A new type of natural bispecific antibody with potential protective effect in Hashimoto thyroiditis.

CONTEXT
As a new antibody concept, natural bispecific antibodies (nBsAbs) have been detected in long-term passive immunization and some diseases, but their potential immunomodulatory role remains unclear. Hashimoto thyroiditis (HT) appears to fulfill the condition for nBsAb production but has not yet been characterized.


OBJECTIVE
The objective of the study was to identify a new nBsAb against thyroid peroxidase (TPO) and thyroglobulin (Tg) in HT patients and to preliminarily explore its immunomodulatory role.


DESIGN, SETTING, AND PATIENTS
Serum samples were obtained from 136 HT patients, 92 diseased controls, and 99 healthy controls for anti-TPO/Tg nBsAb detection. The relationship between anti-TPO/Tg nBsAb and other clinical parameters was also analyzed.


MAIN OUTCOME MEASURES
The anti-TPO/Tg nBsAb was detected using a double-antigen sandwich ELISA. Higher nBsAb levels were found to be associated with decreased inflammation in HT patients.


RESULTS
The prevalence of anti-TPO/Tg nBsAb in HT was 44.9% (61 of 136), significantly higher than that of diseased controls (2.2%, 2 of 92) (P < .0001) and healthy controls (0%, 0 of 99) (P < .0001). HT patients who were nBsAb positive were prone to have significantly lower levels of serum C-reactive protein and TNF-α compared with the nBsAb-negative individuals (P < .05). The serum amyloid A and interferon-γ levels also showed a similar trend in the two groups. The IgG subclass of anti-TPO/Tg nBsAb was IgG4. Further analysis showed a negative correlation between anti-TPO/Tg nBsAb and serum total IgG4 (r = -0.697, P = .025) in IgG4 thyroiditis patients.


CONCLUSIONS
A new type of nBsAb against TPO and Tg in HT patients is identified. Our data also indicate a protective effect of anti-TPO/Tg nBsAb in the pathogenesis of HT and extend prior knowledge about nBsAb in diseases.

N atural bispecific antibody (nBsAb) is a new antibody entity characterized by two distinct antigen-binding sites and natural production in vivo. The asymmetric antibody breaks the traditional view of IgG antibodies as symmetric, homobivalent molecules released from plasma cells (1). The nBsAb was first reported in allergic patients receiving therapeutic injections with two different allergens (2). Several related studies subsequently revealed that an nBsAb is formed by a half-molecule exchange between two different IgG4 molecules under denaturing conditions. This exchange reaction is attributed to the instability of interheavy chain disulfide bonds in the core-hinge region and noncovalent bonds between the third constant (CH3) domains of IgG4 molecules (3)(4)(5).
Previously we demonstrated that nBsAbs can be artificially induced in rabbits immunized with two unrelated antigens (6). We also detected nBsAb in rheumatoid arthritis (RA), a chronic autoimmune disease with a persistent immune response to cyclic citrullinated peptides and globulins (7). These findings imply that nBsAbs can be generated naturally in autoimmune diseases whose internal environment is characterized by chronic and simultaneous stimulation of distinct antigens. However, nBsAbs have been poorly investigated in autoimmune diseases apart from RA, with no reports regarding the possible role of naturally occurring nBsAbs in vivo.
Hashimoto thyroiditis (HT) is a common autoimmune disease characterized by the presence of goiter and specific autoantibodies against the thyroid. It is an ideal model by which to investigate nBsAb in diseases for the following reasons: 1) it is a chronic disease; 2) there is a consistent presence of higher titer autoantibodies in serum, including antithyroid peroxidase antibodies (TPO) and antithyroglobulin antibodies (Tg); and 3) IgG4 is one predominant subclass among anti-TPO and anti-Tg on the basis of currently available data (8,9). As described above, HT fulfills the conditions for nBsAb production. Interestingly, recent data from Japan suggest that a subgroup of HT may be a part of the spectrum of IgG4-related diseases (10 -12). Based on the hypothesis that a high percentage of IgG4 antibodies could become bispecific immunoglobulins through the Fab-arm exchange in IgG4-related disease (13), it is reasonable to speculate that nBsAbs are present in at least the subset of HT cases.
The aim of this study was to identify a new nBsAb against TPO and Tg in HT patients and discuss its potential immunomodulatory role.

Serum samples
A total of 136 HT patients from Beijing Hospital from June 2012 to May 2013 were enrolled in the study. All patients were newly diagnosed and previously untreated. Clinical diagnosis of HT was based on the presence of a firm, rubbery goiter without any other cause (such as Graves' disease), high TSH levels, and serum positive for anti-TPO and anti-Tg antibodies. For comparison, 92 subjects with other thyroid diseases were tested as well (including 23 cases of hyperthyroidism, 20 of hypothyroidism, 22 of thyroid nodules, 20 of goiter, and seven of thyroid cancer). Ninety-nine healthy controls were also included to de-termine the cutoff value. Cases and controls were well matched in age and sex. Sera were stored at Ϫ80°C until analysis. The study was approved by the Ethics Committee of the National Center for Clinical Laboratories and conducted in accordance with the guidelines of the Declaration of Helsinki.
For the bispecific antibody ELISA, 96-well plates (Nunc Maxisorp) were coated with capture antigen Tg (Prospec; 100 L/ well at 5 g/mL) diluted in PBS (pH 7.4) and incubated overnight at 4°C. After blocking for 2 hours at 37°C with 20% newborn calf serum (NCS) in PBS, the plates were washed with washing buffer [0.05% Tween 20 in PBS (PBST)]. As controls, wells precoated with irrelevant antigens (Sm, RNP, Ro, La, Scl-70, Jo-1) and relevant antigen (thyrotropin receptor) were also prepared. Serum samples (100 L/well) were added to the wells and incubated for 1 hour at 37°C, and the plate was shaken at 600 rpm. After three washes, 100 L of the TPO-HRP conjugate diluted 1:500 in PBST-20% NCS (0.5 g/mL) was added to each test well and incubated at 37°C for 1 hour. After extensive washing, 100 L/well tetramethyl benzidine (Sigma-Aldrich) was added and incubated for 30 minutes at room temperature in the dark. Sulfuric acid (0.5 M, 50 L/well) was added to stop the reaction, and ODs were determined at 450 nm with wavelength correction at 630 nm (OD 450/630 nm, ie, subtracting the 630 nm background absorbance from the 450 nm measurement). Samples were assayed in duplicate.
The working concentration of each reagent of the text was determined by chessboard titrations to optimize the assay performance (15). A test was performed to estimate the percentage of antibodies captured at the incubation step. Briefly, serum samples (100 L/well) were incubated in Tg-coated wells, and then the effluent was collected and subjected to ELISA to measure the percentage of unbound analyte (the pretitrated sample was used as a control).
A reciprocal ELISA in which the plates were coated with TPO and the peroxidase-labeled Tg was used as detection antibody was carried out.

Competitive inhibition assay
To exclude the false bispecificity caused by a cross-reaction between TPO and Tg antigens, a competitive inhibition assay was performed. Fifty microliters of nonlabeled TPO or nonlabeled Tg at different concentrations (0 -16 g/mL) in PBS was used as inhibitor and mixed with an equal volume of TPO-HRP conjugate (0.5 g/mL), followed by the measurement of anti-TPO/Tg as described above. The inhibition effect was measured by a decrease in expected color of the pretitrated sample (used as a control).
A similar method was applied to another competitive inhibition assay using the reciprocal ELISA. The competitive inhibition assay in the first capturing stage was also performed, in which the plates were coated with TPO and HRP-Tg was used as detection antibody. In this assay, the inhibitor TPO (0 -16 g/mL) was added with the sample.

Serum rheumatoid factor test
To exclude the false bispecificity caused by rheumatoid factor (RF), we performed quantitative detection of IgG, IgM, and IgA RF in all anti-TPO/Tg nBsAb-positive HT samples using a commercial kit (AESKU). The reference interval for adults was less than 12 U/mL according to the manufacturer's instructions.

Size-exclusion chromatography
To exclude the involvement of antibody aggregates, size-exclusion chromatography was carried out on the BioLogic Duo-Flow System (Bio-Rad Laboratories) using a GE Hiprep26/60 Sephacryl S-200 high-resolution column (GE Healthcare). The anti-TPO/Tg nBsAb-positive serum pool and commercial monomeric IgG (Green Cross China Biological Products) were fractionated on the same column, respectively. Fractions of 1 mL were collected and analyzed for anti-TPO/Tg nBsAb.

IgG subclass determination of anti-TPO/Tg nBsAb
IgG subclass determination of anti-TPO/Tg nBsAb was performed using a new magnetic separation technique. According to the manufacturer's instructions, 100 g each of capture antibodies (mouse antihuman IgG1, mouse antihuman IgG2, mouse antihuman IgG3, mouse antihuman IgG4; Invitrogen) were covalently coupled to 10 mg of surface-activated Dynabeads (Dynabeads Epoxy; Invitrogen) using a Dynabeads coupling kit. The beads coupled with distinct antibodies were then added to independent 1-mL aliquots of sample containing anti-TPO/Tg nBsAb. After 30 minutes of incubation to allow for affinity capture of the targets (IgG1, IgG2, IgG3, IgG4) by the beads, the beads were applied to a magnet. The supernatant was removed and the beads were washed. Then bead-bound targets were eluted off the beads with 2 M KCl. After a step of dialysis against PBS, the four human IgG subclass samples were finally obtained. The IgG subclass of anti-TPO/Tg was determined by the determination of bispecific activity in different samples using the nB-sAb ELISA described above.

IgG4 subclass detection of TPO and Tg autoantibodies
For IgG4 subclass detection, 96-well plates were coated with antigen TPO or Tg (100 L/well at 5 g/mL) diluted in PBS and incubated overnight at 4°C. After blocking for 2 hours at 37°C with 20% NCS in PBS, the plates were washed with 0.05% PBST. Serum samples (100 L/well) previously diluted 100-fold were added to the wells and incubated for 1 hour at 37°C, and the plate was shaken at 600 rpm. After five washes, HRP-conjugated antihuman IgG4 (Sigma-Aldrich; 100 L/well) diluted 1:1000 in PBST-20% NCS was added to each test well and incubated at 37°C for 1 hour. After extensive washing, a colorimetric estimation was carried out as described above.

Serum total IgG4 detection
To investigate whether there is a relationship between anti-TPO/Tg nBsAb levels and serum total IgG4 levels, serum total IgG4 was determined by immunonephelometry (Nephelometer Analyzer II; Siemens) using commercial kits (N Latex IgG4; Siemens). The reference interval for adults was 80 -1350 mg/L according to the manufacturer's instructions.

Inflammatory marker measurements
To analyze the relationship between the inflammation status of HT and anti-TPO/Tg nBsAb, a panel of well-studied inflammatory markers involved in autoimmune diseases, namely C-reactive protein (CRP), serum amyloid A (SAA), TNF␣, and interferon-␥ (IFN␥) were measured.
High-sensitivity CRP was assessed by means of particle-enhanced turbidimetry with a clinical chemistry analyzer (Nephelometer Analyzer II; Siemens). The minimum detectable value was 0.165 mg/L. SAA was quantitatively determined using a solid-phase sandwich ELISA kit (Invitrogen). The quantification of TNF␣ and IFN␥ was carried out using commercial ELISA kits (R&D Systems), also using the quantitative sandwich immunoassay technique. According to the manufacturer's instructions, standards with known content, controls (negative control, positive control, and cutoff calibrator) and samples were tested at the same time. The concentration of a particular marker was calculated with respect to its corresponding standards.

Statistical analysis
All analyses were performed using SPSS 13.0 for Windows and GraphPad Prism version 5.0 (GraphPad Software). For continuous variables, results are expressed as mean Ϯ SD; differences between two groups were analyzed by unpaired t tests. For variables that were not normally distributed, results are expressed as the median and range; differences between groups were assessed using the Mann-Whitney U test. Correlations were determined by Spearman's rank correlation coefficients. Comparisons of proportions were carried out using a 2 test or Fisher's exact probability test. The significant level was set at P Ͻ .05.

Bispecific antibody can be specifically detected in HT patients
We established a double-antigen sandwich ELISA for anti-TPO/Tg nBsAb detection and performed a study about the interfacial reaction kinetics of the ELISA. In our study, serum samples were incubated at 37°C for 1 hour, and the plates were shaken at 600 rpm. Approximately 67% of the antibodies were captured under this condition (Supplemental Figure 1).
Bispecific antibodies response to both TPO and Tg were successfully detected, and the distribution of this new nBsAb among HT, diseased controls, and healthy controls is shown in Figure 1. High titers of anti-TPO/Tg nBsAb were detected solely in HT patients. By contrast, the OD values were extremely low in diseased controls (0.018 Ϯ 0.058) and healthy controls (0.016 Ϯ 0.010). To obtain higher specificity, the cutoff value for positivity was set as 10 SD above the mean OD of the healthy controls, ie, 0.116 [0.016 ϩ (10 ϫ 0.010)]. The prevalence of anti-TPO/Tg nBsAb in HT was 44.9% (61 of 136), significantly higher than diseased controls (2.2%, 2 of 92; P Ͻ .0001) and healthy controls (0%, 0 of 99; P Ͻ .0001). Similar results were achieved using the reciprocal ELISA (Supplemental Figure 2).

Authenticity analysis of anti-TPO/Tg nBsAb
A competitive inhibition assay was carried out to validate the specificity of anti-TPO/Tg nBsAb. Results showed an obvious reduction in bispecific activity of nB-sAb in the presence of nonlabeled TPO at 0.5 g/mL, and the absorbance was almost close to zero when nonlabeled TPO was added at 8 g/mL ( Figure 2). Unlike nonlabeled TPO, bispecific activity was not suppressed, even upon addition of an extremely high concentration of Tg (16 g/mL) (Figure 2), which indicated that bispecific activity was not a result of cross-reactivity.
The competitive inhibition assay using the reciprocal ELISA and the competitive inhibition assay in the first capturing stage were also carried out, and the results of the assays were in line with expectation, further excluding false bispecific activity due to a cross-reaction between TPO and Tg (Supplemental Figures 3 and 4).
To exclude an artifact due to RF, the levels of RF in all anti-TPO/Tg nBsAb-positive samples were quantitatively tested. Results showed that only three patients (3 of 61) were RF positive, and the concentrations of RF were 24 U/mL, 56 U/mL and 102 U/mL, respectively. The corresponding anti-TPO/Tg nBsAb levels (OD 450/630 nm) of these three samples were low (0.94, 0.163, and 0.232, respectively). The nBsAb was further measured in five RFpositive samples from RA patients, and no positive signal was observed, even though the RF concentrations of these samples were up to 2000 IU/mL (data not shown).
To demonstrate that the anti-TPO/Tg nBsAb was monomeric, size-exclusion chromatography was performed. Three distinct peaks were obtained from the serum pool ( Figure 3) and analyzed by nBsAb ELISA. Results are expressed as ELISA end point titers and plotted against the elution time. As expected, nBsAbs were detected only at peak 2, which overlapped with the peak obtained from monomeric IgG.

Levels of anti-TPO/Tg nBsAb were significantly correlated with IgG4 subclass of anti-TPO and anti-Tg
Because we hypothesized that high titer of IgG4 subclass of anti-TPO and anti-Tg might contribute to the formation of anti-TPO/Tg nBsAb, we analyzed the relationship between them. As expected, the levels of anti-TPO/Tg nBsAb were strongly correlated with IgG4 anti-TPO (r ϭ 0.785, P Ͻ .0001) and IgG4 anti-Tg (r ϭ 0.706, P Ͻ .0001) levels. This result further confirms the IgG4 nature of anti-  TPO/Tg nBsAb and suggests that anti-TPO/Tg nBsAb was generated by a half-antibody exchange reaction between IgG4 anti-TPO and IgG4 anti-Tg.

Serum total IgG4 values were inversely related to anti-TPO/Tg nBsAb levels
Of the 136 HT patients, 10 patients (five males, five females) had elevated serum total IgG4 levels (Ͼ1350 mg/ L). Their anti-TPO/Tg nBsAb and serum total IgG4 levels are shown in Table 1. All 10 patients had positive anti-TPO/Tg nBsAb, but their levels of serum total IgG4 showed an unexpected negative correlation with the levels of nBsAbs (r ϭ Ϫ0.697, P ϭ .025).

The distribution of anti-TPO/Tg nBsAb in HT patients with different thyroid functional status
The HT patients were divided into three groups according to thyroid function (16 -18): euthyroidism group (Eu) (n ϭ 78, 11 males, 67 females) whose TSH and FT4 levels are normal; subclinical hypothyroidism group (sH) (n ϭ 32, six males, 26 females) with high levels of TSH and normal levels of FT4; hypothyroidism group (H) (n ϭ 26, five males, 21 females) with high levels of TSH and subnormal levels of FT4. The distribution of anti-TPO/Tg nBsAb in these three groups is shown in Figure 4. The titers of anti-TPO/Tg nBsAb were significantly higher in the sH group and the H group than those in the Eu group (P ϭ .038; P ϭ .002, respectively).

Inflammatory markers analysis
The comparison of the levels of inflammatory markers in the anti-TPO/Tg nBsAb-positive and -negative groups is shown in Table 2. The CRP and TNF␣ levels were significantly lower in the anti-TPO/Tg nBsAb-positive group when compared with those of the negative group (P Ͻ .05). The SAA and IFN␥ levels of the anti-TPO/Tg nBsAbpositive group also displayed a trend toward being lower than those of the negative group, although the differences were not necessarily meaningful.

Discussion
HT is a common autoimmune thyroid disorder, affecting up to 2% of the overall population. Anti-TPO and anti-Tg autoantibodies are a hallmark of HT, and high titers of these can be detected several years before apparent symp- Figure 3. Size-exclusion chromatography. A, Anti-TPO/Tg nBsAbpositive serum pool and commercial monomeric IgG were fractionated on the same column, respectively. Three distinct peaks (peaks 1-3) were obtained from the serum pool, and peak 2 overlapped with the peak obtained from monomeric IgG (peak a). B, All fractions of the three peaks were analyzed by a bispecific antibody ELISA, and the bispecific antibodies were detected only at the expected position (peak 2). toms appear (19,20). Given that HT is a chronic disease and that IgG4 is a predominant subclass among anti-TPO and anti-Tg autoantibodies, it fulfills the conditions for nBsAb production. Thus, it is reasonable to expect that nBsAb against TPO and Tg may naturally exist in HT patients.
In this research, we have established a double-antigen sandwich ELISA for anti-TPO/Tg nBsAb detection and performed a study about the interfacial reaction kinetics of this assay (Supplemental Figure 1). As expected, the new kind of nBsAb was successfully identified in HT patients. To validate the specificity of anti-TPO/Tg nBsAb, we carried out a serial of competitive inhibition assays to exclude a cross-reaction between TPO and Tg (Figure 2 and Supplemental Figures 3 and 4). The results we obtained were consistent with previous research, suggesting that antihu-man TPO and antihuman Tg antibodies present no crossreactivity (21). To further verify the authenticity of anti-TPO/Tg nBsAb, we also carried out an RF test to exclude its interference and size-exclusion chromatography analysis to demonstrate the monomeric IgG nature of anti-TPO/Tg nBsAb. By detecting the bispecific activity, a high prevalence of anti-TPO/Tg nBsAb was observed in HT patients (62 of 136, 45.6%), whereas only two cases were observed in diseased controls (2 of 92, 2.2%) and none in healthy controls.
In humans, it is generally believed that nBsAbs are of the IgG4 subclass and a half-molecule exchange between two structurally unstable IgG4 molecules provides a convincing mechanism for their generation (2-5, 7, 22). However, one recent study showed that other subclasses (IgG1, IgG2, and IgG3) can also participate in the exchange (23). In our study, the bispecific activities of anti-TPO/Tg nB-sAb can be detected only in a pure IgG4 sample from HT serum by magnetic separation technology. This, combined with the strong correlation between the levels of anti-TPO/Tg nBsAb and IgG4 monospecific antibodies (IgG4 anti-TPO and IgG4 anti-Tg), suggests that the predominant IgG subclass of anti-TPO/Tg nBsAb is likely to be IgG4. This finding is consistent with previous reports describing other kinds of IgG4 nBsAbs in patients with allergy or RA (2,7).
One of the interesting findings of this study is the inverse correlation between the levels of anti-TPO/Tg nBsAb and serum total IgG4. Recently IgG4 thyroiditis, a unique subtype of Hashimoto's thyroiditis that shares similar IgG4-related sclerosing features, has been widely recognized to be part of a generalized IgG4-related disease (10 -12, 24 -26). Because serum total IgG4 examination is a noninvasive method and consistent with histological examination, it can be used routinely by clinicians to distinguish IgG4 thyroiditis from non-IgG4 thyroiditis (27). As shown in Table 1, 10 HT patients exhibited an elevation of serum total IgG4 (Ͼ1350 mg/L, the serum criterion of IgG4 related disease). Although these patients were anti-TPO/Tg nBsAb positive, their levels of serum total IgG4 showed an unexpected negative correlation with the levels Figure 4. The distribution of anti-TPO/Tg nBsAb in HT patients with different thyroid functional status. The 136 HT patients were divided into three groups, based on their TSH and FT4 levels: euthyroidism group (Eu) whose TSH and FT4 levels are normal; subclinical hypothyroidism group (sH) with high levels of TSH and normal levels of FT4; and hypothyroidism group (H) with high levels of TSH and subnormal levels of FT4. The differences of anti-TPO/Tg nBsAb titers between groups were assessed using the Mann-Whitney U test. The titers of anti-TPO/Tg nBsAb were significantly higher in the sH and H groups than in the E group (P ϭ .038; P ϭ .002, respectively). of anti-TPO/Tg nBsAb (r ϭ Ϫ0.697, P ϭ .025). There might be extensive half-molecule exchanges in HT patients with high titers of serum total IgG4, resulting in various nBsAb against as-yet-unknown bispecific epitopes. In other words, half-molecule exchange can occur between anti-TPO or anti-Tg and other IgG4 molecules, generating other nBsAbs in addition to anti-TPO/Tg nBsAb, resulting in a reduction in anti-TPO/Tg nBsAb proportion.
HT may take place through three stages, ie, the early stage (euthyroid stage), second, and final stage (hypothyroidism). In general, it is an inconvertible process, and status of thyroid function can provide useful information on disease duration. The fact that the thyroid dysfunction group had increased anti-TPO/Tg nBsAb levels when compared with the euthyroidism group suggests that the generation of nBsAb could be a long-term process, and the exchange mechanism might not operate obviously in the early stages of HT. This is consistent with previous reports indicating that a prolonged immune response is a prerequisite for nBsAb generation (6,7).
Passively injected human IgG4 antibodies have been shown to have antiinflammatory activity via dynamic Fabarm exchange in an animal model (3), but no study has been performed to explore the potential role of nBsAbs in the pathogenesis of human diseases. Our study preliminarily addresses this issue based on results of inflammatory markers analysis. We show evidence of a relationship between anti-TPO/Tg nBsAb levels and inflammation, as determined by levels of CRP, SAA, TNF␣, and IFN␥. These are well-studied inflammatory markers for acute and chronic inflammation of diverse causes, and the changes of their levels can reflect the intensity of inflammation (28 -30). In this study, serum high-sensitivity CRP and TNF␣ levels were significantly lower in anti-TPO/Tg nBsAb-positive HT patients than in negative individuals, and SAA and IFN␥ levels of the anti-TPO/Tg nBsAb-positive group also showed a decreasing trend when compared with those of the negative group. These results provide important evidence to suggest that nBsAb may not solely be a consequence of disease development but rather serve as a protective factor exerting a potent antiinflammatory effect on autoimmunity. One theory has been proposed that the functionally monovalent nBsAb can retain monospecific antibodies from participating in a tissue-destructive inflammation response, which may explain these observations (13). Further research aimed at exploring the emergence of nBsAb on prognosis of disease and validating the pathogenesis of nBsAb in a suitable animal model may provide firm evidence about the functions of nBsAb.
We identified a new type of nBsAb against TPO and Tg in HT patients for the first time. Based on inflammatory markers analysis, we also preliminarily discussed its potential role as an immunomodulator. The results from this study indicate a protective effect of anti-TPO/Tg nBsAb in the pathogenesis of HT and extend prior knowledge about nBsAb in diseases.