Antibodies to TSH-receptor in thyroid autoimmune disease interact with monoclonal antibodies whose epitopes are broadly distributed on the receptor

Abstract

The hyperthyroidism of Graves' disease (GD) is caused by TSH-receptor (TSH-R) stimulating autoantibodies (TSAb), leading to overproduction of thyroid hormones. We present evidence for TSAb interaction with three distinct regions of the TSH-R, one in immediate vicinity of the carboxy terminal serpentine. Three murine monoclonal antibodies (MoAbs 28·1, A9 and 31·7) directed to amino acids 36–40, 147–228 and 382–415 were labelled and tested for their binding to human recombinant TSH-R on solid phase. All MoAbs bound to TSH-R with a K_d of 8–12 nm and showed no competition among themselves. We tested 114 sera from euthyroid controls, 118 TBII positive sera from patients with GD (containing TSAb confirmed by bioassays), 16 TBII positive sera from patients with autoimmune thyroid disease (AIT), who were hypothyroid and had TSH blocking antibodies (TBAb), and 20 patients with AIT, who were hypothyroid but negative for all TRAb. Mid-regional MoAb A9 tracer achieved the highest sensitivity in the GD group (72·0%), whereas C-terminal MoAb 31·7 found most sera positive in the AIT group (87·5%). Surprisingly, the N-terminal MoAb 28·1 had the lowest sensitivity in the GD (10·4%) and AIT group (43·8%). Using a mixture of all three tracer MoAbs did not increase the sensitivity in the GD or AIT group, compared to the best single MoAb alone. Median inhibition of MoAb A9 was significantly (P < 0·001) higher than inhibition of MoAbs 28·1 or 31·7 in the group of GD patients but not in other groups. Almost all patient sera with positive reactivity in the MoAb tracer assays had TBII values in the higher range. However, there were many highly TBII positive sera, which did not show a displacement of the MoAb tracers. We conclude that, contrary to some reports, the binding of TSAb and TBAb to the TSH-R is not restricted to distinct and distant epitopes. The middle part of the TSH-R seems to be more relevant for TSAb binding than the N-terminal part, while a proportion of TSAb autoantibodies also binds to a C-terminal epitope of the TSH-R. The method described here is a TSH independent competitive assay for the detection of TSH-R autoantibodies.

Introduction

Graves’disease (GD) is characterized by thyroid stimulating autoantibodies (TSAb) directed against the thyrotropin receptor (TSH-R) leading to receptor activation in the absence of TSH (see [1,2] for review). In another group of patients with autoimmune thyroid disease (AIT), TSH-R autoantibodies block thyroid activation (TBAb) [3], leading to hypothyroidism. Those autoantibodies with bona fide blocking (no agonistic) activity are very rare [4], but they are useful tools to understand the mechanism of thyroid autoimmunity. Over the last years, a variety of experimental monoclonal antibodies (MoAbs) was raised to the TSH-R by different techniques. Until recently, all failed to show TSAb activity, although some showed TBAb activity in the bioassay, or were able to block TSH action in a ¹²⁵I-TSH competition assay (TBII activity) [5–9]. Finally, three independent groups succeeded in the establishment of stable MoAbs with TSAb activity [10–12], and recently the first human monoclonal antibody with TSAb activity was reported [13].

The precise location of TSAb or TBAb epitopes and the mechanical action of TSH-R stimulation or inhibition is still under debate and sometimes contradicting data are reported. Early reports on the presence of TSAb epitopes at the N terminal and TBAb epitopes at the C terminal part of the extra cellular domain of TSH-R [14–16] are still strongly favoured [17] despite recent reports suggesting that TSAb, TBAb and TSH share epitopes in close vicinity [11,18].

