The thyrotropin receptor (TSHr) is a key protein in the control of thyroid function and a major thyroid autoantigen (1). After cloning the complementary DNA (cDNA) of the TSHr considerable progress in elucidating the structure and function of the TSHr has been made. Analysis of recombinant TSHr proteins expressed in prokaryotic and eukaryotic systems has indicated that post-traslational processing is important for the formation of active receptor. Studies of TSHr glycosylation have shown that a mature form of the receptor containing mainly complex-type sugar residues is principally involved in TSH and TSHr autoantibody (TRAb) binding. Sulfation of the TSHr is also required for efficient recognition and activation of the receptor by TSH. The processing of the TSHr peptide chain into two subunits observed with native receptor has been confirmed using recombinant TSHr. However, the binding site(s) for TSH and TRAb on the TSHr have not been well characterized. The discovery of naturally occurring amino acid mutations of the TSHr confirm the complexity of the hormone and autoantibody binding sites.

TSHr STRUCTURE

The TSH, LH/CG and FSH receptors belong to a subfamily of G protein-coupled receptors and their primary structure, as deduced from their cDNA, predicts the existence of seven segments with hydropathy (in common with all the other G protein-coupled receptors) compatible with transmembrane segments (1, 2). The glycoprotein receptor subfamily (TSH, LH/CG, FSH) share common characteristics that distinguish them from the other G protein-coupled receptors. They contain a signal peptide (20 amino acids for the TSH receptor) and they have a long extracellular aminoterminal domain (398 amino acids for the TSH receptor) with the loose repetition of a motif of 25 residues rich in leucine (1, 2), with six potential N-linked glycosylation sites. Similar leucine-rich motifs are also found in a number of widely different proteins (3, 4) which confer the ability to interact with other proteins.
From site-directed mutagenesis studies it seems that the binding specificity and the effector properties of the glycoproteic receptors are encoded in separate domains of the protein (1); the extracellular N-terminal domain mediates the binding specificities and the portion with the seven transmembrane domains display the effector properties trigging G-protein activation. When aligned, the three glycoprotein receptors show stronger conservation in the transmembrane domains (70% homology) than in the extracellular domain (40% homology). A peculiarity of the TSH receptor is a 51 residue insert at the hinge between the extracellular and the first transmembrane segment with no counterpart in the FSH or LH/CG receptors. A first model of the three dimensional structure of the thyrotropin receptor has been recently proposed (5) based on the analogy with another leucine-rich repeats protein, the ribonuclease inhibitor, which has been crystallized.

PHYSIOLOGY OF THE TSHr

In human thyroid TSH activates both the cAMP and the phospholipase C-diacylglycerol regulatory cascades, although the latter effect requires concentrations of hormones 5 to 10 times higher than the former (1, 6, 7). This activation is mediated by the TSH receptor and it involves the exchange of GDP by GTP and the consequent dissociation of G in its subunit alpha and beta-gamma; the former, to which GTP is bound stimulates the effector enzymes a) adenylyl-cyclase which generates cyclic AMP (cAMP) from ATP and is important for growth and differentiation and b) phospholipase C cascade which stimulates the production of inositol 1,4,5-triphosphate (IP3) and diacylglycerol, which are important for iodination and hormone synthesis (1, 6, 7).
Modifications, over or under production of the natural ligand (TSH), or the presence of agonists or antagonists of the natural ligand (example antibodies or drugs) or alterations of the intrinsic mechanism of the receptor activation may result in diseases.

