The differential expression profile of exosomal proteins in plasma of patients with AITDs was investigated in a case–control study. Patients were consecutively enrolled from the Department of Endocrinology at Zhoupu Hospital; and gender- and age-matched healthy normal controls (NC) were recruited from the Health Check-up Center of the same hospital. To exclude the effects of other interference factors as much as possible, subjects with infectious diseases within 3 months and any other autoimmune and chronic diseases were excluded. The clinical characteristics of all subjects including disease duration, family history, the degree of Grave’s ophthalmopathy, and goiter were collected, as well as the serological data including serum levels of thyroid-stimulating hormone (TSH) receptor antibody (TRAb) (IU/L), thyroid peroxidase antibody (TPOAb) (IU/ml), thyroglobulin antibody (TGAb) (IU/ml), free triiodothyronine (FT3) (pmol/L), free thyroxine (FT4) (pmol/L), and TSH (μIU/ml).

Patients who presented with thyrotoxicosis and were TRAb-positive were diagnosed as GD. Among them, patients who were initially confirmed to have GD and treatment-naive, which reflects the most realistic immune status in the GD state, were assigned to the newly diagnosed GD (NG) subgroup, while patients who had been diagnosed with GD and continued to receive standard antithyroidal therapy at least for 3 years, but still had TRAb greater than 1.75 IU/L, were assigned to the refractory GD (RG) subgroup. Patients who were TGAb positive or TPOAb positive and had high TSH level with clinically obvious hypothyroidism symptoms were diagnosed as HT.

A total of 5 ml of peripheral venous blood from each subject was collected in EDTA anticoagulation tubes. After centrifugation at 2,000 rpm for 5 min at 4°C, 1 ml of the upper clear plasma was aspirated and stored −80°C for future use. To isolate exosomes, all plasma samples were simultaneously thawed to 4°C and centrifuged at 4°C for 10 min at 2,000 g, 20 min at 8,000 g, and 1 h at 20,000 g. At the end of each centrifugation, the supernatants were collected for the next centrifugation. After the final centrifugation, the supernatants were subjected to filtration with 0.22-μm filters to remove any visible precipitates and lipids. The filtrates were then diluted fourfold with phosphate-buffered saline (PBS) in tubes and subjected to ultracentrifugation at 1,500,000 g at 4°C for 2 h after making sure that the weight difference of each tube was less than 0.01 g. The pelleted exosomes were resuspended in 1 ml of PBS and subjected to ultracentrifugation again at the same condition.

The extracted exosomes were lysed with 100 μl of sodium dodecyl sulfate (SDS)-free L3 lysis buffer supplemented with 1× enzyme inhibitor cocktail by pipetting up and down for at least 50 times. After being regulated to a final concentration of 10 mM, samples were incubated in a water bath at 37°C for 45 min. After being cooled to room temperature, samples were treated with 20 mM of iodoacetamide in the dark for 30 min to obtain exosome proteins. Protein concentrations were measured using the Bradford method, and their integrity was examined using polyacrylamide gel electrophoresis.

Protein profile was analyzed using the next-generation label-free quantitative proteomics technology under the DIA (also known as SWATH) mode, which is considered as an ideal differential proteomics analysis and enables accurate and highly repeatable quantification for large amounts of proteins per sample. The DIA analysis pipeline contains three essential steps, namely, spectral library construction, large-sample data acquisition, and data analysis. The spectral library was constructed from samples of interest using data-dependent acquisition (DDA) technique. MaxQuant was then used to carry out database search identification process and obtain all detectable non-redundant high-quality MS/MS spectral information as DIA spectral library. To improve protein identification and quantification and effectively avoid convolution, data analysis was conducted using Spectronaut™, in which iRT peptides were used for retention time calibration and the mProphet scoring algorithm was integrated to accurately reflect the matching level of ion pairs. Then, false-positive control was performed based on the target-decoy model with false discovery rate (FDR) of 1% applicable to SWATH-MS to obtain significant quantitative results. Data quality was evaluated based on intra-group coefficient of variation (CV), principal component analysis (PCA), and quantitative correlation of samples. Quality control samples were inserted intermittently between the continuous original samples to ensure stability and repeatability of the experiment.

Statistical analysis was performed using R software (version 4.0.4) and STATA (12.0, StataCorp, USA). Continuous variables were calculated as mean ± SD. Differences between groups were analyzed using independent samples t-test, and non-normal distribution data were analyzed using the Mann–Whitney U tests. Differences with both p-value and Q value (adjusted for the FDR) <0.05 were considered as statistically significant. Significant differences in proteins or peptides were statistically evaluated using MSstats, a widely used R package from the Bioconductor repository in DIA quantitative experiments with linear mixed-effects model as the core algorithm. The possible functions of the differential proteins between different groups were analyzed using Gene Ontology (GO) enrichment analysis, Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway enrichment analysis, and EuKaryotic Orthologous Groups (KOG) analysis.

