Graves' hyperthyroidism and thyroiditis in HLA-DRB1*0301 (DR3) transgenic mice after immunization with thyrotropin receptor DNA

Abstract

Familial and twin studies in Caucasians have established that the MHC class II allele HLA-DRB1*0301 (DR3) is a strong susceptibility gene in Graves’ hyperthyroid disease (GD). To determine if a DR3 transgene could help establish an animal model for GD, we expressed DR3 molecules in class II-knockout NOD mice (H2A^g7−). DR3⁺g7⁻ mice were given cardiotoxin prior to immunization on weeks 0, 3 and 6 with plasmid DNA encoding human thyrotropin receptor (TSHR). Two groups of mice were also coimmunized with plasmid DNA for IL-4 or GM-CSF. Serial bleeds on weeks 8, 11 and 14 showed that approximately 20% of mice produced thyroid-stimulating antibodies (Abs), and approximately 25% had elevated T₄ levels. In particular, a subset displayed both signs of hyperthyroidism, resulting in approximately 30% with some aspect of GD syndrome. Additional mice had thyroid-stimulating blocking Abs and/or TSH-binding inhibitory immunoglobulins, while most mice showed strong labelling of TSHR⁺ cells by flow cytometry. Interestingly, lymphocytic infiltration with thyroid damage and Abs to mouse thyroglobulin were also noted. Vector controls were uniformly negative. Thus, DR3 transgenic mice can serve as a model for GD, similar to our earlier reports that this allele is permissive for the Hashimoto's thyroiditis model induced with human thyroglobulin.

INTRODUCTION

Thyroid autoimmune diseases are grouped among the most common autoimmune disorders, with a high ratio of female:male prominence in Caucasians [1]. These include Hashimoto's thyroiditis (HT), with destruction of the thyroid follicular cells leading to hypothyroidism, and Graves’ hyperthyroid disease (GD), where autoantibodies to thyrotropin receptor (TSHR) stimulate thyroid function with consequent hypersecretion of thyroid hormones. The familial clustering of both diseases suggests a strong genetic influence in susceptibility, contributed by the major histocompatibility complex (MHC) class II genes and supplemented by cytotoxic T lymphocyte-associated antigen-4 genes in disease development [2]. Among the class II genes, HLA-DRB1*0301 (DR3) has long been well recognized as a susceptibility gene [3]. However, recent genomic analysis of patient population has not provided a satisfactory association between the MHC and susceptibility [2,4].

Over the past few years, successful models of GD in both inbred and outbred mouse strains, induced by a variety of procedures, have been described [5–11]. Two prominent protocols include multiple injections of human TSHR expressed on MHC class II-transfected fibroblasts [5] or as naked complementary DNA (cDNA) [6,9]. Variations along these themes to increase disease incidence and severity have also been reported [7,8,10]. These animal studies have again indicated the strong genetic influence of the MHC on disease susceptibility [10,12,13]. Although GD has been successfully reproduced by the fibroblast injection protocol in MHC-compatible host [8,11], the plasmid cDNA model, which would obviate the problem of MHC disparity, has proven more difficult to replicate. Different laboratories, including ours, have reported varied findings following the original protocol in BALB/c mice [6] in several parameters, such as the induction of TSHR antibodies (Abs), thyroid function as well as thyroid gland pathology [10,11,14,15]. Furthermore, the plasmid injection protocol led to stable Graves’ hyperthyroidism in outbred mice only [9], making it difficult to dissect the disease susceptibility genes or extrapolate to humans.

Recently, we have utilized HLA class II transgenic mice as a HT model to examine DRB1 polymorphism in susceptibility to thyroglobulin (Tg) in the absence of endogenous class II molecules. This strategy demonstrated the important role of DRB1*0301 in susceptibility [16], supporting certain patient studies [2,3]. The DR3 transgene in the class II-knockout mouse on NOD background (H2A^g7–) was more susceptible to human Tg-induced autoimmune thyroiditis than the autoimmunity-prone wild type NOD [17]. We now report on the successful induction of Graves’ hyperthyroidism in DR3⁺g7^– mice by human TSHR plasmid DNA immunization. Furthermore, the coinjection of mouse GM-CSF and IL-4 plasmid cytokine genes to enhance the immune response appeared to have little effect on hyperthyroid incidence.

MATERIALS AND METHODS

HLA-DRB1*0301 transgenic mice

NOD (H2A^g7) class II-knockout mice expressing HLA-DR3 in the absence of endogenous class II molecules were generated and raised in the pathogen-free Immunogenetics Mouse Core facility at the Mayo Clinic prior to shipment [17]. Briefly, the HLA-DRA/DRB1*0301 (DR3) transgene was introduced into class II-negative Ab⁰ mice and backcrossed to B10 mice to create DR3⁺ Ab⁰ mice [16]. The mice were then backcrossed to NOD (H2^g7) mice for several generations (N8) to generate DR3⁺ Ab⁰/NOD mice. DR3 and H2^g7 expression was determined by flow cytometric analysis of peripheral blood leucocytes and polymerase chain reaction, respectively [17,18]. Male and female mice were maintained in a specific pathogen-free facility on acidified water and used at 14–18 weeks of age. Animal care was supervised by veterinarians in accordance with accredited institutional guidelines.

Plasmid DNA immunization

Full-length human TSHR cDNA subcloned in pcDNA 3·1 was used for immunization [11]. Mouse IL-4 and GM-CSF cDNAs were cloned in pNGVL3 (University of Michigan Vector Core, Ann Arbor, MI, USA) and pEF-BOS [19], respectively. All plasmids were prepared using QIAfilter Plasmid Giga kits (Qiagen). Mice (5–6/group) were treated in the muscle of both hind legs with 3·4 µg cardiotoxin (Calbiochem, San Diego, CA, USA) 5 days prior to immunization. On week 0, the hind leg of each mouse was injected with 50 µg TSHR plasmid with or without 50 µg IL-4 or GM-CSF plasmid in saline. Five control mice were injected with 50 µg pcDNA 3·1 vector mixed with 50 µg IL-4 plasmid. Injections were repeated on weeks 3 and 6. Animals were bled at weeks 8, 11 and 14 and sacrificed at week 14 for thyroid histology.

Assays for TSHR Abs

Thyroid-stimulating Ab (TSAb) and thyroid-stimulating blocking Ab (TSBAb) were assayed by using TSHR-expressing CHO cells, essentially as described [11]. JP09 cells were grown overnight from 30 000 seeded cells per well in a flat bottomed 96-well plate. Prior to the assay, the medium was replaced with salt-free, sucrose containing isotonic HBSS supplemented with 0·3% BSA and 0·5 m isobutyl-1-methylxanthine (IBMX, Sigma-Aldrich). For TSAb, test serum (3 µl) was added to each well containing 100 µl salt-free isotonic buffer and incubated at 37°C for 4 h. The cAMP release was measured by radioimmunoassay (RIA, R & D Systems) and expressed as fmols/ml. For TSBAb, the assay was carried out as above, except after 2 h incubation, a suboptimal concentration of bovine TSH (bTSH, 40 µU/ml) was added and the incubation continued for an additional 2 h. All sera were initially tested in single determination, followed by assay of duplicate samples. TSBAbs were calculated as: (1 –[cAMP test serum in presence 40 µU/ml bTSH ÷ cAMP control serum in presence of 40 µU/ml bTSH]) × 100. Values of ≥30% were considered positive.

TSHR Abs were also assessed by flow cytometry using CHO cells expressing the glycophosphatidylinositol-linked extracellular domain of TSHR [20], exactly as described [11]. TSH-binding inhibitory immunoglobulin (TBII) was assayed using the radioreceptor (human) TRAK II kits (BRAHMS AG, Germany) and 100 µl serum (single determination). The results are expressed as percentage inhibition of radiolabelled TSH binding, with ≥10% considered positive.

Thyroid function tests and pathology

Total thyroid hormone (TT₄) was determined by RIA with 20 µl serum (DYNOtest TT4, BRAHMS AG). Serum TSH level was determined by a heterologous disequilibrium double-Ab precipitation RIA, with rat antiserum, ¹²⁵I-labelled rat TSH and mouse serum TSH as standard [21]. Thyroid mononuclear cell infiltration was determined at week 14 by evaluating 50–60 vertical (H&E) sections throughout both lobes (10–15 step levels). Individual pathology scores, based on a scale of 0–4, correlated with the extent of infiltration: 0, no infiltration; 0·5, >0–10%, multiple foci of infiltration without follicular destruction; 1·0, >10–20% infiltration with follicular destruction; and 2·0, >20–40% infiltration with destruction [22].

RESULTS

Hyperthyroidism in TSHR plasmid immunized DR3 transgenic mice

To determine if DR3 transgenic mice could be used to establish a mouse model of GD, mice (5–6/group), were immunized thrice with human TSHR DNA with or without IL-4 or GM-CSF DNA. Serial bleeds were analysed for TSHR Abs and thyroid function. Mice were considered hyperthyroid if elevated levels of both TT₄ and TSAb were present. Measurement of week 14 (terminal) serum TT₄ showed four mice (nos. 2, 4, 5, 9) with a statistically significant, elevated level, in comparison to mean + 3SD of the control group injected with plasmid vector plus IL-4 plasmid (Table 1). Two of these DR3⁺ mice with elevated TT₄ level (nos. 4 and 5) that were immunized with TSHR plasmid also showed statistically significant levels of TSAb and were thus hyperthyroid (Table 1). Two other mice in this group which showed significant enhancement of either TSAb (no. 6) or TT₄ (no. 2) may have transient thyroid stimulation at this point (Table 1), especially when considering the presence of TSAbs also in the week 8 and 11 sera of mouse no. 6 (denoted by asterisk). Because some mice had increased TT₄ levels over control mice, we also determined their TSH levels. No reduction in mTSH concentration was observed in those with elevated TT₄, with or without concomitant TSAbs (Table 1). The lower levels in females compared to males were as expected [21].

Table 1.

Anti-human TSHR antibody production, thyroid function and thyroid pathology in HLA-DR3 transgenic mice after immunization with human TSHR plasmid DNA only or in conjunction with IL-4 or GM-CSF plasmids

Immunogen†	Mouse No.	Sex	TT4 nmol/l‡	TSAbs fmol/ml§	TSBAbs % cAMP release¶	TBII %125I-TSH inhibition††	Flow cytometric analysis‡‡	mTSH mU/l§§	% thyroid infiltration¶¶
Vector + IL-4 control group	19	F	56	760	<30	<10	–	<16	0
	20	F	57	860	<30	<10	–	<16	0
	21	F	63	400	<30	<10	–	<16	>0–10
	22	M	66	900	<30	<10	–	31	0
	23	M	46	1500	<30	<10	–	52	0
Mean + 3SD	–		81	2080					–
TSHR	1	F	59	1240	44	43*	+ +	<16	0
	2	F	81	1130	<30	11	–	<16	0
	3	F	15	1110	<30	<10	+	<16	0
	4	M	96	4080	<30**	36*	+ +	48	>10–20
	5	M	85	6360*	<30	<10	–	36	0
	6	M	76	2440*	<30	<10	–	24	0
TSHR + IL-4	7	F	69	1160	<30	<10	+ +	ND	0
	9	F	86	1400	<30	<10	+ +	<16	>0–10
	10	F	62	960	<30	<10	+ +	ND	>0–10
	11	M	67	1500	<30	<10	+ +	ND	0
	12	M	62	1570	<30	<10	–	ND	>20–40
TSHR + GM-CSF	13	F	52	460	<30	<10	+ +	ND	0
	14	F	78	680	<30	<10	+	ND	0
	15	F	76	1240	<30	<10	+	ND	0
	16	F	79	840	<30	28	+	<16	0
	17	M	59	1720	<30	<10	+	ND	>0–10
	18	M	64	1810	<30	<10	–	60	0

^†HLA-DR3 transgenic Ab°/NOD mice were immunized with 50 µg each of the indicated DNA plasmids on weeks 0, 3 and 6. Sera were collected on weeks 8, 11 and 14 (when mice were sacrificed). The presented data were derived from week 14. All values considered positive were represented in bold.

^‡Total T₄ assessed using a RIA kit. Values above mean + 3SD of control group were considered positive.

^§Thyroid-stimulating Abs were measured by cAMP release from human TSHR⁺ JP09 cells after 4 h. Values above mean + 3SD of control group were considered positive;

^*denotes week 8 and 11 sera also positive.

^¶Thyroid-stimulating blocking Abs were measured by cAMP release from human TSHR⁺ JP09 cells in the presence of 40 µU/ml bTSH. Values ≥30% were considered positive;

^**denotes week 8 and 11 sera positive.

^††TSH-binding inhibitory immunoglobulins were determined with TRAK II kits. Values ≥10% of control group were considered positive;

^*denotes week 8 and 11 sera also positive.

^‡‡Serum-labelled GPI9-5 cells expressing human TSHR-ECD. Negative (–), positive (+) or strongly positive (+ +) denote fluorescent intensity.

^§§Mouse TSH serum level determined by double-antibody precipitation RIA.

^¶¶Extent of mononuclear cell infiltration determined from 50 to 60 sections of thyroid gland. ND, not determined.

