Holger Jaeschke¹, Maren Claus¹, Gunnar Kleinau², Gerd Krause² , Ralf Paschke¹

¹ III. Medical Department, University of Leipzig, Ph.-Rosenthal-Str. 27, D-04103 Leipzig, Germany
² Forschungsinstitut für molekulare Pharmakologie, Robert-Rössle-Str. 10, D-13125 Berlin, Germany

Correspondence:
Ralf Paschke, M.D.
III. Medical Department, University of Leipzig
Ph.-Rosenthal-Str. 27
04103 Leipzig
Phone: 49-341-9713200
Fax: 49-341-9713209
Email: pasr@medizin.uni-leipzig.de

In a previous review (1) on the structure and mechanism of activation of the thyroid stimulating hormone receptor (TSHR), published in 2001, we summarized the evidence in relation to structure and function of the TSHR. Since 2001 several new and important insights in TSHR structure and signal generation have been reported. In this review we will summarize these recent key findings concerning the TSHR research of the last years.
The thyroid stimulating hormone (TSH) as a member of the glycoprotein hormone family is secreted by the pituitary gland and controls thyroid function and proliferation via interaction with the membrane associated TSHR. Binding of TSH results in activation of the G protein-coupled TSHR and in stimulation of second messenger pathways, like cyclic adenosine monophosphate (cAMP), inositolphosphates (IPs) and diacylglycerol (DAG). Intracellular accumulation of these second messengers leads to modification of different physiological properties of thyroid cells.
The TSHR belongs together with the choriogonadotropin/luteinizing hormone receptor (CG/LHR), follicle stimulating hormone receptor (FSHR) and leucin rich repeat containing glycoprotein receptors (LGRs) to the subfamily of glycoprotein hormone receptors (GPHR) (2-4). The TSHR, LHR and FSHR are characterized by a large N-terminal extracellular domain (ECD), a seven transmembrane spanning helix motif (TM1-7) connected by three extra- and three intracellular loops (ECL1-3; ICL1-3) and a C-terminal intracellular domain.
Despite functional characterization of numerous in vivo TSHR mutants less is known about the precise mechanisms of recpetor activation. The understanding is hindered by the lack of information about the three-dimensional structure of the TSHR. One reason is the small number of functionally characterized amino acid exchanges for the large extracellular domain and the extracellular loops of the TSHR. The major part of naturally occurring mutations is located in the TMs (5, www.uni-leipzig.de/~innere/tsh/). The occurrence of in vivo activating mutations revealed first insights into intra- and intermolecular structure-function relationships of the TSHR and led to new approaches for the elucidation of the mechanism of TSHR activation by site-directed mutagenesis and molecular modelling.

Extracellular Domain and Extracellular Loops
The large N-terminal extracellular domain consists of five molecular sections: (i) an N-terminal Cysteine-box-1 (C-b1) (hTSHR: 1-54), (ii) a 230 aa spanning leucin rich repeat motif (LRR) (hTSHR: 55-279), (iii) the central Cysteine-box-2 (C-b2) (hTSHR: 280-314), (iv) a TSHR specific 50 aa insertion (hTSHR: 317-366) and (v) the C-terminal Cysteine-box-3 (C-b3) (hTSHR: 370-410). (6, Figure).



LEGEND: TSHR: cartoon representing a molecular model (6) for the very tight packing of the ectodomain (LRR, Cysteine-box-2,Cysteine-box-3) being located very close to or even in between the extracellular loops of the serpentine domain. A new LRR template based on the Nogo receptor with much higher sequence similarity to the TSHR was introduced, whose ‘scythe blade’ shape also allows an interaction of the hormone parallel to the LRR structure. Furthermore, a new template for Cysteine-boxes -2 and –3 was identified based on a complex structure of the chemokine IL8 and a portion of the N-terminal tail of the IL8 receptor. These findings also support the hypothesis of a disulfide bridge between Cys398/Cys408 (C-b3) either to Cys283/Cys284 (C-b2) or in a reverse manner with Cys408/Cys398. (6, 20). Furthermore, the hydrophilic amino acids Asp403, Glu404 and Asn406 of C-b3, spatially located in proximity to Ser281, are likely to be involved in intramolecular signal transduction from the ECD towards the serpentine domain.

