| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
Autoimmune Disease Unit (S.M.M., B.R.), Cedars-Sinai Research Institute and University of California Los Angeles School of Medicine, Los Angeles, California 90048; and Department of Medical Gene Technology (Y.N.), Atomic Bomb Disease Institute, Nagasaki University Graduate School of Biomedical Sciences, Nagasaki 852-85001, Japan
Correspondence: Address all correspondence and requests for reprints to: Sandra M. McLachlan, Cedars-Sinai Medical Center, 8700 Beverly Boulevard, Suite B-131, Los Angeles, California 90048. E-mail: mclachlans{at}cshs.org
 |
Abstract |
|---|
 Top Abstract I. IntroductionII. Perspective on Other...III. Novel Approaches to...IV. TSHR Structure and...V. Antigen PresentationVI. T Cells and...VII. Th1 vs. Th2...VIII. Genetic vs. Environmental...IX. SummaryX. ConclusionsReferences |
|---|
|
I. Introduction |
|---|
|
TopAbstractI. IntroductionII. Perspective on Other...III. Novel Approaches to...IV. TSHR Structure and...V. Antigen PresentationVI. T Cells and...VII. Th1 vs. Th2...VIII. Genetic vs. Environmental...IX. SummaryX. ConclusionsReferences |
|---|
Recently, models have been developed in which a proportion of animals have some of the immunological and endocrinological hallmarks of Graves’ hyperthyroidism. These models have opened the way to investigating critical issues involved in Graves’ disease, as well as to exploring approaches for immune intervention in the future. Before the models are analyzed, background information is provided on the characteristics of Graves’ disease and the structure of the TSHR, as well as a brief overview of the interactions between immune cells leading to production of antibodies.
A. Clinical and immunological characteristics of Graves’ disease
The clinical features of Graves’ disease include weight loss, hyperkinesis, tachycardia, diffuse goiter, increased levels of serum T4 and/or T3, and suppressed TSH. Other common manifestations are heat intolerance and anxiety and, in women, oligomenorrhea. The immunological hallmark of Graves’ disease (as already mentioned) is the presence of IgG class TSHR autoantibodies [thyroid-stimulating antibodies (TSAbs)] that stimulate thyroid hormone production (reviewed in Refs. 3, 4, 5, 6, 7). Two assays are currently in clinical use for TSHR autoantibodies: 1) inhibition of TSH binding to the receptor [TSH binding inhibition (TBI)]; and 2) a bioassay for TSAb activity measured by a functional response [usually cAMP production by TSHR-expressing cells (thyroid cells or transfected eukaryotic cells)].
In addition to TSHR autoantibodies, approximately 75% of Graves’ patients have autoantibodies to thyroid peroxidase (TPO) (8) and, depending on the assay, 25–55% have autoantibodies to thyroglobulin (Tg) (9). These autoantibodies are more prevalent in Hashimoto’s thyroiditis (reviewed in Ref.10) and, at least for TPO autoantibodies, reflect the underlying thyroid lymphocytic infiltration (e.g., Refs. 11 and 12). Consistent with TPO and Tg autoantibodies, thyroid inflammation in Graves’ disease, indicated by infiltrating T and B lymphocytes and plasma cells, is much less extensive than in Hashimoto’s disease (reviewed in Ref.4). Although recent data confirm the dominant role of T cells rather than antibodies in mediating thyroid destruction and hypothyroidism (13), TPO (and to a lesser extent Tg) autoantibodies remain excellent clinical markers of this process. The typical coexistence of low level thyroiditis with Graves’ hyperthyroidism indicates that thyroid stimulation by TSHR autoantibodies overcomes any thyroid damage associated with thyroid inflammation. In a minority of patients, TSHR autoantibodies that block the stimulatory effects of TSH [TSH blocking antibodies (TBAbs)] give rise to hypothyroidism (e.g., Ref.14).
B. Extrathyroidal manifestations of Graves’ disease
Graves’ ophthalmopathy (GO) develops in 25–50% of Graves’ patients with symptoms including conjunctival injection, chemosis, proptosis, and diplopia (reviewed in Refs. 15 and 16). Although fortunately mild in most cases, in severe cases of GO sight loss may occur consequent to corneal ulceration or optic nerve compression. A subset of Graves’ patients (1–4%), almost always with concomitant GO, develop skin induration. Dermopathy typically affects the pretibial areas [pretibial myxedema (PTM)] and, even more rarely, thickening of the distal phalanges of the hand (acropachy) (reviewed in Refs. 16 and 17). Both GO and PTM are characterized by lymphocytic infiltration of the target tissues, activation of fibroblasts/preadipocytes, glycosaminoglycan accumulation, expansion of fat and, in the orbit, thickening of the extraocular muscles.
The etiology of these distressing extrathyroidal conditions is gradually being elucidated. Because of their association with Graves’ disease, GO and PTM have long been considered to develop as a consequence of autoimmunity to cross-reacting thyroid and orbital autoantigens. Candidate autoantigens included Tg and novel muscle proteins G2 s and D1 (reviewed in Ref.16). However, increasing evidence supports a role for the TSHR: 1) TSHR antibody levels correlate with clinical GO (18); 2) the rare occurrence of GO and PTM in patients with autoimmune hypothyroidism has been associated with extremely high TSHR antibody levels, probably with TBAb activity (19); and 3) multiple studies (many ongoing) regarding expression and function of the TSHR in Graves’ and normal orbital and dermal tissues (e.g., Refs. 20, 21, 22, 23, 24, 25)
Against this background, it seems increasingly likely that TSHR-specific T cells are recruited to the orbit (or skin) and activated to secrete cytokines that induce preadipocyte differentiation, fibroblast proliferation, and glycosaminoglycan production (reviewed in Refs. 15 and 16). Additional local factors, such as dependency, skin trauma or pressure, cigarette smoking, and the anatomical constraints of the bony orbit, play important roles in the manifestations of PTM (26) and GO (reviewed in Refs. 15 and 16). Animal models provide the opportunity to test the role of the TSHR as the major autoantigen in these diseases. Moreover, because of limited available therapeutic options, such models would be invaluable for developing novel immunospecific therapies for GO and PTM.
