Autoantibodies to the TSH Receptor—from discovery to understanding the mechanisms of action and to new therapeutics

Abstract

Prior to 1956, Graves’ hyperthyroidism was thought to be due to high levels of TSH but in that year Adams & Purves demonstrated the presence of a thyroid stimulator in Graves’ sera with a prolonged time course of action (long-acting thyroid stimulator, LATS) quite distinct from TSH. LATS was only present in the serum IgG fraction suggesting it was a thyroid stimulating autoantibody. In 1974 Graves’ IgG was shown to compete with ¹²⁵I-labelled TSH for the TSH receptor providing good evidence that Graves’ hyperthyroidism was caused by TSH receptor autoantibodies. Further breakthroughs occurred in 1989 (TSHR cloning) and 2003 (monoclonal thyroid stimulating autoantibody M22^TM). Subsequently atomic level detail of how TSHR stimulating (2007) and blocking (2011) autoantibodies interact with the TSHR became available. Cryo-EM studies followed (2022–2025) and provide a detailed understanding of how TSHR autoantibodies with different properties function. The human monoclonal autoantibody K1-70^TM with powerful TSH receptor blocking activity is now in clinical trials. It has the expected beneficial effects on Graves’ hyperthyroidism and Graves’ ophthalmopathy and is an exciting new TSHR specific drug.

Prior to 1956, the cause of hyperthyroidism in Graves’ disease was suspected to be due to high levels of thyroid stimulating hormone (TSH). In that year however at the University of Otago, Adams and Purves demonstrated the presence of thyroid stimulating activity quite distinct from TSH in the sera of some patients with hyperthyroid Graves’ disease [1].

A defining feature of the thyroid stimulating activity in these Graves’ sera was a prolonged time course of action in the bioassay used. The peak response time of ¹³¹I release from the thyroid glands of assay guinea pigs being 16–24 hours in the case of Graves’ sera compared with 1.5–3 hours for pituitary TSH.

Subsequently the activity in Graves’ sera was referred to as long-acting thyroid stimulator or LATS. Soon after its discovery, the University of Otago group showed that LATS was present only in the IgG fraction of serum proteins [2] suggesting that LATS was a thyroid stimulating autoantibody.

Subsequently LATS was shown to mimic most, if not all, the effects of TSH [2] and it was proposed [2, 3] that LATS was an antibody to the same thyroid gland component which interacts with TSH i.e. an autoantibody to the TSH receptor. Direct evidence for this concept was published in 1974 [4] when IgG preparations from Graves’ sera were shown to inhibit ¹²⁵I-labelled TSH binding to thyroid membranes (see below).

In 1958, Porter [5] isolated the Fab and Fc fragments of IgG and then showed IgG to be composed of two Heavy (H) chains and two Light (L) chains linked by non-covalent bonds and disulphide bridges [6]. Porter’s studies were quickly applied to IgG with LATS activity [3] and in 1965 I joined Dorrington and Munro’s lab in Sheffield to continue this work.

An example of the resulting studies [7] is show in Fig. 1, illustrating that the isolated L chain of LATS IgG has no activity, isolated H chain a small amount of activity (likely due to contamination with intact IgG), Fab was active but less than expected due to its short half life in the assay mice. Also, when the separated H & L chains were recombined a significant amount of LATS activity was recovered in the reconstituted IgG emphasising that the LATS activity was an integral part of the IgG molecule.

Fig. 1. The thyroid stimulating properties of long-acting thyroid stimulating IgG subunits.

Although the structural features of LATS autoantibody were well established in the 1960s, the nature of LATS autoantigen was unclear although the TSH receptor appeared to be a likely candidate [2, 3].