In this study we used three murine MoAbs with defined epitopes and studied their competition in TSH-R binding with autoantibodies from patients with GD or AIT. By selecting MoAbs to the N-terminal, middle and C-terminal part of the ectodomain, we tried to verify earlier studies on a distinct and different binding site for TSAb and TBAb. We found heterogeneous binding of TBAb to all three regions of the TSH-R. Surprisingly, interaction of TSAb with the N-terminal MoAb was much lower than expected from earlier reports, while binding of TSAb sera to the C-terminal part of the ectodomain was evident.

Materials and Methods

Patient sera

Graves’ disease (GD) sera (n = 118) were obtained from blood donors recruited for the development of in vitro diagnostics (Invent GmbH, Biotechnology Center Hennigsdorf bei Berlin, Germany). This blood donation for the development of in vitro diagnostics was approved by a national ethical committee. Graves’ disease was defined on clinical terms by a physician, and confirmed by antibody detection in the human recombinant TBII assay (DYNOtest^® TRAK human, BRAHMS AG, Berlin). The presence of TSAb was confirmed by bioassay detection (see below). All sera had stimulation indices > 1·5 (compared to a euthyroid control pool) and no TBAb activity.

Sera with hypothyroid autoimmune disease (AIT) were divided into two groups. Those (n = 16) with high TBII and TBAb activity and no agonistic TSAb activity were a kind gift from Dr Daphne Khoo, Singapore General Hospital. These sera are described in detail elsewhere [4]. Sera in the second group (n = 20), classified as Hashimoto's thyroiditis, were TBII negative, negative for TSAb and TBAb, and were selected for their high IgG titre and the presence of anti-TPO and anti-TG autoantibodies as measured by commercial assays (DYNOtest^® TRAK human, DYNOtest^® anti TPOn and DYNOtest^® anti TGn, BRAHMS AG).

A total of 114 control sera were obtained from individuals with no personal or family history of endocrine autoimmune disease. Sera were negative for autoantibodies to TBII, TPO and Tg. Written consent was given by all blood donors.

Monoclonal antibodies used as tracer

Blocking MoAbs 28·1 and 31·7 were kindly provided by Dr Sabine Costagliola, Brussels. MoAb 28·1 is directed to amino acids 36–40 of the TSH-R ectodomain. MoAb 31·7 was produced in the same way and has identical properties as described for the MoAb 15 [9], and is directed towards amino acids 382–415 near the C-terminus of the TSH-R ectodomain, a region known to be involved in TSH and TBAb binding [8]. MoAb A9 is directed to amino acids 147–228 in the mid-region of the TSH-R ectodomain, and is described in detail elsewhere [19].

Purified MoAbs were labelled using acridinium ester as follows. Antibodies (120 µg in 0·4 ml of 200 mm sodium-phosphate buffer, pH 8·0) were incubated for 1 h with 20 µl acridinium ester (1 mg/ml in acetonitril, Hoechst AG, Frankfurt, Germany). Labelled antibodies were purified by HPLC using a hydroxy apatite column (running buffers 1 and 500 mm potassium-phosphate, pH 6·8). The obtained fractions were collected, aliquoted and stored at −20°C.

Preparation of F(ab)₂ fragments

F(ab)₂ fragments were generated using ImmunoPure Fab Preparation Kit (Pierce, Rockford, USA), according to the manufacturer's instructions. Briefly, 1 mg of 31·7 antibody in 1 ml of 25 mm NaAc, pH 4·2, was mixed with 0·1 ml of immobilized pepsin slurry and incubated by shaking at 37°C for 4 h. The mixture was centrifuged, supernatant was collected and neutralized with an equal volume of 200 mm Na-phosphate buffer, pH 8·0 (final pH value 7·5). The crude digest was passed through protein A-sepharose column to adsorb undigested antibodies. The flow through fraction was collected and used for preparation of F(ab)2 tracer as described above for total IgG.