C. Structural features of the TSHR
The TSHR is a member of the G protein-coupled receptor family with seven transmembrane regions. Like the related gonadotropin (LH/choriogonadotropin and FSH) receptors, the TSHR has a large N-terminal ectodomain (397 amino acid residues). However, the TSHR differs from the gonadotropin receptors in that some of the single-chain polypeptide expressed on the cell surface undergoes intramolecular cleavage to form two subunits (A and B) linked by disulfide bonds (Fig. 1
, left panel) (27, 28). Remarkably, the process involves the removal of a looped segment between the A and B subunits (C peptide region; approximately amino acid residues 317–366) (29, 30). The C peptide region is not removed as an intact fragment but by a process of progressive degradation, starting at the upstream cleavage site and continuing downstream (30, 31). Moreover, the cleaved receptor is susceptible to loss of the A subunit by shedding (Fig. 1, right panel), at least in vitro. Suggested mechanisms for shedding include reduction of disulfide bonds linking the A and B subunits (32) as well as continued proteolytic erosion at the N terminus of the B subunit (33). In addition to receptor shedding, another important feature of the TSHR A-subunit is its high degree of glycosylation, about 45% of its mass (34, 35).
|
). Most antibodies to protein antigens recognize conformational epitopes comprising nonlinear amino acid segments that are contiguous with each other in the folded protein. In contrast, T cells interact with antigen that has been degraded into short linear peptides. Cells that present antigen to activate T cells, macrophages, and dendritic cells express major histocompatibility complex (MHC) molecules of class II. APCs internalize antigens (particles or proteins) by phagocytosis or pinocytosis. After internalization, the antigens are processed by proteolysis to produce peptides, 16–20 amino acids long, which bind to the MHC class II groove for presentation to T cells.
|
), some of which are not expressed on resting APCs or T cells (reviewed in Ref.36). Briefly, unprimed APCs express MHC and CD40; naive T cells express the T cell receptor complex and CD28. Peptide presentation by APCs to a T cell (signal 1) up-regulates CD40 ligand (CD40-L) expression on T cells. The interaction between CD40-L (T cell) and CD40 (APC) induces expression of B7 molecules (B7–1/B7–2) on the APC. Binding of B7 with CD28 (T cell) delivers signal 2. CD28 also shares its B7 ligands with CTLA-4 (usually associated with down-regulation). The combination of signal 1 and signal 2 leads to T cell activation. Immune responses can be blocked by inhibition of second signal using antibodies or soluble receptors, or by the absence of costimulatory molecules in knockout mice (reviewed in Ref.37). Other costimulatory molecules, such as inducible costimulator (ICOS) and its ligand, are involved in T cell activation (38). The interaction between T cells and APCs, termed the "immunological synapse," determines the outcome (magnitude and nature) of the T cell response (reviewed in Ref.39).
Infectious organisms induce APCs to secrete proinflammatory cytokines that potentiate antigen uptake, processing, and presentation to T cells. T cells respond by producing IL-2, a growth factor required for T cell survival and proliferation. Multiple cytokines are involved in T and B cell maturation and differentiation. Interferon (IFN)
and IL-12, on the one hand, and IL-4, on the other hand, have mutually exclusive effects on the development of two major T cell subsets, T helper 1 (Th1) and T helper 2 (Th2) (40). Th1 and Th2 cytokines were initially considered to be exclusively associated with cellular and humoral immune responses, respectively. However, cytokines from both subsets are involved in antibody production and are responsible for switching between subclasses of IgG subclass secreted by B cells. For example, in humans, the Th1 cytokine IFN is associated with secretion of IgG1, whereas the Th2 cytokine IL-4 is required for production of IgG4 and IgE (reviewed in Ref.41).
|
II. Perspective on Other Models of Thyroid Autoimmunity |
|---|
|
TopAbstractI. IntroductionII. Perspective on Other...III. Novel Approaches to...IV. TSHR Structure and...V. Antigen PresentationVI. T Cells and...VII. Th1 vs. Th2...VIII. Genetic vs. Environmental...IX. SummaryX. ConclusionsReferences |
|---|
The lack of a spontaneous Graves’ disease model can be appreciated when considering the insights NOD mice have provided into the process leading to type I diabetes (reviewed in Ref.44). Nevertheless, there are discrepancies between autoimmune diabetes in humans and NOD mice. For example, unlike the lack of a gender difference in humans, type I diabetes is more prevalent in female than male NOD mice. Moreover, whereas numerous approaches prevent and cure diabetes in NOD mice, reversing islet inflammation and destruction is extremely difficult in humans. Consequently, for understanding disease pathogenesis and for developing novel immunotherapies in Graves’ hyperthyroidism, appropriate animal models should exhibit the distinctive endocrinological and immunological features of spontaneous disease in humans.
A. Induction of thyroiditis
The conventional approach for inducing autoimmunity involves immunizing animals with protein (antigen) and adjuvant. The classic example is thyroiditis induced in rabbits by injecting rabbit thyroid extracts together with complete Freund’s adjuvant (CFA) (45). Studies of mice immunized with human or murine Tg demonstrated a role for MHC antigens and cytotoxic T cells (but not antibodies) in the development of thyroiditis (reviewed in Ref.46). Conventional immunization with TPO induces thyroiditis in some mouse strains (47), particularly using murine TPO (48). However, unlike spontaneous thyroiditis in chickens (reviewed in Ref.49), hypothyroidism does not develop in induced thyroiditis models. Recently, transgenic mice were generated that express the T cell receptor genes of a TPO-specific human T cell clone; these mice spontaneously develop severe thyroiditis leading to hypothyroidism and weight gain (13).
