| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
Department of Microbiology and Immunology (B.S.P.), College of Medicine, University of Illinois at Chicago, Chicago, Illinois 60612-7344; Division of Endocrinology (R.S.B.), Department of Medicine, Mayo Clinic, Rochester, Minnesota 55901; Division of Molecular Medicine (T.J.S.), Harbor-UCLA Medical Center, and the David Geffen School of Medicine at University of California at Los Angeles, Torrance, California 90822; and Long Beach Veterans Administration Healthcare System (T.J.S.), Long Beach, California 90822
Correspondence: Address all correspondence and requests for reprints to: Bellur S. Prabhakar, Ph.D., Professor and Head, Department of Microbiology and Immunology (MC790), College of Medicine, University of Illinois at Chicago, 835 South Wolcott Avenue, Chicago, Illinois 60612. E-mail: Bprabhak{at}uic.edu
 |
Abstract |
|---|
 Top Abstract I. IntroductionII. Risk Factors for...III. Immunological Basis for...IV. Autoimmune Responses to...V. TSH Receptor as...VI. Evolution of Autoimmune...VII. Graves’...VIII. Genetic and Environmental...IX. Orbital AutoimmunityX. Orbital AutoantigensXI. Role of Orbital...XII. Role of GAGs...XIII. Role of Orbital...XIV. Future PerspectiveReferences |
|---|
|
I. Introduction |
|---|
|
TopAbstractI. IntroductionII. Risk Factors for...III. Immunological Basis for...IV. Autoimmune Responses to...V. TSH Receptor as...VI. Evolution of Autoimmune...VII. Graves’...VIII. Genetic and Environmental...IX. Orbital AutoimmunityX. Orbital AutoantigensXI. Role of Orbital...XII. Role of GAGs...XIII. Role of Orbital...XIV. Future PerspectiveReferences |
|---|
Pathological findings in the thyroid reflect the hyperthyroidism associated with GD. A salient feature of thyroid pathology is hypertrophy and hyperplasia of the parenchyma. A change in the thyroid epithelium is a common finding, with the appearance changing from cuboidal to columnar with papillary infoldings. In addition, these cells demonstrate enlarged Golgi, increased numbers of mitochondria, and vacuolation of the colloid. These changes in the thyroid are often accompanied by a diffuse cellular infiltration consisting of lymphocytes and plasma cells. Occasionally, the infiltration can be more focal and resemble changes seen in Hashimoto’s thyroiditis.
|
II. Risk Factors for Developing Graves’ Disease (GD) |
|---|
|
TopAbstractI. IntroductionII. Risk Factors for...III. Immunological Basis for...IV. Autoimmune Responses to...V. TSH Receptor as...VI. Evolution of Autoimmune...VII. Graves’...VIII. Genetic and Environmental...IX. Orbital AutoimmunityX. Orbital AutoantigensXI. Role of Orbital...XII. Role of GAGs...XIII. Role of Orbital...XIV. Future PerspectiveReferences |
|---|
A. Genetic factors
Increased incidence of GD among members of a family indicates that genetic factors might play an important role in determining susceptibility to GD (1, 2). Studies involving twins indicate a higher degree of concordance, suggesting that the genetic factors might be a major contributor (3). Earlier studies have shown that patients with GD express human leukocyte antigen (HLA)-B8 more often than the control subjects without the disease (4). Other studies have shown that the risk of developing the disease is higher among individuals with an MHC (major histocompatibility complex) class-II haplotype of HLA-DR3 (5, 6) or with DQA1*0501 haplotype (7, 8). In contrast, the expression of HLA DR ß1*07 appears to confer protection (9). However, there is racial variation in the association of MHC haplotypes with increased risk for the development of GD (9). Two recent reviews on genetic susceptibility to GD have summarized a large number of studies to date (10, 11). The general conclusions one can draw from studies to date are that multiple genetic factors appear to contribute to the risk of developing GD and the penetrance of the disease is about 30% in monozygotic twins. Some of the susceptibility genes are involved in the generation of immune responses and/or associated with X chromosome, whereas others appear to be specific to GD (e.g., GD-1, -2, and -3).
How could these gene products play a role in protecting against or enhancing the risk for the development of the disease? Hyperthyroidism in GD is mediated by autoantibodies to TSHR protein, and the protein antigens are T cell-dependent antigens. Therefore, CD4+ T cells, most likely, play an important role because they provide the necessary help for autoantibody production. If this were true, then the most effective antigen presentation would be in the context of MHC class-II molecules, and thus they may play a critical role in the pathogenesis of GD. MHC class-II molecules form peptide binding grooves, the specificity of which is determined by their amino acid sequence. This allows binding of antigenic peptides of a certain length and with specific sequences. T cells, bearing cognate receptors, in turn recognize these MHC-peptide complexes and undergo activation. Thus, MHC molecules play a very critical role in the antigen presentation and thus in the T cell repertoire generation.
Early in the thymocyte development, immature T cells migrate from bone marrow into the thymus where they undergo maturation and selection. Although T cells that react with self-antigens expressed on the thymic epithelium are deleted, some self-reactive T cells escape negative selection because many tissue-specific antigens may not be expressed on the thymic epithelium. However, cells capable of interacting with exogenous antigens and some cells that can interact with self-antigens are positively selected and migrate to the periphery (12). Because a given MHC molecule can bind only a limited number of peptides with certain sequences, they largely determine the repertoire of T cells that are either eliminated in the thymus (negative selection) or allowed to migrate to the periphery (positive selection). Because individuals expressing HLA DR3 and DQA1*0501 haplotypes have an increased risk of developing the disease, it is possible that they might fail to present certain tissue-specific pathogenic peptides in their peptide binding groove and thus prevent negative selection of self-reactive T cells that express the cognate T cell receptor (TCR) on their surface. These self-reactive T cells could encounter the antigen in the periphery and undergo activation. In contrast, HLA DR ß1*07, which confers protection, might bind pathogenic peptides and facilitate negative selection of self-reactive T cells, expressing cognate TCRs, in the thymus and prevent an autoimmune response. Thus, MHC gene products can differentially affect both the repertoire selection in the thymus and subsequent autoimmune responses in the periphery.