In the groups immunized additionally with IL-4 or GM-CSF DNA, only one mouse (no. 9, TSHR + IL-4) was positive for elevated level of TT₄, suggesting no obvious contribution by either cytokine (Table 1). Because the group size for each cytokine was limited by the availability of transgenic animals, it is not possible at this time to conclude if the cytokines played any role. Thus, all data for TSHR DNA immunized mice were combined to facilitate comparison with the control group. In sum, TSHR DNA immunization led to 3/17 (∼20%) mice positive for TSAbs and 4/17 (∼25%) mice positive for elevated TT₄ levels. Two of these four were positive for both, giving a total of 5/17 (∼30%) mice with one or both signs of hyperthyroidism.

Additional TSHR Ab responses in DR3 transgenic mice

To characterize further the anti-TSHR response, analysis of serial bleeds from weeks 8, 11 and 14 showed four mice (no. 1, 2, 4 and 16) with TBII activity (Table 1). When the sera were assayed for TSBAb, only mouse no. 1 was positive at week 14, while mouse no. 4 was positive at weeks 8 and 11. TSAb and TSBAb levels were also examined in select mice over time. In mouse no. 6, week 8 serum was positive for TSAb, but no TSBAbs were detected at any time point. Figure 1 presents the data for the two hyperthyroid mice. Mouse no. 4 was positive for TSBAb, but negative for TSAb at weeks 8 and 11; at week 14, TSAbs were observed, whereas the TSBAbs had declined. TSAb levels in mouse no. 5 were high, but the TSBAb levels were low in all weeks tested.

Fig. 1.

Relative TSAb and TSBAb levels in DR3 transgenic mice at intervals after genetic immunization with human TSHR. Sera from weeks 8, 11 and 14 for mice with hyperthyroidism (nos. 4, 5) were evaluated. TSAbs were measured by cAMP production of TSHR⁺ CHO cells cultured with sera. Values greater than the mean + 3SD of control mice were considered positive (as indicated by dotted line). TSBAb levels were determined by cAMP production of TSHR⁺ CHO cells in the presence of bovine TSH. Values of ≥30% were considered positive (as indicated by dotted line).

Anti-TSHR response of sera from immunized mice was also analysed by flow cytometry with TSHR-ECD⁺ cells; 70% were positive (Table 1). Two mice, no. 1 and no. 4, had TSAb and TBII. An example of the differences in fluorescent intensity of labelled cells is shown (Fig. 2c,d).

Fig. 2.

Flow cytometric analysis of TSHR⁺ cells for TSHR Abs from sera of TSHR DNA-immunized mice. TSHR-ECD⁺ GPI9-5 cells were labelled with sera (1 : 5), followed by FITC-conjugated antimouse IgG. (a) labelling with TSHR-specific monoclonal Ab A10 (positive control in bold). (b) sera from individual control mice nos. 19–23, overlaid in histogram. Serum from a control mouse (no. 20) with the highest background labelling was used for subsequent comparisons. (c) week 8 serum from mouse no. 3 representing (+) binding in bold. (d) week 8 serum from mouse no. 4 representing (+ +) binding in bold.

Thyroid pathology

The thyroids of all animals were examined for pathology. Foci of perivascular infiltration were observed in some mice; 2 mice (nos. 4 and 12) had infiltration resulting in thyroid destruction (Table 1). Figure 3 shows the typical mononuclear cell infiltration involving >10–20% of the thyroid consistent with autoimmune thyroiditis (mouse no. 4). At the higher magnification (200×), follicular destruction is clearly visible. Mouse no. 12 also showed thyroid destruction involving >20–40% of thyroid. Interestingly only this animal had a low level of mouse Tg autoantibodies (1 : 50), whereas all others had none (data not shown). Three other TSHR-immunized mice displayed minimal but significant infiltration of >0–10%. But, because of their NOD autoimmunity-prone background, with one vector control mouse (no. 21) also showing similar infiltration, it is unclear how much of this low level of thyroiditis was attributable to TSHR and/or cytokine stimulation.

Fig. 3.

Mononuclear cell infiltration in the thyroid of TSHR DNA-immunized mice. Thyroid sections from (a) vector control mouse, (b) mouse no. 4 with infiltration involving >10–20% of the gland (original magnification ×100) and (c) at higher magnification showing destruction of follicular architecture by infiltrating cells (×200).