The largest structural characteristic of the extracellular domain is the leucine rich repeat (LRR) motif (2-4). The LRRs are not only a direct interaction partner for the ligand, but also have an essential role for receptor function. Mutagenesis of the N-terminal part of the TSHR and CG/LHR has shown that activity of the TSHR can be directly influenced by changes in the extracellular domain (7, 8). Based on the crystal structure of LRR containing proteins only the structural properties of the LRRs in the middle of the ECD sequence of the GPHRs were determined (9, 10) To evaluate an optimal structure template for the LRR Kleinau et al. (6) determined the LRR motif based on the Nogo receptor (11) as the best matching motif for the LRR for all three human GPHRs. To investigate the influence of the ECD for TSHR activation in more detail, Zhang et al. (12) deleted the ECD. This resulted in a strong constitutive activation of the receptor. A similar experiment described the loss of constitutive activity of the naturally occurring mutations Ile486Phe (ECL1) and Ile568Thr (ECL2) after deletion of the ECD, which implies that the ECD likely interacts with regions of the ECLs as a linked inverse agonist to maintain the inactive receptor conformation (13). In addition, these findings suggest that the structure of the ECD alternates from a tethered inverse agonist to an agonist in the process of receptor activation. Another experimental approach for the CG/LHR suggests that a structure within the ECD serves as an agonist for the ECLs as part of the transmembrane domain after agonist binding or mutational alterations (14). These findings underline the importance of the ECD and the ECLs in the process of TSHR activation and support a model in which the ectodomain acts as a silencer of the serpentine domain of the receptor. The current research is mainly focused on the identification of residues or epitopes in the ECD and ECLs, respectively, which are involved in the maintenance of the inactive conformation of the TSHR. Ser281 is the only position in the ECD, which is affected by constitutively activating in vivo mutations (15-18), and it was intensively characterized together with the surrounding residues (C-b2). The highly conserved Ser281 is situated in the hinge region between the LRR motif of the ECD and the TMs like the homologous positions Ser277 in the CG/LHR and S273 in the FSHR. Mutagenesis of amino acids Pro280, Cys283 and Cys284 in the vicinity of Ser281 also led to constitutive activation of the TSHR (19). In addition, recent studies on the corresponding position Ser277 in the CG/LHR (TSHR: Ser281) by substitution of all other 19 amino acids suggest that Ser277 is an integral part of a loop like epitope that may be involved in stability and signal generation of the CG/LHR (14, 20). In fact, 15 substitutions at position Ser277 in the CG/LHR led to constitutive activation of the receptor. Taken together, the C-b2 epitope 279YPSHCC284 can probably act as an intramolecular switch for receptor activation.
In another approach a systematic search with fragments of the ECD in the protein structure database (PDB) was performed (6). Based on sequence similarities two new structural templates were identified. First, a new LRR template based on the Nogo receptor with much higher sequence similarity to the TSHR was introduced, whose ‘scythe blade’ shape allows also an interaction of the hormone parallel to the LRR structure. Second, a new template for C-b2 and C-b3 was identified. This template showed homologous properties to the structure of the chemokine IL8 and to a portion of the N-terminal tail of the IL8 receptor. In this way sequence similarities of C-b2 to the chemokine IL8 and C-b3 to the IL8 receptor, respectively, were determined. The very tight packing of LRR, C-b2 and C-b3 with the extracellular loops also supports the hypothesis of disulfide bridges between Cys398/Cys408 (C-b3) or between one or both of these cysteins with Cys283/Cys284 (C-b2) and/or Cys408/Cys398. (6, 21). Furthermore, the hydrophilic amino acids Asp403, Glu404 and Asn406 of C-b3 were likely to be involved in intramolecular signal transduction from the ECD towards the serpentine domain (6). Substitution of these residues by amino acids with an opposite charge and the smallest nonpolar amino acid alanine, respectively, leads to constitutive activation of the TSHR in three of five mutants (Asp403Ala, Glu404Lys, Asn406Ala). Residues Asp403 and Asn406 are highly conserved in GPHRs and alanine substitutions at these positions lead to constitutive activation of the receptor. Interestingly, Glu404, a specific amino acid for the TSHR, showed no differences in basal cAMP accumulation when mutated to alanine, whereas Glu404Lys revealed a strong basal cAMP accumulation. The authors suggest a spatial proximity of the epitope Asp403-Asn406 (C-b3) to the Ser281 epitope (C-b2) based on combined data from homology models and functional data. It is important in this context that these two epitopes in the ECD are the only ones that have been reported to lead to constitutive activation of the TSHR by in vivo mutations (Ser281Thr/Ile/Asn) or mutagenesis (Asp403-Asn406). Based on sequence similarities between the three members of the GPHRs it has been suggested that mutations of the CG/LHR and FSHR at positions corresponding to Asp403-Asn406 in the TSHR also cause constitutive activity (Figure).