More recently, a number of reports have suggested an association between cytotoxic T lymphocyte antigen-4 (CTLA4) and GD. The CTLA4 protein exists in multiple isoforms due to genetic polymorphism in the first exon (13). This is often linked with another polymorphism of AT dimers at the 3'-untranslated region of the third exon, which is thought to be associated with increased risk of GD (14, 15). Similarly, other associations with genetic polymorphism in CTLA4 have been reported (16, 17, 18). How could this molecule be linked with increased susceptibility to GD? T Lymphocytes constitutively express a costimulatory molecule called CD28 on their surface. During the initiation of an immune response, the T cells receive their first signal through their TCR interaction with the MHC-peptide complexes on the antigen-presenting cells (APCs). However, naive lymphocytes require a second signal to undergo activation that is delivered through CD28 upon its interaction with specific ligands, B7.1 and B7.2 (i.e., CD80 and CD86, respectively), expressed on the surface of APCs. Subsequent to activation, T cells up-regulate CTLA4 expression on their surface. The CTLA4, which is structurally related to CD28, not only interacts with CD80 and CD86, but does so with a 100-fold higher affinity. Because of higher affinity, CTLA4 competes with CD28 for binding to CD80 and CD86 and causes down-modulation of T cell responses due to negative signaling and/or lack of continued positive signaling because of competition with CD28 for ligand binding. A defective interaction of CTLA4 with its ligands could result in a generalized defect in the down-modulation of immune responses. Therefore, although the observations on CTLA4 polymorphism are of considerable interest, at this time it is not clear how a defect in this gene could specifically enhance immune response to TSHR and susceptibility to GD.
B. Environmental factors
Factors such as genetic background, age, and gender partially account for the development of GD. As with other autoimmune diseases, an environmental factor has long been suspected in the etiology of GD. If environmental factors such as bacteria are involved, how would such an agent initiate an autoimmune response to a specific self-antigen? Cells other than APCs do not express MHC class-II and thus are not capable of presenting antigens to CD4+ T cells. Similarly, most cells show either little or no expression of costimulatory molecules and thus fail to provide the second signal to antigen-specific T cells, which results in either a failure to activate T cells or induction of anergy. There are a number of potential mechanisms by which an environmental agent could trigger an autoimmune response. These include induction of an inflammatory response leading to the production of proinflammatory cytokines that can cause enhanced/aberrant expression of MHC class-II and costimulatory molecules. This could lead to an otherwise innocuous presentation of self-peptide by MHC molecules into an effective antigen presentation and cause activation of antigen-specific T cells. Thus, cytokine production and/or an imbalance in cytokines caused by infection could lead to the initiation of an autoimmune response. The role of cytokines in GD has been discussed in a recent review (19), and the role of cytokines as it relates to GO will be discussed later. Microbial infections can also cause overexpression (e.g., heat shock proteins) and/or altered expression of certain self-proteins (altered self), which will either provide the necessary strength of signal or be perceived as foreign (20).
Many microbial products can act as either B cell or T cell mitogens and cause nonspecific activation of lymphocytes, including self-reactive lymphocytes. Once self-reactive lymphocytes are activated, they can continue to persist and expand through antigenic stimulation from self-antigens. This is because the strength of antigen presentation required to sustain an anti-self response is lower than that required to initiate an anti-self immune response. A number of bacteria-derived proteins have now been identified as superantigens and linked to Kawasaki disease (21), rheumatoid arthritis (22), and other autoimmune diseases (23). Although the exact mechanism of their action is not known, superantigens have been postulated to facilitate development of autoimmune diseases by: 1) activation of a diverse population of T cells with different fine specificities, including self-reactive T cells; and/or 2) polyclonal activation of B cells, including those that produce autoantibodies. Polyclonal activation of B cells resulting in production of antibody has been shown for Mycoplasma arthriditis mitogen (24) and toxic shock syndrome toxin-1 (25). Production of autoantibodies is thought to be mediated by bridging of superantigen to CD4+ T cells and B cells. Superantigens have been identified from culture supernatants from Yersinia enterocolitica (26) and Y. pseudotuberculosis (27), and these infections have been linked to Reiter’s syndrome, reactive arthritis, GD, and ankylosing spondylitis (28). Although we have a much better understanding of superantigens and the mechanisms by which they activate lymphocytes, we know very little about the role of Yersinia or Yersinia-derived superantigens and/or mitogens in different disease states including GD. It is possible that Yersinia-derived T cell superantigens and B cell mitogens might act independently or in combination to activate T cells and/or B cells, resulting in preferential expansion of B cells recognizing cross-reactive epitopes on TSHR and Yersinia. Such B cells will not only produce autoantibodies reactive with TSHR, but could also serve as efficient APCs that could present TSHR peptides to self-reactive T cells and perpetuate anti-TSHR responses.
Molecular mimicry is one of the most commonly invoked and studied mechanisms for the induction of autoimmunity. A finite probability exists for two proteins sharing common epitopes formed either by a primary sequence of identical amino acids or by noncontiguous amino acids adjacent by virtue of three-dimensional conformation, and resulting in molecular mimicry. After infection, the host mounts a brisk response against the microbe, and this response might become directed against a self-antigen that can be sustained even after the clearance of the microbial agent due to continued antigenic stimulation by the endogenous host protein. This response then can expand due to epitope spreading and eventually cause the disease. Molecular mimicry has been insinuated in other examples of autoimmunity including HLA-B27 and Klebsiella in ankylosing spondylitis (29), myelin basic protein, and mouse hepatitis virus in experimental allergic encephalomyelitis (30), cardiac myosin, and coxsackie virus B4 in myocarditis (31, 32) and others (33). Primary amino acid sequence homology between self-peptides and the foreign antigens as well as conformational similarities have been demonstrated (30, 32, 33).
C. Y. enterocolitica and GD
Due to seasonal trends and geographic variations in the incidence of GD (34, 35), it has been suggested that infectious agents might be involved in triggering the breakdown of tolerance to TSHR in patients with GD. In addition to the reports of increased susceptibility to infections in individuals who fail to secrete certain ABO blood group antigens into the saliva (36, 37), serological evidence for a recent bacterial or viral infection has been reported in a high number of newly diagnosed patients with GD (38).