Homogenates of Graves’ thyroid tissue absorbed or neutralised LATS activity and much to the surprise of this investigator, most of the neutralising activity was present in the homogenate supernatant after centrifugation at 105,000 g rather than in the sediment containing the thyroid cell membrane material [8]. Fractionation of the soluble fraction of the thyroid homogenates by gel filtration separated the material into 4 main fractions (Fig. 2) with LATS absorbing activity i.e. LATS autoantigen or LAA present only in fraction III. Serum albumin co-eluted in fraction III suggesting that the molecular weight of the LATS absorbing activity was similar to serum albumin i.e. a 4S protein molecular weight about 50,000.

Fig. 2. Chromatography of the soluble fraction of thyroid homogenate on Sephadex G-200 showing LATS autoantigen to be of a similar molecular weight to serum albumin.

The 4S LAA material was further purified, characterised and its interaction with LATS studied in detail [9]. Also, it was shown [9] to be distinct from thyroglobulin or thyroid microsomal antigen and it was clear to me at the time that the ability of purified preparations of LAA to neutralise LATS activity in mice had potential for specifically neutralising and inactivating LATS activity in patients. Such a potential treatment is considered in detail later in this review.

The relationship between LAA and the putative TSH receptor however was not clear in 1970, particularly as pituitary TSH (human or bovine) activity could not be absorbed by particulate or soluble preparations of thyroid homogenates in the same way LATS activity could [10].

More than 10 years later however when affinity labelling studies with ¹²⁵I-TSH were used to determine TSHR structure [11] it became clear that the A subunit of the TSH receptor and the 4S LAA were one and the same.

Although many key observations were made using the bioassays based on stimulation of ¹³¹I release from the thyroids of guinea pigs or mice, the assays were always technically demanding, imprecise and unsuitable for analysing large numbers of patient samples.

In 1974 however there was a major conceptual and technical breakthrough when methodologies based on binding of ¹²⁵I-labelled TSH to thyroid membrane preparations were described [4]. The materials needed for these receptor assays were difficult to prepare and serum immunoglobulin preparations rather than serum were needed. Consequently, the assays could only be carried out in specialised research laboratories.

Despite this limitation, extensive studies on TSH receptor antibody (TRAb) activity in patient serum immunoglobulin preparations measured by inhibition of ¹²⁵I-labelled TSH binding to thyroid membranes were carried out in the 1970s. TRAb activity was detectable in almost all untreated hyperthyroid Graves’ disease sera using this method. These studies were able to show a good relationship between TRAb levels and various parameters of thyroid function including early ¹³¹I uptake and TSH control of thyroid function [12, 13]. Also, the effects of treatment for Graves’ disease on TRAb levels were clearly evident with patients treated by surgery showing the lowest prevalence, patients treated by ¹³¹I a much higher prevalence and patients treated with antithyroid drugs an intermediate prevalence [12].

Although TRAb stimulate the TSHR in untreated Graves’ disease and when present in most patients with treated Graves’ disease, TRAb with blocking (i.e. TSHR antagonist) activity were described by Endo et al. in 1978 [14] in occasional hypothyroid patients. Subsequent studies showed that TRAb can transition from stimulating to blocking type activity (and vice versa) during the disease and treatment course although TRAb with blocking activity are uncommon and transition in the type of TRAb activity infrequent [11].

These studies carried out in the 1970s using the early receptor assays were repeated and extended when improved receptor assays and reliable bioassays became available for routine use 30 years later [15-19]. The rigorous early studies were confirmed well.

Studies in the 1970s with detergent solubilised TSH receptors provided the basis for improved receptor assays. In particular, ¹²⁵I-bovine TSH binding to detergent solubilised porcine TSHR was inhibited well by TRAb but the interaction showed virtually no non-specific interference from healthy blood donor IgG or untreated serum [20, 21].

