimmune response
The induction of MHC class II expression on TFC is critically dependent on γ-IFN released by the lymphocytic infiltrate. In response to this and other cytokines, especially IL-1 and TNF-α, TFC themselves secrete an increasingly well identified array of cytokines, which in turn would be expected overall to amplify the intrathyroidal autoimmune response (Ajjan et al., 1996; Takiyama et al., 2002). This array includes chemokines which will further facilitate the localization of the infiltrate to the thyroid gland (Armengo et al., 2003; Kemp et al., 2003). Other soluble proinflammatory mediators released by TFC include reactive oxygen metabolites, prostaglandins and nitric oxide (Weetman et al., 1992; Kasai et al., 1995). TFC cell surface molecules upregulated by cytokines include CD40, ICAM-1 and several complement membrane attack complex regulatory proteins, especially CD59 (Weetman et al., 1989; Tandon et al., 1994; Metcalfe et al., 1998) The latter prevent complement-mediated lysis of TFC, despite the evidence of widespread complement attack on TFC in AITD, but do not impede the proinflammatory effects of sublethal complement activation, which can alter the metabolic function of TFC and cause cytokine release (Weetman et al., 1989).
Why are TFC destroyed in autoimmune hypothyroidism?
Because complement attack is clearly not a major pathway for thyroid destruction in autoimmunity, attention has turned to other mechanisms (Fig. 4). Cytokines and other soluble mediators can impair TFC function in a number of ways, but do not appear to be directly cytotoxic (Ajjan et al., 1996). Disruption of thyroid epithelial integrity by IL-1 may be important in exposure of sequestered autoantigens, such as TPO, which might then become accessible to TPO antibodies and T cell-mediated immune responses (Nilsson et al., 1998).
ICAM-1 expression by TFC is induced by cytokines and increases the binding of CD8+ cells and subsequent cytotoxicity, and perforin-expressing T cells with cytotoxic potential accumulate within the thyroid in AITD (Weetman et al., 1990; Wu et al., 1994). Most attention recently has focused on the Fas (CD95) pathway with the description of novel Fas ligand (CD95L) expression on TFC from patients with AITD, which in turn gave rise to the notion that autocrine or paracrine interaction with CD95 on TFC (itself upregulated by IL-1) would lead to suicide or fratricide of the thyroid cells (Giordano et al., 1997). A clearer example of the crucial role of TFC in the autoimmune process is difficult to imagine!
This hypothesis has nonetheless created some controversy, with arguments ranging from the technical aspects of CD95L detection to more theoretical concerns that any such expression could actually protect TFC from lymphocyte recognition by causing death of any Fas-expressing autoreactive T cells, a mechanism known as immunological privilege (Baker, 1999). A further argument against the fratricide/suicide hypothesis is the enhanced expression of apoptosis regulatory proteins such as Bcl-2 by TFC, at least in Graves’ disease, although it is also accepted that attacking lymphocytes can mediate apoptosis through several pathways involving TNF, CD95 or TRAIL receptors, all of which are on thyroid cell membranes and might be less susceptible to interference from Bcl-2 (Bretz & Baker, 2001). Overall, it seems highly likely that death receptor-mediated apoptosis plays a significant role in thyroid cell destruction, but its relative importance compared to other mechanisms (Fig. 4), the phase of disease in which it is prominent, the separate mechanisms involved in Graves’ disease, in which TFC are not destroyed, and the possibility of immune privilege are all outstanding questions. Answers to these will be another step in understanding the complex cellular immune response in AITD.