Immunotherapy With Apitopes Blocks the Immune Response to TSH Receptor in HLA-DR Transgenic Mice

Abstract

We have combined major histocompatibility complex–binding assays with immunization and tolerance induction experiments in HLA-DR3 transgenic mice to design apitopes (antigen-processing independent epitopes) derived from thyrotropin receptor (TSHR) for treatment of patients with Graves’ disease (GD). A challenge model was created by using an adenovirus-expressing part of the extracellular domain of the thyrotropin receptor (TSHR₂₈₉). This model was used to test whether current drug treatments for GD would have an impact on effective antigen-specific immunotherapy using the apitope approach. Furthermore, selected peptides were assessed for their antigenicity using peripheral blood mononuclear cell samples from patients with GD. A mixture of two immunodominant apitopes was sufficient to suppress both the T-cell and antibody response to TSHR when administered in soluble form to HLA-DR transgenic mice. Tolerance induction was not disrupted by current drug treatments. These results demonstrate that antigen-specific immunotherapy with apitopes from TSHR is a suitable approach for treatment of GD.

Graves’ disease (GD) is a common autoimmune disorder affecting the thyroid gland. The disease has a prevalence of 0.5% to 2.0%, with a strikingly rapid increase in incidence (1, 2); moreover, ~90% of people affected by GD are women. Stimulating autoantibodies bind to the thyrotropin receptor (TSHR) and cause the release of T4 and T3 (3). These stimulating autoantibodies are predominantly IgG isotype and, therefore, depend on CD4⁺ T-cell help for their generation (4). TSHR is found on thyroid epithelial cells, orbital fibroblasts, and adipose tissue (5). This, in part, explains why people with GD can develop Graves’ orbitopathy (GO). Almost all patients with GO have anti-TSHR antibodies, implying that the immune response to TSHR underpins the pathogenesis of both GD and GO (6). In addition, antibodies to the insulin-like growth factor I receptor are implicated in the pathogenesis of GO (7).

A recent Cochrane report comparing radioiodine therapy with antithyroid medication for GD (8) showed that GO developed and worsened in 36% of radioiodine-treated patients compared with 19% of patients treated with methimazole (MMI). Normal thyroid function was not seen in patients receiving radioiodine treatment, 95% of whom were hypothyroid. By contrast, 94% of individuals treated with MMI retained thyroid function; however, this group was more likely to experience a disease relapse. These data demonstrate that the current treatments for GD do not adequately prevent progression to GO or prevent relapsing disease while retaining normal thyroid function. This is primarily because neither of these treatments for GD controls the underlying immunopathology of the disease. We propose that an antigen-specific immunotherapy based on CD4⁺ T-cell epitopes from TSHR will provide such a therapy. This antigen-specific immunotherapy is designed to functionally deprive B cells of the T-cell help required for production of stimulating autoantibodies through tolerance induction in TSHR-specific T cells.

It is clear that antigen-specific immunotherapy with intact antigen can be applied to hypersensitivity conditions, notably in allergic disorders. Indeed, allergic desensitization with intact allergens has been an effective means of controlling allergy for more than a century (9, 10). Attempts to apply this to autoimmune diseases, including multiple sclerosis (11), type 1 diabetes (12), and GD (13), have been hampered by induction of pathogenic antibodies or cytotoxic T cells. We have argued that these complications can be avoided by designing tolerogenic CD4⁺ T-cell epitopes to replace intact antigens (14). Indeed, peptides based on T-cell epitopes have been broadly effective in experimental models of hypersensitivity conditions and have now translated through to clinical trials of allergic diseases (15, 16) and autoimmune diseases, including systemic lupus erythematosus (17), type 1 diabetes (18, 19), and multiple sclerosis (20).

Effective therapeutic peptides must follow two rules. Primarily, they must bind to major histocompatibility complex (MHC) class II molecules in the same conformation as the naturally processed T-cell epitope to engage disease related T cells (21). Second, they must be highly soluble because insoluble peptides are immunogenic, whereas soluble derivatives of the same peptide are tolerogenic (22). Peptides mimicking naturally processed CD4⁺ T-cell epitopes are defined as antigen processing independent epitopes, or apitopes. Administration of apitopes induces type 1 regulatory (Tr1)-like T cells with immunosuppressive properties (23–26); such an approach has the potential for treatment of GD and prevention of GO. Here we demonstrate that TSHR-derived apitopes profoundly suppress both CD4⁺ T-cell and T-dependent antibody responses to TSHR in mice transgenic for the GD-associated HLA class II DRB1*0301 gene.

Materials and Methods

Mice

HLA DR3 transgenic mice used in this study were bred at Charles River Laboratories (Margate, United Kingdom) and at InnoSer (Lelystad, Netherlands). HLA DR3-tg founder mice were obtained from Gunter Hämmerling (German Cancer Research Center, Heidelberg, Germany) (27). Briefly, a 6-kb NdeI fragment of an HLA DRA genomic clone in pUC and a 24-kb ClaIxSalI fragment of cos 4.1 containing the B gene were coinjected into fertilized eggs from (C57BL/6xDBA/2)F1 donors mated with C57BL/6 males. The DR4tg mouse strain was originally created by Lars Fugger et al. (28). Endogenous MHC class II deletion was achieved by breeding onto the I-A-β knockout (B6;129S2-H2dlAb1-Ea/J) mouse. Both female and male mice were included in every experimental group. For the establishment of the adenovirus expressing a part of the TSHR_ECD (Ad-TSHR) model, 6-week-old female BALB/cJOlaHsd mice (Harlan Laboratories, Venray, Netherlands) were purchased.