Furthermore, the TSHR preparations could be stabilised by freeze-drying as could ¹²⁵I-labelled bovine TSH and in 1982 these two reagents were incorporated into a kit (Smith kit) for TRAb measurement [20, 21]. The procedure used 50 μL of neat serum and PEG to separate receptor bound and free ¹²⁵I-TSH. The kits were assembled in my laboratory in the University of Wales College of Medicine in Cardiff and the first kits shipped to Japan in 1982. These were well received and RSR Ltd was formed in the same year to organise Smith kit manufacture, sales and shipping. The kits were quickly adopted for TRAb measurement in many laboratories throughout the world providing a rapid, reproducible, sensitive, specific and inexpensive manual isotopic assay for TRAb.

The studies on detergent solubilised TSH receptor preparations which lead to development of the Smith kit for TRAb measurement also facilitated investigations of TSH receptor structure. In particular, affinity labelling using derivatives of bovine TSH labelled with ¹²⁵I [11]. These investigations showed that the TSH receptor in detergent solubilised thyroid membranes consists of two subunits linked by a disulphide bridge(s). One subunit (A, molecular weight 50,000) is water soluble and forms the binding site for TSH and TRAb on the outside surface of the cell membrane. The other (B) subunit (molecular weight 30,000) penetrates the lipid bilayer. The non-covalent bonds between the A and B subunits are readily dissociated at neutral pH and when thyroid membrane preparations are reduced the A subunit is released [11].

The isoelectric point of the TSH receptor A subunit is about 5 [9] and it will have a negative charge under physiological conditions. TRAb have isoelectric points in the region of 8–10 (irrespective of the autoantibodies being stimulating or blocking), TSH about 9 and they will have a positive charge under physiological conditions. Consequently, charge-charge interactions will be of major importance in the binding of TRAb or TSH to the TSH receptor A subunit.

The TSHR structure derived from the studies described above agrees well with subsequent studies commencing with cloning of the TSHR in 1989 [22-25], the crystal structures of the TSH LRD in complex with TSHR autoantibodies in 2007 and 2011 [26, 27] and ligand free TSHR LRD in 2019 [28]. Cryo-EM structures of the TSHR in complex with monoclonal TSH receptor autoantibodies [29] are also in agreement with the structure determined prior to TSHR cloning.

Cloning of the human TSHR showed that it was a 764 amino acid protein with a large (414 amino acids) extracellular domain (ECD), a slightly smaller (369 amino acids) transmembrane domain (TMD) which crossed the cell membrane 7 times and a short C-terminal arm (81 amino acids).

Once the amino acid sequence of the TSHR was known, a series of peptides were prepared [30] and used to analyse a panel of monoclonal antibodies to different parts of the receptor [31, 32]. These included MAb 4E31 which binds to the C-terminal end of the TSHR, most distant from the TRAb and TSH binding site [33, 34].

The binding location and high affinity for the TSHR characteristic of 4E31 enabled plastic surfaces coated with 4E31 to immobilise detergent solubilised TSHR in such a way that the captured receptor could still bind TSH or TRAb. This permitted the development of a new (second) generation of assays for TRAb based on inhibition of ¹²⁵I-labelled TSH binding to TSHR coated plastic tubes [33, 35] or inhibition of TSH-biotin binding to TSHR coated ELISA plate wells [34]. The TRAb ELISA based on detection of TSH-biotin bound to the TSHR with streptavidin-peroxidase [34] was a particular advance providing the first non-isotopic receptor assay suitable for routine TRAb measurement.

The second generation of TRAb assays based on inhibition of labelled TSH binding to plastic tubes or ELISA plates were much less cumbersome than the earlier method based on separation of bound and free ¹²⁵I-TSH by PEG precipitation and centrifugation.

This had advantages for routine measurement of TRAb as well as in attempts to produce monoclonal antibodies to the TSH binding site and TRAb binding site on the TSHR. The aim of these studies was to produce monoclonal antibodies with the characteristics of patient serum TSHR autoantibodies. In particular the MAbs needed to inhibit labelled TSH binding to the TSHR and (a) to stimulate cyclic AMP production in TSHR-expressing cells or (b) block the activities of TSH and the stimulating activities of patient sera TSHR autoantibodies.