MoAb assay principle

The assay is based on the human recombinant TBII assay (TRAK human, BRAHMS AG), except that the tracer is a labelled MoAb instead of labelled TSH. TSH-R coated tubes were produced as described [20]. Sera (100 µl) were added in duplicate to TSH-R coated tubes. To this were added 200 µl buffer containing 100 mm Hepes-NaOH, pH 7·5, 0·5% Triton-X 100, 1% BSA, 20 mm EDTA, 0·5 mm N-ethyl-malemide and 0·1 mm leupeptin. After 2 h incubation under shaking at room temperature (RT), tubes were washed twice with 2 ml washing buffer (8 mm Tris-HCl, 60 mm NaCl, 0·02% Tween-20, pH 7·5). Then, 300 µl tracer (1 ng MoAb, ∼200 000 rlu) in buffer (100 mm Hepes-NaOH, pH 7·0, 50 mm NaCl, 0·5% Triton-X 100, 0·5 mg/ml mouse IgG) were added to tubes and incubated for 1 h at RT by shaking. Tubes were washed three times with 2 ml washing buffer and detection was performed in a luminometer (Berthold, Germany). The MoAb binding inhibiting activity of the sera was presented as inhibition index (InI) calculated as:

$$\text{InI}\left(\text{\%}\right)=[1-(\text{count rate for the test serum/mean count rate for control sera}\left)\right]\times 100\text{.}$$

Calculation of K_d of MoAbs

Different concentrations of labelled MoAbs were added in 0·3 ml buffer (20 mm Hepes-NaOH, pH 7·5, 150 mm NaCl, 1% BSA, 10% glycerol) to TSH-R coated tubes (DYNOtest TRAK human). Tubes were incubated for 24 h at room temperature, washed four times with 2 ml washing buffer, and bound RLU were measured in a luminometer. Non-specific binding, determined in excess of unlabelled purified MoAb, was subtracted. The dissociation constant (K_d) was calculated at 50% saturation by non-linear regression analysis (Prism 4, Graphpad Software).

Bioassays for TSAb and TBAb detection

Data on the biological activity of all sera were obtained from at least one of two established bioassays. Either the conventional cAMP bioassay using TSH-R expressing JP09 cells [21] or a modified bioassay were ligand binding to the TSH-R induces a luciferase reporter gene [22,23] was used.

TBII assay

TBII were measured by the human recombinant TBII assay (DYNOtest^® TRAK human, BRAHMS AG) following the manufacturer's instructions. This assay is calibrated with human autoantibodies [20]. Data are expressed in percentage inhibition of TSH binding, or alternatively in international units (IU/l), where 1 IU/l represents about 10% inhibition of TSH binding, 8 IU/l about 50% inhibition and 40 IU/l (the highest standard) > 85% inhibition [24].

Definition of cut-off and statistical analysis

To obtain the optimal decision threshold level for positivity, receiver-operating characteristic (ROC) analysis was performed [25]. Sensitivity/specificity pairs were calculated by varying the decision threshold levels over the entire range of inhibition by MoAbs. Sensitivity (the true-positive results) was calculated from the 134 patients with GD or AIT, while specificity (the true-negative results) was calculated from 114 healthy controls. Statistical analysis was performed using Kruskall–Wallis anova or Mann–Whitney rank sum analysis for comparison of the autoantibody levels in the different groups determined with one assay, and Wilcoxon's signed rank test for comparing median values obtained in different assays. Method comparison was performed by Spearman's rank analysis.

Results

After labelling with acridinium ester, all three MoAbs showed high affinity binding to TSH-R on solid phase. The K_d of all three MoAbs was obtained by saturation experiments. Figure 1 shows as example the binding of MoAb 28·1. The K_d was calculated by non-linear regression analysis and was 8·6 nm (95% confidence interval: 6·8–10·4 nm). The K_d for MoAb A9 was 11·1 nm (95% CI: 7·2–15·0), and for MoAb 31·7 the K_d was 11·9 nm (95% CI: 9·7–14·0), respectively (figures not shown).

Fig. 1.

Determination of K_d of MoAbs. The K_d of all MoAbs was determined by saturation studies as described in the Methods and calculated by non-linear regression analysis (Prism 4, Graphpad), as shown here for MoAb 28·1. The insert shows the conventional Scatchard plot.

This specific binding was maintained in the presence of euthyroid control serum. While the binding of MoAb 28·1 was about as effective as the binding of labelled bovine TSH (10·9% of total versus 12·3% of total, Fig. 2a), MoAb A9 gave a lower specific binding (7·98%) of total tracer added (200 000 rlu) (Fig. 2b), and the lowest specific binding with the highest background was seen for MoAb 31·7 (4·57%) (Fig. 2c). In the presence of excess unlabelled antibody (50 µg/ml), this specific binding was suppressed almost completely, resulting in relative light units similar to the background seen in the absence of TSH-R. Addition of bovine TSH (2 U/ml) resulted in no inhibition of MoAb 28·1 binding (7% compared to control serum), a moderate inhibition of MoAb A9 binding (37%) and complete reduction of MoAb 31·7 binding to background.

Fig. 2.

Binding of labelled MoAbs to TSH-R coated tubes. Binding of acridinium ester-labelled MoAbs to TSH-R coated tubes. Amount of total tracer added per tube was 200 000 rlu. No receptor: tubes without TSH-R were used to estimate non-specific binding of tracer. Control serum: specific binding of MoAb tracer in the presence of healthy control sample. 50 µg/ml antibody: excess amount of unlabelled MoAb to estimate maximal inhibition. TSH: inhibition by bovine TSH. s1–s3: inhibition by sera from patients with Graves’ disease containing stimulating autoantibodies to the TSH-R (TSAb). b1–b3: inhibition by sera from patients with hypothyroid autoimmune disease containing blocking autoantibodies to the TSH-R (TBAb). (a) MoAb 28·1 (amino terminal part of TSH-R ectodomain). (b) MoAb A9 (mid-region). (c) MoAb 31·7 (carboxy terminal part of TSH-R ectodomain). (d) Mixture of all three MoAbs. In all panels, the same human sera were used.

Binding of all three labelled MoAbs to the TSH-R could be inhibited to some extent by TBII positive sera from patients with GD or hypothyroid autoimmune disease. Figure 2 shows data as relative light units of six sera with high TBII activity (> 90% inhibition, not shown), three of those with TSAb activity (s1–s3) and three with TBAb activity (b1–b3). Also, to mimic the polyclonality discussed for GD sera, we combined all three MoAb tracers (Fig. 2d).

Inhibition of TSH binding of the unlabelled MoAbs in the coated tube TBII assay resulted in no (8%) TBII activity of MoAb 28·1, weak (42%) TBII activity of MoAb A9, and moderate (66%) TBII activity of MoAb 31·7 at high antibody concentrations (150 µg/ml, data not shown).

All three MoAbs showed no competition between themselves, as assessed by addition of 15 µg/ml unlabelled antibody to 2·5 ng labelled tracer (Table 1).

Table 1.

Competition of MoAbs among themselves

	Unlabelled MoAb 28·1	Unlabelled MoAb A9	Unlabelled MoAb 31·7
Labelled MoAb 28·1	70%	0%	1%
Labelled moab A9	0%	89%	1%
Labelled MoAb 31·7	0%	1%	79%

Data show percentage inhibition of binding of 2·5 ng labelled MoAb with a 100-fold higher concentration of unlabelled MoAb.

To examine the influence of non-specific hindrance in the experiment, we produced F(ab)₂ fragments of one MoAb and compared the inhibition of sera on labelled F(ab)₂ fragments and total IgG. Figure 3 shows the data for five controls and 10 positive patients. There was no difference between total IgG and F(ab)₂ fragments.

Fig. 3.

No difference between labelled total IgG and labelled (Fab)₂ fragments. The same sera were tested with either labelled total IgG MoAb 31·7 or F(ab)₂ fragments of this MoAb. There was no difference in displacement of both tracers by autoantibodies in the serum.

To assess the antibody heterogeneity in more patients, we tested a total of 268 sera [114 controls, 118 sera from patients with GD, 16 sera from TRAb positive and 20 from TRAb negative patients with autoimmune hypothyroidism (AIT)] and compared the results with the TRAK human assay. Data were expressed as percentage inhibition of antibody binding (MoAbs) or bTSH binding (TRAK human). (Fig. 4).

Fig. 4.

Inhibition of MoAb tracer by sera from patients with Graves’ disease. A total of 268 sera was tested for their inhibitory effect on the three MoAbs to the TSH-R. control: 114 control sera from blood donors with no personal or family history of endocrine autoimmune disease. GD: 118 sera form patients with GD containing TBII and TSAb. AIT: 16 sera from TBII and TBAb positive patients with autoimmune hypothyroidism. AIT control: 20 sera of patients with autoimmune hypothyroidism without TBAb or TBII. (a) MoAb 28·1 (amino terminal part of TSH-R ectodomain). (b) MoAb A9 (mid-region). (c) MoAb 31·7 (carboxy terminal part of TSH-R ectodomain). (d) Labelled bovine TSH was used as a tracer instead of MoAb, essentially the human recombinant TBII assay (LUMItest TRAK human). Data were expressed as percentage inhibition of antibody binding (a–c) or bTSH binding (d). The median is indicated by a solid line; the ROC calculated cut-off (see Table 2) is indicated by the dotted line.

In three assays (A9, 31·7, bTSH) the group of GD patients showed significantly (P < 0·0001) higher inhibition of tracer binding compared to healthy or AIT controls. However, this was not significant with MoAb 28·1. In all four assays the group of TRAb positive AIT patients showed (including MoAb 28·1) significantly (P < 0·0001) higher inhibition of tracer binding compared to healthy or AIT controls. Between the two TRAb positive groups, the TRAb positive AIT patients showed significantly (P < 0·001) higher inhibition of tracer binding in three assays (28·1, 31·7, bTSH), but not with MoAb A9, than the GD patients. No significant differences were seen between the two control groups for all four tracers.

Between the MoAbs, median inhibition seen with MoAb A9 was significantly higher in the GD group than inhibition seen with MoAb 28·1 or MoAb 31·7 (P < 0·001, Kruskal–Wallis anova). However, there was no significant difference between the MoAb tracers in the TRAb positive AIT group.

ROC plot analysis to define the cut-off was carried out with the 114 control sera for specificity and the 134 TRAb positive GD and AIT patients’ sera for sensitivity. To compare all tracers, individual cut-offs were selected to obtain a specificity of 99·0%. The cut-off for this specificity was used to calculate the number of positive sera in the group of GD patients and TRAb positive AIT patients. Table 2 summarizes the data. The data for the TRAK human are given as comparison. The area under the curve (AUC) was highest for MoAb A9 (0·90). This tracer also achieved the highest sensitivity in the GD group (72·0%, P < 0·001 by χ² test versus all other MoAbs), whereas MoAb 31·7 found most sera positive in the AIT group (87·5%, P < 0·01 by χ² test versus all other MoAbs). Surprisingly, the N-terminal MoAb 28·1 had the lowest sensitivity in the GD and TRAb positive AIT group. Using a mixture of all three tracer MoAbs did not increase the sensitivity in the GD or AIT group, compared to the best single MoAb alone.

Table 2.

Results of ROC plot analysis

Tracer used	MoAb 28·1	MoAb A9	MoAb 31·7	mix MoAb	bTSH
AUC	0·60 (0·52–0·67)	0·90 (0·86–0·94)	0·78 (0·72–0·84)	0·75 (0·68–0·81)	1·0
Cut-off (% inhibition)	16	10	12·5	14	11·5
Positve GD sera	10·4%	72·0%	44·1%	23·2%	100%
Positive AIT sera	43·8%	68·8%	87·5%	61·5%	100%

AUC: area under the curve. Cut-off: clinical decision threshold at which all assays obtain 99% specificity. Positive GD sera: percentage of Graves’disease sera above the cut-off. Positive AIT sera: percentage of sera from patients with autoimmune hypothyroidism (containing TRAb) above the cut-off.

Only 7·6% of GD but 31·3% of AIT sera were positive in all three assays, and about one-third of GD patients were positive in two assays. Overall, the contribution of MoAb 28·1 to identifying GD patients was lowest, while MoAb A9 detected the highest number of GD patients.

There was only a weak, but significant correlation between TSAb and TBII values. Similar weak correlations were obtained with TSAb and MoAb tracer data (r-values between 0·28 and 0·41). The correlation between TBII and MoAb tracer data was stronger (r-values between 0·47 and 0·60). Almost all sera with positive reactivity in the MoAb tracer assays had high TBII values. However, there were also many highly TBII positive sera, which did not show a displacement of the MoAb tracers. The sera, which competed with all three MoAb tracers, had higher TBII values compared to sera competing with none or only one MoAb tracer (Fig. 5).

Fig. 5.

Correlation between TBII data and positive MoAb assays in GD patients. Correlation between the TBII data and the number of positive MoAb assays for each serum. 0: Serum is negative in all three MoAb assays. 1: Serum is positive in only one of three MoAb assays. 2: Serum is positive in a combination of two MoAb assays. 3: Serum is positive in all three moab assays.

Discussion

In Graves’ disease (GD), autoantibodies to the TSH-R (TRAb) bind to several parts of the TSH-R ectodomain and interact with the binding sites of the natural ligand TSH. This property is used for the detection of TRAb by radio receptor assays, employing either conventional porcine thyroid membranes in liquid phase [26] or modern coated tube technology with human recombinant antigen [20]. Despite a variety of proposals for the further subclassification of TRAb, the simplified concept of TSAb and TBAb contributing to the TBII activity measured in those assays is generally accepted (see [1,27] for a detailed discussion).

The precise location of TSAb or TBAb epitopes on the TSH-R is less certain. Early reports on the presence of TSAb epitopes at the N-terminal and TBAb epitopes at the C-terminal part of the extra cellular domain of TSH-R [14–16] are still strongly favoured [17,28], despite recent studies indicating a closer proximity of the TSAb and TBAb binding sites [11,18].

With the availability of coated tube TSH-R assays, it is now possible to assess binding of autoantibodies in a technically reproducible way with relatively low background. We tried to elucidate the binding of autoantibodies to different parts of the TSH-R by competition studies with labelled murine MoAbs to distinct epitopes on the TSH-R instead of bovine TSH. If successful, this would also allow to distinguish TSAb in a simplified in vitro assay without the need for cumbersome bioassays.

We used murine MoAbs to three parts of the TSH-R ectodomain: MoAb 28·1 is directed to amino acids 36–40 at the N-terminal part of the ectodomain, where the majority of TSAb are supposed to bind. MoAb A9 (amino acids 147–228) represents an epitope in the middle of the ectodomain (within the horseshoe structure of the leucine-rich repeats). Finally, MoAb 31·7, identical in properties to the reported MoAb 15 [9], is directed to amino acids 382–415 of the TSH-R. This region is in the direct vicinity of the serpentine, and comprises two tyrosine residues at 385 and 387, whose sulphation was shown to be crucial for TSH binding [9]. Furthermore, the C-terminal part of the ectodomain is considered an important area for the binding of TBAb.

All three MoAbs showed high affinity binding to the TSH-R coated on tubes with a K_d in the range of 8–12 nm. This specific binding was inhibited by a proportion of sera from patients with GD containing TBII and TSAb or patients with autoimmune hypothyroidism containing TBII and TBAb. No inhibition was seen by sera from patients with autoimmune thyroid disease, which were TBII negative, and negative for TSAb or TBAb in the bioassay, although the latter group had similar total IgG concentrations to the former groups.