B. Conventional immunization with TSHR protein and adjuvant
After the cloning of the TSHR, numerous attempts were made to induce Graves’ hyperthyroidism by conventional approaches. Rabbits and mice immunized with human TSHR expressed in bacteria and/or in insect cells developed antibodies that reacted with receptor preparations (e.g., Refs. 50, 51, 52, 53). Serum antibodies and murine monoclonal antibodies from these immunized animals provided invaluable reagents for immunohistochemistry, Western blotting, and immunoprecipitation. Nevertheless, despite using different TSHR preparations, a variety of mouse strains and different adjuvants, none of these approaches induced antibodies with TSAb activity as in Graves’ patients (summarized in Table 1). Because of the possibility that the human TSHR may be unsuitable for inducing stimulatory autoantibodies and hyperthyroidism in mice, the murine TSHR ectodomain was cloned and expressed in insect cells (54, 55). Again, even when using purified murine TSHR ectodomain and adjuvant, conventional immunization failed to induce Graves’-like disease in mice.
|
Because SCID mice lack mature T and B cells, both human tissue xenografts and infiltrating lymphocytes survive in these recipients (reviewed in Refs. 56 and 61). Autoantibodies to Tg and TPO develop in SCID mice engrafted with Graves’ or Hashimoto blood lymphocytes, thyroid lymphocytes, or thyroid tissue (62, 63, 64, 65, 66). Pitfalls with SCID mouse models include the loss of thyroid autoantibodies 8–10 wk after engraftment, probably because of a rapid decline in T cell function (61), as well as variability between individual animals (62, 67). Nevertheless, SCID mice opened the way to studying several aspects of thyroid autoimmunity such as lymphocyte homing (68), regulatory (CD8+) cells (69), and T cell characterization in thyroid organoids (70).
Some mice xenografted with Graves’ thyroid tissue developed TSAb activity and transient hyperthyroxinemia (66). Moreover, mice that received TSHR-specific T cell lines together with thyroid grafts developed TSAb activity, and the thyroid grafts increased in size, but serum T3 levels were unchanged (71). Simultaneous xenotransplantation of Graves’ thyroid tissue and autologous bone marrow cells induced higher TSAb titers and elevated T4 levels (72). However, wider application of these approaches is restricted by the need to develop T cell lines before surgery, MHC matching of T cell lines and grafts (71), and the difficulty of obtaining bone marrow cells.
|
III. Novel Approaches to Induce Graves’ Disease |
|---|
|
TopAbstractI. IntroductionII. Perspective on Other...III. Novel Approaches to...IV. TSHR Structure and...V. Antigen PresentationVI. T Cells and...VII. Th1 vs. Th2...VIII. Genetic vs. Environmental...IX. SummaryX. ConclusionsReferences |
|---|
A. Cells stably expressing the TSHR
Graves’ disease models using this approach have used different cell types (Fig. 3).
|
2. Hamster Shimojo model.
Outbred hamsters repeatedly injected with TSHR-expressing Chinese hamster ovary (CHO) cells together with the adjuvants alum and pertussis toxin developed TSHR antibodies including TBI activity. Moreover, 30% of animals had elevated T4 levels and goiters as well as thyroid lymphocytic infiltration (78). Of interest, mRNA for MHC class II was detectable in the CHO cells by PCR. Unlike inbred mice, however, the hamsters resemble human populations in being genetically different from one another. Consequently, it is possible that MHC differences between individual hamsters and the injected TSHR-expressing CHO cell line play a role in stimulating allogeneic responses.
3. B cells and human embryonic kidney cells.
Murine B cells (M12 cells) stably expressing the TSH holoreceptor (human or mouse) were injected on multiple occasions into BALB/c mice, which have the same MHC (H2-d) as the M12 line (79). In a second approach, mice of the same strain were injected with xenogeneic human embryonic kidney (HEK) 293 cells expressing the TSHR ectodomain, alone or together with soluble TSHR ectodomain protein and a Th2 adjuvant, cholera toxin B. In both approaches, mice had TSHR antibodies detectable by ELISA after 1 month and TBI activity after 4 months. TSAbs, hyperthyroidism, and goiter developed by 5–6 months, followed later by focal necrosis and thyroid lymphocytic inflammation. Human TSHR and mouse TSHR were equally effective at inducing disease. However, as discussed later, the thyroid histology in these models (79) does not correspond to that in Graves’ disease. Using TSHR-expressing B cells or HEK293 cells, disease incidence was first reported to be 100% (79). However, in a subsequent study, only 50% of mice became hyperthyroid using TSHR-HEK293 cells (80). Of interest, immunization with TSHR-ectodomain protein and cholera toxin B was equally effective (79). These findings represent the sole report of Graves’ disease induction not involving in vivo expression of the TSHR.
4. Insight into the Shimojo approach.
The success of the Shimojo model for inducing TSAbs poses an intriguing question: How does the immune response differ when the antigen is presented as a cell-associated molecule vs. a soluble, purified protein? This question was addressed for TPO because of the availability of native human TPO and a panel of human monoclonal TPO autoantibodies that define the immunodominant region (IDR) recognized by patients’ autoantibodies (81).
The qualitative nature of induced antibodies was assessed by comparison of AKR/N mice injected with fibroblasts coexpressing TPO and MHC class II with mice immunized with purified TPO and adjuvant. Only TPO-fibroblast-injected mice developed antibodies that resembled human TPO autoantibodies in terms of their high affinity [dissociation constant (Kd) 
10–10 M] and restricted epitopic recognition of the TPO IDR (82). To date, no similar comparison has been made for TSHR antibodies, namely animals injected with TSHR and MHC class II-positive fibroblasts vs. the same mouse strain immunized with TSHR protein plus adjuvant. Nevertheless, these findings for TPO suggest that TSHR antibodies generated by the Shimojo protocol may have restricted epitopes and higher affinities compared with TSHR antibodies induced by conventional immunization that do not induce Graves’-like hyperthyroidism.