Y. enterocolitica has been postulated to play a role in the induction of GD via molecular mimicry (39, 40, 41). This speculation is supported by studies reporting the presence of antibodies against Y. enterocolitica in a high proportion of patients (i.e., 72–81%) with autoimmune thyroid disease (42, 43, 44, 45). In addition, Y. enterocolitica has been shown to have binding sites for TSH (46), and antibodies isolated from GD patients could inhibit TSH binding to Yersinia (47) and react with a 64-kDa protein, which was thought to be the bacterial protein to which TSH binds (47). Moreover, antibodies raised against thyroid membranes have been shown to bind Y. enterocolitica (48), and outer membrane proteins (YOPs) encoded by a virulence plasmid of Yersinia have also been linked to the induction of GD (49). Wenzel et al. (50) reported that 72% of patients with GD and 81% of patients with recurrent GD had antibodies to YOPs, whereas only 35% of unmatched controls had such antibodies, and YOPs could induce antibodies in rabbits that recognized components of thyroid epithelial cells. However, data reported by Arscott et al. (51) suggested that there was no unique serological reactivity against YOPs or other Yersinia membrane antigens and that any causal relationship between Yersinia infection and GD may be related to activation of cross-reactive T cells leading to breakdown of self-tolerance to TSHR. Furthermore, Ebner et al. (52) reported that rats immunized with Y. enterocolitica-purified YOPs developed thyroiditis. In this regard, T cell-mediated immunity toward Y. enterocolitica has also been reported in patients with GD (53). Burman et al. (54) reported that TSHR-specific cDNA probe could hybridize to DNA from Y. enterocolitica under low stringency conditions, indicating that they may share stretches of common sequences. However, to date, a causal relationship between Yersinia infection and the development of GD has not been fully established (41, 42).
D. Cross-reactivity of Y. enterocolitica proteins with TSHR
Several years ago, Seetharamaiah et al. (55, 56) expressed the extracellular domain of human TSHR (ETSHR) using a baculovirus expression system. ETSHR protein was purified to homogeneity using HPLC, and upon refolding it could bind TSH (55) and anti-TSHR antibodies in patient sera (56). Mice immunized with ETSHR developed some clinical features characteristic of GD, including high levels of antibodies to the TSHR, with a concomitant elevation in T4 levels (57).
Using this recombinant ETSHR, cross-reactivity of Y. enterocolitica with ETSHR was explored. Initial studies showed that antibodies produced in rabbits and mice to purified ETSHR reacted with envelope preparations from Y. enterocolitica. This cross-reactivity was specifically blocked by purified ETSHR or an envelope preparation of Y. enterocolitica. Moreover, antibodies reactive with ETSHR were readily induced in mice by immunizing with Y. enterocolitica, but not with other Gram-negative or Gram-positive bacteria. These studies provided the first direct evidence that immunization with Y. enterocolitica can lead to the production of antibodies capable of reacting with TSHR (39).
Subsequently, the envelope protein(s) responsible for cross-reactivity to ETSHR was identified using anti-Yersinia antibodies that specifically bound to the ETSHR protein (50-kDa band) on a Western blot. The two low-molecular weight Yersinia envelope proteins (5.5 kDa and 8 kDa) that cross-reacted with the ETSHR, TSHR-cross-reactive protein (TSHR-CRP), were gel purified (40) and used for immunizing CBA/J, C57BL/6 and BALB/c mice. Sera obtained from these mice recognized both ETSHR protein and TSHR-CRP in an ELISA and Western blot (58). Subsequently, it was shown that TSHR-CRP was mitogenic for spleen cells (58) from C3H/HeJ mice, which are lipopolysaccharide nonresponders, and was pronase-sensitive (58). Cell proliferation studies using B cell- and T cell-enriched populations showed that these protein(s) are mitogenic to B cells, and not to T cells, and could induce production of high levels of IL-6 and significant amounts of IgG and IgM by B cells. These results indicate that TSHR-CRPs, which contain epitopes that cross-react with TSHR, represent highly effective B cell mitogens, and the mitogenic activity was not due to lipopolysaccharide contamination. Alignment of peptide sequences showed that TSHR-CRP is homologous to bacterial lipoproteins (LP). Although, all Gram-negative bacteria produce LP, cross-reactivity of Y. enterocolitica LP with TSHR was unique (59). Moreover, LP was capable of inducing high levels of IL-6 and IL-8 production by human B cells. These observations have significant implications for understanding the role of molecular mimicry in the breakdown of self-tolerance and generating autoimmunity to TSHR (59).
In a recent review, Benoist and Mathis (60) raise some important issues to consider with regard to the role of microbial agents in promoting T cell-mediated autoimmunity. Although a number of animal models have provided convincing evidence that a given autoimmune disease can be induced by a particular microbe, they argue that many of the associations are less convincing based upon five different criteria they propose for "air-tight" cases (which are modifications of previously presented concepts). Although they raise some important points with regard to the role of molecular mimicry for us to consider, their perspective is restricted to issues involving T cell epitope recognition and diseases in which T cells, and not B cells, act as effector cells. Moreover, it should be noted that molecular mimicry is only one of several potential mechanisms that a microbe can use to induce autoimmunity. As discussed in Section II.B, microbial agents can cause immunological perturbations through their mitogenic activity, ability to alter self-proteins and/or induce proinflammatory cytokines, and direct up-regulation of MHC gene products and costimulatory molecules (bystander effects). Needless to say, as in all other diseases, including metabolic disorders, cancer, etc., it is not easy to directly demonstrate the role of environmental factors in autoimmunity. As argued by Benoist and Mathis (60), more experimental evidence is needed to fully establish the role of a given microbial agent in the initiation of autoimmunity and autoimmune diseases. With the availability of an animal model, now it might be feasible to investigate the potential immunomodulatory effects of Y. enterocolitica and/or the consequence of molecular mimicry that it exhibits with TSHR on the development of GD.