To provide a hypothesis for interactions between the ECD and ECLs (12, 13) more knowledge about the structural and functional properties of these regions is necessary. In contrast to the CG/LHR and FSHR, only very few functionally characterized in vivo and in vitro mutations in the extracellular loops of the TSHR are available. Most of them are in vivo mutations: Thr477Ile (ECL1), Ile486Phe, Met (ECL1), Ile568Thr (ECL2), Asn650Thr (ECL3), Val656Phe (ECL3) and del658-661 (ECL3) (5). Only position Asp474 in the TSHR has been intensively characterized by mutagenesis (22). The ECLs seem to be important for both, the interactions with the ECD and for signal transduction towards the serpentine domain. Agretti et al. (23) generated a receptor harboring the inactivating mutation Thr477Ile in the ECL1 and the activating mutation Pro639Ser in the 6th transmembrane segment (5, 24), resulting in a dominant effect of the activating Pro639Ser mutation. Interestingly, Thr477Ile was characterized by an impaired cell surface expression and Pro639Ser showed a cell surface expression comparable to the wt TSHR. However, the double mutant Thr477Ile/Pro639Ser was expressed on the cell surface like the single mutant Pro639Ser. This suggested that not only signal pathways but also structural properties between the ECL1 and TM6 were affected, which are necessary for correct folding and trafficking to the cell surface. In a second approach these authors combined two constitutively activating mutations (Ile486Phe in the ECL1 and Pro639Ser in the ECL2), which led to an increased basal cAMP accumulation. This observation was also reported by Kosugi et al. (25) and Angelova et al. (26) for the CG/LHR. Much more is known about the ECLs of the CG/LHR and FSHR, especially for the ECL3. An alanine scan of the ECL3 including residues Lys580-Lys590 of the FSHR (TSHR: An650-Lys660) identified several residues which are crucial for activation of the Gs and Gq signal pathway and for hormone binding (27, 28). Based on the highly conserved amino acid sequence of the ECL3 within the GPHRs it is likely that the ECL3 also plays an important role in signal generation and hormone binding for the CG/LHR and TSHR. Only for position Lys660 (unpublished data) in the TSHR and Lys583 in the CG/LHR (corresponding to Lys590 in the FSHR) functional data are available (29). Comparing the results regarding cell surface expression, cAMP and IP synthesis and hormone binding similar effects of this position could be determined within these GPHRs. In addition to the ECL3 of the FSHR, Ryu et al. (30) characterized the ECL2 of the CG/LHR by an alanine scan. Also for the ECL2 several residues, which are involved in signal generation and hormone binding, were identified. Taken together, these data can be used for refining the three dimensional models of the receptors to determine residues which are interaction points between the ECD and the transmembrane domain.