B. Transient TSHR expression
In addition to injection of stably transfected cells, there has been much recent interest in inducing Graves’ hyperthyroidism by immunization of plasmid or adenovirus vectors with transient in vivo TSHR expression (Fig. 4).
|
Subsequently, three groups were unable to reproduce these findings: TSHR antibody levels were low and thyroiditis was not observed in female BALB/c mice vaccinated im with TSHR-DNA (77, 85, 86). The lack of antibody responses and thyroiditis could be explained by one or more of the following factors: different immunization protocols (single vs. multiple immunization sites, with or without cardiotoxin pretreatment); subtle differences between substrains bred separately for many years (BALB/cJ vs. BALB/cAnCrlBR); and differences in animal housing (conventional vs. specific pathogen free). A recent study emphasizes that the same protocol used to vaccinate BALB/cAnCrlBR mice had variable outcomes in animals maintained in different conventional housing units (presumably exposed to different microorganisms) and fed different diets. Thus, genetic immunization in Brussels induced thyroiditis and orbital changes but no TSAbs (87), whereas the same TSHR-DNA vaccination approach in Cardiff induced TSHR antibodies and TSAbs but no thyroiditis (88).
Modified DNA vaccination protocols induce TSHR antibodies and, to a lesser extent, hyperthyroidism in female BALB/c mice. Intradermal, rather than im, injection of TSHR plasmid DNA induced TSHR antibodies in 70% of mice, and 30% had elevated T4 levels (89). Moreover, as described in more detail later (Section V), by diverting plasmid TSHR expression to the lysosome, the majority of BALB/c mice developed TSHR antibodies and about 25% became hyperthyroid (90).
2. DNA vaccination of other strains.
In outbred mice housed in a conventional facility, 25% of females developed TSAbs, hyperthyroidism, and goiter 2–4 wk after the third TSHR-DNA vaccination (91). Among 30 male mice, only one developed subclinical hyperthyroidism, and one was hypothyroid. As in BALB/c mice (see above), thyroid lymphocytic infiltrates (that were phenotyped) developed in genetically immunized outbred mice (91).
In addition to outbred mice, im genetic immunization induces TSHR antibodies in other non-BALB/c strains. C57BL/6 mice maintained in pathogen-free conditions developed TSHR antibodies (but not hyperthyroidism) after vaccination with TSHR-DNA (92). Of particular interest are studies performed in mice lacking murine MHC class II and, instead, expressing the allele HLA-DR3 (DRB1*0301) associated with Graves’ disease in Caucasians (reviewed in Ref.93). Some HLA-DR3 transgenic mice on a mixed background (50% C57BL/10 and 50% CBA, C57BL/6 and 129) developed TSHR antibodies, but all were euthyroid (94). However, TSHR-DNA vaccination induced TSHR antibodies and hyperthyroidism in 30% of HLA-DR3 transgenics on the NOD background (95). These observations, together with those described for BALB/c mice above, indicate the importance of genetic and environmental factors in the outcome of TSHR-DNA vaccination.
3. Adenovirus encoding the TSHR.
Intramuscular injection of a replication-deficient adenovirus vector encoding the human TSHR-cDNA is an efficient approach for inducing Graves’-like hyperthyroidism in mice. In the original report, after three injections of TSHR-adenovirus, most BALB/c mice developed TSHR antibodies detectable by TBI, and 50% had TSAb activity and became thyrotoxic (86). In contrast, only 25% of C57BL/6 mice became hyperthyroid and DBA/1J, DBA/2J, CBA/J, and SJL/J mice remained euthyroid (86, 96). Importantly, as determined using chimeric TSH-LH receptors for detection, TSHR antibodies in BALB/c mice recognized epitopes similar to autoantibodies in Graves’ sera. Thyroid glands from hyperthyroid animals were hyperplastic, but there was no lymphocytic infiltration. This adenovirus model has been applied to hamsters (78). Moreover, unlike the marked interlaboratory variability of TSHR-DNA vaccination, the TSHR-adenovirus approach has been confirmed in two laboratories (97, 98). Finally, as will be described later (Section IV), the TSHR-adenovirus model has been modified and optimized (97, 99).
4. Dendritic cells transfected with TSHR-adenovirus.
Dendritic cells are the most potent APCs and are a prerequisite for the initiation of immune responses. Instead of injecting the adenovirus im, dendritic cells were isolated from mouse bone marrow, expanded using the cytokines IL-4 and granulocyte macrophage colony stimulating factor, and then infected with TSHR-adenovirus. After two to three injections of TSHR-expressing dendritic cells, 70% of BALB/c mice produced TSHR antibodies detectable by TBI, and 35% had TSAb activity, elevated serum T4, and goiter (100). Again, no thyroid lymphocytic infiltration was detected. This immunization protocol demonstrated the efficacy of dendritic cell presentation of TSHR to naive T cells.
C. Mice transgenic for a monoclonal TSAb (mTSAb)
Two human monoclonal antibodies (B6B7 and 101–2) isolated by Epstein-Barr virus transformation of Graves’ peripheral blood lymphocytes had weak TSAb activity when used at high concentrations (23 µg IgG/ml of B6B7) (101). The Ig genes encoding the heavy and light variable regions of these antibodies were determined (102), expressed in eukaryotic vectors (101), and their epitopes have been analyzed (103).
Recently, transgenic mice were generated with the variable region genes of antibody B6B7 expressed together with the constant region of human IgM (104). Hyperthyroidism developed in 68% of these transgenic mice, as reflected in elevated serum levels of T4 and reduced TSH, as well as increased thyroid uptake of technetium pertechnetate. Thyroid tissue in these mice was hyperplastic but lymphocytic infiltration was absent. T4 levels correlated positively with the level of human IgM in serum, demonstrating that hyperthyroidism was determined by the TSAb concentration. It was assumed that the isotype change (IgG to IgM) would not influence the antibody activity. However, the IgM pentamer of B6B7 likely converts the low-affinity monomeric parent IgG to a high-avidity antibody that, at high concentrations, induces hyperthyroidism. These transgenic mice provide a valuable model for studying B cell tolerance to the TSHR (Section VI).
