Transgenic mice expressing the human HLA-DRB1*03:01 (DR3) allele, kindly provided by Dr. Chella David (Mayo Clinic), were generated by co-injection of HLA-DRA and HLA-DRB1*03:01 genes into (C57BL/6 × DBA/2) F1 × C57BL/6 embryos backcrossed onto a B10 genetic background (20). These mice express the human HLA-DR3 transgene and lack endogenous mouse MHC class II molecules. Age-matched female mice (5–8 weeks old) were used for all experiments. Animals were housed under conventional conditions. All animal care and experimental procedures were conducted in accordance with the Guidelines for Animal Experimentation of Wakayama Medical University and were approved by the Institutional Animal Care and Use Committee.

A known GD-associated epitope peptide corresponding to amino acid residues 78–94 of the human TSH-R (ISRIYVSIDVTLQQLES), referred to as p37, was synthesized (8, 9). A mutated version of this peptide, termed 37m (ISRIYVSIDATLSQLES), containing two substitutions at positions critical for DR3 binding (position 5: Valine to Alanine; position 8: Glutamine to Serine), was also generated as previously described, to reduce binding affinity to the T cell receptor (10). Additionally, an HLA-DR3 high-affinity binder peptide (TSH-R 132–150, termed p10: GIFNTGLKMFPDLTKVYST) and a control peptide (TSH-R 109–124, termed p8: RNTRNLTYIDPDALKE) were included in the study ( Table 1 ). Peptide sequences were confirmed and purity (>90%) was verified by reverse-phase HPLC (Peptide Institute, Inc., Osaka, Japan and CHI scientific, Inc., Maynard, MA, USA).

A recombinant adenovirus expressing the extracellular domain of human TSH receptor (amino acids 1–289; Ad-TSH-R289) was generated for GD induction. PCR was performed using the pSVL plasmid encoding full-length human TSH-R (1–764), kindly provided by Dr. Refetoff (University of Chicago). The primers used were: Forward: 5′-CACCATGAGGCCGGCGGACTTG-3′ and Reverse: 5′-TTACTGATTCTTAAAAGCACAGCA-3′. PCR amplification was followed by cloning into the pENTR/D-TOPO vector, and subsequent recombination into the pAd/CMV/V5-DEST vector using the Gateway system (Thermo Fisher Scientific, Waltham, MA, USA). The constructed plasmid was transfected into 293A cells, and a single adenoviral plaque was isolated and expanded. A control adenovirus lacking the TSH-R insert: pAd/CMV/V5-DEST was similarly prepared and used as a control. Viral titers: plaque forming unit (PFU) were determined by OD260 or TCID₅₀ assays. Adenoviruses were purified by CsCl gradient ultracentrifugation and dialyzed in PBS (8). Expression of TSH-R (1–289) was confirmed by Western blot using a monoclonal anti–TSH-R antibody (clone A10, Advanced Targeting Systems, San Diego, CA).

GD was induced by intramuscular injection of 100 μL phosphate-buffered saline (PBS) containing 1 × 10⁹ particles of Ad-TSH-R289 into the quadriceps muscle. As a control, the same amount of Ad-CMV-DEST was administered. Three weeks prior to Ad-TSH-R289 immunization, mice received subcutaneous pretreatment with TSH-R peptide p37, 37m, p10, p8, or PBS. Two immunization protocols were employed: a single-dose protocol and a step-up protocol ( Figure 1 ). In the single-dose protocol, mice were subcutaneously injected with 50 μg of each peptide dissolved in 50 μL PBS ( Figure 1A ). In the step-up protocol, mice received subcutaneous injections of each peptide in an escalating dose regimen: 0.05 μg on day 0, 0.5 μg on day 3, 5 μg on day 8, and 50 μg on day 18, for a total of four injections ( Figures 1B, C ). All peptides were dissolved in 50 μL PBS. All mice were sacrificed five weeks after Ad-TSH-R289 immunization, and serum, thyroid glands, and spleens were collected for analysis.

To deplete Tregs, a subgroup of mice received 500 µg of anti-CD25 antibody (clone PC61.5, Rat IgG1λ; eBioscience, San Diego, CA, USA) intraperitoneally 4 days prior to Ad-TSH-R289 immunization.

Serum FT4 levels were measured using an ELISA kit (ENZAPLATE-FT4, Siemens Healthcare Diagnostics, Tokyo, Japan; normal range: 0.79–2.00 ng/dL). Anti–TSH-R antibodies were measured using a third-generation TRAb assay (Tosoh Corporation, Tokyo, Japan; normal range: <2.0 IU/L).

Mouse serum IL-2, IL-4, IL-10, and IFN-γ levels were measured using ELISA kits (Thermo Fisher Scientific, Waltham, MA, USA). Serum (12.5 μL) was diluted to 100 μL with assay diluent and processed according to the manufacturer’s instructions.

Splenocytes were stained with the following antibodies: CD4-FITC (RM4-5), CD8-PE (53-6.7), IFN-γ–PerCP-Cy5.5 (XMG1.2), IL-4–PE (11B11), FoxP3–PE-Cy5 (FJK-16s), CD44-PE (IM7), CD62L–PerCP-Cy5.5 (Mel-14), CD25-PE (PC61.5), and PD-1–PE-Cy7 (J43). Intracellular staining for FoxP3 was performed using the FoxP3/Transcription Factor Staining Buffer Set according to the manufacturer’s protocol (eBioscience, San Diego, CA, USA). Samples were analyzed using a FACSCalibur™ (Becton Dickinson, New Jersey, USA).