Transmembrane Domain
The transmembrane domains of GPHR consist of seven helices connected by three extracellular and three intracellular loops. Based on sequence alignments of the G protein-coupled receptors a homology of more than 70% could be determined (2-4). New insights into structural relationships of the TMs were possible with the description of the x-ray crystal structure of the bovine rhodopsin with a high resolution (31). Due to the high homology of GPCRs within their TMs this structure model is used as a template for many members of the large superfamily of G protein-coupled receptors. In addition, this model allowed many groups to provide new important insights into the architecture and side chain orientations of the TMs.
The TMs of the TSHR are characterized by a large number of constitutively activating in vivo mutations. Most of them are located in TM6 between residues 629 and 639 (5). Furthermore, naturally occurring mutations were identified in TM2, TM3, TM5 and TM7. The high number of in vivo mutations underlines the importance of the TMs with regard to stabilization of the inactive receptor conformation. In particular, interactions between TM5-TM6 and TM6-TM7 seem to be involved in maintaining the native receptor conformation. Neumann et al. (32) have identified a hydrogen bond between Asp633 (TM6) and Asn674 (TM7) by a combined approach of mutagenesis and modeling guided by naturally occurring mutations. This finding was supported by Kosugi et al. (33) and Lin et al. (34) which have characterized the homologous position Asp578 in LHR (corresponding to Asp633 in TSHR) by mutagenesis. A breakdown of this interaction between TM6 and TM7 leads to a change in receptor conformation and finally to a constitutive activation of the receptor. A recent report demonstrated that residue Met389 in the LHR (corresponding to Met453 in TSHR) is essential to maintain the inactive receptor conformation by interactions with the highly conserved residues Ile460 in TM3 (Ile515 in TSHR), Met571 in TM6 (Met626 in TSHR) and Tyr623 in TM7 (Tyr678 in TSHR) (35). Due to the fact that mutation of the homologous residue Met453 in the TSHR also causes constitutive activity, it is likely that a similar network of interactions exists in the TSHR (36-39).

Intracellular Loops
A common feature of all GPCRs is their interaction with G-proteins and subsequently the activation of downstream signaling cascades. Although many studies have focused on the molecular mechanisms of G-protein activation, conserved structure motives participating in the processes of G-protein recognition and selectivity have not been identified yet (40, 41). However, studies on several GPCRs revealed that residues, which are involved in G-protein coupling and selectivity, are primarily localized at the transmembrane/cytoplasmic borders between TM3/ICL2, TM5/ICL3 and ICL3/TM6 (42). Similarly, ICLs 2 and 3 of the TSHR were found to be important for recognition and selective activation of G_αs and G_αq (1, 43). Initially, studies on chimeric TSHRs containing homologous sequences of the α1- and β2 adrenergic receptors revealed the impact of the ICL2 for cAMP- and IP signaling (44). Kosugi et al. demonstrated the importance of the middle part of the ICL2 for G_s activation and identified amino acids 528-532 as important for G_q mediated IP production. Mutagenesis studies of the LH and FSH receptor also provided evidence for an important functional role of several amino acid residues within ICL2 (45, 46). However, only single selected residues in ICL2 were substituted in these studies. To further investigate the influence of this domain on TSHR signaling, deletion studies and alanine-scanning mutagenesis were carried out (43). Deletions of four to five residues and their corresponding multiple alanine substitutions were introduced into ICL2. Residues I523-D530, comprising mainly the N-terminal half of ICL2, appeared to be critical for G_s- and G_q-mediated signaling. A single alanine substitution screening within ICL2 revealed hydrophobic residue M527 in particular and to lesser extents F525, R528, L529 and D530 as residues that selectively abolished or strongly impaired G_q-activation. Further, double mutants between residues in ICL2 and 3 suggested interactions between these loops in the vicinity of Phe525 and Thr607, indicating a conformational cooperation between ICLs 2 and 3 during G_q-activation by TSHR.

