Thyroglobulin antibody (TgAb) is a diagnostic serological marker of Hashimoto's thyroiditis (HT). The pathogenesis of HT progression from euthyroidism to hypothyroidism is still not clear. Epitope recognition patterns of TgAb have been shown to be different in individuals who are euthyroid or who have clinical disease. The aim of our study was to investigate the role of thyroglobulin (Tg) epitope specificities in HT progression. Sera from 107 patients with newly diagnosed HT were collected and divided into three groups: patients with hypothyroidism (H, n = 39), subclinical hypothyroidism (sH, n = 31) and euthyroidism (Eu, n = 37). A panel of Tg murine monoclonal antibodies (mAb: PB2, 5E6, 1D4, 5F9, Tg6) and a hircine pAb (N15) were employed as the probe antibodies to define the antigenic determinants recognized by HT sera on competitive enzyme-linked immunosorbent assays (ELISAs). Eight of 39 sera samples in H and seven of 31 in sH inhibited PB2 binding, respectively, whereas none did in Eu. The ratio of sera samples, inhibiting PB2 binding in Eu, was significantly lower than that in H (P = 0·011) and in sH (P = 0·008). For N15, five of 39 sera samples in H, six of 31 in sH and 15 of 37 in Eu inhibited its binding, respectively. The ratio of sera samples, inhibiting N15 binding in Eu, was significantly higher than that in H (P = 0·013). Our study demonstrated that HT patients in different thyroid functional status exhibited different Tg epitope recognition patterns. Epitope patterns of TgAb might be used as a prediction marker of HT progression.
Hashimoto's thyroiditis (HT) is an organ-specific autoimmune disease caused by multiple factors including immunological activity, environmental exposure and genetic susceptibility. Patients with HT characteristically generate antibodies against thyroglobulin (Tg), one of the major thyroid autoantigens, and serum thyroglobulin antibody (TgAb) is a diagnostic hallmark of HT.
In clinical practice, HT patients with TgAb may manifest various clinical features and have different thyroid functional status, such as euthyroidism and subclinical, even overt, hypothyroidism. It is still not clear which are the important factors in the determination of HT progression. Our previous studies have demonstrated that immunological properties of TgAb such as immunoglobulin (Ig)G subclasses , titres and avidity  might be involved in HT progression, which suggested that humoral response was also important in the pathogenesis of HT. The epitope recognition pattern of autoantibodies is another important component of immunological properties, therefore we assumed that it might also play a role in HT progression.
Analysis of epitope recognition patterns is a feasible strategy to investigate the role of TgAb in the pathogenesis of autoimmune thyroid diseases. Earlier studies have studied Tg epitope recognition patterns in patients with autoimmune and non-autoimmune thyroid diseases , such as HT, Graves' disease, non-toxic goitre and thyroid carcinoma . It has been shown that TgAb in sera from healthy subjects and non-toxic goitre patients exhibit a non-restriction epitope recognition pattern, while sera TgAb from thyroid carcinoma individuals, HT and Graves' disease patients preferentially recognize one or more certain epitopes . Although some researchers have focused on the different Tg epitope specificities recognized by sera TgAb between patients with HT and non-HT, none of them were concerned about the epitope specificities in HT patients with different thyroid functional status. The aim of our study was to investigate the role of Tg epitope recognition patterns in the pathogenesis of HT progression.
Materials and methods
A total of 107 patients with newly diagnosed HT in Peking University First Hospital were collected in the current study. None of the patients had evidence of hereditary and acquired variations in the concentration of thyroxine-binding globulin. There was no evidence of other autoimmune diseases which may influence the determination of tetraiodothyronine, including systemic lupus erythematosus, rheumatoid arthritis, type 1 diabetes mellitus and pernicious anaemia. None of the patients had evidence of co-existent pregnancy or tumour. According to thyroid function, all the 107 patients with TgAb were divided into three groups: patients with hypothyroidism (H) (n = 39, six males, 33 females), subclinical hypothyroidism (sH) (n = 31, three males, 28 females) and euthyroidism (Eu) (n = 37, one male, 36 females).
This study complied with the Helsinki Declaration and was approved by the Ethics Committee of Peking University First Hospital. All the patients gave written informed consent.
Detection of thyroid function
Sera samples were collected at diagnosis and kept frozen at −80°C until use. Chemiluminescence immunoassays were used to detect total triiodothyronine (TT3), total tetraiodothyronine (TT4) and thyroid stimulating hormone (TSH) (ADVIA Centaur; Bayer Healthcare Diagnostics, Tarrytown, NY, USA). TgAb was detected by electrochemiluminescence immunoassays (Cobas e 601 Analyzer; Roche Diagnostics, Indianapolis, IN, USA).
Determination of saturated dilution on sera TgAb
Saturated dilution of each serum was determined by antigen-specific enzyme-linked immunosorbent assay (ELISA). Briefly, 96-well microtitre plates (Costar, Data Packaging Corporation, Spencer, MA, USA) were coated with 0·5 μg/ml Tg (Calbiochem Merck KGaA, Darmstadt, Germany) in 0·05 mol/l carbonate–bicarbonate buffer, pH 9·6 for 1 h and were blocked with 3% bovine serum albumin (BSA; Sigma, St Louis, MO, USA). Each well contained 100 μl in all the steps and all incubations were carried out at 37°C. The plates were washed three times with phosphate-buffered saline (PBS) containing 0·1% Tween-20 (PBST) between stages. After the washing steps, sera samples were diluted from 1:6·25 to 1:25 600 with PBST and incubated in duplicate for 30 min. Every plate contained a positive control, a negative control and a blank. A horseradish peroxidase-conjugated goat anti-human immunoglobulin (Jackson ImmunoResearch Laboratories, Inc., Baltimore Pike, PA, USA), 1:5000 dilution, was subsequently employed for antibody detection. Ortho-phenylenediamine (OPD) diluted in citrate buffer containing 0·1% hydrogen peroxide was used as substrate/chromogen mixture. The reaction was stopped by the addition of 1 mol/l hydrochloric acid. A Varioskan Flash Multimode Reader (Thermo Fisher Scientific Inc., Waltham, MA, USA) was utilized to measure optical density (OD) values at 490 nm. The actual OD was calculated as the OD difference in the presence and absence of Tg (non-specific binding, NSB). As shown in Fig. 1, the saturated serum dilution was determined as the intersection of the plateau and the linear part on the reverse sigmoid curve.
Cross-inhibition study of the probe antibodies
A panel of Tg murine monoclonal antibody (mAb) and hircine pAb were employed as the probe antibodies (clone numbers: PB2, 5E6, 1D4, 5F9, Tg6, N15, respectively; Santa Cruz Biotechnology, Inc., Santa Cruz, CA, USA). All the mAbs are IgG2a/IgG2b subclasses and their immunogen is the full-length human Tg molecule. N15 is an affinity purified goat polyclonal antibody raised against a peptide mapping within the first 50 amino acids near the N-terminus of human Tg (Protein Accession no. P01266).
Because there was little information on Tg epitopes recognized by the panel of commercial antibodies, the cross-inhibition experiments between any two types of the mAbs and the pAb were conducted to exclude the possibility that these antibodies recognized the same epitopes on human Tg molecules. First, biotinylation of the commercial antibodies was performed using the Lightning-Link (LL) biotin conjugation kit (Innova Biosciences, Babraham, Cambridge, UK). Briefly, 1 μl of LL-modifier reagent was added for each 10 μl of antibodies. Each antibody sample (with added LL-modifier) was then pipetted directly onto the lyophilized material and resuspended gently. After the vials were left standing for 3 h at room temperature, 1 μl of LL-quencher reagent was added for every 10 μl of antibodies used. The set of biotinylated commercial antibodies could be used 30 min later.
Further, cross-inhibition assays were performed. Briefly, microtitre plates were coated with 1 μg/ml Tg and blocked with 3% BSA and the non-biotinylated PB2 was incubated for 1 h. Following the washing steps, the set of biotinylated antibodies, diluted to give an absorbance of 1–1·5 at 490 nm, was added, and then streptavidin–horseradish peroxidase (Jackson ImmunoResearch Laboratories, Inc.) was employed. The subsequent procedures were the same as described above. The percentage inhibition of the biotinylated commercial antibody binding by non-biotinylated PB2 was calculated as: [(OD in the absence of PB2 – NSB) – (OD in the presence of PB2 – NSB)] / (OD in the absence of PB2 – NSB) × 100%. Greater than 70% inhibition was regarded as complete inhibition and 35–70% as partial inhibition. The cross-inhibition assays between the other non-biotinylated antibodies and the set of biotinylated commercial antibodies were performed similarly.
Competitive inhibition assays between the probe antibodies and HT sera samples
The six commercial non-biotinylated antibodies mentioned above were employed as probe antibodies to define the antigenic determinants recognized by these HT sera on competitive ELISAs, as described previously . The procedure was similar to that described above. Briefly, the microtitre plates were coated with 0·5 μg/ml Tg and blocked with 3% BSA. Serum samples were diluted to the saturated dilution as mentioned above. For the competitive inhibition assays, the six probe antibodies (mAb: PB2, 5E6, 1D4, 5F9, Tg6, pAb: N15) were added, diluted to give an absorbance of 1–1·5 at 490 nm in the absence of sera. Following the washing steps, a horseradish peroxidase conjugated goat anti-mouse/rabbit anti-goat immunoglobulin (H + L) (Jackson ImmunoResearch Laboratories) was employed. The percentage inhibition of mAbs and the pAb binding by HT sera was calculated as: [(OD in the absence of serum – NSB) – (OD in the presence of serum – NSB)] / (OD in the absence of serum – NSB) × 100%. Complete inhibition and partial inhibition were defined as above.
Statistical analysis was performed using the spss version 13·0 (SPSS, Chicago, IL, USA) statistical package. Comparisons were carried out by the Mann–Whitney U-test, independent-sample t-test and χ2 test. A P-value under 0·05 was considered statistically significant.
Demographic data of HT patients in the three groups
Overall, there were no significant differences in age and gender distribution in patients in the H, sH and Eu groups. Thyroid functional status and TgAb levels in the three groups are summarized in Table 1. TT3 and TT4 levels in H were significantly lower than those in the other two groups (P <0·001, respectively), and TSH levels in the H and sH groups were significantly higher than those in Eu (P <0·001 and P <0·001, respectively). TT4 levels in sH was also significantly lower than those in Eu (P = 0·04). There was a higher TgAb median in H than in sH (1251 versus 1039 IU/ml) and in sH than in Eu (1039 versus 481 IU/ml); however, these differences were not statistically significant.
Table 1. Demographic data, thyroid functional status and thyroglobulin antibody (TgAb) levels of the patients in different groups.
(n = 39)
(n = 31)
(n = 37)
Numbers expressed as mean ± standard deviation, median (interquartile range). Reference value range: total triiodothyronine (TT3): 0·92–2·79 nmol/l; total tetraiodothyronine (TT4): 58·1–140·6 nmol/l; thyroid stimulating hormone (TSH): 0·35–5·5 mIU/l. H: hypothyroidism; sH: subclinical hypothyroidism; Eu: euthyroidism. aP <0·05 versus sH. bP <0·05 versus Eu.
Cross-inhibition between any two types of the mAbs and the pAb
Whether or not the six probe antibodies reacted with the same epitope on Tg was investigated by cross-inhibition experiments. The binding ability of each antibody to Tg was decreased slightly before and after the biotin conjugation (Fig. S1), and the reagents of the LL biotin conjugation kit and the biotinylated irrelevant antibody did not recognize Tg (Fig. S2). As shown in Fig. 2, there was no cross-inhibition between PB2 and N15, and partial inhibition could be found between PB2 and 1D4, 5E6 and 1D4, 5E6 and N15, 5E6 and 5F9, Tg6 and 1D4, Tg6 and N15 and Tg6 and 5F9.
To further exclude the possibility that the probe antibodies recognized similar epitopes on Tg, but with different affinities, the inhibition results between each probe antibody and HT sera were analysed carefully, and three serum samples (S1, S2, S3) were shown as examples. We assumed that if any two types of the probe antibodies recognized the same epitope with different affinities, the inhibition tendency of the probe antibody binding by different serum samples should be consistent. As shown in Fig. 2, PB2 binding could be inhibited partially by 1D4, and they might recognize the same epitope with different affinity. However, all three serum samples selected from HT patients could partially inhibit PB2 binding, and only one of three could inhibit 1D4 binding (Fig. 3a). The inhibition tendency of PB2 and 1D4 binding by the same three samples was distinct, which indicated that the epitope recognized by PB2 might be partially different from 1D4. Similar comparisons of the inhibition binding results by the same three sera samples were made between any two types of the commercial antibodies with reciprocal partial inhibitions (Fig. 3b,c). This indicated that there was a low possibility that the six commercial antibodies reacted with the same epitopes on the Tg molecule, and all of them could be employed as probe antibodies to determine the epitope recognition patterns of sera TgAb from HT patients.
Inhibition ratios and the percentage inhibition of the probe antibodies binding by HT sera
As shown in Table 2, eight of 39 serum samples in H and seven of 31 in sH inhibited PB2 binding, respectively, whereas none did in Eu. The ratio of sera samples, inhibiting PB2 binding in Eu, was significantly less than that in H (P = 0·011) and in sH (P = 0·008), respectively. When we pooled H and sH groups together (thyroid dysfunction group), the significant difference was reinforced (P = 0·002). The median percentage inhibition of PB2 binding in sH was significantly higher than that in Eu (P = 0·007), and there was no significant difference in the H and Eu groups (Fig. 4). Sera samples from the patients with thyroid dysfunction also had a significantly higher percentage inhibition of PB2 binding than those from euthyroid patients (P = 0·022).
Table 2. The inhibition ratios of the probe antibodies binding by Hashimoto sera in different groups.
The inhibition numbers of different probe antibodies binding n(%)
H: hypothyroidism; sH: subclinical hypothyroidism; Eu: euthyroidism; TD: thyroid dysfunction (H and sH groups together). aP <0·05 versus sH. bP <0·05 versus Eu.
For N15, five of 39 sera samples in H, six of 31 in sH and 15 of 37 in Eu inhibited its binding, respectively. The ratio of sera samples, inhibiting N15 binding in Eu, was significantly higher than that in H (P = 0·013) and in the thyroid dysfunction group (P = 0·004), and the median percentage inhibition of the H and sH groups was significantly lower than that of Eu (P = 0·010, P = 0·036, respectively). Significantly higher ratios of sera TgAb inhibiting 5E6 and 1D4 binding were found in sH than in H, respectively (P = 0·033, P = 0·033); however, no significant differences existed in sH and Eu, and in H and Eu. No statistical differences were noted in the inhibition ratios and the percentage inhibition of the other probe antibodies' binding among the three different thyroid function groups.
Associations between the epitope recognition patterns of TgAb and TSH levels of HT patients
The patients with sera TgAb inhibiting PB2 binding had a higher median TSH level than those without (16·01 versus 6·09, P = 0·019). Moreover, the median TSH levels were significantly lower in the patients with sera TgAb inhibiting N15 binding than those without (3·26 versus 8·26, P = 0·003). No relationship was observed between the TSH levels and the percentage inhibition of PB2 binding (r = 0·161, P > 0·05) and of N15 binding (r = −0·235, P = 0·020), respectively. No significant difference in TSH levels was found between patients with and without sera TgAb inhibiting the binding of the other probe antibodies. Furthermore, the sH group was divided into two parts according to TSH levels, with a cut-off value of 10 mIU/l. There were no significant differences in epitope recognition patterns between the group with 5·5 mIU/l < TSH < 10 mIU/l (TSH normal range: 0·35–5·5 mIU/l) and the group with TSH > 10 mIU/l.
TgAb is a diagnostic serological marker of HT, and high titres of TgAb have been found in approximately 90% of HT patients . Its role is still unclear in the induction of HT. Until now, little has been known about HT progression from euthyroidism to hypothyroidism. Our previous study showed that the immunological properties of TgAb such as IgG subclass distribution and avidity might be involved in HT progression.
The epitope recognition pattern of TgAb is also an important immunological property. In the current study, sera TgAb could inhibit binding of the six probe antibodies, which indicated that TgAbs were generated by polyclonal activation of B cells in HT. Moreover, it has been well proven that the antigen determinants of Tg are different in various diseases such as AITD, multi-nodular goitre and thyroid carcinoma . TgAbs produced in the course of AITD have a more restricted reactivity pattern than that in healthy subjects [8-10], therefore McLachlan et al. proposed that epitope recognition patterns of TgAb could be used to distinguish the individuals who are euthyroid or who have clinical disease . The present study focused on the epitope specificities of TgAb in HT, a relatively restricted organ-specific disease. It was interesting that the epitope recognition patterns of TgAb were also different among the HT patients with different thyroid function status. Only sera TgAb in patients with thyroid dysfunction could recognize the epitope that PB2 identified, and HT patients with sera TgAb inhibiting N15 binding might have a greater possibility of being euthyroid. Geld et al. thought that the differences in binding specificity may influence antibodies' pathogenic potential . Therefore, we speculated that the immunogenic and pathogenic potential of all the Tg epitopes might be not equivalent, and TgAb reacting to certain restricted epitopes, namely the disease-causing epitopes, was more likely to be pathogenic and capable of inducing inflammatory response . Our data favoured the view that some Tg epitopes might be more important than others in the pathogenesis of autoimmune thyroid disease ; a dominant epitope might promote the development of HT. Autoantibodies to Tg might not merely be a simple epiphenomenon derived from thyrocytes destruction but, rather, might play an important role in the trigger for autoimmune thyroid response.
It is conceivable that a euthyroid stage of HT exists, and that progression to a full-blown disease stage is a matter of time . Therefore, a prediction marker of HT progression might be useful to monitor HT patients. The TgAb epitopic specificity pattern might be an important phenotypical marker for predicting HT progression, but only if the conservative epitopic fingerprint phenomenon is confirmed. In Okosieme's study , it was reported that a small number of schoolgirls in Sri Lanka showing AITD-like epitope recognition patterns had high TgAb activity, which persisted in the 5-year follow-up; TPOAb epitope conservation was also observed for serum TPOAb in a cohort of women with postpartum thyroiditis over a 1-year period [17, 18]. Jaume's study found remarkable conservation in the epitopic autoantibody fingerprints to the TPO immunodominant region in 10 of 12 individuals followed over a period of 13–15 years . Therefore, epitope fingerprints might be a characteristic of thyroid antibodies. As only sera TgAb in patients with thyroid dysfunction could recognize the epitope that PB2 identified, we speculated that HT patients with TgAb reacting with an identical epitope partially or completely recognized by PB2 might be at high risk of developing thyroid functional failure, and analysis of the epitope recognition pattern of TgAb might be helpful in predicting HT progression in antibody-positive individuals. A longitudinal follow-up study is needed to provide more evidence for the potential implications of using Tg epitope specificities as prediction markers of HT progression.
Regarding the importance of TgAb, Chen et al. have reported that breaking B cell self-tolerance occurred first for Tg and subsequently for TPO, and TgAb arose first followed later by TPOAb in HT . Hence, it is important to elucidate the nature and location of pathogenic TgAb epitopes. It has been proved that the epitopes on Tg recognized by human autoantibodies are conformational and restricted to certain parts of the molecule. Analysis of the epitopes recognized by PB2 using Tg fragments (fusion proteins or digestion products)  might be helpful in identification of the location of some important Tg epitopes in future.
In our study, the six commercial antibodies were used as probes to determine the epitope recognition patterns of TgAb in HT, because these antibodies did not react with the same epitopes on Tg based on the cross-inhibition results. Previous observations on human and murine TgAb have revealed that there might be an incomplete overlap between the two species probe antibodies [21, 22], and TgAb from immunized mice might recognize the epitopes outside human antigenic determinants . Further studies are needed to confirm whether Tg epitope specificities are phenotypical markers for predicting HT progression by using human monoclonal Tg autoantibodies.
In conclusion, our study demonstrates that HT patients with different thyroid functional status exhibit different Tg epitope recognition patterns. Epitope specificities of TgAb might be helpful in predicting HT progression. More studies on thyroid autoantibodies might provide a different way to investigate the pathogenesis of HT.
This work was supported by the National Natural Science Foundation (no. 30800530) and Beijing Natural Science Foundation (no. 7082096).
The authors have declared no conflicts of interest concerning this paper.