All animals were used in experiments at 6 to 10 weeks of age. All animal studies were approved by the Ethical Committee for Animal Experiments at Hasselt University and performed with the highest standards of care in a pathogen-free facility.

Antibodies, proteins, and peptides

Human recombinant extracellular domain of TSHR (TSHR_ECD, AA19-417) was produced in a Trichoplusia ni larval expression system by using the Chesapeake PERL technology PERLXpress (Savage, MD). Peptides were synthesized by GL Biochemistry Ltd (Shanghai, China) or by Severn Biotech Ltd (Severn, Kidderminster Worcs, United Kingdom) with a C-terminal amide (>95% purity) or by PolyPeptide Laboratories (Strasbourg, France) as HCl-salt (technical batch > 95% purity). Peptide sequences are described in Table 1 and cocktails were designed as follows: peptide 4KG: KKGNLPNISRIYVSIDVTGKK; peptide 5DK: KKKKYVSIDVTLQQLESHKKK; peptide 9BN: GLKMFPDLTKVYSTD; ATX-GD-459: an equal-parts mixture of peptides 4KG, 5DK, and 9BN; ATX-GD-59: an equal-parts mixture of peptides 5DK and 9BN.

Table 1.

Characteristics of Peptides Derived From TSHR

Peptide Name	Sequence	Number a	Solubility	Immunogenicity in HLA-DR3 Transgenic	Presentation by Live APC	Presentation to Hybridoma by Fixed APC	Antigenicity PBMC From Patients
4	LRTIPSHAFSNLPNISRIYVSIDVTLQQL	64–92	−	+	+	−	+
4K	NLPNISRIYVSIDVT	74–88	−	+	+	+	+
4KG	KKGNLPNISRIYVSIDVTGKK	74–88b	+	+	+	+	+
5	ISRIYVSIDVTLQQLESHSFYNLSKVTHI	78-107	−	+	+	−	+
5D	IYVSIDVTLQQLESH	81–95	−	+	+	+	+
5DK	KKKKYVSIDVTLQQLESHKKK	82–95b	+	+	+	+	+
9	TGLKMFPDLTKVYSTDIFFILEITDNPYM	136–164	−	+	+	NT	+
9B/9BN	GLKMFPDLTKVYSTD	137–151	+	+	+	NT	+

DR3tg mice were immunized with TSHR-derived 30-mer peptides, and peptides 4, 5, and 9 induced strong peptide-specific splenocyte proliferation. The IL-2 production of TSHR-specific hybridoma T-cell clones was measured on presentation of peptides 4KG and 5DK by live or fixed APCs. Isolated PBMCs from patients with GD were cultured with peptides 4, 4K, 4KG, 5, 5D, 5DK, 9, and 9BN, and peptide-specific proliferation (stimulation index > 2) was determined by using 3H-thymidine incorporation.

Abbreviations: +, positive assay response/peptide property; −, negative assay response/peptide property; NT, not tested.

^a

UniProtKB database (https://www.uniprot.org/) protein P16473.

^b

Core sequence amino acid numbers in the TSHR sequence: these peptides were modified to optimize solubility.

Peptide characterization

Immunogenicity was tested by detecting antigen-specific T-cell proliferation in DR3tg mice or by measuring IL-2 secretion (R&D Systems, Oxon, United Kingdom) of T-cell hybridoma clones specific for TSHR using live or paraformaldehyde-fixed antigen-presenting cells (APCs) (29). T-cell hybridomas were prepared as described previously (21). For in vivo immunogenicity, DR3tg mice were immunized subcutaneously in the base of the tail with 30 nmol parental peptide (nonmodified 4K, 5D and 9B or 5D and 9B) emulsified in 100 µL complete Freund’s adjuvants [CFAs; 2 mg/mL Mycobacterium tuberculosis H37RA (Difco Laboratories, Detroit, MI) in incomplete Freund adjuvant (Difco)]. Ten days after immunization, draining lymph nodes and spleens were harvested. Lymph node and spleen single-cell suspensions were prepared and cultured in X-vivo 15 medium supplemented with 2 mM l-glutamine, 50 U/mL penicillin, and 50 U/mL streptomycin (all Lonza, Verviers, Belgium) in 96-well flat-bottom plates. To investigate antigen-induced T-cell proliferation, 0.5 × 10⁶ cells per well were cultured (200 µL per well) for 72 hours with TSHR_ECD in a concentration range of 0 to 25 μg/mL or with 12.5 μg/mL purified protein derivative (PPD; immunization control; Statens Serum Institut, Copenhagen, Denmark). Peptide-induced T-cell proliferation was measured using ³H-thymidine incorporation.

The Apitope^® (Diepenbeek, Belgium) property for 5DK and 4KG peptides (binding MHC class II molecules directly without APC uptake and intracellular loading to the MHC class II molecule) was defined by using antigen-specific T-cell hybridomas. The peptides were presented in culture by fresh or formaldehyde-treated [fixed in 0.5% paraformaldehyde at a pH of 7 (Merck, Darmstadt, Germany) for 5 minutes at room temperature] VAVY (HLA-DR3) or BM14 (HLA-DR4) cells. After 48 hours, peptide-induced IL-2 production by the antigen-specific T-cell hybridoma clone was measured by ELISA (mouse IL-2 ELISA kit, R&D Systems).

Peptide solubility was measured from a stock of 20 mg/mL in dimethyl sulfoxide, where the test peptide was diluted in PBS to obtain concentrations of 4, 2, and 1 mg/mL. Peptide solutions were left at room temperature for 16 hours to allow any precipitate to form. To separate precipitated/colloidal peptide from truly dissolved solute, peptide solutions were centrifuged at 20,800 g for 10 minutes. The absorbance of the peptide solution before and after centrifugation was measured at 205, 280, and 320 nm on the NanoDrop spectrophotometer (Thermo Fisher, Aalst, Belgium), and the peptide concentration was calculated by the Nanodrop 2000 software.