DISCUSSION

We demonstrate in this report that the DRB1*0301 transgene, expressed in a murine class II-negative NOD strain, was sufficient as a susceptibility allele for a GD model. As reported previously [18], this NOD strain lacks endogenous class II molecules and cannot respond to mouse or human Tg immunization in the absence of the DR3 transgene. In the presence of DR3, about 30% of mice developed Graves’-like hyperthyroidism or some sign of the disorder at week 14 after immunization with human TSHR plasmid DNA. TSAbs were detected as early as week 8, 2 weeks after the third immunizing dose. As we also assayed for TBII and TSHR Ab by flow cytometry at week 8, we were unable to perform TT₄ until week 14 when more serum could be obtained at sacrifice. At week 14, both TSAbs and elevated TT₄ were detected in approximately 12% of DR3⁺ mice, with another mouse showing TSAbs but TT₄ below the mean + 3SD of control group. We did not observe simultaneous presence of TSAbs and TSBAbs over time in the hyperthyroid mice (weeks 8–14, Fig. 1); either TSBAbs were not present or they had declined when the TSAb level increased (mouse no. 4, week 14), similar to the findings in many outbred NMRI mice [9].

Although mice that developed TSAbs and/or elevated TT₄ were males rather than females, the number of animal is too small to assess if gender preference exists in this transgenic strain. No gender preference has been observed in this strain for Tg-induced autoimmune thyroiditis [17,18]. The hyperthyroid male mice also did not display suppressed TSH level, as reported for hyperthyroid female mice in the NMRI strain [9]. Since female mice of several tested strains normally have much lower TSH level than males [21], it may be more difficult to show reduced TSH level beyond the normal fluctuation in male mice. We were also limited by the number of DR3 transgenic mice available at a given time for comparison of coimmunization with cytokine plasmids. In an attempt to promote TSAb production, we selected plasmids encoding IL-4, a Th2 cytokine, rather than IL-12, a proinflammatory cytokine favouring Th1 development. And again, constrained by the availability of DR3⁺ mice, we included the IL-4 plasmid in the vector control group, reasoning that, if mice given vector + IL-4 plasmids were negative, then vector alone would have no effect. Indeed, all such vector controls were negative in all assays. Apart from the presence of high levels of TSHR Abs detectable by flow cytometry, we observed no obvious influence attributable to IL-4 or GM-CSF, a Th1/Th2 cytokine, similar to reports in BALB/c [10] and outbred CD-1 [23] mice given GM-CSF or M(macrophage)-CSF and IL-4 or IL-12, respectively. Cytokine manipulation by the use of interferon-γ knockout BALB/c mice also did not alter the Th1/Th2 bias after genetic immunization [14,15].

In terms of pathology, no follicular changes relating to hypertrophy were noted, similar to the findings in BALB/c [11,23] and outbred CD-1 [23] mice. Cellular changes were reported in outbred NMRI mice [9] and two inbred strains, BALB/c and C57BL/6 but not several other haplotypes, using adenovirus-expressed TSHR as immunogen [10]. In contrast to hypercellularity, some DR3 transgenic mice displayed foci of mononuclear infiltration, as seen in BALB/c mice [6]. But since one vector control had similar low grade infiltration (> 0–10%) which could have stemmed from its NOD background (approximately 5% wildtype NOD develop thyroiditis at about 6 months [24]), we cannot be certain if the mild infiltration in three TSHR-immunized mice was related to immunization or background infiltration for this strain. However, two mice did display thyroiditis accompanied by tissue damage (nos. 4 & 12, Table 1); of all the animals, only one (no. 12) had low titres of mouse Tg Abs and one (no. 4) was hyperthyroid. This extent of thyroiditis could reflect the influence of TSHR and/or cytokine immunizations. It is not uncommon to observe GD patient thyroids with foci of thyroid inflammation [25].

Clearly, none of the murine models of Graves’ hyperthyroid disease induced by the several protocols of plasmid DNA immunization in inbred or outbred strains is ideal. While DNA immunization eliminates the need for syngeneic cells as a vehicle, the use of outbred strains makes genetic analysis of susceptibility factors difficult [9]. And, although adenovirus-TSHR immunization appears to be the most successful across several, yet not all, inbred strains [10] and in hamsters [26], special animal housing for infectious agent may be required. For our study, we selected the DR3⁺g7^– transgenic mice because of previous findings that the autoimmunity-prone NOD background enables them to become more susceptible to thyroiditis induction with human Tg [17,18] than mice on B10 background [16]. Here, we demonstrate further that GD-like parameters may also be induced in these mice. Our findings support the important role of DR3 molecules in humans for the development of both HT and GD [2,3]. Armed with the possible utility of this humanized strain for a GD model, we have begun exploring ways to improve the incidence of aproximately 12%. Studies are underway to include other cytokine plasmids (such as IL-12) and electroporation to enhance TSHR incorporation as reported in other systems [27]. Moreover, using this DR3 transgenic strain, we have identified several pathogenic human Tg epitopes based on computer-predicted, DR3-binding motifs [28]. Similar strategy may be applied to predicting and determining important TSHR epitopes in the future.