So far, constitutively activating mutations have only been identified in the ICL3 of the TSHR, but not in the ICL1 and 2 or the C-terminus (47). Both constitutively activating point mutations in ICL 3 Asp619Gly and Ala623Ile are localized in the C-terminal part of the loop and at the ICL3/TM6 junction. Further, the activating deletion mutation del613-621 also includes the C-terminus of ICL3. Mutagenesis studies on hybrid angiotensin II AT₁ and AT₂ receptors showed, that the N-terminal part of ICL3 is primarily responsible for G_q recognition. Further, for the AT1 receptor it was shown that residues in the C-terminus of ICL3 also contribute to Gq activation. Investigation of chimeric TSHR containing homologous sequences of the α1- and β2-adrenergic receptors revealed the importance of N- and C-terminal amino acids in ICL3 for G_q mediated signaling (48). In this work Kosugi et al. showed that substitution of residues in the N- and C-termini of ICL3 abolished Gq mediated IP formation, whereas mutation of amino acids in the middle portion had no effect on TSHR signaling. In contrast, substitution of amino acids 617-620 resulted in increased basal activity regarding to the G_s – cAMP pathway. Based on the constitutively active in vivo deletion del613-621, different deletion- and alanine substitution mutants were characterized (49). Thereby, it has been shown that not the loss of a specific group of mino acids residues is decisive for the constitutive activation of the TSHR. However, shortening of ICL3 leads to a relative movement of TM6 towards the cytoplasm enabling critical transmembrane portions to interact with G_αs. This assumption is supported by recent work of Janz et al. (50) demonstrating interactions of transducin with hydrophobic residues in TM6 following the activation of Rhodopsin. Moreover, the conserved Asp619 at the ICL3/TM6 junction was found to be necessary for maintaining the inactive conformation of TSHR and of CG/LHR as well (49). It has been suggested that Asp619 is involved in a helix-capping structure, indicating that the inactive receptor state is stabilized by interactions between residues within the helix of ICL3/TM6 junction and adjacent structures.
In addition to constitutively activating and inactivating in vivo mutations, several polymorphisms were identified in the TSH receptor gene. The most frequently found polymorphisms are Ile606Met (ICL3) and Ala703Gly, Gln703Glu and Asp727Glu in the C-terminal tail. Several studies produced controversial results regarding the association between these polymorphisms and thyroid diseases. Gabriel et al. (51) found Asp727Glu to be an important factor in the pathogenesis of toxic multinodular goiter, whereas Ile606Met, Ala703Gly and Gln703Glu had no effect. In addition, Asp727Glu was also postulated to be involved in the development of Graves' Disease (52). In contrast, other studies found no correlation between Asp727Glu or other polymorphisms and the appearance of thyroid disorders (53, 54, 55). In the context of the wild type TSHR, the polymorphism Asp727Glu does not seem to be functionally important. Interestingly, both basal and TSH stimulated cAMP levels of the constitutively activating mutant Ala593Asn could be significantly reduced by generation of the Ala593Asn/Asp727Glu double mutant (56). This finding suggests possible unknown properties of TSH receptor polymorphisms which should be further investigated.

Conclusions
The identification of activating and inactivating mutations has revealed first insights into TSHR activation and intramolecular interactions. The combination of site-directed mutagenesis and molecular modelling represents a valuable tool for understanding of structure-function relationships and will lead to a continuous refining of the 3D structure of the TSHR. This improved THSR model will have the potential for rational design of new therapeutical compounds, the development of TSHR agonists, antagonists or superagonists and can help to understand the molecular pathogenesis of thyroid diseases.