Critical differences between induced and spontaneous mouse models of Graves’ disease with implications for antigen-specific immunotherapy in humans

Abstract

Graves’ hyperthyroidism, a common autoimmune disease caused by pathogenic autoantibodies to the thyrotropin (TSH) receptor, can be treated but not cured. This single autoantigenic target makes Graves’ disease a prime candidate for antigen-specific immunotherapy. Previously, in an induced mouse model, injecting TSHR A-subunit protein attenuated hyperthyroidism by diverting pathogenic TSHR antibodies to a non-functional variety. Here we explored the possibility of a similar diversion in a mouse model that spontaneously develops pathogenic TSHR autoantibodies, NOD.H2^h4 mice with the human(h) TSHR A-subunit transgene expressed in the thyroid and (shown here) the thymus. We hypothesized that such diversion would occur following injection of “inactive” hTSHR A-subunit protein recognized only by non-pathogenic (not pathogenic) TSHR antibodies. Surprisingly, rather than attenuating the pre-existing pathogenic TSHR level, in TSHR/NOD.H2^h4 mice inactive hTSHR antigen injected without adjuvant enhanced the levels of pathogenic TSH-binding inhibition (TBI) and thyroid stimulating antibodies, as well as non-pathogenic antibodies detected by ELISA. This effect was TSHR-specific as spontaneously occurring autoantibodies to thyroglobulin and thyroid peroxidase were unaffected. As controls, non-transgenic NOD.H2^h4 mice similarly injected with inactive hTSHR A-subunit protein unexpectedly developed TSHR antibodies, but only of the non-pathogenic variety detected by ELISA. Our observations highlight critical differences between induced and spontaneous mouse models of Graves’ disease with implications for potential immunotherapy in humans. In hTSHR/NOD.H2^h4 mice with ongoing disease, injecting inactive hTSHR A-subunit protein fails to divert the autoantibody response to a non-pathogenic form. Indeed, such therapy is likely to enhance pathogenic antibody production and exacerbate Graves’ disease in humans.

INTRODUCTION

Graves’ hyperthyroidism is directly caused by pathogenic autoantibodies to the TSHR that mimic the stimulatory effects of TSH (reviewed in (1). The disease can be treated but there is no cure for the underlying autoimmune process. Thyroid ablation with radioiodine, the most common therapy in the USA (2), results, almost inevitably, in permanent hypothyroidism requiring life-long thyroid hormone ingestion. Thiourea drugs are effective at inhibiting thyroid hormone synthesis (3) and can be used for years but in the majority of cases the disease recurs when the drugs are discontinued.

Several novel therapeutic approaches for human Graves’ disease are being tested including small molecule inhibitors of TSHR function (4,5), monoclonal TSHR antibodies that block the function of thyroid stimulating autoantibodies (TSAb)(6), and inhibitors of components of the adaptive immune system, such as rituximab which target B-lymphocytes (7,8). Most important, however, even if successfully introduced into the pharmacopeia, these approaches may treat, but will not cure Graves’ disease, and non-specific immunological inhibitors have potentially severe side effects.

Antigen-specific immunotherapy has long been attempted for autoimmune conditions such as multiple sclerosis and type 1 diabetes mellitus but clinical trials have been disappointing. An editorial on the limited therapeutic efficacy of myelin basic protein peptide immunotherapy for multiple sclerosis was ascribed to the wide spectrum of antigens targeted by the immune system in this disease, leading to the suggestion that in attempts to cure autoimmune diseases “It may be advantageous to focus on rarer diseases ...... where the immune response in largely limited to a single antigen” (9,10). There is no need to search for a rare disease. Graves’ disease is one of the most common autoimmune diseases affecting humans with a prevalence of ~1% (11). Moreover, it is the prime example of an autoimmune disease directly caused by autoimmunity to a single autoantigen, the thyrotropin receptor (TSHR).

In mice with induced hyperthyroidism, a number of novel therapeutic approaches have been attempted, including a shift from Th1 to Th2 CD4+ T-cells (or vice versa) induced by various agents (12-14); blockade of tumor necrosis factor family ligand inhibitors (BAFF and APRIL) (15); anti-CD20 monoclonal antibody (rituximab)(16); immunoproteasome inhibition (17); small molecule antagonism of the TSHR (4) and injection of purified, recombinant TSHR protein (18).