Production of such mouse thyroid stimulating MAbs was challenging but in 2002 we produced the first three mouse TSHR MAbs with strong thyroid stimulating activity using TSHR DNA immunisation [36]. Also a hamster MAb with strong thyroid stimulating activity was described [37]. This success was a culmination of research by many different laboratories and represented an important milestone in the thyroid field [38].

Subsequently we produced a further four thyroid stimulating mouse MAbs [39] and other laboratories were able to confirm our original findings [40, 41]. Also we were able to produce a mouse MAb to the TSHR (RSR-B2) which acted as a power antagonist of the TSHR stimulating activities of TSH and of patient serum autoantibodies [42].

At the time [36, 42], it was proposed that TSHR mouse MAbs with the characteristics of patient serum autoantibodies ie inhibiting TSH binding to the TSHR at low concentrations and potent agonists (or occasionally antagonists) of the TSHR, had considerable potential as reagents for new assays for TRAb, alternatives to recombinant TSH and probes for TSHR structure and function. In the case of the mouse MAb with powerful TSHR antagonist activity (RSR-B2) in vivo applications when tissues containing the TSHR need to be made unresponsive to stimulating autoantibodies or to TSH were proposed including treatment of TSHR mediated eye signs of Graves’ disease [42].

The success in producing potent mouse MAbs to the TSHR in 2002 was however eclipsed by a further breakthrough shortly afterwards. In particular, peripheral blood lymphocytes isolated from a 19 year old Welsh male with Graves’ disease (and high serum TRAb levels) were immortalised using EBV, screened for TRAb activity, a positive well fused with a heterohybridoma cell line and the cells recloned four times to produce a hybridoma secreting M22^TM [43].

M22^TM is an IgG1 λ human monoclonal autoantibody to the TSHR with thyroid stimulating activity 3,000× more potent than the donor serum IgG. Also, M22^TM is 10–100 times more active than the mouse MAbs with thyroid stimulating activity described above. High concentrations of M22^TM have no effect on CHO cells expressing the FSH receptor or LH receptor emphasising the high degree of specificity of M22^TM for the TSHR [44]. The potential applications of M22^TM were similar to those of the thyroid stimulating mouse MAbs described above but more so.

One of the original challenges of the 1960s to produce a pure LATS preparation had been achieved. A crystal structure of M22^TM Fab at 1.65 Å resolution was quickly obtained [45] providing atomic level detail.

M22^TM also provided the way forward to understanding the nature of the LATS antigen (i.e. the TSHR) at the atomic level. In particular, early studies [10] showed that LATS-LATS antigen complexes were heat stable in solution whereas LATS antigen (i.e. TSH receptor A subunit) itself was unstable.

As soon as high affinity TSHR MAbs, particularly M22^TM became available we could proceed to prepare stable M22^TM TSH receptor A subunit complexes, purify them to homogeneity, prepare crystals and determine the crystal structure of the complex. A 2.55 Å resolution structure was obtained [26]. This showed that the TSH LRD (amino acids 22–260) comprises a curved helical tube and M22^TM Fab clasps its concave surface at 90° to the tube length axis (Fig. 3). A large interface (2,500 Å) is buried in the complex and there an extensive network of ionic, polar and hydrophobic bonds. Atomic level understanding of how a TSHR autoantibody interacts with the TSH LRD had been achieved.

Fig. 3. Crystal structure of the TSHR LRD in complex with M22 Fab at 2.55 Å resolution.

(A) shows M22 heavy and light chains interacting with the entire concave surface of the LRD (aa 22–260).

(B) Is the structure in (A) rotated by 90°. It shows the curved helical tube structure of the LRD with M22^TM clasping the concave surface at 90° to the tube length axis.

Reproduced with permission of the copyright holder, RSR Ltd.

The high affinity and specificity of M22^TM for the TSHR enabled visualisation of the TSHR in human thyroid tissue sections by immunohistochemistry based on M22^TM-biotin. These studies showed positive staining of the entire external surface of thyroid follicular epithelial cells (Fig. 4) consistent with the location of the TSHR on the basal membrane [46].