|
III. Immunological Basis for GD |
|---|
|
TopAbstractI. IntroductionII. Risk Factors for...III. Immunological Basis for...IV. Autoimmune Responses to...V. TSH Receptor as...VI. Evolution of Autoimmune...VII. Graves’...VIII. Genetic and Environmental...IX. Orbital AutoimmunityX. Orbital AutoantigensXI. Role of Orbital...XII. Role of GAGs...XIII. Role of Orbital...XIV. Future PerspectiveReferences |
|---|
Although, imbalances in T cell ratios have been proposed for the regulation of anti-TSHR antibody production (79, 80), other studies have reported contradicting results (81, 82, 83). Because of the unavailability of purified TSHR, the antigen specificity of T cells could not be determined, and thus it is difficult to assess the significance of changes in the total T cell ratios. More recent studies have analyzed the TCR gene utilization in lymphocytes infiltrating the thyroid gland (84, 85). The rationale for these studies was that cells infiltrating the thyroid gland are likely to be thyroid antigen-specific and that utilization of a restricted number of TCR genes by these infiltrating T cells would suggest that a limited number of T cell epitopes are involved. Data from these studies showed that early after onset of GD there is preferential usage of certain TCR variable (V)
genes (84). However, this study and others also showed that there was considerable variation in TCR usage among different patients (84, 86). This is not surprising, considering the fact that T cells recognize antigens in the context of MHC molecules. Because different patients might have different HLAs, they need different TCRs to recognize the antigenic peptide (87). The HLA variation, coupled with the lack of information on antigen specificity of infiltrating cells, makes the interpretation of these data difficult. A putative role for T cells in GO will be discussed later.
Passive transfer of Igs from GD patients to experimental animals causes increased thyroid hormone production (88). However, efforts to induce anti-TSHR antibody response using thyroid membranes have largely failed (89). Because of the low abundance of TSHR on thyroid cells (103 to 104 receptors per cell) (90) as well as instability of the TSHR protein (91), it has been very difficult to purify the TSHR from thyroid membranes. Thus, to date, it has not been feasible to fully evaluate the importance of the idiotypic network for the regulation of TSHR-specific antibody production. Based on studies published thus far, there exists little or no evidence to suggest that an idiotypic network plays a critical role in the regulation of immune responses to TSHR.
Recently, considerable efforts have been made to understand the role of apoptosis in both Hashimoto’s thyroiditis and GD. It appears that thyroid-infiltrating lymphocytes in patients with Hashimito’s thyroiditis express little or no Fas but exhibit high levels of Bcl-2, whereas their thyrocytes show decreased Bcl-2 and increased Fas as well as Fas ligand. This would render thyrocytes susceptible to Fas-mediated apoptosis and facilitate thyroid destruction. However, in contrast, patients with GD show increased Bcl-2 and decreased Fas expression on their thyrocytes, while exhibiting increased Fas and decreased Bcl-2 on their lymphocytes. This would lead to apoptotic death of infiltrating lymphocytes and sparing of thyroid from the effects of inflammatory responses. Recent papers (92, 93, 94) on this topic have discussed the role of apoptosis in thyroid autoimmunity; therefore, we will refrain from a detailed discussion here.
|
IV. Autoimmune Responses to Thyroid Autoantigens |
|---|
|
TopAbstractI. IntroductionII. Risk Factors for...III. Immunological Basis for...IV. Autoimmune Responses to...V. TSH Receptor as...VI. Evolution of Autoimmune...VII. Graves’...VIII. Genetic and Environmental...IX. Orbital AutoimmunityX. Orbital AutoantigensXI. Role of Orbital...XII. Role of GAGs...XIII. Role of Orbital...XIV. Future PerspectiveReferences |
|---|
In 1989, a cDNA-encoding dog TSHR was cloned (95) followed by cloning of human and rat cDNAs (96, 97, 98). This resulted in numerous studies aimed at understanding the structure-function relationship of the protein aimed at the determination of various binding sites for TSH as well as functionally different antibodies (99, 100, 101, 102, 103). Many initial studies used synthetic peptides that were derived from the protein sequence predicted from the cDNA sequence and provided some insights into the potential antibody binding epitopes (99, 100, 101, 102, 103). Initial efforts to express the protein in bacteria resulted in limited yields of protein that was not folded properly. Subsequent studies using an insect cell expression system resulted in higher yields of both glycosylated and nonglycosylated proteins representing the ETSHR. Studies utilizing these protein preparations clearly showed that the ETSHR was sufficient for the TSH and patient autoantibody binding, and that glycosylation was required for autoantibody recognition (104, 105). Although a small fraction of the protein produced in insect cells acquired appropriate conformation, as indicated by TSH and autoantibody binding, this expression system was unsuitable for the production of TSHR with native conformation because the protein remained largely aggregated. Therefore, many studies, aimed at determining the functionally relevant epitopes on TSHR were conducted using two major approaches involving transfection of mammalian cells with various cDNA constructs. One involved construction of TSHR/LH/choriogonadotropin receptor chimeras in which select segments of TSHR were replaced with the corresponding segments from the LH/choriogonadotropin receptor. Studies from this approach resulted in the identification of regions of TSHR that are critical for TSH and blocking and stimulatory antibody binding. Another approach has been to use deletion mutants and/or site-directed mutagenesis to identify the functionally relevant regions of TSHR. These studies revealed that certain glycosylation sites and cysteine residues are absolutely essential for proper folding of the receptor. Because of limitations associated with each of the approaches mentioned above, no single approach has yielded conclusive results on the overall structure-function relationship of the TSHR. However, collectively, these studies have led to several important conclusions. Both TSH and autoantibodies can primarily bind to the ETSHR, and the TSH binding epitope consists of discontinuous amino acids that come together due to protein folding. Antibodies against one or more epitopes can inhibit TSH binding or TSH-mediated activation of cells by affecting a step subsequent to TSH binding. The stimulatory autoantibody binding sites reside predominantly at the N terminus of the protein, whereas the blocking antibodies bind primarily to the C terminus. Glycosylation of the receptor is important for proper folding of the receptor and autoantibody binding. However, it is not clear whether autoantibodies directly interact with the sugar moieties or with epitopes that are formed as a result of proper glycosylation of the receptor protein. A number of excellent reviews (99, 100, 101, 102, 103, 106) have dealt with the structure-function relationship of TSHR; therefore, we will not discuss this issue any further in this paper.
|
V. TSH Receptor as an Autoantigen |
|---|
|
TopAbstractI. IntroductionII. Risk Factors for...III. Immunological Basis for...IV. Autoimmune Responses to...V. TSH Receptor as...VI. Evolution of Autoimmune...VII. Graves’...VIII. Genetic and Environmental...IX. Orbital AutoimmunityX. Orbital AutoantigensXI. Role of Orbital...XII. Role of GAGs...XIII. Role of Orbital...XIV. Future PerspectiveReferences |
|---|
In addition to the subunit structure of TSHR, other important immunological factors must be required to induce an autoimmune response. Although self-reactive T cells might escape negative selection in the thymus, they can be kept in an inactive state through a number of different peripheral tolerance mechanisms (12). For an effective antigen presentation, not only does the antigenic peptide have to be presented by the MHC molecule, but the strength of signal is also very important. This is referred to as the first checkpoint, is demonstrated in animal models of diabetes, and is a consequence of suboptimal presentation of antigens (110, 111). Even if an appropriate signal is delivered, it has to be accompanied by a second signal to activate naive T cells. It is generally accepted that self-peptides do not efficiently activate APCs (i.e., lack of costimulation) and thus can induce T cell anergy rather than activation (111, 112). An even more important mechanism is the engagement of CTLA4, which down-modulates the immune response (113, 114, 115, 116, 117, 118). More recently, another molecule called PD1 (programmed cell death) has been shown to suppress cytokine production through cell cycle arrest and induce T cell anergy (119, 120, 121, 122, 123, 124). Yet another mechanism that can help maintain tolerance in the face of appropriate antigen presentation is phenotypic skewing of helper T cells (125, 126, 127, 128). There are two types of helper T cells, namely Th1 and Th2. Cytokines produced by Th2 cells play a more important role in antibody production than the cytokines produced by Th1 cells. Therefore, if Th1 cells instead of Th2 cells are activated, it is unlikely that it will result in a pathogenic antibody response against the TSHR. If appropriate self-reactive T cells are activated, still they can be deleted by activation-induced cell death (129, 130, 131, 132, 133). Self-antigens are not readily cleared, and therefore they can provide repetitive and continuous stimulation to T cells, causing them to undergo activation-induced cell death through Fas ligation by Fas ligand. Therefore, yet unknown perturbations in the immunoregulatory mechanisms of the host might contribute to the pathogenesis of GD.
|
VI. Evolution of Autoimmune Response to TSHR |
|---|
|
TopAbstractI. IntroductionII. Risk Factors for...III. Immunological Basis for...IV. Autoimmune Responses to...V. TSH Receptor as...VI. Evolution of Autoimmune...VII. Graves’...VIII. Genetic and Environmental...IX. Orbital AutoimmunityX. Orbital AutoantigensXI. Role of Orbital...XII. Role of GAGs...XIII. Role of Orbital...XIV. Future PerspectiveReferences |
|---|
TSHR is a member of the large family of guanine-nucleotide binding (G) protein-coupled receptors. The full-length TSHR cDNA consists of an open reading frame encoding 764 amino acids, including a 21-amino acid signal peptide, a large extracellular domain, seven transmembrane domains, and a short cytoplasmic tail (95, 96, 97, 98). The ETSHR forms the binding site for TSH on the outside surface of the thyroid cell membrane, and the cytoplasmic domain interacts with the regulatory subunits of adenylate cyclase. Initially, high levels of the nonglycosylated ETSHR were produced using a baculovirus expression system (55). This protein was purified to homogeneity using reversed phase HPLC. The purified ETSHR protein was refolded and shown to specifically bind 125I TSH in a dose-dependent manner (56).
A. Early studies on pathogenesis of experimental GD
The insect cell-derived protein was used to successfully produce antibodies capable of blocking TSH binding to native TSHR (55). These studies suggested that the ectodomain was sufficient for TSH binding, albeit at a lower affinity, and for the induction of TSHR-specific antibodies. To identify a strain of mice susceptible to the development of hyperthyroxinemia, C57Bl/6J, BALB/CJ, B10BR/SgSnJ, and SJL/J mice were immunized with ETSHR (57). Sera obtained after five immunizations from different strains of mice were TBII-positive, albeit at different levels of significance (57), whereas only sera collected on d 55 from BALB/CJ mice showed significant elevation in T4 hormone compared with sera obtained from the corresponding control mice. However, the T4 elevation was transient and indicated that immunization with ETSHR had failed to break tolerance to self-TSHR and that immunization using TSHR with native conformation may be required.
To address this, BALB/CJ mice were primed with the ETSHR and then challenged with solubilized thyroid membrane. Relative to controls, immunized mice showed high levels of TBII activity and significantly elevated T4 (57). The [131I] uptake by thyroid glands of test mice was found to be higher than in control mice, suggesting that the elevations in free T4 were not a consequence of gland destruction. Histopathological examination of thyroid glands from hyperthyroid mice showed morphological alterations characteristic of hyperactive glands, including hydropic and subnuclear vacuolar changes, as well as focal scalloping of the colloid within isolated acini. No inflammatory activity, interstitial fibrosis, acinar atrophy, or glandular destruction was found, and this further suggested that the elevated T4 levels were due to hypersecretion. These results showed that ETSHR can prime the animal and allow development of antibodies that can cause hyperthyroidism. The limitation of this study was that the mice became euthyroid within weeks after cessation of TSHR injection, again indicating failure to break tolerance to self-TSHR.
In an attempt to develop an animal model for GD, Marion et al. (134) used affinity-purified human TSHR from a human thyroid cell clone designated GEJ to immunize B10S (H-2s), B10 (H-2b), B10G (H-2q), B10 D2 (H-2d), and B10BR (H-2k) male and female mice. The male and female H-2s, male H-2b, and female H-2q mice exhibited lymphocytic infiltration in their thyroid glands. Immunized mice showed transient marginal elevation in T3. Because no data on serum TBII, TSBAb, and thyroid-stimulating antibody (TSAb) activities were provided, it was not possible to conclude whether these mice developed hypothyroidism or hyperthyroidism.
Other investigators have attempted to develop an animal model for autoimmune GD (hyperthyroidism) and thyroiditis (hypothyroidism) using recombinant TSHR. Costagliola et al. (135) used a fusion protein consisting of ETSHR and maltose binding protein (MBP-ECD). Sera or IgG from immunized mice did not show any thyroid-stimulating activity (TSAb), but showed significant TBII and TSBAb activity and lower serum T4 levels compared with mice immunized with MBP alone. Histological examination of thyroids revealed extensive vascularization and an atypical lymphoblastoid infiltration. In a subsequent study, they extended these observations to male and female mice (136). In another study, they immunized female NOD (H2g), CBA (H2k), and C57BL (H2b) mice with MBP-ECD (137). No significant serum TBII, TSAb, and TSBAb were detected in any of the strains of mice. Although there was no lymphocytic infiltration of the thyroid glands of CBA or C57BL mice, all NOD mice had severe thyroiditis. Based on these studies, the authors concluded that H2d mice develop thyroiditis and TBII/TSBAb, H2g mice develop thyroiditis in the absence of functional TSHR antibodies, whereas H2b and H2k mice are resistant. Furthermore, they showed the ability of TSHR-primed splenocytes to transfer thyroiditis to naive recipients (138). However, neither the antibody properties nor the hormonal perturbations were consistent with findings in GD. In many of these studies, although immunization with MBP alone had no significant effect, the influence of the MBP component of MBP-ECD on the induction of thyroiditis was not evaluated. One possible explanation is that because more recent studies have shown that glycosylated ETSHR is required for stimulatory antibody recognition (and presumably for its induction), antibodies raised against the nonglycosylated ECD produced in bacteria might not have been sufficient to cause GD.
Both BALB/c and NOD mice were immunized with MBP-ECD, and unfractionated T cells or CD4+-enriched cells obtained from these mice were stimulated in vitro with the MBP-ECD protein and then adoptively transferred to the corresponding strain of naive recipients. The recipients showed TBII activity in their sera that persisted for 12 wk. NOD mice showed marginal hypothyroidism, whereas most BALB/c mice were euthyroid with few showing some elevation in T4 levels (138). Approximately two thirds of BALB/c mice showed accumulation of adipose tissue and infiltration of mast cells in the orbital tissue. Collectively, these studies showed that multiple factors determine the outcome of immunization with TSHR, and suggested the polygenic nature of the disease (139).
It is apparent that an effective antibody response (i.e., IgG) against the TSHR protein would require help from T cells that have the same antigen specificity (140, 141, 142). This phenomenon is called linked recognition and would require antigen presentation by professional APCs and activation of antigen-specific T cells before autoantibody production by B cells with specificity for the same antigen. However, it is interesting to note that the epitopes recognized by T cells and B cells need not be part of the same protein but have to be physically linked (140, 141, 142). This is important for subsequent antigen presentation by B cells because these cells will internalize the antigen through receptor-mediated endocytosis, process the antigen, and present specific peptides on their surface leading to subsequent antigen-specific T cell activation. This principle has been exploited in the development of an effective vaccine against Haemophilus influenzae type B. The protective antibodies against this bacterial infection are directed against the bacterial polysaccharides. Adults readily produce such protective antibodies, whereas infants are not as efficient. Therefore, the bacterial polysaccharide is linked to the tetanus toxoid against which the infants mount an effective response. In this case, B cells that bind to the polysaccharides can readily produce antibodies with help from T cells specific for peptides from tetanus toxoid because of the linked recognition. In light of this, what is the consequence of using a self-antigen (TSHR) that is fused to a foreign antigen (MBP) to elicit a response against the self-antigen? It is well known that a foreign antigen is likely to be more immunogenic than a self-antigen such as TSHR. If this were the case, it is likely that the immune response will be largely directed against the MBP. In the absence of a clear demonstration of TSHR specificity of the T cells in MBP-TSHR immunized mice, the possibility of MBP-specific T cells providing help to TSHR-specific B cells, through linked recognition, cannot be ruled out.
Carayanniotis et al. (143) attempted to develop an animal model by immunizing C57Bl/10, B10.BR/SgSn, C3H/He, SJL, DBA/I, and BALB/c mice with an ETSHR protein produced in insect cells. All strains of mice showed strong specific T cell proliferative responses. Sera from SJL and BALB/c mice showed good TBII activity. However, there were no significant changes in serum TSH, T4, and iodine uptake. Authors suggested that their inability to induce hyperthyroidism is most likely due to the inappropriate folding of the ETSHR protein. Vlase et al. (144) immunized female BALB/c and CBA mice with human TSHR-ECD produced in Escherichia coli and in insect cells, respectively. Similar to Carayanniotis et al. (143), these investigators failed to induce either hyperthyroidism or thyroiditis. Subsequently, studies from a number of laboratories have shown that insect cells are not capable of removing the human signal sequence from the protein. Because the cDNA used for protein expression in these two studies encodes for a human signal peptide consisting of 21 largely hydrophobic amino acids at the N terminus, it likely affected the conformation of the protein that may be needed for pathogenic antibody production. In another study, Vlase et al. (145) used refolded ectodomain of mouse TSHR produced in insect cells to immunize BALB/c mice. Immunized mice developed considerable TBII and TSBAb activities and exhibited lower levels of T3, but not T4, and higher levels of TSH.
B. Recent studies on the pathogenesis of experimental GD
Shimojo et al. (146) immunized female AKR/N (H-2K) mice with fibroblasts expressing both MHC class-II and a full-length human TSHR. Most of these mice exhibited good serum TBII activity. About 20% of these mice developed hyperthyroidism as evidenced by significantly elevated serum T4 levels. This was similar to earlier findings of severe hyperthyroidism when BALB/c mice were inoculated with ETSHR followed by thyroid membrane preparations (57). Based on these and other results (146, 147, 148), the authors have suggested that expression of MHC class-II on cells that express TSHR can result in the induction of stimulatory anti-TSHR antibodies in a small proportion of animals. The data presented in this study and an earlier study (57) may also suggest that some of the pathogenic epitopes might be present on the native TSHR. A similar study from another group (149) confirmed these observations by showing that approximately 25% of the immunized animals developed hyperthyroidism and one mouse developed hypothyroidism with elevated TSH levels. Mice with hyperthyroidism showed thyroid hypertrophy without lymphocytic infiltration. When mice were immunized with cells expressing TSHR and MHC class-II along with alum as an adjuvant, they showed increased prevalence of the disease (approximately 45%), with a shorter duration required for the onset. However, when TSHR along with complete Freund’s adjuvant was used, approximately 30% of the animals developed the disease with a slower onset. In this study, no apparent difference was found between male and female mice (149).
Additional studies showed that several different strains of mice with different non-MHC background but expressing H-2K haplotype were able to produce TBII antibodies upon immunization with cells expressing class-II and TSHR proteins (148). Unlike the earlier report with AKR mice (147), these studies showed that TBII activity could be induced in CBA and C3H mice using fibroblasts expressing TSHR without the class-II protein. A subsequent study showed that the N-terminal half of the ectodomain might be sufficient to induce TBII activity in AKR mice, but the entire ectodomain was needed to induce TSAb activity in approximately 20% of the animals. To carry out systematic studies on the molecular mechanisms, a higher proportion of animals need to develop the disease, and that may require a more susceptible strain of mice, different antigen preparation, and/or immunization schedule (150, 151).
Results from studies using fibroblasts stably transfected with human TSHR and MHC class-II have been used to argue that aberrant class-II expression is required for disease induction (146). Although this might be correct, there are no conclusive data to support this conclusion. The contention that aberrant expression of class-II is necessary has been based largely on finding that MHC class-II is expressed in both thyroid and islets of affected patients (152). Subsequent studies showed that expression of class-II molecules or exogenous antigens on ß-islet cells using a rat insulin promoter can result in varying outcomes, including development of spontaneous diabetes, and thus lent support for the initial observation (153). This seems reasonable in a T cell-mediated autoimmune disease that results in the destruction of the target tissue because the effector T cells need to recognize the self-peptide/MHC complex on the target tissue. It is unlikely that such an aberrant expression of class-II is required for the induction of an autoantibody response. This proposition is based on the current understanding of T cell and B cell interactions required for an effective antibody production (Fig. 1
). The initial step in an immune response against a T cell-dependent antigen is phagocytosis of the antigen in the periphery by professional APCs, most likely dendritic cells (DCs). Once they pick up the antigen, DCs migrate and begin processing the antigen in the T cell zone of the regional lymph node. Here, antigen-specific T cells, which enter the lymph node through the afferent lymphatics, are trapped by the DCs with relevant peptide-MHC class-II on their surface. Subsequently, as they circulate through the lymph nodes, antigen-specific B cells are trapped where they are activated by antigen-specific T cells to produce antibodies. Antibodies enter the circulation and react with the TSHR, causing hyperthyroidism. In this scenario, it is likely that the availability of appropriately folded protein to B cells might be more important than the expression of class-II molecules on the cells used for immunization. Moreover, a large number of studies have clearly shown that there is little or no priming of naive T cells in the absence of costimulation. Because these RT cells most likely lacked both adhesion and costimulatory molecules (i.e., B7.2), it is not apparent how these cells could have served as efficient APCs to prime naive T cells required for optimal anti-TSHR responses.
|
-chain and a ß-chain. Each of the chains consists of two domains. RT cells constitutively do not express class-II molecules and are transfected with a cDNA that can encode a chimeric class-II molecule. The -1 and -2 and ß-2 domains of this chimeric class-II were derived from H-2K haplotype, and the ß-1 domain was derived from H-2D (155). These cells were originally constructed to identify peptide binding sites on class-II and were used in vitro as APCs to test for proliferative responses of T cells primed in vivo by direct antigen inoculation. However, it is not clear whether the chimeric class-II molecule expressed on RT cells could have served as an alloantigen (due to ß-1 domain derived from H-2d) in AKR mice (H-2 k haplotype) leading to activation of alloreactive T cells rather than TSHR-specific T cells, that can provide nonspecific help. This might explain a lack of T cell infiltration in thyroids of mice with severe hyperthyroidism. Nevertheless, these studies are very important because they clearly showed that hyperthyroidism can be induced in experimental animals using a recombinant TSHR. Moreover, they showed that the nature of immune response is directly dependent on the antigen and/or the adjuvant used for inoculation. Thus, immunization using TSHR with native conformation might result in the production of stimulatory antibodies that could then cause hyperthyroidism.
Other studies have reported using cDNA vaccination to induce pathogenic antibodies against the TSHR. These studies yielded somewhat contradictory results depending upon the strain of mice used for immunization. BALB/c mice immunized with the cDNA developed antibodies to TSHR (156). Only one mouse contained TSAb, but all mice, irrespective of their autoantibody status, were euthyroid. Examination of the thyroid glands from these mice revealed nondestructive thyroiditis, with B cell infiltration. When a similar experiment was carried out using NMR outbred mice, all mice developed antibodies to TSHR detected by fluorescence-activated cell sorter (157). Nine of 30 males showed mild hypothyroidism with TSBAb and lower T4, whereas four of 30 females showed signs of hyperthyroidism with elevated T3 and T4 levels and TSAb activity with a concomitant decline in TSH levels. Another study showed a heterogeneous response upon vaccination of wild-type and interferon-
(IFN-) -/- BALB/c mice with TSHR cDNA (158). Further refinement of this approach and the use of female mice might lead to the development of an animal model in which a higher proportion of mice develop GD.
In a more recent study (159), different strains of mice were immunized with adenovirus expressing either TSHR or ß-galactosidase three times at 3-wk intervals, and mice were observed for 8 wk after the third immunization. Fifty-five percent of female and 33% of male BALB/c mice and 25% of female C57BL/6 mice developed hyperthyroidism characterized by the presence of TBII and TSAb activity and elevated T4 levels in the sera. Thyroid glands of affected mice showed hypertrophy with no cellular infiltration, and the extraocular muscles were normal.
Another study reported successful induction of Graves’-like disease in nearly 100% of female BALB/c mice using either mouse or human TSHR proteins (160). BALB/c mice, known to generate Th2-type responses, were used because antibody-mediated autoimmune diseases are currently thought to be driven by Th2 helper T cells. Female BALB/c mice were used because of their expected enhanced susceptibility to GD. The protein was either expressed on the surface of syngeneic or xenogeneic cells or produced in a soluble form. Irrespective of the antigen used for immunization, nearly 100% of the mice developed hyperthyroidism characterized by the presence of high levels of TBII, TSAb, T4, and T3. This study was the first to use mouse TSHR to induce the disease in mice. Mice were equally susceptible to immunization with either a full-length human TSHR expressed on 293 human cells (xenogeneic cells that are histoincompatible) or purified ectodomain of the human TSHR. This showed that the cells expressing the TSHR need not act as APCs and that all necessary T and B cell epitopes are present on the ectodomain. Similarly, mice immunized with m12 cells (a B lymphoblastoid cell line with H-2D MHC haplotype, same as that of BALB/c mice) and expressing either mouse or human TSHR developed the disease with a similar time of onset, frequency, and severity. In addition, some groups of mice showed a reduction in body weight and thyroid infiltration of lymphocyte. In this model, development of the disease is a slow and progressive process with antibodies against TSHR that can be detected only by ELISA, appearing at approximately 4–6 wk after initiation of antigen administration. This is followed by the appearance of TBII activity by d 60–80 and TSAb by d 150, with clinical onset of the disease by d 250 or later.
Why do the pathogenic antibodies (i.e., TSAbs) develop late in the immune response and result in a delay in the onset of the disease? Although the answer remains uncertain, it is most likely due to the time required for affinity maturation and/or repertoire spreading eventually leading to the production of pathogenic antibodies. Although the reasons for the delay are not known, we speculate that TSAbs might have to interact with TSHR with a relatively higher affinity, and/or with specific epitopes to mediate their TSH agonist activity. It is well known that affinity maturation is largely dependent upon somatic hypermutation of the Ig genes, which in turn is dependent on DNA replication and cell division. This would imply slow and continuous evolution of anti-TSHR antibody responses toward the production of higher affinity antibodies. Alternatively, these results suggest that antibody specificity might gradually spread from one that is restricted against immunodominant epitopes to include cryptic epitopes. This later phenomenon of epitope spreading has been clearly shown in other autoimmune diseases (161, 162, 163, 164). In this scenario, the initial immune response against a self-antigen is mostly directed against the dominant epitopes (operationally defined) and is not pathogenic. However, as the immune response evolves, it may become directed against pathogenic cryptic epitopes. Should immunodominant epitopes become pathogenic, deleterious consequencess would be anticipated. Therefore, for the survival of the species, it is imperative that the pathogenic epitopes remain cryptic. Efforts are under way to use a receptor preparation that has the stimulatory antibody binding sites but is devoid of the immunodominant regions of the receptor, which are irrelevant for TSAb binding or induction, to determine whether the disease can be induced within a shorter duration. Our unpublished studies suggest repertoire spreading as indicated by the change in the TCR Vß and the TSHR peptide specificity of TSHR reactive T cells during the evolution of the immune response. Further studies are required to determine whether both B cell repertoire spreading and somatic hypermutation contribute to the evolving immune response against the TSHR.
|
VII. Graves’ Ophthalmopathy |
|---|
|
TopAbstractI. IntroductionII. Risk Factors for...III. Immunological Basis for...IV. Autoimmune Responses to...V. TSH Receptor as...VI. Evolution of Autoimmune...VII. Graves’...VIII. Genetic and Environmental...IX. Orbital AutoimmunityX. Orbital AutoantigensXI. Role of Orbital...XII. Role of GAGs...XIII. Role of Orbital...XIV. Future PerspectiveReferences |
|---|
; Ref. 165). Although some patients experience only mild ocular discomfort, 3–5% of patients suffer from intense pain and inflammation with double vision or even loss of vision. In addition, a small percentage of patients with GD have clinically apparent pretibial dermopathy, a diffuse or nodular thickening of the skin on the anterior lower leg (Fig. 3). These skin changes occasionally occur on other parts of the body, often after local trauma. Although eye and pretibial skin changes are clinically obvious in the minority of patients with GD, subtle eye and skin changes can be detected using orbital or dermal ultrasonography, or other sensitive detection techniques, in the majority of patients (166, 167). Conversely, although approximately 10% of patients with GO do not have hyperthyroidism, the majority have laboratory evidence of thyroid autoimmune disease, including the presence of antibodies directed against thyroid antigens such as TSHR or thyroid peroxidase (168). Regardless of whether hyperthyroidism occurs first, the signs and symptoms of GO become manifest in 85% of patients within 18 months (169).
|
|
B. Pathological findings
Many of the clinical symptoms and signs of GO can be explained on a mechanical basis by an increase in the volume of both the orbital fatty connective tissues and the extraocular muscle bodies (170). This expanded tissue volume within the orbit in GO can be measured in computed tomography scans taken of GO patients’ orbits (171). Some patients appear to have enlargement of extraocular muscles as the predominant anatomical change. Others exhibit little eye muscle involvement but show significantly increased orbital adipose tissue volume. Histological examination of extraocular muscles reveals intact muscle fibers that are widely separated by edematous connective tissues (172). These perimysial tissues contain excess glycosaminoglycans (GAGs) that appear composed predominately of hyaluronan and chondroitin sulfate (173). A similar accumulation of GAGs, with more profound adipose tissue expansion, is apparent in the fatty connective tissue compartments of the posterior orbit (Fig. 4; Ref. 174). It remains likely that both the accumulation of GAGs and the expansion of the adipose tissues contribute to the increase in orbital tissue volume characteristic of GO. However, further quantitative studies are needed to reliably partition the net effects imposed by hyaluronan accumulation and de novo adipogenesis in GO.
|
|