Of these approaches, only purified, recombinant TSHR protein was both antigen-specific and targeted the immune system with the goal of inducing tolerance to the TSHR (18). However, success with this approach was limited. In the induced Graves’ disease model using adenovirus expressing the TSHR A-subunit, prior injection of A-subunit protein attenuated the development of hyperthyroidism but was ineffective in reversing hyperthyroidism once established (18). Disease attenuation occurred with eukaryotic, not prokaryotic, A-subunit protein but, contrary to expectation, it was not associated with reduced TSHR tolerance. Instead, there was a diversion from bioactive to non-functional TSHR Ab. Antibody diversion has been used to treat experimentally induced myasthenia gravis in rats by injecting pathologically irrelevant epitopes on the cytoplasmic domains of the acetylcholine receptor (19).

Graves’ disease develops spontaneously in humans. Therefore, in the present study we focused on a mouse model that spontaneously develops pathogenic TSHR antibodies, namely NOD.H2^h4 mice with the human (h)TSH receptor A-subunit transgene targeted to the thyroid (TSHR/NOD.H2^h4)(20). In the present study, hTSHR A-subunit transgenic mice and their non-transgenic NOD.H2^h4 littermates were injected two or three times with hTSHR A-subunit protein in the absence of adjuvant. Our findings were most unexpected and emphasize the critical importance of using a spontaneous model, rather than an induced model, in attempting to apply autoantigen-specific treatment to deviate pathogenic TSHR antibodies.

MATERIALS AND METHODS

Mouse strains and injection of TSHR A-subunit protein or ovalbumin

Transgenic NOD.H2^h4 mice expressing the human TSH receptor A-subunit (TSHR/NOD.H2^h4)(20) and non-transgenic NOD.H2^h4 littermates were bred at Cedars-Sinai Medical Center. Mice of the TSHR/NOD.H2^h4 strain have been cryopreserved by the Mutant Mouse Regional Resource Center under the designation NOD.Cg_Tg(TG_TSHR)51.9Smcl. Wild type BALB/cJ mice were purchased from Jackson Laboratories (Bar Harbor, ME).

The novel transgenic strain was derived by crossing BALB/c mice expressing low levels of the hTSHR A-subunit in the thyroid and thymus (21-23) with NOD.H2^h4 mice, repeated backcrossing of the transgenic progeny to wild-type NOD.H2^h4 for 8 generations (N8)(20). TheTSHR/NOD.H2^h4 mice used in the present study were from the N10 and N11 generations (>99.9% NOD.H2^h4 genome). To emphasize that the transgene encodes the human TSHR A-subunit, this strain will be referred to as ‘hTSHR/NOD.H2^h4 ‘mice.

hTSHR/NOD.H2^h4 (n=37) and non-transgenic NOD.H2^h4 littermates (n=30) were injected subcutaneously on the back with hTSHR A-subunit protein (prepared as described below) or saline. At the indicated internals, mice aged 8 weeks received three injections of hTSHR A-subunit protein (4μg, 10μg and 10μg), and in another series, mice aged 12 weeks received two injections of 10μg hTSHR A-subunit protein (Fig. 1). It must be emphasized that the hTSHR A-subunit protein was injected without adjuvant. As controls, BALB/c mice aged 8 weeks (5 males, 5 females) were injected three times at the indicated intervals with hTSHR A-subunit protein (4μg, 10μg and 10μg) without adjuvant. Additional BALB/c mice were studied without hA-subunit protein injections. As a further control, a separate group of non-transgenic NOD.H2^h4 mice (6 males, 5 females) were injected subcutaneously on the back with ovalbumin (OVA; Sigma Chemical Co,) following the same time intervals and doses as for A-subunit protein.

Figure 1.

Protocol for injecting hTSHR A-subunit protein in the absence of adjuvant.

From the age of 9 weeks, all mice (hTSHR/NOD.H2^h4, non-transgenic NOD.H2^h4 and BALB/c) were provided with sodium iodide supplemented drinking water (0.05% NaI). Blood was drawn at 17 and 25 weeks and mice were euthanized after 41 weeks (Fig. 1). These time intervals correspond to 8, 16 and 32 weeks exposure to NaI. Mouse studies were performed in accordance with the guide-lines of the Institutional Animal Care and Use Committee at Cedars-Sinai Medical Center and were carried out with the highest standards of care in a pathogen-free facility.