Splenocytes were isolated on sacrifice as described previously (7–9). Splenocytes (5 × 10⁵ cells/well) were cultured in 96-well round-bottom plates in RPMI 1640 medium supplemented with 25 mM HEPES, 2 mM L-glutamine, 100 U/mL penicillin, 100 µg/mL streptomycin, and 10% FCS (Gibco BRL). Cells were incubated for 3 days with or without TSH-R peptides (5 µg/mL). BrdU was added during the final 8 hours, and incorporation was measured using a chemiluminescent assay (Roche Diagnostics, Mannheim, Germany). Results were expressed as a stimulation index (SI): BrdU uptake in peptide-stimulated wells divided by uptake in unstimulated wells.

Thyroid glands and orbital tissues were fixed in 10% formalin and embedded in paraffin at Applied Medical Research (Osaka, Japan). Sections (5 μm) were stained with hematoxylin and eosin.

Data were analyzed using Mann–Whitney U tests for comparisons between two groups. For comparisons among multiple groups, the Kruskal–Wallis test was used. Spearman’s rank correlation coefficient (Rs) was used to assess correlations. All analyses were performed using JMP software version 17 (SAS Institute Inc.). A P-value <0.05 was considered statistically significant.

Following immunization with Ad-TSH-R289, a subset of mice with elevated serum FT4 levels (>2.0 ng/dL) and positive for TRAb (>2.0 IU/L) showed histopathological features consistent with Graves’ disease (GD), including thyroid enlargement, colloid ballooning, colloid vacuolization, and thyrocyte hyperplasia ( Supplementary Figures 1A, B ). Among a total of 54 mice, 11 (20%) had elevated FT4 levels and 19 (35%) tested positive for TRAb ( Supplementary Table 1 ). In contrast, mice immunized with the control adenovirus vector, Ad-CMV-DEST, did not show these changes ( Supplementary Figures 2A, B ).

Pretreatment with a single dose of TSH-R peptides did not result in significant changes in serum FT4 levels, TRAb titers, or the proportion of splenic Tregs (CD4⁺CD25⁺FoxP3⁺) compared to the Ad-TSH-R289 group ( Figures 2A–C ) ( Supplementary Figures 3A, B ). Interestingly, the Th1/Th2 cytokine ratio was higher in the p37 group than in both the Ad-TSH-R289 and 37m groups ( Figure 2D ). Correlation analysis revealed that TRAb levels positively correlated with the proportions of naïve (CD4⁺CD44^lowCD62L^high) and memory (CD4⁺CD44^highCD62L^high) T cells, and inversely correlated with the frequencies of Tregs and effector T cells (CD4⁺CD44^highCD62L^low) ( Table 2A ). Conversely, Treg frequency was positively associated with the Th1/Th2 ratio, effector T cells, and serum IFN-γ levels, and negatively associated with naïve and memory T cells ( Table 2A ).

To enhance the efficacy of peptide-based tolerance induction, a step-up dosing protocol was evaluated. Among the experimental groups, mice receiving step-up administration of p37 up to 50 μg exhibited the most effective suppression of FT4 elevation, with levels maintained within the normal range ( Figure 3A ). This group also showed greater suppression of TRAb titers compared to the group receiving p37 up to 5 μg ( Figure 3B ). Additionally, the p37 (up to 50 μg) group demonstrated a more pronounced expansion of splenic Tregs ( Figure 3C ), a reduced frequency of CD4⁺PD-1⁺ T cells ( Figure 3D ), and an increased frequency of CD8⁺PD-1⁺ T cells ( Figure 3E ). In the step-up protocol, FT4 levels positively correlated with the frequencies of CD4⁺PD-1⁺ and naïve T cells, and negatively with CD8⁺PD-1⁺ T cells ( Table 2B ). Similarly, TRAb titers showed positive correlations with CD4⁺PD-1⁺ and naïve T cells, and negative correlations with CD8⁺PD-1⁺ T cells and Tregs. In contrast, Treg frequency was positively associated with CD8⁺PD-1⁺ T cells, and inversely correlated with CD4⁺PD-1⁺ and naïve T cells ( Table 2B ).

Finally, the efficacy of step-up versus single-dose peptide immunization was compared. Although the differences did not reach statistical significance, the step-up protocol with p37 up to 50 μg consistently demonstrated superior control of serum FT4 levels and TRAb suppression ( Figures 4A, B ). This group also showed a more robust increase in the frequency of splenic Tregs ( Figure 4C ).

We investigated the role of Tregs in mediating the therapeutic effect of peptide pretreatment in GD ( Figures 5A, B ). Splenic Treg depletion was confirmed by flow cytometric analysis four days after administration of anti-CD25 antibody ( Supplementary Figures 4A–D ). Mice pretreated with the previously identified immunomodulatory peptide 37m were compared to those additionally administered anti-CD25 antibodies prior to Ad-TSH-R289 immunization to deplete Tregs. Administration of anti-CD25 antibodies effectively reduced the Treg population in splenocytes and led to a significant increase in serum IFN-γ levels (P<0.05); however, FT4 and TRAb levels remained unchanged ( Figures 6A–C ). Splenocyte proliferative responses to TSH-R peptides p37, p8, and p10 were not significantly different between the anti-CD25 antibody–treated group and the untreated group ( Figures 6D–F ). Although not statistically significant, the control peptide p8 exhibited a higher stimulation index in the anti-CD25 antibody–treated group ( Figure 6E ).

Finally, mice were divided into TRAb-positive and TRAb-negative groups and analyzed for the frequencies of splenic Tregs, CD4⁺PD-1⁺, and CD8⁺PD-1⁺ T cells ( Figures 7A–C ). TRAb-positive mice exhibited a significantly reduced Treg population (P<0.001) ( Figure 7A ), an increased proportion of CD4⁺PD-1⁺ T cells (P = 0.004) ( Figure 7B ), and a decreased frequency of CD8⁺PD-1⁺ T cells (P<0.001) ( Figure 7C ).