RECOMBINANT THYROTROPIN RECEPTOR

The TSHr is present in very low numbers on the surface of thyreocytes, which has made the receptor difficult to clone and, even after the cDNA sequence had been published, difficult to produce in high numbers and to purify (8). Various methods of expressing the recombinant TSHr in different systems to obtain larger quantities of the protein for purification and production of both polyclonal and monoclonal antibodies have been investigated. Low levels expression of some G-protein receptors have been obtained in Escherichia Coli. However, the full length TSH receptor has not been successfully expressed in this system. Large amounts of TSH extracellular domain in prokaryotic systems can be produced but most protein is present in inclusion bodies and the denaturated non-glycosylated product does not bind TSH (8). Despite extensive studies neither the full-length nor the extracellular domain of the TSHr has been successfully expressed in Yeast (8). A third system using the baculovirus system failed to produce full-length or the extracellular part of the TSH receptor or a low yeld was obtained (8).
Eukaryotic expression systems such as Chinese hamster ovary (CHO) cells, 293 human embryonal kidney cells, L cells, and a transformed myeloma cell line SP65, have been used to produce stably transfected cell lines expressing the TSHr (8). In each case the recombinant receptor produced is expressed on the cell surface, is functional for hormone (TSH) binding, is coupled to cAMP, is highly glycosylated and is able to bind TSH receptor antibody (TRAb). The amount of receptor produced in CHO stably transfected with the cDNA of the human TSHr or in transient expression in COS cells is not sufficient for purification because the trypsine treatment necessary to detach cell causes proteolytic cleavage, even if an adaptation of CHO cells grown in suspension has been reported. Until very recently, constructs encoding the complete extracellular amino-terminal domain (ECD) alone did not yield bioactive material capable of binding TSH with high affinity. Costagliola et al. (9) created a chimeric cDNA construct encoding the ECD of the TSHr fused to the signal for addition of glycosylphospahatidylinositol from the Thy-1 gene which directs efficient expression of the ECD at the plasma membrane of transfected CHO cells (9). Treatment of these transfected cells with a specific phospholipase C released a soluble 80 kDa molecule which neutralizes the antibodies from Graves' patients. Whereas it does not bind TSH when released from the cells after incubation with phospholipase C in free form, the soluble ECD displays clear TSH binding activity when it is released as a complex with a monoclonal antibody recognizing a conformational epitope of the ECD (9). These observations together with those from Osuga et al (10) showed that the complete ECD of the TSHr require additional signal sequences to be correctly targeted to the plasma membrane in a native form. In the holoreceptor, the signaling is probably carried out by the serpentine portion of the receptor itself. It is likely that the Glycosylphosphatidylinositol-anchoring peptide (9) provide adequate substitutes for this targeting, allowing a significant proportion of the ECD molecules to undergo normal glycosylation and maturation, during the journey through the membrane system of the cell.
The soluble ectodomain could be released from the cells by treatment with a GPI-phospholipase C and purified to apparent homogeneity by chromatography (11). This soluble ectodomain purified in a functionally competent conformation allows direct studies of its interaction with TSH and autoantibodies and open the way to structural studies.

TSH RECEPTOR AS A TARGET OF AUTOIMMUNITY

To increase our understanding of the processes involved in the pathogenesis of autoimmune thyroid disease it is important to understand the structure of the TSHr and especially the sites of interaction between the receptor, TSH and TRAb. TSH receptor antibodies can be classified as: a) TBII which inhibits the binding of TSH to the receptor b) TSAb which stimulates cAMP production and are responsible for growth and hyperfunction of thyrocytes characteristics of Graves' disease and c) TBAb which inhibit TSH mediated cAMP accumulation and are the cause of some cases of hypothyroidism in Hashimoto's thyroiditis and idiopathic myxedema.
There have been several reports using chimeras of full-length TSHr with segments of the LH-CG receptor extracellular domain exchanged for the corresponding regions in the TSHr extracellular domain expressed in CHO cells. The reports have concluded that the binding sites for TSH and TRAb are not identical but appear to overlap and cover most of the length of the extracellular domain of the receptor (8).
Another approach to studying the binding sites of the TSHr has involved the use of synthetic peptides corresponding to regions of the TSHr extracellular domain and the effects of these peptides on stimulation of cAMP production by TSH and TRAb has been investigated. Polyclonal antibodies have been raised to synthetic TSHr peptides in both rabbit and chicken while bacterially expressed fusion proteins of the TSHr extracellular domain and the TSHr extracellular domain expressed in the baculo virus system have also been used to immunize rabbits. There are conflicting results as to whether antibodies do or do not inhibit TSH binding to the TSHr. Some of these studies reported that the binding sites for TRAb with TSH antagonistic activity were at the C-terminal segment of the TSHr extracellular domain whereas TRAb with TSH agonistic activity bound to the N-terminal part of the TSHr extracellular domain (8).
The production of human monoclonal antibodies to the TSHr has proved to be very difficult (8). Isolation of Epstein-Barr virus transformed, IgG expressing B cell lines from patients with autoimmune thyroid disease with TSH agonist and TSH antagonist activity has been reported but these preparations do not inhibit TSH binding to TSHr. To obtain a true TSHr stimulating monoclonal antibody, several animal models of Graves' disease have been generated in recent years. Murine monoclonal TSHr antibodies generated with these models have been shown to recognize the native conformation of the TSHr, but all have been without thyroid stimulating activities. Recently Ando et al (12) isolated a TSHr-stimulating monoclonal antibody that had a marked thyroid-stimulating activity at nanogram concentrations. This antibody recognized a conformational epitope. By using genetic immunization Costagliola et al (13) were able to produce a monoclonal antibody with thyroid stimulating activity and surprisingly this antibody was very effecting in detecting the purified ectodomain in hTSHr on Western Blot. The mere existence of monoclonal antibodies directed against the TSHr and capable of activating it tells us that there is no need for cooperation of multiple immunoglobulins with different recognition specifities to achieve stimulation of the receptor in Graves' disease. The question, however, remains whether recognition of different epitopes, overlapping or not, would similarly result in receptor activation. Further experiments will be needed to determine the relation if any between the epitope identified here and those of autoantibodies from Graves' patients. If there is structural relation between them, monoclonal antibodies may constitute tools allowing development of in vitro binding assays capable of differentiating autoantibodies with TSAb from those simply displaying TBII activity.