TSHR A_subunit protein

Recombinant human (h)TSHR A-subunit protein (amino acid residues 22-289) secreted by Chinese Hamster Ovary cells (CHO) with an amplified transgenome (24) was purified from culture supernatants by affinity chromatography (25) and dialyzed against 10 mM Tris, pH 7.4, 50 mM NaCl. Two different conformational forms of this recombinant hTSHR protein can be purified separately from the conditioned culture medium based on their reciprocal recognition by Graves’ patients’ autoantibodies and mouse mAb 3BD10; ‘active’ h TSHR A-subunits only by the former and ‘inactive’ h TSHR A-subunits only by the latter (26). To preclude the possibility of passive TSHR antibody neutralization, we used inactive hTSHR-289, hereafter referred to as ‘hA-subunit protein’.

TSHR antibody assays

TSHR antibodies were measured using three assays:-

a) ELISA

The assay for ELISA TSHR antibodies (IgG class) was reported previously (27). ELISA wells were coated with hA-subunit protein (‘inactive’ form, described above; 5 μg/ml) and incubated with test sera (1:100 dilution). The positive controls were serum from BALB/c mice immunized with hTSHR A-subunit adenovirus (for example (28) and the monoclonal antibody (McAb) 3BD10 (25). Antibody binding was detected with horseradish peroxidase-conjugated mouse anti-IgG (A 3673, Sigma Chemical Co., St. Louis, MO) and the signal was developed with o-phenylenediamine and H₂O₂. Data are reported as the optical density (OD) at 490 nm.

b) TSH binding inhibition (TBI) assay

TBI levels were measured in 25 μl mouse serum using a clinical assay kit (Kronus Inc, Star ID). The data are reported as the % inhibition of ¹²⁵I-TSH binding to the TSH holoreceptor.

c) Bioassay for TSAb

A bioassay was used to measure cAMP generation by Chinese hamster ovary (CHO) cells expressing the human TSHR (27,29). To permit testing 10 μl mouse serum per well, we used the following approach: 25 μl test mouse serum + 75 μl normal human serum (‘carrier protein’) was precipitated with 300 μl 20% polyethylene glycol 4000 (PEG; Sigma-Aldrich) in water. Because of the large amount of ‘carrier’ protein needed for PEG precipitation, we used normal human serum (one batch from a single donor in all assays) as described previously (for example (29). The negative controls were sera from BALB/c mice; for the positive control, sera from normal BALB/c mice were supplemented with human monoclonal TSHR M22 (30). Negative and positive control samples were precipitated using PEG in the same way using normal human serum as carrier protein. The pellets were resuspended in 240 μl Ham's F12 medium supplemented with 10 mM Hepes, pH 7.4, 1 mM isobutylmethylxanthine and 0.3% bovine serum albumin.

Duplicate aliquots (110 μl) were applied to human TSHR-expressing CHO cell monolayers in 96 well plates. After incubation (90 min, 37°C), the medium was aspirated, intracellular cAMP was extracted with ethanol, evaporated to dryness and resuspended in 0.2 ml Dulbecco's PBS. Aliquots (12 μl) were assayed using the LANCE cAMP kit (PerkinElmer, Boston MA). TSAb activity was expressed as a percentage of cAMP values attained with PEG precipitated IgG from BALB/c mice.

Autoantibodies to thyroglobulin (Tg) and thyroid peroxidase (TPO)

Tg was isolated from murine thyroid glands as described (31). ELISA wells (Immulon 4HBX, Thermo Scientific, Rochester NY) were coated with mouse Tg (1.5 μg/ml) and incubated with test sera (duplicate aliquots, 1:100 dilution). Antibody binding was detected with horse radish peroxidase-conjugated goat anti-mouse IgG (A3673, Sigma Chemical Co., St. Louis MO), the signal developed with o-phenylenediamine and the reaction stopped using 20% H₂SO₄. The negative control was serum from 8 week old NOD.H2^h4 mice on regular water; the positive control was serum from BALB/c mice immunized with mouse thyroglobulin and complete Freund's adjuvant (28). TgAb data are presented as the optical density (OD) at 490 nm.

TPOAb were measured by flow cytometry using CHO cells stably expressing mouse-. TPO as previously described (31). A cell-based assay was used rather than ELISA because NOD.H2^h4 mice do not recognize human TPO (28) and mouse TPO generated in eukaryotic cells is not available. Sera (diluted 1:50) were incubated with mouse TPO-CHO cells; binding was detected with fluorescein isothyocyanate-conjugated affinity purified goat anti-mouse IgG (A16073, Life Technologies, Carlsbad, CA). Cells staining with propidium iodide (1 μg/ml) were excluded from analysis. The negative control for IgG class antibody binding to mouse TPO-CHO cells was serum from 8 week old NOD.H2^h4 mice. Positive controls were mouse monoclonal antibodies #15 and #64 to human TPO (32), kindly provided to us by Dr. J Ruf (Marseille, France), that also recognize mouse TPO (31,32). Flow cytometry was performed (10,000 events) using a FACScan with CELLQUEST Software (Becton Dickinson, San Jose, CA). Data are reported as the geometric mean (Geo Mean).

Antibody response to ovalbumin (OVA)

Antibodies to OVA were measured by ELISA as previously described (33). Briefly, ELISA plates were coated with OVA (1μg/ml) in 20mM Tris-HCL (pH 7.5). Duplicate serum aliquots (diluted 1:100) were applied to the wells and antibody binding was detected with horse radish peroxidase-conjugated goat anti-mouse IgG (A3673, Sigma Chemical Co., St. Louis MO). The signal was developed with o-phenylenediamine and the reaction stopped using 20% H₂SO₄. The positive control was mouse monoclonal antibody HYB 099-01 obtained from CF1 × BALB/c mice immunized intra-peritoneally with native OVA (33). The “cut-off” point for positivity was established from the binding (mean + 2SD) to OVA-coated ELISA plates of sera from NOD.H2^h4 mice on NaI for 16 weeks without OVA injections.

Intrathymic expression of the endogenous mouse TSHR and the transgenic human TSHR A-subunit

Thymuses (32-day old mice) from hTSHR/NOD.H2^h4 (n=3) and wild-type NOD.H2^h4 mice (n=3) were stored in RNAlater (LifeTechnologies, Carlsbad, CA). Quantitative (q) RT-PCR was performed essentially as previously described (23). Tissue was homogenized with QIAshredder columns (QIAGEN, Valencia, CA). Total RNA was prepared using RNeasy Plus Mini kit (QIAGEN, Valencia, CA). The mRNA samples were treated with TURBO DNase (LifeTechnologies, Carlsbad, CA) to remove genomic DNA. Reverse transcription was performed with AffinityScript QPCR cDNA Synthesis Kit (Agilent Technologies, Cedar Creek, TX) using oligo(dT) and random primers. Quantitative real-time PCR (qPCR) was performed using the FastStart SYBR Green Master mix (Roche, Basel, Switzerland) with 2.5% of cDNA (20μl final volume). Reactions were run on an iCycler Thermal Cycler with iQ5 Real Time PCR Detection System module (BioRad Laboratories, Hercules, CA). An initial denaturation step at 95 C (10 min) was followed by denaturation at 95 C (30 sec), and annealing and extension at 55 C (30 sec) for 40 cycles. Relative gene expression levels were calculated using the comparative Ct method (ΔΔCt), according to the Pfaffl model (35), using Bio_Rad iQ5 2.0 software. Samples were tested in triplicate; parallel controls lacked reverse transcriptase. Data were normalized to mouse β–actin. For the mouse TSHR and mouse β-actin, we used mouse TSHR RT2 qPrimer Assay for Mouse TSHR and RT2 qPrimer Assay for Mouse Actb (QIAGEN, Valencia, CA). Primers for the hTSHR A- subunit were as follows:- sense 5′ GCAAGAAACACCTGGACTCTTAA3′, hTSHR A-subunit antisense 5-GGTGGTGATGGCTAGTCTGA-3′. Primers for the mouse TSHR were designed to avoid overlap with A-subunit of human TSHR in transgenic animals.

Statistics

Significant differences between responses in different groups were determined by Mann Whitney rank sum test or, when normally distributed, by Student's t test. Multiple comparisons were made using analysis of variance (ANOVA). Tests were performed using SigmaStat (Jandel Scientific Software, San Rafael, CA). Statistical results and the specific tests performed are provided in the appropriate figure legends.

RESULTS

Injecting hTSHR A-subunit protein induces non-pathogenic TSHR antibodies NOD.H2^h4 mice and enhances generation of these antibodies in NOD.H2^h4 with the hTSHR A-subunit transgene

Before attempting to attenuate the spontaneous development of pathogenic TSHR antibodies, we examined the effect of human TSHR A-subunit protein injections on non-pathogenic TSHR antibodies, which are measured by ELISA. As a control, these antibodies were not induced in non-transgenic NOD.H2^h4 mice injected with saline (Fig. 2A). Unexpectedly in the absence of adjuvant, TSHR hA-subunit protein injected subcutaneously did induce TSHR ‘ELISA’ antibodies in these non-transgenic mice, detectable at 16 and 32 weeks (Fig. 2B). In NOD.H2^h4 mice with the human TSHR A-subunit transgene (hTSHR/NOD.H2^h4), TSHR ELISA antibodies were detected at low levels in a few saline injected animals at 16 and 32 weeks (Fig. 2C), consistent with our former observations (20). Relative to saline controls, TSHR ELISA antibody levels were significantly higher (at 16, but not 32 weeks) in mice injected with TSHR hA-subunit protein, again without adjuvant (Fig. 2D). Similar observations were made for mice that received 3 injections of A-subunit protein (from 8 weeks of age) or 2 injections (from 12 weeks of age). No significant differences were observed between males and females (data not shown).