A total of seven patients with newly diagnosed GD, five patients with refractory GD, 10 patients with HT, and seven healthy controls were enrolled in the study. The clinical and serological data of those subjects are summarized in Table 1 . The mean age of the total patients with GD, HT, and healthy controls were 41.4, 35.8, and 37.4 years, respectively. The treatment duration in refractory GD patients was 5.2 ± 1.6 (mean ± SD) years with an average TRAb level of 15.3 IU/L, while the patients with newly diagnosed treatment-naive GD had slightly higher TRAb with an average level of 17.7 ± 6.9 IU/L. All patients who have HT had elevated TSH, TPOAb, and TGAb levels.

The MS data of 29 samples were acquired, and quantitative statistics of each sample using Spectronaut™ are shown in Supplementary Table 1 . Qualification of these data based on CV, PCA and quantitative correlation of samples indicates that these data are in good quality, stability, and repeatability ( Supplementary Figure 1 ).

A total of 8,251 peptides and 1,319 proteins were detected. Among these proteins, 733 proteins were unchanged, 11 proteins were downregulated, and 197 proteins were upregulated in the total GD group compared with the NC group, with MAP1S (log₂ FC = 4.669, p = 0.009, Q = 0.004) and VAMP8 (log₂ FC = 3.216, p = 0.003) being the most significantly upregulated, and RSU1 (log₂ FC = −6.797, p = 0.001, Q = 0.010), ACTB (log₂ FC = −4.795, p < 0.001), and CXCL7 (log₂ FC = −4.674, p < 0.001) being the most significantly downregulated. The top 10 upregulated and top 10 downregulated proteins are listed in Table 2 , and volcano plot of the differential proteins is shown in Supplementary Figure 2 .

The subgroup analysis of the GD patients showed that the NG group had more differential proteins than the RG group compared with the NC group. Specifically, screening all differential proteins with Q value (adjusted p-value) <0.5 indicates that 127 proteins were downregulated, and VAMP8 (log₂ FC = 3.216, p = 0.003, Q = 0.02) was the most upregulated in the NG group ( Figure 1A ). Euclidean distance and hierarchical clustering of differential proteins showed almost no significant difference between the RG group and NC group except that only PTN12 (log₂ FC = −2.154, p < 0.001, Q = 0.006) was significantly downregulated. The analysis of differential proteins between the NG and RG groups found that the levels of CORO1C (log₂ FC = 7.185, p = 0.003, Q = 0.006), BTK (log₂ FC = 6.394, p = 0.017, Q = 0.034), and PRG2 (log₂ FC = 6.245, p = 0.016, Q = 0.012) were significantly higher in the RG group than in the NG group ( Figure 1B ).

Further analysis of differential proteins showed that in the HT group, seven proteins including HGFL (log₂ FC = 2.766, p = 0.001, Q = 0.036), FAK1 (log₂ FC = 2.213, p < 0.001, Q = 0.015), FBN1 (log₂ FC = 1.745, p < 0.001, Q = 0.017), PTN12 (log₂ FC = 1.624, p < 0.001, Q = 0.021), C1QB (log₂ FC = 1.048, p < 0.001, Q = 0.015), and C1QC (log₂ FC = 1.029, p < 0.001, Q = 0.008) were upregulated. Moreover, nine proteins such as PSMF1 (log₂ FC = −3.591, p < 0.001, Q = 0.012), PXL2B (log₂ FC = −2.622, p = 0.001, Q = 0.037), CYTM (log₂ FC = −1.609, p < 0.001, Q = 0.021), and TRFE (log₂ FC = −1.009, p = 0.001, Q = 0.016) were significantly downregulated ( Table 3 and Figure 1C ). Differential proteins between all AITDs patients and NC were shown in cluster map ( Figure 1D ). The NG group and the NC group showed the most significant difference, while the RG group and the NC group showed the least significant difference.

KEGG analysis found that the differential proteins were mainly concentrated in the immune system and endocrine system, which is consistent with the etiological background of AITDs ( Figure 2A ). GO analysis found that the differential proteins between the GD and NC groups mainly manifested in cellular process, metabolic process, and biological regulation ( Figure 2B ). KOG analysis found that the differential proteins between the GD and NC groups were mainly enriched in posttranslational modification, protein turnover, and chaperones ( Supplementary Figure 3A ). Further pathway enrichment analysis revealed that most of the differential proteins were enriched in the metabolic pathways and glycolysis/gluconeogenesis pathways ( Supplementary Figure 3B ). Similarly, KEGG analysis of the differential proteins between the HT and NC groups also found that they were mainly enriched in the immune system ( Figure 3A ). GO analysis results showed that these differential proteins were mainly concentrated in biological regulation, metabolic process, biological process regulation, and immune system among other pathways ( Figure 3B ). KOG analysis found that these proteins were mainly enriched in the signal transduction mechanisms ( Supplementary Figure 4A ), and further KEGG analysis revealed that the exosome proteins in HT patients were related to other autoimmune diseases such as systemic lupus erythematosus (SLE) ( Supplementary Figure 4B ). Comprehensively, the functional analysis suggested that the exosomal differential proteins in patients with AITDs were mainly immune-related.