POST-TRANSLATIONAL MODIFICATIONS OF THE TSHr GENE

Glycosylation

The extracellular domain of the hTSHr contains six potential N-linked glycosylation sites and has been shown to be heavily glycosylated with approximately 35 kDA of carbohydrate residues contributing to the overall molecular weight when expressed in CHO cells (1). However when expressed in E Coli the TSHr is unglycosylated and has been found to be incapable of both high affinity TSH binding and autoantibody binding (8).
The full-length TSHr expressed in CHO cells has been shown to consist of two species of full-length receptor, one of approximately 100 kDA and the other approximately 120 kDA. Pulse labeling of L cells expressing the TSHr showed that the 100 kDA product was produced first and was shown to be the precursor for the upper band of 120 kDA. Some of the upper band, mature receptor, then appeared to be cleaved into two subunits. The upper band contained complex-type carbohydrate residues with a high content of sialic acid. The lower band contained predominantly high mannose type carbohydrates (8).

Sulfation

Sulfation of tyrosines is a late post-transcriptional modification taking place in the trans-Golgi network and affecting a wide spectrum of membrane or secreted proteins. Recently, sulfation of tyrosine residues of the N-terminal extension of three GPCRs belonging to the chemokine or chemoattractant receptor family has been demonstrated. In CCR5 tyrosine sulfation was required for high affinity recognition of the receptor by its natural agonist. Similarly to the situation described recently in CCR5, Costagliola et al. (14) demonstrated that the TSHr, as it is present at the cell surface, is sulfated on tyrosines in a motif located downstream of the C-terminal cysteine cluster. Sulfation of one of the two tyrosines in the motif is mandatory for high affinity binding of TSH and activation of the receptor. Site directed mutagenesis experiments indicate that the motif, which is conserved in all members of the glycoprotein hormone receptor family, seems to play a similar role in the LH or FSH receptors.

TSH RECEPTOR MUTATIONS

G protein-coupled receptor naturally occurring mutations can be cause of diseases. Depending on the nature of the mutation (somatic, germline), and on its localization in the protein, and in the case of dominant diseases differences in genetic background as well as environmental factors, can be responsible for different phenotypes.

TSHr gain-of-function mutations

Any molecular lesion leading to constitutive activity of the cAMP cascade (TSH receptor, G protein, cyclase, protein kinase) could be responsible for the growth and hyperfunction typical of thyroid adenoma. After somatic mutations impairing GTPase activity of Gs-alpha had been found in some of these benign tumors it was logical to study the TSH receptor gene. In the first study from the group of Vassart (15, 16), nine out of eleven tissues studied were shown to harbor an activating TSHr mutation. Other studies have confirmed this observation, describing mutations in other residues (17, 18). All the mutations found are heterozygous, as expected from gain-of-function mutations with dominant effect, and confined to the adenomatous tissue.
Recently we (19) reported that similarly to solitary toxic thyroid adenoma, activating TSHr mutations are present in single hyperfunctioning nodules (either adenomas or hyperplastic nodules) within toxic multinodular goiters in which nonfunctioning nodules also coexist.

TSHr loss-of-function mutations

Mutations that inactivate the thyrotropin receptor protein can cause thyrotropin resistance, resulting in either hypothyroidism or euthyroidism depending on the completeness of the defect (18). When transfected in COS cells the mutated thyrotropin receptors showed no or a reduced biological activity (18).

STRUCTURE-FUNCTION RELATIONSHIP OF THE TSHr, AS DEDUCED FROM THE STUDY OF ACTIVATING AND INACTIVATING MUTATIONS

The majority of activating mutations of the TSHr gene have been studied by transient expression in COS cells. When mutant receptors are transiently expressed from recombinant constructs in COS cells the result is a constitutively stimulation of cAMP accumulation (18).
Interestingly these experiments clearly show that the wild type TSH receptor displays easily measurable constitutive activity (18). When comparing all G protein-coupled receptors not all show significant basal activity. For example while the TSH receptor shows a measurable constitutive activity when expressed in COS cells the LH/CG receptor displays little constitutive activity if any (18). In agreement with a current model for G-protein coupled receptor activation, this observation suggests that the unliganded TSH receptor would be less constrained than others when in the inactivate state (18). This model has two important physiopathological consequences: 1) a minor structural alteration caused by a mutation determines a destabilization which is responsible for a phenotype and the diversity of mutations able of increasing the constitutive activity of the TSHr is surprisingly high with respect to other receptors 2) the existence of a basal tonic activity opens the possibility of regulating them negatively by so called inverse agonists (20).
Some mutants (two with modified residues in the extracellular loops, one in the third intracellular loop and one in the sixth transmembrane segment) activate also the phospholipase C-dependent cascade (18). The different mutant receptors show a different level of expression when transfected in COS cells in identical conditions; to compare their efficacy we derived a specific constitutive activity (basal cAMP/receptor number); some mutants, though expressed at low levels at the surface, cause strong stimulation of the cAMP cascade (I486F, T632I, C672Y). Besides, for many mutants a higher affinity for the ligand is observed. Most mutants respond to stimulation by TSH by further increasing both cAMP and inositolphosphate accumulation, but the magnitude is highly variable, some mutations behaving as if they were fully activated (e.g. mutant I486F for cAMP) or displaying very little stimulability (e.g. mutation C672Y for inositol-phosphates). Other mutants (N670S for example) clearly show a dissociation in the ability of the receptor to respond to bTSH for the Gs-alpha and Gq-alpha dependent regulatory cascades, favoring the idea of the existence of multiple active conformations of the TSH receptor, with differential capabilities to couple to Gs-alpha and Gq-alpha.
A current favored model for GPCR activation holds it that a structural constraint is responsible for the maintenance of unliganded receptors in the inactive state. This model was elaborated from the observation that a variety of aminoacids substitutions, first in the third intracellular loop of adrenergic receptors, then in transmembrane helices of other receptors could activate them in the absence of agonists. A series of experimental observations suggest that the extracellular domain of the TSHr could contribute in keeping its serpentine portion inactive. 1. Aminoacid substitution in the first (Ile 486) and second extracellular loops (Ile 568) are amongst the strongest activating mutations identified. 2. The TSH receptor can be activated by a limited proteolytic treatment by trypsin which removes an epitope (residues 354-359) of the extracellular domain. 3. The group of Kosugi (21) has demonstrated significant increase in constitutive activity of a deleted mutant lacking residues 339-367. These observations are compatible with a model in which the unliganded inactive conformation of the receptor would be stabilized by interactions between the extracellular N-terminus and the extracellular loops. Ruptures of these interactions would activate the serpentine portion while increasing the affinity of the extracellular domain for TSH binding. According to the model, unliganded receptors would exist as an equilibrium between a closed inactive conformation and an open active conformation lacking the interaction between the loops and the N.-terminal domain (22). The concentration of the latter would be responsible for the constitutive activity of the wtTSHr. Binding of TSH to the extracellular domain would activate the receptor by stabilizing the open conformation. The model does not exclude that an interaction of TSH with the extracellular loops contributes to the stabilization of the active conformation of the serpentine portion as suggested by some experiments with LH receptor (22).

REFERENCES