Overall, the inhibition of MoAb binding by patients’ sera supports the concept that TSAb binding to the TSH-R is heterogeneous. However, we did not find a predominant binding of patient autoantibodies at the N-terminal part. While inhibition of the N-terminal MoAb 28·1 by TSAb was present in only 10·4% of GD sera tested, we found a clear interaction between TSAb positive GD sera and the middle part (MoAb A9, 72%) and also the C-terminal region (MoAb 31·7, 44·1%) of the TSH-R ectodomain. This was confirmed with a second N-terminal MoAb (IRI-SAb 1 [12]) which, although stimulating in the bioassay, interacted with less than 10% of GD sera on TSH-R tubes (not shown). This lack of GD sera to interact with a defined N-terminal epitope was surprising. On the other hand, TBAb containing AIT sera showed a clear displacement of all three tracers. Although strongest with the C-terminal MoAb 31·7, this displacement was also evident with the N-terminal MoAb 28·1 (53·8%).

Although we performed this study with total IgG, we do not attribute the observed differences to non-specific effects. Several arguments reason against this. First, for a few sera we tested a direct comparison between F(ab)₂ fragments and total IgG, and could see no difference. Secondly, a non-specific effect should be evident with all MoAb tracers and could not explain the lack of displacement seen with MoAb 28·1. Thirdly, although an IgG MoAb may be of considerable size, it may not be as ‘rigid’ as often presumed, but shows rather ‘flexible’ properties. Evidence for this can be seen in a variety of sandwich assays, which fitted two or three IgGs on small peptides of about 20–30 amino acids. This can also be seen with the TSH-R, where it is possible to fit up to four different MoAbs on the molecule without any loss of TSH binding (S. Costagliola, personal communication).

As seen by ROC plot analysis, the discrimination between control sera and TRAb positive patients sera was most prominent with MoAb A9, and the sensitivity in our selected cohort of GD patients was as high as 72%. The A9 epitope in the middle part of the TSH-R ectodomain may therefore represent an area of the receptor, which is important for the binding of TSAb, TBAb and TSH [11,18]. Also of interest, although we found no competition of the three MoAbs between themselves, a combination of all three had no additive effect in the system, as it neither improved the signal-to-noise ratio nor the number of positive GD sera.

The patients’ sera used in this study were selected initially for the presence of TBII, and therefore do not represent an unbiased cohort. A correlation between TBII data and MoAb tracer data showed the strongest tracer displacement with those sera, which had also very high TBII values. However, not all sera with high TBII values were positive in the MoAb assays. It is therefore not justified to explain the reactivity in the MoAb assays with a strong humoral response to the TSH-R, irrespective of whether or not the sera contain TSAb or TBAb. In particular, the lack of interaction with the MoAb 28·1 is puzzling, because in terms of assay technology this tracer worked most effectively.

Also, it should be noted that the interaction of TSAb (but also TBAb) with labelled MoAbs is clearly not as prominent as seen with labelled TSH. It should be taken into account that the latter assay is the product of an extensive optimization process. As we saw no additional clinical benefit compared to the TRAK human assay, this optimization was not performed for the MoAb tracers described here.

While this manuscript was under revision, we were able to affinity purify TRAb from the serum of patients with GD [29]. After indirect labelling, these purified TRAb could be used as tracer. Contrary to the MoAbs described here, the interaction of purified TRAb tracer with patients’ serum was practically identical to that of bTSH, and the sensitivity for GD was 99%.

We conclude that the binding of TSAb and TBAb to the TSH-R may not be restricted to distinct and distant epitopes. The far N-terminal part of the TSH-R seems to be less important for autoantibody binding than the middle part, and also a proportion of TSAb (and TBAb) antibodies show interaction with a C-terminal epitope of the TSH-R. The method described here is a TSH-independent competitive assay for the detection of TSH-R autoantibodies. However, it does not allow discrimination between TSAb and TBAb.