Peripheral blood mononuclear cell specificity tests

Peripheral blood mononuclear cell (PBMC) collection from patients with GD received ethical approval in the United Kingdom (12/NE/0101) and Belgium (B403201316268). Isolated PBMCs from patients with GD were cultured at 4 × 10⁶ cells per milliliter (RPMI-1640 supplemented with 20 mM HEPES, 50 U/mL each of penicillin and streptomycin, 2 mM l-2 glutamine (all Lonza, Verviers, Belgium), and 5% heat-inactivated male AB serum (Sigma-Aldrich, St. Louis, MO) with 30-mer and 15-mer peptides in a 24-well plate in a 37°C incubator with 5% CO₂. At days 4, 6, and 8, cell cultures were mixed, three 100-µL aliquots were transferred to a 96-well plate, and 1 µCi ³H-thymidine (PerkinElmer, Waltham, MA) was added to each well for the next 18 hours, followed by harvesting and counting in a 1450 Wallac MicroBeta Counter (PerkinElmer). A positive response to peptide is defined as stimulatory index ≥ 2.

MHC class II binding

In silico predictions for HLA DR binding epitopes in TSHR were performed by using the NetMHCII 2.2 server and the Immune Epitope DataBase consensus method. The HLA-DR binding affinity of the peptides was measured in the Reveal assay (ProImmune, Oxford, United Kingdom) (Table 2). In this cell-free competition assay, the test peptide competes with a known DR haplotype–specific high-affinity peptide for binding to soluble HLA-DR molecules with a read-out of binding of a labeled antibody specific for the control peptide/HLA-DR complex. The IC50 for each test peptide was generated after 12 different concentration dose-response curves competing with a fixed concentration of the DR-specific peptide. The use of different competitor control peptides with different binding affinity for each HLA-DR molecule prevents comparison of peptide-binding affinity between HLA-DR haplotypes but enables ranking of the peptides for a certain HLA haplotype.

Table 2.

Relative Peptide-MHC Class II Binding Affinity of TSHR-Derived Peptides

MHC Class II Allele	Peptide-MHC Binding in Affinity Order, High to Low
HLA-DRA*01:01 HLA-DRB1*01:01	9BN > 5DK
HLA-DRA*01:01 HLA-DRB1*15:01	4KG ≥ 9BN > > 5DK
HLA-DRA*01:01 HLA-DRB1*03:01	5DK > > 9BN > 4KG
HLA-DRA*01:01 HLA-DRB1*04:01	5DK > > 4KG > 9BN
HLA-DRA*01:01 HLA-DRB1*11:01	5DK > > 9BN ≥ 4KG

Each peptide was included in a cell-free competition assay competing for binding to soluble HLA-DR molecules in the presence of a fixed concentration of a known control, high-affinity peptide. Test peptide concentration ranged from 0 to 300 μM, and 12 different concentrations were included to generate the IC50 in micromolars. Because the control peptide is HLA-DR molecule–specific, the IC50 cannot be compared between HLA-DR haplotypes but allows a ranking of test peptides within the individual HLA-DR haplotype assay.

Ex vivo tolerance experiments

DR3tg mice were injected with the peptide cocktail at 45 pmol, 450 pmol, and 4500 pmol of ATX-GD-459 or ATX-GD-59 subcutaneously in the flank region on days −15, −13, and −11, respectively, followed by three injections of a top dose of 33.75 nmol when ATX-GD-459 was administered or a top dose of 45 nmol when ATX-GD-59 was administered on days −8, −6, and −4 (dose escalation schedule). ATX-GD-459 includes the peptides 4KG, 5DK, and 9BN, and ATX-GD-59 consists of the peptides 5DK and 9BN. Control mice received PBS or a control peptide, which was a soluble, HLA-DR3–binding, TSHR-irrelevant peptide from another project. Group sizes ranged from 9 to 11, and equivalent numbers of female and male mice were included in each group. The peptides were administered by using PBS as vehicle. For antigen-challenge experiments, mice were immunized with 30 nmol parental peptide (nonmodified 4K + 5D + 9B or 5D + 9B) in CFA as described above (see "Peptide characterization").

Adenoviral model for GD

Six-week-old DR3tg mice were immunized by intramuscular injection with an adenoviral vector expressing the truncated version of Ad-TSHR [TSHR 1-289(Viraquest, North Liberty, IA) or β-galactosidase (Ad-LacZ)] (30). Access to the Ad-TSHR289 construct was provided by Basil Rapoport and Sandra McLachlan, Thyroid Autoimmune Disease Unit, Cedars-Sinai Research Institute and University of California, Los Angeles, California. A total of 10¹⁰ adenoviral particles were injected at weeks 0 and 3. Five weeks after the first immunization, the experiment was terminated. Spleens were collected to assess TSHR-specific T-cell proliferation, and serum was collected before and 2 and 5 weeks after the first immunization to measure anti-TSHR antibody levels.

To validate the adenoviral GD model using MMI, DR3tg mice were immunized with 10¹⁰ Ad-TSHR viral particles at week 0 and week 3 and received vehicle treatment or MMI treatment (Sigma-Aldrich, Bornem, Belgium; n = 8) starting on the day of the first immunization and continuing for 4 weeks until the end of the experiment. MMI was dissolved in sterile water to give a dose of 500 µg per mouse per day and was administered subcutaneously by ALZET osmotic pumps (model 1004, 0.11 µL/h; Charles River, L’Arbresle, France). To validate the model with methylprednisolone, DR3tg mice were immunized with 10¹⁰ Ad-TSHR viral particles at week 0 and week 3 and received vehicle treatment or MPred treatment (Sigma-Aldrich, Bornem, Belgium) starting 3 days before the first immunization and continued for 4 weeks until the end of the experiment. MPred was dissolved in sterile water and administered subcutaneously by ALZET osmotic pumps (model 1004, 0.11 µL/h) at a dose of 7 mg/kg/d.