Fig. 4. Immunohistochemistry evaluation of TSHR MAb M22. Human thyroid ×100 stained with biotinylated M22 (2.5 μg/mL) showing positive staining of the entire external surface of thyroid follicular epithelial cells consistent with the location of the TSHR on the basal membrane. Reproduced with permission of the copyright holder, RSR Ltd.

The binding characteristics of M22^TM IgG and Fab to the TSHR including TSHR preparations immobilised on plastic surfaces or beads via mouse MAb 4E31 were particularly important in the next phase of TSHR autoantibody assay development.

Firstly, M22^TM Fab labelled with biotin could be used instead of TSH-biotin in a new ELISA for TRAb which showed high levels of sensitivity and specificity and improved robustness [15].

At the same time, we were working with Roche scientists on development of a rapid fully automated electrochemiluminescence assay for TRAb based on ruthenium labelled M22^TM, 4E31-biotin, detergent solubilised TSHR and streptavidin coated beads [47]. Once bound to the TSHR, M22^TM does not dissociate. This is a critical feature essential for the Elecsys system (and other fully automated systems) where vigorous wash steps by special solutions are required.

After optimisation, the Roche Elecsys TRAb was evaluated in two extensive studies [16, 17] in which the assay achieved a performance level the same or better than other assays commercially available at that time. Elecsys TRAb takes approximately 30 minutes to obtain a reliable, reproducible result and does not require special skills It was adopted for routine use by many clinical laboratories worldwide and quickly became the method of choice for TRAb detection, a position it holds today more than 15 years after introduction.

We have been able to use M22^TM labelled with ¹²⁵I, as a ligand in competition with individual patient TRAb for the TSHR, to estimate the TRAb affinities and concentrations in the sera [48]. Affinities for the TSHR were similar in 15 Graves’ sera studied ranging from 3–7 × 10¹⁰ L/mol and TRAb concentrations ranged from 50–500 ng/mL of serum. Blocking antibody sera (n = 4) had similar affinities (3–7 × 10¹⁰ L/mol) to the 15 sera with stimulating activity but were present at much higher concentrations (1.7–27 μg/mL).

Further to our success in producing M22^TM we were able to use the same procedures to prepare a human monoclonal autoantibody with powerful TSH receptor antagonist activity (5C9) [49]. The lymphocyte donor was a 27 year old Italian female diagnosed with severe hypothyroidism, high levels of TRAb (260 U/L) with TSHR blocking activity. The 5C9 produced has a high affinity for the TSHR (4 × 10¹⁰ L/mol) and its binding is inhibited by patient serum TSHR autoantibodies (with stimulating or blocking activity). It does not stimulate cyclic AMP production in CHO cells expressing the TSHR and lowers the constitutive activity of the cells.

The antagonist properties of 5C9 in relation to thyroid stimulating autoantibodies and TSH have potential for in vivo applications similar to those we have proposed for the blocking mouse MAb RSR-B2 (see above). Further, 5C9’s effect on TSHR constitutive activity could be useful in suppressing TSHR activity in thyroid cancer tissues and/or metastases.

Our studies on 5C9 however were diverted by developments with monoclonal TSHR autoantibodies produced from the lymphocytes of a 54 year old Japanese female patient with hypothyroidism and high TRAb levels (160 U/L) [50]. We obtained two different TSHR monoclonal autoantibodies from the same lymphocyte preparation. One TSHR MAb (K1-18^TM, IgG1 κ) was a strong stimulator of cyclic AMP production in TSHR-transfected CHO cells. The other (K1-70^TM, IgG1 λ) blocked stimulation of the TSHR by TSH, M22^TM, K1-18^TM and patient sera with thyroid stimulating activity [50]. V region gene analysis indicated that K1-18^TM and K1-70^TM used the same V region germline gene but different D and J germline genes as well as having different light chains. Consequently the two autoantibodies have evolved separately from different B cell clones. This proves that a patient can produce a mixture of blocking and stimulating TSHR autoantibodies at the same time. Further studies on the activities of K1-18^TM, K1-70^TM and M22^TM have been reported by Tagami and colleagues [51, 52].