D. Models of Graves’ ophthalmopathy
A model of GO was developed involving the adoptive transfer of splenocytes from immunized mice (87). This protocol was based on the previous induction of thyroiditis in naive BALB/c mice by transfer of splenocytes (activated in vitro with antigen) from BALB/c mice immunized conventionally with TSHR plus fusion protein (105). Building on the hypothesis that GO is caused by TSHR-specific T cells, BALB/c and NOD mice were genetically immunized with TSHR-DNA or (as before) with a TSHR fusion protein and adjuvant. Splenocytes ex vivo were cultured with the TSHR-fusion protein and injected into unimmunized recipients. These studies confirmed that thyroiditis could be transferred to both mouse strains. Moreover, orbital pathology comparable to that in humans, developed in BALB/c (but not in NOD) mice. These orbital changes included accumulation of adipose tissue, edema, dissociation of muscle fibers, TSHR immunoreactivity, and infiltration by lymphocytes and mast cells (87). Similar findings were reported in BALB/c and outbred (CD-1) mice genetically immunized against the TSHR or the eye muscle protein G2 s (106). In contrast to these studies, no extraocular muscle abnormalities have been detected in other models of Graves’ hyperthyroidism, e.g., the TSHR-adenovirus model (86).
The splenocyte adoptive-transfer model appeared to open the way to further studies of GO. However, despite using the same protocol, these findings have not been reproduced in the same BALB/c substrain in a different location (Cardiff vs. Brussels) (88). Unexpectedly, no thyroiditis or orbital changes were observed, even using the same water, bedding, and food obtained from Brussels. Moreover, misleading artifacts were noted in extraocular muscles as well as misinterpretation of ectopic thymus as thyroid lymphocytic infiltration (88). These findings emphasize the difficulty of reproducing models in conventional (non-pathogen-free) housing.
An unexpected observation was made in transgenic mice overexpressing adiponectin, a protein secreted by adipocytes. Transgenic animals were generated to investigate the long-term effects of elevated adiponectin on insulin sensitivity. Remodeling of fat depots in older mice (12 months) led to selective enlargement of the interscapular and orbital fat pads. Proliferation of orbital fat pushed the orbit away from its bony structure and created marked proptosis, keratopathy, and, ultimately, corneal ulceration. Individual orbital muscles were separated by the expansion of adipose tissue. Despite the potential implications for GO, orbital tissue in these animals had no lymphocytic infiltration, and TSHR antibodies were undetectable in multiple assays (binding to TSHR-expressing cells, TBI and TSAbs) (107). Although not a definitive model of autoimmune ophthalmopathy, these transgenic mice provide the opportunity to study noninflammatory orbital fat proliferation that may be a component of GO in humans.
E. Overview of mouse models for Graves’ hyperthyroidism
1. Summary.
In 1952, Rose and Witebsky (108) suggested criteria for establishing the autoimmune etiology of an autoimmune disease. The presence of autoantibodies to a defined antigen (the TSHR) have long been recognized to fulfill two of the Rose-Witebsky postulates, and direct proof was provided by transplacental transfer of TSHR antibodies leading to neonatal Graves’ disease (e.g., Refs. 109 and 110). However, the final hurdle to the Rose-Witebsky criteria, namely that "an analogous autoimmune response be identified in an experimental animal" (111), was only overcome in 1996 when the Shimojo model of Graves’ hyperthyroidism (73) was described.
Since that time, a number of protocols have been tested to induce Graves’ disease in animals (summarized in Table 2). These studies have generated highly diverse observations, ranging from the detection of TSAbs without alteration in thyroid hormone levels to TSAbs associated with increased serum T4 and T3 levels with reciprocal suppression of TSH. Essential clinical and immunological features for a useful Graves’ disease model are thyrotoxicosis, goiter, and TSAbs. Moreover, as in human disease, chronically overstimulated thyroid epithelial cells are cuboidal or columnar with increased intracellular vacuolation. These findings, consistent with high secretory activity, contrast with flattened thyroid epithelium in euthyroid animals (Fig. 5, B vs. A). Of note, in the approach using TSHR-expressing B cells or HEK293 cells, thyroid histology involves "hypertrophy and enlargement of colloid with thinning of the thyroid epithelium" (79), an appearance inconsistent with other mouse models of Graves’ disease, as well as with human Graves’ disease.
|
|
Observations can be categorized into three groups. 1) Thyroid lymphocytic infiltrates are absent in mice made hyperthyroid using the Shimojo approach (73, 75), following intradermal vaccination with TSHR-DNA (89), injection of TSHR-adenovirus-transfected dendritic cells (100), and im injection of TSHR-adenovirus (86, 97). 2) Thyroiditis without hyperthyroidism has been reported in TSHR-DNA-vaccinated BALB/c mice in one study (84) [but not in three other studies (77, 88, 112)], as well as in HLA-DR3 transgenic mice on a mixed C57BL/10 background (94) (Fig. 5, panel D vs. panel C). 3) Both thyroiditis and hyperthyroidism are reported after im TSHR-DNA vaccination of outbred mice (91) and HLA-DR3 transgenics on the NOD background (95), as well as after injection of TSHR-expressing B cells or HEK293 cells in BALB/c mice (79). It should be noted that assessing thyroid lymphocytic infiltration in mice is complicated by the frequent occurrence of ectopic thymus, and the proximity of parathyroid glands, which may be mistaken for dense lymphocytic infiltrates (113). Indeed, as noted above, a recent study points to possible misidentification of thyroid lymphocytic infiltrates in some previous reports (88).
In Graves’ disease, focal thyroiditis is often present, but lymphocytic infiltration is less extensive than in Hashimoto’s thyroiditis (114, 115). As discussed above, autoantibodies to other thyroid antigens, particularly TPO, are also present in many Graves’ patients. An unanswered question, therefore, is whether thyroiditis in human Graves’ disease is more closely associated with TPO (and perhaps Tg) autoantibodies than with autoimmunity to the TSHR. The clinical syndrome of Graves’ disease is dependent on a humoral immune response and not thyroid lymphocytic infiltration. Transient neonatal Graves’ disease arises from transplacental transfer of maternal TSAbs to the fetus (e.g., Ref.109) without the requirement for lymphocytic involvement in the thyroid. Also, in thyroiditis-prone NOD.H-2h4 mice, TSHR-adenovirus immunization induced TSHR antibodies and hyperthyroidism without enhancing thyroiditis or autoantibodies to Tg (116). These data support the possibility that the variable thyroiditis component in Graves’ patients involves autoimmunity to TPO (and perhaps also to Tg) rather than to the TSHR. On the other hand, a cell-mediated immune response to the TSHR may occur at a later stage of the disease process. Such a late event would be compatible with the long time course required for development of thyroiditis, TSAbs, and hyperthyroidism in the B cell/HEK293 cell model for Graves’ disease (79).
3. Advantages and disadvantages of different mouse models.
The Shimojo approach provides an in vivo model with many features of Graves’ disease, including TSHR antibodies with TSAb activity and elevated T4 in 25% of mice (73, 75, 76), TSAb epitopes like those in Graves’ patients (117), and a role for genetic factors other than MHC class II (74) (see below). One limitation is restriction to a particular MHC (H2-k). More important, because RT4.15HP fibroblasts constitutively express a costimulatory molecule (B7–1; Fig. 2B), they induce a potent, nonspecific activation of antibodies, T cells, and cytokine production (118). These responses preclude detailed in vitro dissection of the cellular immune mechanisms in these animals. On the positive side, however, TSAbs and hyperthyroidism develop within approximately 3 months, a shorter time frame than the 6 months required for Graves’ disease induction with TSHR-expressing B cells (79).
The reason for the greater efficacy of TSHR-expressing B cells vs. TSHR-expressing fibroblasts (incidence of hyperthyroidism 100% vs. 25%, respectively) is not known but may relate to different mouse strains, BALB/c vs. AKR/N, respectively. Both B cells and fibroblasts express MHC class II as well as the B-7 molecule required for T cell activation (79, 118). However, other molecules present on B cells (but not fibroblasts) may provide additional activation signals for T cells. Because of the differences between human and mouse MHC, it is not clear why injecting mice with TSHR-expressing human HEK293 cells is as effective as injecting TSHR-expressing murine B cells for inducing Graves’ hyperthyroidism. Of interest, these studies with HEK293 cells showed that the TSHR ectodomain alone, without the serpentine region of the receptor, is sufficient to induce TSHR antibodies resembling those in Graves’ disease (79).
Intramuscular vaccination with TSHR-DNA in a plasmid is less effective than injecting TSHR-expressing fibroblasts or B cells for inducing TSHR antibodies and hyperthyroidism, at least in the inbred BALB/c strain. The problems associated with TSHR-DNA vaccination include lack of reproducibility in different laboratories (Ref. 84 vs. Refs. 77 ,85 ,86 , and 88) and the likely effects of conventional vs. pathogen-free housing facilities. However, DNA vaccination has greater power than the Shimojo model for studying a variety of mouse strains and, as described later (Section VI), permits in vitro analysis of memory T cell responses to the TSHR (85).
TSHR-adenovirus immunization combines the advantages of naked DNA vaccination without the disadvantages of the Shimojo approach. Mice of any strain can be immunized with TSHR-adenovirus to investigate induction of TSHR antibodies and hyperthyroidism. Moreover, the induction of TSHR antibodies, goiter, hyperthyroidism, and thyroid hyperplasia in BALB/c mice has been reproduced in three different laboratories (86, 97, 98). Also, unlike the Shimojo approach, splenocytes from TSHR-adenovirus-immunized mice can be used to analyze memory T cell responses (119) (Section VI).
|
IV. TSHR Structure and Antibody Epitopes |
|---|
|
TopAbstractI. IntroductionII. Perspective on Other...III. Novel Approaches to...IV. TSHR Structure and...V. Antigen PresentationVI. T Cells and...VII. Th1 vs. Th2...VIII. Genetic vs. Environmental...IX. SummaryX. ConclusionsReferences |
|---|
A. TSHR shedding and induction of Graves’ disease
As mentioned above, some TSHRs on the cell surface undergo intramolecular cleavage to form disulfide-linked A and B subunits, a process that may be followed by A subunit shedding (Fig. 1, right panel). In addition, the epitope for TSAbs in Graves’ disease is partially obscured in the wild-type TSHR but is exposed on the same TSHR ectodomain tethered to the cell surface by a glycosylphosphatidylinositol anchor (120). These observations raised the possibility that the shed A subunit, rather than the full-length, cell surface TSHR, initiates or enhances immune responses to the TSHR that lead to hyperthyroidism.
This hypothesis was tested in the TSHR-adenovirus Graves’ model utilizing two different forms of the TSHR. If correct, immunization with the free A subunit, but not a TSHR engineered to prevent cleavage and A subunit shedding (Fig. 6A), should preferentially generate TSAbs with consequent hyperthyroidism. Indeed, this hypothesis was confirmed. BALB/c mice immunized with A subunit adenovirus, unlike animals injected with noncleaving TSHR-adenovirus, developed elevated T4 levels (Fig. 6B) (97). Remarkably, however, TBI levels were not significantly different between the two groups (Fig. 6C). On the other hand, TSAb levels were significantly higher, and TBAb activities were significantly lower, in A subunit adenovirus-injected mice than in animals immunized with adenovirus expressing the noncleaving TSHR (Fig. 6, D and E). These findings were confirmed recently: adenovirus expressing the TSHR A subunit was superior to adenovirus encoding the holoreceptor in inducing hyperthyroidism in BALB/c mice (98).
|
B. Antibody titer and TSAbs vs. TBAbs
Other than TSAbs, autoantibodies that activate rather than inhibit antigen function are a rare phenomenon, no doubt requiring a highly specific epitope(s). The relatively common occurrence of TSAbs may also be related to the susceptibility of the TSHR to be activated by mutations (largely within the membrane-spanning regions) (122). TSAb titers in Graves’ patients’ sera are typically very low (34, 123, 124, 125, 126). Moreover, the functional balance between TSAb and TBAb activities is known to be related to antibody titer (109). The relationship between TSAb epitope(s) and titer is therefore a highly relevant issue.
1. Genetic immunization with TPO.
Comparing TSHR and TPO antibodies induced by different immunization protocols is potentially informative for the pathogenesis of Graves’ disease because of the frequent concurrence of TPO antibodies in this condition. BALB/c mice were studied for antibodies induced by injecting TPO-plasmid, TPO-adenovirus, or dendritic cells infected with TPO-adenovirus (127). TPO antibody levels were highest after TPO-adenovirus immunization, intermediate after TPO-transfected dendritic cells transfer, and lowest by TPO-plasmid vaccination (Fig. 7A).
|
2. Shimojo model.
The dynamics of TSHR antibodies have been studied in the Shimojo model of Graves’ disease. AKR/N mice injected with fibroblasts, coexpressing the TSHR and MHC class II, developed TSHR antibodies by 7–8 wk. The majority of individual animals had either TSAbs or TBAbs, and these patterns were maintained throughout the 17–24 wk of the study (130). In a small proportion of mice, TSAbs and TBAbs appeared to cycle over time.
3. Low-dose TSHR-adenovirus.
Immunization with TSHR A subunit-adenovirus induces hyperthyroidism in 65–85% of BALB/c mice (97) (Fig. 6B) and is more effective than adenovirus expressing the full-length TSHR (50% of mice) (86). Nevertheless, antibodies in these mice have higher TBAb levels than most Graves’ patients. A possible explanation for this difference between the Graves’ mouse model and human disease was a greater degree of antigenic stimulation in the model after immunization with a high dose of TSHR-adenovirus. To address this possibility, BALB/c mice immunized with the standard high (1011) viral particle dose were compared with animals receiving injections of medium (109) and low (107) doses of viral particles. Not surprisingly, mice receiving lower viral doses generated lower TSHR antibody titers on ELISA as well as (more importantly) lower TBAb activity (Fig. 7B). Remarkably, however, and consistent with the hypothesis, TSAb levels and the incidence of hyperthyroidism (80%) remained comparable regardless of viral dose. Thus, low-dose TSHR A subunit adenovirus immunization provides an induced model with a high prevalence of hyperthyroidism and a TSHR antibody profile more closely resembling autoantibodies in human Graves’ disease.
Together, the studies on TPO and TSHR provide important insight into human disease. Higher TPO antibody levels are accompanied by decreased recognition of IDR epitopes (Fig. 7A). Similarly, increasing TSHR antibody levels diverts the balance away from TSAbs and toward TBAbs (Fig. 7B). Despite different endpoints, these observations indicate that increased immune stimulation leads to higher antibody titers and epitopic spreading. Unlike the very low concentrations of TSHR antibodies in Graves’ disease (34, 123, 124, 125, 126), patients with hypothyroidism caused by TBAbs have much higher TSHR antibody titers (e.g., Refs. 120 and 131) as well as epitopic differences (described below).
C. TSHR antibody epitopes in polyclonal sera
1. Human autoantibodies.
Patients’ TSHR autoantibodies are polyclonal, although commonly restricted in terms of IgG subclass and light chain type (132, 133). There are several reasons for the difficulty of identifying the precise TSHR amino acids involved in the epitopes of these functional autoantibodies. First, TSH and TSHR autoantibody epitopes comprise discontinuous sequences of the polypeptide chain that are contiguous in the folded protein under native conditions (134, 135). Second, autoantibodies preferentially recognize the glycosylated TSHR (136, 137), although it is likely that the glycan component is not part of their epitopes. Rather, complex glycan is acquired after correct TSHR folding and trafficking to the cell surface. Incidentally, a complication for some assays is nonspecific serum IgG binding to the heavily glycosylated A subunit (138). A third reason, mentioned above, is the extremely low serum concentration of TSHR autoantibodies (34, 123, 124, 125, 126), typically nanograms per milliliter (139) compared with micrograms per milliliter of TPO autoantibodies.
Two approaches have been used to analyze the epitopes of TSHR autoantibodies in humans (Fig. 8A), namely synthetic peptides (typically 20 amino acids long) encompassing the receptor ectodomain (residues 22–417) and chimeric receptors with sections of the TSHR replaced by homologous regions of the LH receptor (LHR). In studies from different laboratories, Graves’ sera bound to different peptides without consistent recognition of particular linear epitopes (reviewed in Ref. 5). In contrast, investigations using chimeric receptors expressed in eukaryotic cells provided more reproducible insight into the binding sites of TSHR autoantibodies (Fig. 8B). In general, despite some overlap, TSAbs interact mainly with N-terminal components of the ectodomain (e.g., Refs. 135 and 140). Moreover, the TSHR A subunit neutralizes TSAb activity in all sera tested (141) and can be used to affinity-enrich the TSAb component from Graves’ sera (138). Chimeric receptor studies indicated that TBAbs interacted primarily with C-terminal components of the TSHR ectodomain (e.g., Refs. 135 and 140). However, unlike TSAbs, the epitopic range of some TBAbs can be much broader, also extending into the TSHR N-terminal region (141). Thus, TBAb activity in some patients’ sera can be adsorbed, at least in part, by the purified A subunit. The TSH binding site is primarily composed of components in the leucine-rich repeats in the midregion of the TSHR (142, 143). Nevertheless, there is epitopic overlap between TSH and TSAbs or TBAbs (Fig. 8B). For this reason, and taking into account the larger size of an antibody (150 kDa) vs. TSH (30 kDa) and the TSHR ectodomain (60 kDa), it is not surprising that both TSAbs and TBAbs are measured by the TBI assay.
|
). The N-terminal cysteine-rich peptide is the immunodominant epitope in mice vaccinated with TSHR-DNA or TSHR-adenovirus (92), as well as in rabbits and mice conventionally immunized with TSHR-protein and adjuvant (144, 145, 146, 147). Moreover, the folding of this N-terminal peptide is crucial for recognition by human TSHR autoantibodies (35, 92, 148). In other Graves’ models, serum antibodies interact primarily with peptides in the vicinity of the TSHR cleavage region; such epitope recognition occurs with TSHR-fibroblast-immunized mice (Shimojo model) (75), TSHR-adenovirus-immunized hamsters (149), and (although less well recognized than N-terminal peptide 22–41) TSHR-adenovirus-immunized mice (92). It is of interest that immunodominant epitopes of antibodies to other antigens are located at the amino terminus (150, 151) and the carboxy terminus (152). Incidentally, peptide 97–116 recognition by some Shimojo mice (75) is likely to be nonspecific because of similar observations in sera of mice injected with control fibroblasts or vaccinated with control DNA or control adenovirus (92). The likely relationship between linear epitope recognition and TSAb activity in these sera could only be determined subsequent to mTSAb isolation (see below).