The prophylactic effect of ATX peptide treatment was tested by injecting DR3tg mice subcutaneously in the flank region with ATX-GD-459 or ATX-GD-59 and control treatment according to the dose escalation schedule. The control treatment was PBS or a soluble HLA-DR3–binding peptide from another project. Treatment was followed by an intramuscular injection with 10⁹ to 10¹⁰ Ad-TSHR or Ad-LacZ on week 0 and week 3. Different doses of viral particles were tested to determine whether thyroid-stimulating antibodies could be generated in HLA-DR3 mice. Serum samples were collected at different time points, and the experiment was terminated 5 weeks after the first immunization.

Comedication of ATX-GD-459 and MMI was tested by injecting DR3tg mice with ATX-GD-459 peptide or control peptide according to the dose escalation schedule with a 15-nmol top dose for each peptide (total, 45 nmol). The control peptide was a soluble, HLA-DR3–binding peptide from another project. All mice were injected intramuscularly with 10⁹ Ad-TSHR viral particles at week 0 and week 3. Starting on the day of the first peptide treatment until termination of the experiment, mice were treated with a daily dose of 500 µg MMI via subcutaneous administration using ALZET osmotic pumps (model 1004, 0.11 µL/h). New osmotic pumps were implanted after 4 weeks according to the same procedure. Serum samples were collected at different time points, and the experiment was terminated 5 weeks after the first immunization.

Comedication of ATX-GD-459 and propranolol was tested by injecting DR3tg mice with ATX-GD-459 peptide or control peptide according to the dose escalation schedule with a 15-nmol top dose for each peptide (total, 45nmol peptides). Starting on the day of the first peptide treatment until termination of the experiment, mice received daily intraperitoneal injections of vehicle or treatment with freshly dissolved propranolol hydrochloride at a dose of 10 mg/kg/d (Sigma-Aldrich, Bornem, Belgium). All mice were injected intramuscularly with 10¹⁰ Ad-TSHR viral particles at week 0 and week 3. Plasma samples were collected at different time points, and the experiment was terminated 5 weeks after the first immunization. After the last dose, plasma samples were collected 1 hour after the dose and analyzed for plasma levels of propranolol by liquid chromatography/tandem mass spectrometry (Anacura, Evergem, Belgium).

Detection of anti-TSHR antibodies

Ninety-six–well ELISA plates (Corning Costar; Sigma-Aldrich, Bornem, Belgium) were coated overnight at room temperature with TSHR_ECD protein in PBS (0.5 µg/mL). After blocking with 1% BSA in PBS, mouse sera were tested at 1:50 or 1:500 by incubation for 1 hour at room temperature. Mouse anti-TSHR antibody [Abcam, Cambridge, United Kingdom; catalog no. ab6047; RRID: AB_305256 (31)] was used as a positive control. Antibody binding was then detected with horseradish peroxidase–conjugated goat anti-mouse IgG [Abcam, catalog no. ab6789; RRID: AB_955439 (32)]. To detect anti-TSHR antibodies of IgG1, IgG2b, and IgG2c isotypes, horseradish peroxidase–conjugated rat anti-mouse IgG1 [Southern Biotech, Birmingham, AL; catalog no. 1144-05; RRID: AB_2734757 (33)], goat anti-mouse IgG2b [Abcam, catalog no. 97250; RRID: AB_10695945 (34)], and goat anti-mouse IgG2c [Abcam, catalog no. 97255; RRID: AB_10680258 (35)] antibodies were used, respectively. The signal was developed with tetramethylbenzidine (TMB) and optical density was measured in a plate reader at 450 nm (Tecan Benelux, Mechelen, Belgium).

Detection of hyperthyroidism

The presence of thyroid-stimulating antibodies in mouse serum was detected by using lulu* cells, which express both the human TSHR and a cAMP-responsive luciferase reporter construct, as described previously (36). Cells were kindly provided by Professor M. Ludgate (Cardiff University, Cardiff, United Kingdom). In brief, the cells were seeded in Ham F12 medium with 10% charcoal stripped serum (Sigma-Aldrich, Bornem, Belgium) and then incubated the next day for 4 hours at 37°C in Ham medium containing 5% polyethylene glycol and 10% test serum. The human stimulatory antibody 08/204 was used as a positive control for cAMP production [National Institute for Biological Standards and Control, Hertfordshire, United Kingdom; catalog no. 08/204; RRID: AB_2734758 (37)]. After washing and air drying, cells were lysed by using luciferase cell culture lysis buffer (Promega, Leiden, Netherlands). The cAMP responsive luciferase production was measured by a luciferase reporter assay (Promega) according to the manufacturer’s instructions, and light emission was measured with a FLUOstar omega luminometer (BMG Labtech, Ortenberg, Germany).

Total T4 was measured in undiluted mouse serum by using the CBI mouse/rat thyroxine ELISA kit (Calbiotech, Spring Valley, CA) according to the manufacturer’s instructions. T4 values were computed from standards in the kit and are expressed as micrograms per deciliter.

Statistical analysis

Differences between the groups were analyzed by ANOVA for comparison between groups using Prism software (GraphPad Software Inc., La Jolla, CA), followed by Bonferroni post hoc testing. Sex was not considered a factor in the statistical analysis of the data.