The high affinity of K1-70^TM Fab for the TSHR (2 × 10¹⁰ L/mol) enabled us to determine the crystal structure of K1-70^TM in complex with the TSHR LRD (as we had done for M22^TM Fab) at 1.9 Å resolution [27]. The structure showed that K1-70^TM Fab binds to the concave surface of the LRD forming a large interface (2,565 Å) with an extensive network of ionic, polar and hydrophobic interactions. The binding arrangements in the TSHR LRD—K1-70^TM Fab complex are similar to those observed in the TSHR LRD—M22^TM Fab complex, however K1-70^TM clasps the concave surface of the LRD in the opposite orientation (rotated 155°) to M22^TM Fab. Also, K1-70^TM Fab binds more N-terminally than M22^TM Fab.

More details of the interactions between K1-70^TM and full length TSHR determined by cryo-EM are described later in this review as well as cryo-EM studies of M22^TM—TSHR interactions and 5C9^TM–TSHR interactions.

The studies described above suggested that K1-70^TM was a suitable monoclonal antibody for blocking the effects of TSHR stimulating autoantibodies in vivo and appropriate pre-clinical studies were carried out [46, 53]. Also, recombinant K1-70^TM IgG was produced in CHO cells following good manufacturing practice and formulated at 10 mg/mL accordingly with a view to proceeding with a Phase 1 clinical trial [54]. Also, we were able to supply K1-70^TM formulated for the clinical trial to the Mayo Clinic for compassionate use in a patient with advanced follicular thyroid cancer and Graves’ disease with high levels of stimulating TSHR autoantibodies (TSAb) and severe ophthalmopathy [55].

The patient received her first dose of K1-70^TM (18 mg im) in April 2017. K1-70^TM was then administered at 3 weekly intervals with the dose increased to 40 mg im and then to 120 mg im and this blocked her high TSAb activity. Thereafter, doses of K1-70^TM were adjusted according to measurements of serum TSAb activity with the dose reduced to 60 mg im when serum TSAb became undetectable and increased to 120 mg im when TSAb started to increase.

Soon after the first dose of K1-70^TM, the patient reported improvement in her eye symptoms. After 4 months of K1-70^TM her CAS improved from 6/7 to 0–1/7 and she was able to have eye surgery to correct her diplopia [55]. GO symptoms remained under control while on K1-70^TM therapy and when K1-70^TM was withdrawn before ¹³¹I therapy the patient’s eye disease deteriorated. This resolved quickly after prednisone support and resumption of K1-70^TM. Therapy with K1-70^TM also appeared to attenuate at least some foci of FTC growth [55].

Overall, these observations in a single patient suggested that blocking TSHR stimulation with K1-70^TM can be an effective treatment for GO and may also benefit select patients with FTC and Graves’ disease.

The UK phase 1 clinical trial of K1-70^TM is now complete [54]. This showed IV administration to be the preferred route. A single dose of 50 mg or 150 mg caused thyroid hormone and TSH levels to progress to hypothyroid ranges. Also, there were improvements in the symptoms of Graves’ disease (reduced tremor, improved sleep, improved mental focus, reduced toilet urgency) and GO (reduced exophthalmos measurements, reduced photosensitivity).

Maximum serum concentrations of about 15 μg/mL and about 45 μg/mL respectively of K1-70^TM were observed after about 2 hours of administration of 50 mg or 150 mg (Fig. 5A). Thereafter the serum concentrations declined with a half life of about 3 weeks.