Conformational TSHR antibody epitopes have been analyzed in two of these mouse models. In the Shimojo model, mice were injected with fibroblasts expressing chimeric TSH-LHRs (Mc 1 + 2, Mc 2 and Mc 4) (Fig. 8A) previously used to characterize TSHR autoantibodies in Graves’ patients (140). Injection of fibroblasts expressing chimeric receptors with N-terminal substitutions Mc 1 + 2 (residues 9–165) or Mc 2 (residues 90–165) could not induce TSHR antibodies or hyperthyoidism. Fibroblasts expressing Mc 4 (substitution of C-terminal residues 262–370) did induce TBI antibodies, but information was not provided regarding hyperthyroidism (117).
In mice immunized with TSHR-adenovirus, TSAb activity was analyzed using different chimeric receptors, TSHR-LHR-6 and TSH-LHR-8 (the latter being similar to Mc 1 + 2) (Fig. 8A). With TSHR-LHR-6, the N-terminal region of the TSHR (residues 1–260) remains intact. This chimeric receptor responded to TSAbs in the mouse sera (86). In contrast, TSH-LHR-8 (N-terminal residues 1–160 replaced) was unresponsive to TSAbs. These findings in both mouse models are consistent with data from human autoantibodies and reinforce the importance of the TSHR N-terminal region for recognition by TSAbs in Graves’ patients (Fig. 8B).
D. mTSAbs from immunized animals
1. Isolating mTSAbs.
Monoclonal antibodies were initially generated by fusing a B cell line (such as a myeloma) with splenic lymphocytes from immunized mice (153). The technique has been standardized and reproduced for many antigens including the autoantigens Tg (154) and TPO (155). Many TSHR monoclonal antibodies have been generated that are invaluable for immunoprecipitation and Western blotting (e.g., Refs. 156 and 157). However, the successful isolation of functional mTSAbs was only possible after novel approaches had been developed to induce Graves’ hyperthyroidism in animals.
Even with these models, isolating mTSAbs has been extremely difficult and only recently accomplished (158, 159, 160, 161). In addition to dogged persistence, the problems were overcome in two ways (reviewed in Ref. 162): first, by selecting donor animals with very high TSHR antibody (TBI) activity (158, 160) and/or hyperthyroidism (159, 161); and second, by screening for TBI, TSAb, or binding to TSHR-expressing cells rather than by ELISA (which may preclude detecting functional antibodies). mTSAbs have been isolated from TSHR-DNA-vaccinated mice (158, 160, 161) or TSHR-adenovirus-immunized hamsters (159). Moreover, despite the greater difficulties in isolating human monoclonal antibodies, an extremely potent human mTSAb has been cloned from a Graves’ patient (163).
2. Stimulating activities of mTSAbs.
mTSAbs can be compared based on the IgG concentration required to stimulate cAMP generation by TSHR expressing eukaryotic cells in the sensitive NaCl-free system (Table 3). Low concentrations (200 ng/ml IgG) of some mouse or hamster mTSAbs increased basal cAMP levels 2- to 3-fold (158, 159, 160). One murine mTSAb (IRI-SAb3) is as potent as the single human monoclonal TSAb (hTSmAb1) (163), requiring only 10 ng/ml to induce maximal stimulation comparable to TSH (161). Fab prepared from mTSAbs were as active as their parent mouse IgGs (158, 161).
|
).
3. TBI, TBAbs, and affinities.
In addition to TSAb, all mTSAbs reported to date (except IRI-SAb1) had TBI activity (161). Although some TSAbs, such as MS-1, can inhibit TSH-induced cAMP stimulation (159), this observation does not imply clinically significant TSH blocking activity, a rare cause of hypothyroidism. As discussed for TSAbs (141, 164), an elementary pharmacological principle is that a weak (or partial) agonist is also an antagonist. Therefore, it is not possible to conclude that a monoclonal antibody or polyclonal serum (e.g., Ref. 165) contains functionally significant TBAb activity unless TSAb activity is absent.
Before monoclonal antibodies were available, it was difficult to evaluate TSAb affinities for the TSHR. That low concentrations (nanograms to micrograms) of TSHR (A subunits or TSHR ectodomain) neutralized TBI (34), and TSAb (125, 141) activities in patients’ sera suggested a high affinity of TSHR autoantibodies for the TSHR. However, it has now been shown unequivocally that, like the single human mTSAb (163), mTSAbs from immunized animals (TSHR-DNA or TSHR-adenovirus) can be high-affinity antibodies with dissociation constants in the nanomolar or lower range (158, 160, 161, 166).
4. mTSAb epitopes.
Murine mTSAbs do not bind to nonglycosylated 35S-labeled TSHR transcribed in vitro (163). These data do not necessarily imply that TSAb epitopes contain glycan; for reasons described previously, abnormal polypeptide folding is a more likely explanation. Furthermore, unlike nonstimulating monoclonal antibodies, the epitopes of which have been identified by peptide scanning (e.g., Refs. 149 ,156 , and 157), mTSAbs do not recognize linear epitopes (158, 159, 161). These data indicate that serum antibody components in immunized mice or hamsters that bind to TSHR peptides (Fig. 8C) are unlikely to have TSAb activity.
Competition assays can test the relationship between the epitope of a monocl