Results

Peptide selection

The genes most strongly associated with GD map within the HLA class II region of the MHC. Association of HLA class II DRB1, DQB1, and DQA1 loci with GD reveals a predisposing effect of DR3 (DRB1*03-DQB1*02-DQA1*05) and a protective effect of DR7 (DRB1*07-DQB1*02-DQA1*02) (38). We therefore chose to test the immunogenicity of peptides from the human TSHR_ECD in HLA-DR3 transgenic mice. Studies were also conducted in HLA-DR4 transgenic mice to identify immunogenic, pan-DR–binding epitopes. Mice were immunized with overlapping 30-mer peptides from the TSHR_ECD; this revealed five immunodominant regions from which peptides 4, 5, and 9 were selected as immunodominant because these peptides also were found to be immunodominant in humans (see Table 1 for details). MHC binding predictions indicated that epitopes within these 30-mer peptides were capable of binding to HLA-DRB1*03:01 and other DR alleles; therefore, it seemed likely that we would identify pan-DR–binding epitopes within these regions.

T-cell hybridomas were generated from HLA-DR3 and HLA-DR4 transgenic mice immunized with 30-mer peptides or TSHR_ECD. Fine specificity analysis of peptides 4 and 5–specific hybridomas was undertaken using 15-mer peptides extending across the 30-mer sequence by single–amino acid shifts. This mapped immunodominant epitopes within peptides 4 and 5 to the 4K and 5D sequences, respectively (Table 1). Identification of peptide 9B arose from screens of overlapping 15-mer peptides using freshly isolated lymphocytes from HLA-DR3 transgenic mice immunized with TSHR_ECD because T-cell hybridomas were not available.

Peptide optimization

Previous studies emphasized the importance of peptide solubility in antigen-specific immunotherapy (22). Initial studies showed that peptides 4K and 5D were poorly soluble in aqueous solution, whereas peptide 9B was soluble. Insoluble peptides were modified at their N- and C-termini with double or triple lysine motifs (KKK) with or without a glycine spacer (KKG or GKK). The optimal modification was assessed by comparing the solubility and antigenicity of the resulting peptide analogs in vitro. Peptides were modified with double lysine and a glycine spacer for peptide 4K (4KG) and the triple lysine motif for 5D (5DK) (Table 1). 9BN is an unmodified version of the 9B sequence that was soluble in its native form. We have previously shown that peptides behaving as antigen-processing independent epitopes, apitopes, are tolerogenic (20). The ability of peptides 4KG and 5DK to function as apitopes was confirmed by their ability to stimulate IL-2 production from relevant T-cell hybridomas when the peptides were presented by paraformaldehyde-fixed APC (Table 1).

Peptide-MHC binding properties

MHC-peptide competition binding assays were conducted by Proimmune Ltd. These revealed relative binding affinities for the selected peptides (Table 2). Among the five HLA-DR alleles tested, peptide 4KG bound to four alleles but failed to bind with measurable affinity to HLA-DR1. Both peptides 5DK and 9BN bound all five alleles, although with varying relative affinities. These results confirm the in silico analysis, which predicted that these peptides were pan-DR binders.

Tolerance induction

Previous studies have shown that subcutaneous injection of soluble CD4⁺ T-cell epitopes induces immunologic tolerance in mice (20, 26). HLA-DR3 transgenic mice received a dose escalation of peptides followed by three injections at the highest dose. A mixture of peptides 4KG with 5DK and 9BN (ATX-GD-459) was compared with a mixture containing 5DK and 9BN alone (ATX-GD-59). Mice were then immunized with the relevant parental 30-mer peptides in CFA (i.e. for ATX-GD-59 studies mice were immunized with parent peptides 5 and 9) (Table 1), before measurement of splenocyte and lymph node T-cell responses to TSHR_ECDin vitro. As shown in Fig. 1, results of positive control immunizations (filled symbols) show that the parent peptides induced a strong immune response to the TSHR_ECD in both draining lymph node and spleen when the parent peptides were injected in CFA. However, both ATX-GD-59 and ATX-GD-459 were capable of suppressing the immune response to TSHR_ECDin vitro when mice were pretreated with soluble peptides before to immunization.

Figure 1.

ATX-GD-59 treatment reduces TSHR-induced proliferation in HLA-DR3 transgenic mice. (A and B) DR3tg mice were pretreated with ATX-GD-459 (an equal-parts mixture of peptides 4KG, 5DK, and 9BN) (n = 10) or PBS (n = 10) according to the dose escalation schedule, with a 34-nmol top dose for each peptide. Animals received an immunization of 50 µg of each parental peptide 4K, 5D, and 9B in CFA and, after 10 days, lymph nodes (LNs) and spleens were harvested to assess the TSHR-specific proliferation. Data represent mean ± SEM of stimulation index (SI) values for the control-treated mice (filled symbols) and peptide-treated mice (open symbols). (C and D) DR3tg mice were pretreated with ATX-GD-59 (an equal-parts mixture of peptides 5DK and 9BN) (n = 9) or control peptide (n = 10) according to the dose escalation schedule, with a 45-nmol top dose for each peptide. Animals received an immunization of 50 µg of each parental peptide 5D and 9B in CFA and, after 10 days, LNs and spleens were harvested to assess the TSHR-specific proliferation. Data represent mean ± SEM of SI values for the control-treated mice (filled symbols) and peptide-treated mice (open symbols). Two-way ANOVA was used to measure overall treatment effects on T-cell proliferation; P values are provided in the graphs. Bonferroni post hoc testing was used; significant differences are indicated in the graphs (*P < 0.05; **P < 0.01; ***P < 0.001; ****P < 0.0001). The average percentage reduction in T-cell proliferation induced by peptide treatment is shown. Data shown are representative of two independent and reproducible experiments.