Fig. 5. (A) K1-70^TM serum concentration time course in Graves’ patients after a single IV injection. Reproduced with permission of the copyright holder, RSR Ltd. (B) Changes in serum TSAb activity in 3 Graves’ patients following a single 75 mg dose of K1-70^TM IV. Reproduced with permission of the copyright holder, RSR Ltd.

The values for serum K1-70^TM concentrations can be compared with the 0.05–0.5 μg/mL of TRAb seen in the serum of patients with Graves’ disease [48]. Consequently, administration of 50 mg or 150 mg K1-70^TM IV will result in a rapid and prolonged excess of K1-70^TM over the patient’s endogenous thyroid stimulating autoantibodies, blocking their stimulating activity (Fig. 5B).

The UK phase 1 trial showed K1-70^TM to be safe, well tolerated and produced the expected pharmacodynamic effects with no immunogenic responses. A Japanese phase 1 trial is also complete [56] and the results are in good agreement with the UK phase 1 trial. In particular a consistent pharmacokinetic/pharmacodynamic relationship, no immunogenic responses and promising safety profile. A phase 2 trial is now in progress.

Further to determining the crystal structures of the TSHR LRD in complex with M22^TM [26] and in complex with K1-70^TM [27], we have been able to use cryo-EM to determine the structure of full length TSHR in complex with K1-70^TM [29].

In these experiments the full length human TSHR complexed with K1-70^TM Fab was detergent solubilised, purified to homogeneity and analysed by cryo-EM [29].

The structure (global resolution 3.3 Å) is a monomer with all three domains (LRD, HR and TMD) visible (Fig. 6). The ECD composed of the LRD and HR is positioned on top of the TMD extracellular surface. Extensive interactions between the TMD and ECD are observed in the structure and their analysis provides an explanation of the effects of various TSHR mutations on TSHR constitutive activity and ligand induced activation.

Fig. 6. Cryo-EM structure of the TSHR in 2 views rotated 180°. The structure is a monomer; all three domains are visible: leucine-rich repeat domain (LRD), hinge region (HR) and transmembrane domain (TMD). The TSHR extracellular domain (ECD, composed of the LRD + HR) is positioned on top of the TMD extracellular surface. Extensive interactions between the TMD and ECD are observed in the structure and their analysis provides an explanation of the effects of various TSHR mutations on TSHR constitutive activity and on ligand induced activation. Reproduced with permission of the copyright holder, RSR Ltd.

The TSH receptor transitions between inactive and active states as shown in Fig. 7A. In its inactive state the TSHR LRD is positioned closer to the cell membrane. To achieve its active state, the receptor LRD rotates about the hinge region (HR) in a motion that moves the LRD away from the cell membrane.

Fig. 7A. Cryo-EM structure of full length TSHR bound to K1-70^TM (blocking autoantibody). Reproduced with permission of the copyright holder, RSR Ltd.

K1-70^TM only interacts with the N terminal part of the LRD and as such it can bind to active and inactive states of the receptor without clashing with the membrane. Its powerful blocking action on TSHR stimulation by TSH, M22^TM or patient serum TRAb is due to it preventing binding of the stimulators to the LRD consistent with K1-70^TM not influencing TSHR constitutive activity [29].

We have also obtained cryo-EM structures of the blocking type human monoclonal autoantibody 5C9 [49] in complex with the full length TSHR and the stimulating autoantibody M22^TM in complex with full length TSHR. 5C9^TM binds from the 3rd leucine rich repeat in the LRD to the hinge region (Fig. 7B) and in so doing it blocks binding of TSH, M22^TM and patient TRAbs to the LRD. When bound to the inactive TSHR, the 5C9^TM heavy chain CDR3 interacts with the receptor’s hinge loop and locks the receptor in the inactive conformation. This results in inhibition of TSHR constitutive activity and the activity of TSHR activating mutations as observed experimentally [49].

Fig. 7B. Cryo-EM structure of full length TSHR bound to 5C9^TM (blocking autoantibody). Reproduced with permission of the copyright holder, RSR Ltd.