Figure 2.

TSHR antibodies measured by binding to hTSHR-A-subunit coated ELISA plates after 8, 16 and 32 weeks in non-transgenic NOD.H2^h4 mice (A and B) and hTSHR transgenic NOD.H2^h4 mice C and D). Data are shown as OD 290 nm in ELISA for individual mice (males and females) injected 3 times with saline (panels A and C, open circles) or with hTSHR A-subunit protein (B and D; 2 injections hatched circles; 3 injections solid circles). All injections of hTSHR A-subunit protein were performed without adjuvant. The numbers of mice in each group ranged between 13 and 23 animals in each group. Dotted line: mean + 2SD in non-transgenic NOD.H2^h4 mice. Values significantly different from saline injected mice for the same time interval : *p=0.010; ** p=0.002; ^ p=0.015 (Rank sum tests).

These changes were specific for non-pathogenic TSHR ELISA antibodies. Consistent with the known characteristics of NOD.H2^h4 mice, TgAb, first detectable at 8 weeks, continued to be positive at 16 and 32 weeks in virtually all animals whether non-transgenic or hTSHR transgenic, and regardless of whether or not injected with TSHR A-subunit (Fig. 3). Similarly, TPOAb (measured after 16 and 32 weeks) remained detectable in most mice and were not altered by A-subunit protein injection (Fig. 4).

Figure 3.

TgAb after 8, 16 and 32 weeks in A, non-transgenic NOD.H2^h4 mice and B, hTSHR transgenic NOD.H2^h4 mice. Data are shown as OD 290 nm in ELISA for individual mice (both males and females). Saline injections are shown as open circles; hTSHR A-subunit protein injected twice (hatched circles) or 3 times (solid circles) without adjuvant. The “cut-off” level for TgAb positivity is shown as a dashed line (mean + 2SD in BALB/c mice).

Figure 4.

TPOAb after 16 and 32 weeks in A, non-transgenic NOD.H2^h4 mice and B, hTSHR transgenic NOD.H2^h4 mice. Males and females were studied. Data are shown for individual mice as geometric means. For both panels, saline injections are shown as open circles; TSHR A-subunit protein injected twice as hatched circles or three times as solid circles. Injections of hTSHR A-subunit protein were performed without adjuvant. The “cut-off” level for TPOAb positivity is shown as a dashed line (mean + 2SD in BALB/c mice).

Intrathymic expression of the hTSHR A-subunit transgene and the mouse TSHR

The level of intrathymic expression of the TSHR has been shown to be a susceptibility factor in human Graves’ disease (36,37). It should be noted that the hTSHR/NOD.H2^h4 mouse is not a global transgenic; the transgene was targeted to the thyroid using the tissue-specific thyroglobulin promoter. We, therefore studied intrathymic expression of both the human TSHR A-subunit and the endogenous mouse TSHR. As determined by quantitative (q) RT-PCR, both non-transgenic NOD.H2^h4 and transgenic hTSHR/NOD.H2^h4 mice expressed the endogenous mouse TSHR in the thymus but only intrathymic mRNA from hTSHR/NOD.H2^h4 transgenic mice was positive for the hTSHR A-subunit (Fig. 5). These data are consistent with our previous observations in low expressor TSHR A-subunit BALB/c mice (23) from which the transgene was transferred to the NOD.H2^h4 background.

Figure 5.

Intrathymic expression of the transgenic human TSHR A-subunit and the endogenous mouse TSHR in transgenic TSHR/NOD.H2^h4 and non-transgenic NOD.H2^h4 mice. qPCR are shown as normalized fold expression relative to mouse β-actin of triplicates (mean + SD); Un, undetectable.

Responses to hTSHR A-subunit protein injected without adjuvant occur in BALB/c mice and autoimmune prone NOD.H2^h4 mice do not exhibit over-reactivity

The surprising finding of non-pathogenic TSHR A-subunit antibody induction in NOD.H2^h4 mice injected with hA-subunit protein without adjuvant (Fig. 2) raised the possibility that this autoimmune prone mouse strain was unduly susceptible to loss of tolerance to the TSHR compared to other mouse strains or to the TSHR specifically. We, therefore, performed two additional series of hA-subunit injections without adjuvant according to the protocol outlined in Fig. 1 in non-autoimmune prone BALB/c mice, and with an irrelevant antigen, ovalbumin (OVA), in NOD.H2^h4 mice.

All BALB/c mice (males and females) developed surprisingly high levels of TSHR ELISA antibodies at 16 and 32 weeks following injection of hTSHR A-subunit protein in the absence of adjuvant (Fig. 6A). After 16 weeks, mean TSHR ELISA antibody ELISA OD values were 2.25 ± 0.09 in BALB/c compared with 0.71 ± 0.26 in wild-type NOD.H2^h4 and 0.22 ± 0.05 in TSHR/NOD.H2^h4 transgenic mice. The significantly lower values in hTSHR/NOD.H2^h4 transgenic mice than BALB/c and wild-led type littermates (p< 0.05, ANOVA) reflects greater tolerance in the transgenics, consistent with the intrathymic expression of the hTSHR A-subunit in this strain (Fig. 5). Incidentally, a few TSHR A-subunit protein injected BALB/c mice developed low levels of TgAb (Fig. 6B, left panel). In contrast, TgAb did not develop in BALB/c mice also exposed to NaI but without TSHR A-subunit protein injections (Fig. 6B, right panel).

Figure 6.

Upper panel: A. hTSHR ELISA antibodies in BALB/c mice injected three times with TSHR A-subunit protein and exposed to NaI water for 16 and 32 weeks; B. Tg antibodies in the same BALB/c mice (left panel) and in BALB/c mice that were not injected with TSHR A-subunit protein (right panel) after 16 and 32 weeks on NaI water. Data are shown as OD 290 nm for individual mice (males and females). Dotted line for hTSHR ELISA antibody: mean + 2SD for in non-transgenic NOD.H2^h4 mice; for TgAb: mean + 2 SD in BALB/c mice.

Lower panel: C: NOD.H2^h4 mice injected three times with OVA antibodies were tested for OVA antibodies after 8 and 16 weeks on NaI water. Data are shown as OD 290 nm for individual mice (males and females). Dotted line: mean + 2 SD in NOD.H2^h4 mice on NaI for 16 weeks. D. Comparison of 16 week values for hTSHR antibody ELISA OD values (mean + standard error) in non-transgenic NOD.H2^h4 mice injected three times with hTSHR A-subunit protein (from mice shown in Fig. 2) versus mean OD values for non-transgenic NOD.H2^h4 mice injected three times with OVA (values from Fig. 6B). ELISA binding OD values for the monoclonal (McAb) used in each assay are included, namely 3BD10 for hTSHR A-subunit protein (25) and HYB 099-01 for OVA (33); * p=0.017 (Rank sum tests).

Turning to the outcome of OVA injection without adjuvant, OVA Ab were induced at low levels in a few NOD.H2^h4 mice (Fig. 6C). In assays standardized using McAb specific for each protein, induced TSHR ELISA antibody levels were much higher in non-transgenic NOD.H2^h4 mice than OVA Ab levels induced in the same strain (Fig. 6D).

Pathogenic TSHR antibodies are enhanced by hTSHR A-subunit protein injection (without adjuvant) only in hTSHR/NOD.H2^h4 transgenic mice

The pathogenicity of TSHR antibodies was first tested using in the clinical TSH binding inhibition (TBI) assay. This study was limited to female mice because males develop very high levels of TSH that cannot be distinguished from TBI (20). Whether or not injected with hA-subunit protein, sera from all non-transgenic NOD.H2^h4 mice were TBI negative, except for one mouse that was borderline positive at 16 weeks (Fig. 7A). In contrast, some sera from saline-injected transgenic TSH/NOD.H2^h4 mice were TBI positive (2/6 mice at 16 weeks; 4/5 mice at 32 weeks). Moreover, only in the transgenic mice, TSHR hA-subunit protein injections enhanced TBI levels (Fig. 7B) and also increased the number of TBI positive mice (5/10 at 16 weeks and 9/11 at 32 weeks). It should be noted that despite the high levels of non-pathogenic TSHR antibodies detected by ELISA that developed in BALB/c mice injected with TSHR hA-subunit protein (Fig. 6A), none developed TBI activity (7.9 ± 2.2 % inhibition of TSH binding, Mean ± SE).

Figure 7.

TSHR binding inhibition (TBI) in female mice after 16 and 32 weeks in A, non-transgenic NOD.H2^h4 mice and B, hTSHR transgenic NOD.H2^h4 mice. Data are shown as % inhibition for individual mice after 3 saline injections (open circles) or 2 injections (hatched circles) or 3 injections (solid circles) of hTSHR A-subunit protein in the absence of adjuvant. The dotted line represents the cut-off for non-transgenic NOD.H2^h4 mice (mean + 2SD) at 16 weeks. Values above this line are considered positive.

In a second approach, sera from female mice were tested for TSHR antibodies in a TSAb bioassay, the ‘gold standard’ for pathogenicity. TSAb was positive in a single non-transgenic NOD.H2^h4 mouse injected with hA-subunit protein (Fig. 8A). In contrast, TSAb activity was enhanced in 5/10 hTSHR/NOD.H2^h4 mice injected with TSHR hA-subunit protein in the absence of adjuvant (Fig. 8B).

Figure 8.

TSAb activity in females after 16 and 32 weeks in A) non-transgenic NOD.H2^h4 mice and B) hTSHR transgenic NOD.H2^h4 mice. Data are shown as % control values for individual mice. Saline injections are shown as open circles; hTSHR A-subunit protein injected twice (hatched circles) or 3 times (solid circles) in the absence of adjuvant. The dotted line represents the cut-off for non-transgenic NOD.H2^h4 mice at 16 weeks. Values above this line are considered positive.

It should be emphasized that the hA-subunit protein used for injection was in the “inactive” state, that is the form that is unable to neutralize pathogenic TSHR antibodies detected by the inhibition of TSH binding to its receptor (TBI) or thyroid stimulating antibody activity (TSAb).

DISCUSSION

The availability of a mouse model (hTSHR/NOD.H2^h4 mice) that spontaneously develops pathogenic TSHR autoantibodies with properties identical to those observed in human Graves’ disease provides a unique opportunity to investigate immunotherapeutic approaches to attenuate or reverse the generation of these antibodies. Previous studies in the induced TSHR A-subunit adenovirus model showed that injecting human (h)TSHR A-subunit protein without adjuvant prior to adenovirus immunization attenuated hyperthyroidism by deviating TSHR antibodies from functional to inactive antibodies (18). Turning to the spontaneous hTSHR/NOD.H2^h4 model, in addition to pathogenic TSHR autoantibodies measured using clinical assays (TBI and TSAb), these mice also develop non-pathogenic TSHR antibodies detectable by ELISA. The human TSHR antigen on ELISA plates is of the ‘inactive’ form, not recognized by pathological autoantibodies and only by non-pathogenic antibodies. In the present study we used these hTSHR/NOD.H2^h4 mice to test the hypothesis that injecting inactive hTSHR A-subunit protein would enhance generation of non-pathogenic TSHR antibodies detected by ELISA and thereby deviate the spontaneous development of pathogenic TSHR antibodies to these innocuous antibodies.

The data obtained did not fulfill our hypothesis, but, on the other hand, provided unexpected information that will be important in future studies in both animals and humans. Surprisingly, injecting inactive hA-subunits into hTSHR/NOD.H2^h4 transgenic mice boosted, rather than attenuated, the generation of pathogenic TSHR antibodies. These changes were specific for the TSHR because the levels of antibodies to autologous Tg and TPO were unchanged following injection of hTSHR A-subunit protein.

A second observation with heuristic relevance in our study relates to the non-transgenic NOD.H2^h4 parental strain. ‘Inactive’ hTSHR A-subunit protein injected without adjuvant led to the generation of high levels of non-pathogenic (but not pathogenic) antibodies detected by ELISA. However, BALB/c mice, that are not autoimmune disease-prone, behave in a similar manner to NOD.H2^h4 mice in developing high levels of non-pathogenic TSHR antibodies when challenged with heterologous human TSHR A-subunit protein in the absence of adjuvant. Moreover, these findings were unexpected because previous studies of BALB/c mice were performed by injecting TSHR protein together with a variety of adjuvants (reviewed in (38). On the other hand, heterologous ovalbumin injected into NOD.H2^h4 mice led to a much lower specific antibody response in comparison to injection with inactive hTSHR A-subunit protein.

These data point to the hTSHR A-subunit having unique antigenic properties rather than the NOD.H2^h4 mice being ‘globally’ autoimmune prone or having multiple immune defects. In particular, NOD.H2^h4 mice do not develop autoantibodies to the endogenous mouse TSHR (20). The relatively low OVA antibody response versus the strong response to the hTSHR A-subunit in this mouse strain may relate to glycosylation, which plays a role in antigen uptake and internalization (for example (39). Thus, OVA has a single N-linked glycosylation site (32) whereas the hTSHR A-subunit has 5 N-linked glycosylation sites (40).

Focusing on the outcome of injecting hTSHR A-subunit protein without adjuvant into non-transgenic NOD.H2^h4 and transgenic hTSHR/NOD.H2^h4 mice, our data provide insight into the specificity of B cells in these two strains (Fig. 9). The injected hTSHR A-subunit protein will activate T cells in wild-type NOD.H2^h4 mice and stimulate memory T cells in the transgenics. T cells are critical for providing “help” to B cells to generate antibodies. However, the outcome with respect to TSHR antibodies depends on the B cell epitopes recognized. It should be emphasized that the two forms of TSHR A-subunits (‘active’ and ‘inactive’) have the identical amino acid composition and tertiary conformation but differ in their quaternary structure (41). Evidence suggests that the former is a trimer, the latter a dimer, distinguished by differential recognition by pathogenic and non-pathogenic TSHR antibodies (42). For this reason, we suggest that the human TSHR A-subunit expressed by the transgene (absent in the non-transgenics) influences the B cell precursor repertoire. We hypothesize that both transgenic and non-transgenic NOD.H2^h4 mice have precursor B cells for non-pathogenic ELISA-type TSHR antibodies. In contrast, in hTSHR/NOD.H2^h4 mice, the in vivo thyroidal expression of conformationally intact TSHR A-subunit by the transgene is responsible for selecting B cells specific for the epitopes of bioactive/pathogenic TSHR antibodies. It is also possible that, as suggested for humans (43), intrathymic hTSHR A-subunit expressed in hTSHR/NOD.H2^h4 mice plays a role in selecting pathogenic-type B cells. Injecting ‘inactive’ TSHR A-subunit protein stimulates TSHR specific T cells and B cell precursors for ELISA-type TSHR antibodies in non-transgenics and, not surprisingly, expands these B cells in transgenic hTSHR/NOD.H2^h4 mice. Remarkably, injecting ‘inactive’ TSHR A-subunit protein, a related but ‘incorrect’ protein, expands (with help from memory T cells) memory B cells specific for TBI/TSAb type pathogenic TSHR antibodies.

Figure 9.

Schematic representation of likely B cell precursors and their expansion in A) non-transgenic NOD.H2^h4 mice and B) hTSHR transgenic NOD.H2^h4 mice in response to injecting hTSHR A-subunit protein (inactive form) without adjuvant. Not shown are antigen-specific memory T cells in hTSHR transgenic NOD.H2^h4 mice which will be activated by the injected hTSHR A-subunit protein and provide critical help for antibody production by B cells.

B cells with affinity for self antigens (like the transgenic hTSHR) are tolerized by a number of mechanisms including receptor editing and anergy (functional unresponsiveness) rather than deletion as for self-reactive T cells (44). Using hen egg lysozyme-specific transgenic mouse models, Goodnow and colleagues demonstrated that self-reactive B cells were not eliminated when this antigen was expressed by thyroid cells (45). Similarly, in our spontaneous model, B cells specific for the intrathyroidal human TSHR A-subunit remain in the repertoire and can be expanded by the “cross-reacting” antigen, the inactive hTSHR A-subunit.

Our observations have important implications for antigen-specific immunotherapeutic approaches to deviate pathogenic TSHR antibodies using an inappropriate form of hTSHR A-subunit protein. It is likely that Graves’ patients cannot be treated by injecting “inactive” TSHR A-subunit protein because, as in hTSHR/NOD.H2^h4 mice, stimulation of memory T cells and B cells specific for pathogenic antibodies will lead to exacerbation of hyperthyroidism. Further, Graves’ orbitopathy involves both TSHR antibodies and cytokines, the latter secreted in response to TSHR-specific T cell activation (reviewed in (46). For these reasons, injecting inactive” hTSHR A-subunit protein is precluded as an antigen-specific treatment to divert production of pathogenic antibodies and ameliorate hyperthyroidism in Graves’ disease.

Abbreviations

qPCR

quantitative real-time PCR

TBI

Inhibition of TSH binding to the TSHR

Tg

thyroglobulin

TPO

thyroid peroxidase

TSAb

thyroid stimulating antibody

TSH

thyrotropin

TSHR

thyrotropin receptor

hTSHR

human thyrotropin receptor