Ad-THSR model

To assess the ability of ATX-GD-59 and ATX-GD-459 to induce tolerance in a relevant model of anti-TSHR antibodies, HLA-DR3 transgenic mice were immunized with adenovirus encoding a truncated version of the TSHR_ECD equivalent to the A-subunit, as described previously (30, 39). Preliminary studies revealed that the peak of antibody production occurred at week 2 after the first adenovirus administration, with heightened levels of anti-TSHR antibodies detected through week 5 after the second immunization at week 3. Anti-TSHR antibody levels fell consistently after week 5, making further immunizations uninformative. We did not detect hyperthyroidism in the HLA-DR3 mouse upon side-by-side comparison with the original BALB/c mouse strain used (30). We were able to reproduce long-term elevated anti-TSHR antibody responses, remaining high at week 10, with raised T4 levels in the BALB/c mouse model. The relatively poor antibody response to Ad-TSHR in the HLA-DR3 transgenic mice resulted from differences in background genetics because (BALB/c x DR3Tg)F1 mice maintained a high antibody titer through to week 10; however, T4 levels were not raised in this F1 strain (data not shown).

The validity of the Ad-THSR model of anti-THSR antibody production was confirmed by demonstrating that methimazole suppressed steady-state T4 levels without changing anti-TSHR antibody levels (Fig. 2A and 2B). Furthermore, MPred suppressed anti-THSR antibodies and, as expected, drastically reduced the cellularity of both thymus and spleen (Fig. 2C–2E). These data demonstrate that the Ad-TSHR model of anti-TSHR antibody production behaves as expected in response to commonly used drugs for treatment of GD.

Figure 2.

Validation of adenoviral GD model using MMI and MPred. (A and B) DR3tg mice were immunized with 10¹⁰ Ad-TSHR at week 0 and 3 and received vehicle (n = 8) or MMI (n = 8) administered by subcutaneously implanted osmotic pumps. Serum was collected 2 weeks after the first immunization, and anti-TSHR IgG levels and T4 levels were analyzed by ELISA. One-way ANOVA was used to measure overall differences in anti-TSHR IgG levels and T4 levels. Bonferroni post hoc testing was used; significant differences are indicated in the graphs (*P < 0.05; **P < 0.01; ***P < 0.001; ****P < 0.0001). (C–E) DR3tg mice were immunized with 10¹⁰ Ad-TSHR at week 0 and 3 and received vehicle treatment (n = 10) or MMI treatment (n = 10). Serum was collected 2 weeks after the first immunization, and anti-TSHR IgG levels were analyzed by ELISA. Thymus weight and total number of splenocytes were determined on termination of the experiment. Each dot represents data from one mouse, and the mean ± SEM is shown per group. One-way ANOVA and Bonferroni post hoc testing were used; significant differences are indicated in the graphs (*P < 0.05; **P < 0.01; ***P < 0.001; ****P < 0.0001).

Tolerance induction in the Ad-TSHR model

HLA-DR3 transgenic mice received a dose escalation of peptides ATX-GD-459 or ATX-GD-59 before immunization with adenovirus encoding the TSHR_ECD. As shown in Fig. 3, tolerance induction with either peptide mixture led to suppression of antibodies to TSHR. This demonstrates that peptide immunotherapy with TSHR-derived peptides can prevent the antibody response to the native protein by suppressing the T-cell help required for immunoglobulin class switching and B-cell differentiation associated with IgG production (Fig. 3A–3F). Furthermore, analysis of immunoglobulin isotypes revealed that antigen-specific immunotherapy with TSHR-derived peptides suppressed class switching to isotypes supported by both Th1 (IgG2c) and Th2 (IgG1) responses (40) (Fig. 3G and 3I).

Figure 3.

Prophylactic ATX-GD-59 treatment reduces anti-TSHR antibody levels. (A–C) DR3tg mice were injected subcutaneously in the flank region with 15 pmol, 150 pmol, and 1500 pmol ATX-GD-459 (n = 9) or control treatment (n = 7) on days −15, −13, and −11, followed by three injections of 15 nmol of each peptide (total 45 nmol/dose) of ATX-GD-459 or by PBS control treatment on days −8, −6, and −4 (dose escalation schedule). Mice were injected intramuscularly with 10⁹ Ad-TSHR or Ad-LacZ on two occasions (days 0 and 21). Blood was collected (A) before and (B) 2 and (C) 5 weeks after the first immunization to measure anti-TSHR total IgG levels by ELISA. (D–I) DR3tg mice were injected subcutaneously in the flank region with ATX-GD-59 (n = 11) or control peptide (n = 11) according to the dose escalation schedule, with a 22.5-nmol top dose for each peptide, followed by intramuscular injection of 10¹⁰ Ad-TSHR or Ad-LacZ viral particles (n = 7) on week 0 and week 3. Blood was collected (D) before and (E) 2 and (F) 5 weeks after the first immunization to measure anti-TSHR total IgG levels by ELISA. Blood samples of week 2 were used to measure anti-TSHR (G) IgG1, (H) IgG2b, and (I) IgG2c antibodies. Each dot represents data from one mouse, and the group mean ± SEM is indicated. One-way ANOVA was used to measure overall differences in anti-TSHR IgG levels. Bonferroni post hoc testing was used; significant differences are indicated in the graphs (*P < 0.05; **P < 0.01; ***P < 0.001; ****P < 0.0001).

Comedication does not abrogate tolerance induction in the Ad-TSHR model

The reciprocal influence of ATX-GD-459 and MMI treatment on disease parameters was investigated (Fig. 4). Ad-TSHR immunization evoked the production of anti-TSHR antibodies in control peptide–treated mice but not in ATX-GD-459–treated mice. ATX-GD-459 treatment reduced the average anti-TSHR antibody production by >90%. In contrast, MMI treatment alone had no effect on anti-TSHR antibody levels (Fig. 4A). The effect of MMI and ATX-GD-459 on serum T4 levels was investigated; we found that MMI treatment significantly reduced T4 levels, making the mice hypothyroid, whereas prophylactic treatment with ATX-GD-459 did not (Fig. 4B). More importantly, ATX-GD-459 treatment had no influence on the T4-reducing capability of MMI, whereas MMI cotreatment did not influence the antibody-reducing capability of ATX-GD-459 peptide treatment (Fig. 4A and 4B).

Figure 4.

Comedication of ATX-GD-459 and MMI or propranolol in an adenoviral-based GD model. (A and B) DR3tg mice were injected subcutaneously with a 15-nmol top dose of each ATX-GD-459 peptide or control peptide according to the dose escalation schedule. Starting together with peptide treatment, MMI or vehicle treatment was administered via subcutaneously implanted osmotic pumps. Mice were injected intramuscularly with 10⁹ Ad-TSHR viral particles on weeks 0 and 3. Experimental groups were composed as follows: vehicle + control peptide, n = 9; MMI + control peptide, n = 9; vehicle + ATX-GD-459, n = 9; and MMI + ATX-GD-459, n = 10. Serum samples were collected 2 weeks and 5 weeks after the first immunization to measure anti-TSHR (A) IgG and (B) T4 levels, respectively. (C–E) DR3tg mice were injected subcutaneously with a 15-nmol top dose of each ATX-GD-459 peptide (total 45 nmol) or control peptide according to the dose escalation schedule. Starting together with peptide treatment, propranolol or vehicle treatment was administered via daily intraperitoneal injections, followed by intramuscular injection with 10¹⁰ Ad-TSHR viral particles on weeks 0 and 3. (C) Serum samples were collected 2 weeks after the first immunization to measure anti-TSHR IgG. Plasma samples were collected at the end of the experiment to measure (D) T4 and (E) propranolol levels. Experimental groups were composed as follows: vehicle + control peptide, n = 10; propranolol + control peptide, n = 10; vehicle + ATX-GD-459, n = 10; and propranolol + ATX-GD-459, n = 10. Each dot represents data from one mouse, and the group mean ± SEM is indicated. One-way ANOVA and Bonferroni post hoc testing was used; significant differences are indicated in the graphs (*P < 0.05; **P < 0.01; ****P < 0.0001)

Patients with GD often use β-blockers to fight the adrenergic symptoms of hyperthyroidism; therefore, the effect of comedication with ATX-GD-459 and the β-blocker propranolol was investigated. ATX-GD-459 treatment reduced the anti-TSHR antibody production in Ad-TSHR immunized DR3tg mice by >90%. Propranolol treatment alone had no effect on anti-TSHR antibody levels and, more important, did not influence the antibody-reducing capability of ATX-GD-459 peptide treatment (Fig. 4C). As expected, neither propranolol nor ATX-GD-459 treatment affected serum T4 levels upon Ad-TSHR immunization (Fig. 4D). Although no clinical effect of propranolol could be observed in this animal model, high plasma levels of propranolol were confirmed in propranolol-treated animals (Fig. 4E). Taken together, these comedication data indicate that our antigen-specific peptide therapy and either MMI or propranolol do not hamper the independent functions of these different therapeutic approaches and can be safely combined in the adenoviral-based model of GD.

Patient PBMC response to TSHR epitopes

A study was undertaken with PBMC samples from patients displaying a positive proliferative response to TSHR_ECD. This population (n = 31) was partially enriched for patients expressing DRB1*03:01 and included 64% with this allele and 36% without. The frequency of DR3 in patients with GD varies between 40% and 55% compared with the general population (20% to 30%) (41). The results showed that 26% of patients responded to peptide 5DK, whereas 40% responded to peptide 9BN. Among these patients, 80% of 5DK responders shared the DRB1*03:01 allele; similarly, 60% of 9BN responders expressed DRB1*0301. These data indicate that a majority of responders express DRB1*03:01, which validates our use of mice expressing this human class II allele as a test bed for these apitope peptides.

Discussion

Our results show that it is possible to design tolerogenic peptides from TSHR that can be used to prevent the induction of anti-TSHR antibodies in HLA-DR3 transgenic mice. This description of tolerogenic peptides provides the rationale for the use of antigen-specific immunotherapy for the control of GD.

The ideal tolerogenic peptide for an autoimmune disease should have the following features. It should (1) display high solubility so as to avoid entrapment and destruction at the site of injection (22); (2) target endogenous tolerogenic DC in the recipient (42); (3) mimic the naturally processed antigen when bound to MHC II (21); (4) "switch off" pathogenic T cells by inducing apoptosis or anergy; (5) suppress secretion of inflammatory cytokines while promoting secretion of anti-inflammatory cytokines, such as IL-10; (6) induce linked and bystander suppression (43); and (7), most important, be safe for repeated administration to patients.

The peptides described in this article satisfy many of the key features of an effective therapeutic peptide. Peptide 9BN was sufficiently soluble so as not to require modification. Peptides 4K and 5D were relatively insoluble in aqueous buffer. Hence, the peptides were modified at the N- and C-termini with hydrophilic amino acids so as to improve their solubility while retaining their MHC-binding and antigenic properties. We have recently shown that soluble T-cell epitopes selectively target tolerogenic DC in vivo (42) following subcutaneous injection, thereby explaining their ability to induce tolerance to the parent protein antigen.

Not all exogenous peptide antigens bind back to the MHC in the same conformation as was formed after antigen processing of the native antigen (44). Peptides that are presented to T cells specific for the naturally processed antigen by fixed APCs are referred to as antigen-processing independent epitopes or apitopes. Peptides 4K and 5D were shown to behave as apitopes in vitro. We were unable to generate T-cell hybridomas specific for peptide 9B; however, the fact that this peptide induces tolerance against the naturally processed epitope from TSHR in vivo indicates that it does behave as an apitope.

The immunologic mechanism by which the TSHR epitopes described in this paper induce tolerance and suppress antibody responses has not been characterized. Previous studies have shown, however, that analogous peptides induce anergy and/or generate IL-10–secreting anergic cells, with a Tr1-like phenotype, depending on the strength of signal associated with the epitope used for tolerance induction (26, 45). Further work is required to assess the precise mechanism by which these TSHR-derived peptides induce suppression.

Our previous work has shown that antigen-specific immunotherapy with peptides induces both linked (46) and bystander (47) suppression. These mechanisms depend on suppression of local antigen presentation through secretion of IL-10 (25, 43) and/or induction of anergy or deletion in critical helper T cells. Although this has not been formally demonstrated in this study, the evidence implies that our peptide mixtures induce linked suppression. Evidently the application of two soluble epitopes from the TSHR is able to suppress the response to any remaining epitopes within the TSHR_ECD and is consistent with a linked suppressive mechanism.

In this study, we compared two mixtures of peptides ATX-GD-459 and ATX-GD-59. Importantly, both preparations were equally good at inducing tolerance to the TSHRECD or the TSHR domain expressed in vivo as an adenoviral construct. This implies that peptide 4K is not required for effective immunotherapy in this model or that peptides 5DK and 9BN are capable of inducing linked suppression so as to prevent priming of the immune response to peptide 4.

This study was conducted primarily in HLA-DR3 transgenic mice based on the association between this disease and HLA-DR3. Our peptides were selected, however, to be pan-DR binding; indeed, 5DK and 4KG have been shown to suppress the response to TSHR_ECD in HLA-DR4 transgenic mice (data not shown). This demonstrates not only that the peptides can bind different class II MHC molecules but also that they can induce tolerance via these molecules in vivo.

Our epitope mapping studies build on previous work in both HLA-DR3 transgenic mice and PBMC samples from patients and controls. Inaba et al. (48) measured the response of CD4⁺ T cells to overlapping peptides following immunization of HLA-DR3 transgenic mice with human TSHR protein and identified peptide 78-94 as a dominant epitope. This epitope (ISRIYVSIDVTLQQLES) partially overlaps with the two distinct epitopes, 4K and 5D, identified in our studies. In a further study, the same group used an MHC-peptide binding approach to map putative epitopes and correlated these with clinical data; they found that peptide 132-150 (GIFNTGLKMFPDLTKVYST) was a strong candidate (49). Peptide 9BN partially overlaps with this sequence. Taken together with our confirmatory human PBMC studies, the published epitope mapping studies strongly support our identification of relevant human T-cell epitopes that will have utility in the immunotherapy of GD.

The HLA-DR3 transgenic mouse used in this study is not currently suitable as a model of GD in humans, other than for investigating HLA-DR3–restricted T-cell responses and tolerance induction. Analysis of an F1 cross between the HLA-DR3 transgenic and BALB/c mice reveals that this is due to the background genes of the transgenic mouse. Further backcrossing of the DR3 transgenic onto the BALB/c background should be investigated. Despite deficiencies of the DR3 model as it currently stands, we demonstrated unequivocally that drugs currently used as pharmacologic interventions in GD do not interfere with the ability of our peptide mixtures to induce tolerance. This provides proof that these different interventions do not have overlapping mechanisms that could interfere with the conduct of future clinical trials of antigen-specific immunotherapy.

In conclusion, we have used HLA-DR transgenic mice and human PBMC samples to map immunodominant pan-DR–binding epitopes in TSHR. These have been designed as apitopes and tested for tolerance induction in the HLA-DR3 transgenic mouse. The results show that ATX-GD-59 is able to suppress the immune response underpinning the immune pathology of GD and warrants further investigation through clinical trials in patients with this disease.

Abbreviations:

Abbreviations:
 
• AD-TSHR

  adenovirus expressing a part of the extracellular domain of human thyrotropin receptor

 
• APC

  antigen-presenting cell

 
• CFA

  complete Freund’s adjuvant

 
• GD

  Graves’ disease

 
• GO

  Graves’ orbitopathy

 
• MHC

  major histocompatibility complex

 
• MMI

  methimazole

 
• MPred

  methylprednisolone

 
• PBMC

  peripheral blood mononuclear cell

 
• PPD

  purified protein derivative

 
• TMB

  tetramethylbenzidine

 
• TSHR

  thyrotropin receptor

 
• TSHR_ECD

  extracellular domain of human thyrotropin receptor

Acknowledgments

The authors want to thank Dorien Feyaerts, Kristina Rekstyte-Matiene, and Sarah Aerts for excellent technical work and Professors Basil Rapoport, Marian Ludgate, and Paul Banga for helpful discussions and for providing reagents.

Financial Support: The development of ATX-GD-59 is part of the DAVIAD project (www.daviad.eu), cofinanced by the European Commission in the 7th Framework Programme, FP7-HEALTH-2013-INNOVATION-1, 602779. The DAVIAD consortium comprises Apitope (as coordinator), Quintiles, and KWS Biotest Ltd. This work was supported by grants from Agentschap voor lnnovatie door Wetenschap en Technologie (IWT 120512), Belgium.

Disclosure Summary: L.J., K.V., and A.J. were employees of, and D.C.W. served as a consultant to, Apitope International NV, and K.F.M. was an employee of Apitope Technology (Bristol) Ltd. at the time of this study.