M22^TM has important interactions with the C terminal part of the LRD. Consequently it cannot interact with the receptor in its inactive state as this would result in a clash with the membrane (Fig. 7C).

Fig. 7C. Cryo-EM structure of full length TSHR bound to M22^TM (stimulating autoantibody). Reproduced with permission of the copyright holder, RSR Ltd.

M22^TM can however bind to the active state of the receptor as in this conformation, the LRD is further away from the membrane and M22^TM does not clash with the membrane when bound. Once bound to the active state structure, M22^TM holds the receptor in its active state i.e. it cannot return to the inactive state with M22^TM bound. This results in the observed prolonged activation.

These cryo-EM studies provide important insights into how different TSH receptor autoantibodies interact with the TSHR and cause their different effects.

The ability of M22^TM and K-70^TM to form stable complexes with the TSHR LRD (aa 22–260; TSHR260) facilitated purification of the complexes and determination of their crystal structures (see above). However, free TSHR LRD (i.e. not bound to autoantibody) has poor stability and until recently purification attempts had not succeeded beyond those with LATS autoantigen (TSHR A subunit) in 1970 [9].

Purification and crystallisation of some other unstable GPCRs such as the β2-adrenergic receptor has been achieved after systematically mutating every residue and identifying which mutations improved thermostability. Combined mutations improved stability further [57].

We used a similar approach to stabilise the free TSHR LRD (TSHR260) and introduction of six mutations resulted in a preparation approximately 900 times more thermostable than wild type TSHR260 [28]. The six mutations did not affect the binding of human TSHR monoclonal autoantibodies or patient serum TSHR autoantibodies. Also when the six mutations in TSHR260 were incorporated into full length TSHR stimulation by TSH or human monoclonal autoantibodies M22^TM and K1-18^TM was not affected.

Thermostable TSHR260 (TSHR260 STABL^TM) was purified to homogeneity without ligand, crystallised and the structure solved at 2.83 Å resolution. The unbound TSHR260 STABL^TM structure and the structures of wild type TSHR260 bound to M22^TM and to K1-70^TM are remarkably similar. This suggests that neither the mutations nor the binding of M22^TM or K1-70^TM change the rigid leucine-rich repeat domain structure of TSHR260 [28].

Now a pure stable preparation of the TSHR LRD is available which interacts well with TSH receptor autoantibodies. TSHR260 STABL^TM can be coated onto plastic surfaces and used to absorb out patient serum TRAb. This provides a useful specificity test for TSHR autoantibody assays and proof of principle for the concept of using TSHR260 STABL^TM linked to a solid phase for specific absorption of TRAb circulating in patients by plasmapheresis.

Also, TSHR260 STABL^TM has potential for use in targeted in vivo therapies aimed at specifically reducing circulating TRAb. For example TSHR260 STABL^TM linked to a molecule which will direct the complex formed with TRAb to a particular degradation pathway [58, 59].

In addition TSHR260 STABL^TM has possibilities in applications with CAR-T cell therapies which target TRAb production [60, 61].

In conclusion, there has been considerable progress in understanding the characteristics of thyroid stimulating autoantibodies since their discovery in 1956. Pure highly active preparations of LATS (TRAb) and LATS autoantigen (TSHR) have been obtained and their structures and how they interact with each other have been determined at the atomic level. Fully automated assays for TRAb are now used routinely worldwide in the diagnosis and management of Graves’ disease. Pure stable TSHR260 (LATS autoantigen) is a promising reagent for use in specific absorption of patient TRAb, targeting TRAb to particular degradation pathways and in CAR-T cell therapies. The TSH receptor blocking monoclonal autoantibody K1-70^TM is an exciting new TSHR specific drug which should soon be available for clinical use.

Disclosure Statement

I am a director and shareholder of the RSR group of companies which develop and manufacture diagnostics and therapeutics for autoimmune diseases.

Funding Statement

This study did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors.