The major cause of hyperthyroidism in adults is Graves' disease in which an autoimmune reaction is mounted against the thyroid gland and auto-antibodies are produced that recognize and stimulate the TSH receptor. This may explain why the initial description by the group of Leclère of a family showing segregation of thyrotoxicosis as an autosomal dominant trait in the absence of signs of autoimmunity was met with skepticism (5). Re-investigation of this family together with another family from Reims identified two mutations of the TSH receptor gene, which segregated in perfect linkage with the disease (6). A series of additional families have been studied since and, surprisingly, they each showed a different mutation of the TSHr gene (7-16). [For a complete list of TSH receptor gene mutations with their functional characteristics, see “Glycoprotein-hormone receptor information system database”, GRIS, at http://gris.ulb.ac.be/(17)]. The functional characteristics of these mutant receptors confirm that they are constitutively stimulated (see below). This new nosological entity, hereditary toxic thyroid hyperplasia (HTTH), sometimes called Leclère’s disease, is characterized by the following clinical characteristics: autosomal dominant transmission; hyperthyroidism with a variable age of onset (from infancy to adulthood, even within a given family); hyperplastic goiter of variable size, but with a steady growth; absence of clinical or biological stigmata of auto-immunity. An observation common to the cases described to-date is the need for drastic ablative therapy (surgery or radioiodine) in order to control the disease, once the patient has become hyperthyroid. The autonomous nature of the thyroid tissue from these patients has been elegantly demonstrated by grafting in nude mice (18). Contrary to tissue from Graves'disease patients, HTTH cells continue to grow in the absence of stimulation by TSH or TSAb.

The prevalence of hereditary toxic thyroid hyperplasia is difficult to estimate at the present time. It is likely that many cases have been (and still are) mistaken for Graves'disease. This may be explained by the relative insensitivity and lack of specificity of thyroid stimulating antibody assays, together with the high frequency of the other thyroid auto-antibodies (anti-thyroglobulin, anti-thyroperoxidase) in the general population. It is expected that wider knowledge of the existence of the disease will lead to better diagnosis. This is not a purely academic problem, since presymptomatic diagnosis in children of affected families may prevent the developmental or psychological complications associated with infantile or juvenile hyperthyroidism. A country-wide screening for the condition has been performed in Denmark. It was found in one out of 121 patients with juvenile thyrotoxicosis (0.8%; 95% CI: 0.02-4.6%), which corresponds to one in 17 patients with presumed non-autoimmune juvenile thyrotoxicosis (6%; 95% CI:0.15-28.69) (19).

Sporadic toxic thyroid hyperplasia

Cases with toxic thyroid hyperplasia have been described in children born from unaffected parents (20-22;22-26). Conspicuously, congenital hyperthyroidism was present in most of the cases and required aggressive treatment. Mutations of one TSH receptor allele were identified in the children, but were absent in the parents. As paternity was confirmed by mini- or microsatellite testing, these cases qualify as true neomutations. When comparing the aminoacid substitutions implicated in hereditary and sporadic cases, for the majority, they do not overlap (see table 1; and “Glycoprotein-hormone receptor information system database”, GRIS, at http://gris.ulb.ac.be/(17)) Whereas most of the sporadic cases harbor mutations that are also found in toxic adenomas, most of the hereditary cases have "private" mutations. Although there may be exceptions, the analysis of the functional characteristics of the individual mutant receptors in COS cells, and the clinical course of individual patients, suggest an explanation for this observation: "sporadic" mutations seem to have a much stronger activating effect than "hereditary" mutations. From their severe phenotypes, it is likely that newborns with neomutations would not have survived, if not treated efficiently. On the contrary, from inspection of the available pedigrees, it seems that the milder phenotype of patients with 'hereditary" mutations has only limited effect on reproductive fitness. The fact that "hereditary" mutations are rarely observed in toxic adenomas is compatible with the suggestion that they would cause extremely slow tissue growth and, accordingly, would rarely cause thyrotoxicosis. If this explanation holds true, one may predict that mutations of the hereditary type may be found in the older patients with toxic adenoma.

Soon after mutations of Gs had been found in adenomas of the pituitary somatotrophs (27), similar mutations (also called Gsp mutations) were found in some toxic thyroid adenomas and follicular carcinomas (28-31). The mutated residues (Arg201, Glu227) are homologous to those found mutated in the ras proto-oncogenes: i.e. the mutations decrease the endogenous GTPase activity of the G protein, resulting in a constitutively active molecule. Toxic adenoma was found to be a fruitful source of somatic mutations activating the TSHr, probably because the phenotype is very conspicuous and easy to diagnose (32). Most of the mutations are located in the third cytoplasmic loop or in the adjacent sixth transmembrane segment of the receptor (see figure 16a-1). The clustering reflects the pivotal role of this portion of the molecule in the activation mechanisms (33;34) and has allowed to identify several molecular locks implicated in the activation mechanism (35). However aminoacid substitutions were found over most of the serpentine portion of the receptor (32;36-40) and even in the extracellular aminoterminal domain (41). [For a complete list of TSH receptor gene mutations with their functional characteristics, see “Glycoprotein-hormone receptor information system database”, GRIS, at http://gris.ulb.ac.be/(17)].

Figure 1a. Schematic representation of the TSH receptor with the position of activating mutations indicated. For references and a complete list of mutations see text and http://gris.ulb.ac.be/.

Figure 1b. Schematic representation of the TSH receptor with the position of activating mutations indicated. For references and a complete list of mutations see text and http://gris.ulb.ac.be/.

Despite some dispute about the prevalence of TSH receptor mutations in toxic adenomas (which may be due to different origin of patients (42;43) or different sensitivity of the methodology) we conclude that in countries with a moderate shortage of iodine, activating mutations of the TSH receptor are the major cause of solitary toxic adenomas, accounting for up to 80% of the cases (40;44). In contrast to previous results (42), activating mutations of the TSH receptor have been found in 70% of autonomous adenomas in a Japanese population (45). In some patients with a multinodular goiter and two zones of autonomy at scintigraphy, a different mutation of the TSH receptor was identified in each nodule (46-49). This indicates that the pathophysiological mechanism responsible for solitary toxic adenomas can be at work on a background of multinodular goiter and may be responsible for some of the autonomous zones appearing late in the evolution of these patients. The independent occurrence of two activating mutations in a patient may seem highly improbable at first. However, the multiplicity of the possible targets for activating mutations within the TSH receptor (at least 20 different residues) makes this less unlikely. It is also possible that a mutagenic environment is created in glands exposed to a chronic stimulation by TSH, resulting in H2O2 generation (50). Finally the involvement of TSH receptor mutations in thyroid cancers has been implicated in a limited proportion of follicular thyroid carcinoma selected for their high basal activity of adenylylcyclase (51;52), and one case with mutations in both the TSH receptor and the Ki-Ras oncogene has been reported (53).

The majority of the activating mutations of the TSH receptor have been studied by transient expression in COS cells. By the built-in amplification of the transfected construct, this system makes it possible to detect even slight increases in the constitutive activity of the TSH receptor. An important observation has been that the wild type receptor itself displays significant constitutive activity (6;54). This characteristic is not unique to the TSH receptor (55-57) but, interestingly, it is not shared by its close relative the LH/CG or the FSH receptors (55-57). The effect of activating mutations must accordingly be interpreted in terms of "increase in constitutive activity".Most aminoacid substitutions found in toxic adenomas and/or toxic thyroid hyperplasia share common characteristics: (1) they increase the constitutive activity of the receptor towards stimulation of adenylyl-cyclase; (2) with a few notable exceptions (38), they do not display constitutive activity towards the inositolphosphate/diacylglycerol pathway; (3) their expression at the cell surface is decreased (from slightly to severely); (4) most, but not all of them, keep responding to TSH for stimulation of cAMP and inositolphosphate generation, with a tendency to do so with decreased EC50; (5) they bind 125I bovine TSH with an apparent affinity higher than the wild type receptor. There is no simple relationship between the position of the mutations or the nature of the aminoacid substitution and their functional characteristics. Mutations found in transmembrane segments II, III, VI and VII and the third cytoplasmic loop all have similar phenotypes; they involve aminoacids belonging to all classes (charged, polar, hydrophobic), with substitutions not necessarily involving a shift to another class. Mutations involving Ile486 and Ile568 in the first and second extracellular loops, respectively, and Pro639 in transmembrane segment VI are exceptional in that, in addition to stimulating adenylylcyclase, they cause constitutive activation of the inositolphosphate pathway (38). Three additional mutations deserve special mention because of their unexpected nature or location: the four aminoacid deletion (residues 658-661) in the third extracellular loop (44), the nine aminoacid deletion in the third intracellular loop (39) and the substitutions at serine 281 in the aminoterminal extracellular domain (41).There is no direct relation between the level of cAMP achieved by different mutants in transfected COS cells and their level of expression at the cell membrane (58). This means that individual mutants have widely different "specific constitutive activity" (measured as the stimulation of cAMP accumulation/receptor number at the cell surface). While this specific activity may tell us something on the mechanisms of receptor activation, it is not a measure of the actual phenotypic effect of the mutation in vivo. Indeed, one of the relatively mild mutations, observed up to now only in an HTTH family (Cys672Tyr), is among the strongest according to this criterion. It would be logical to expect the best correlation to be found between the phenotype and the actual level of cAMP achieved, irrespective of the level of receptor expression. However, differences between the effects of the mutants in COS cells and thyrocytes in vivo may render these correlations a difficult exercise. Indeed, there is presently no convincing correlation between the nature of the mutations and the actual phenotype, whether in germline or somatic mutations.According to a current model for GPCR activation, the receptor would exist under at least two interconverting conformations: R (silent conformation) and R* ( the active forms) (59). The unliganded receptor would shuttle between both forms, the equilibrium being in favor of R. Binding of the ligand, to the slit between the transmembrane segments (for biogenic amines) and/or residues of the N-terminal segment or extracellular loops (for neuropeptides), is believed to stabilize the R* conformation. The resulting R to R* transition is supposed to involve a conformational change modifying the relative position of transmembrane helices. In turn, this would translate into conformational changes of the cytoplasmic domains interacting with trimeric G proteins. Seminal studies with the adrenergic receptor a1b have shown that a variety of aminoacid substitutions in the C-terminal portion of the third intracellular loop led to their constitutive activation (60). The observation that all aminoacid substitutions at Ala293 were effective in activating the receptor led to the concept that the silent form of GPCRs would be submitted to a structural constraint, requiring the wild type primary structure of the third intracellular loop. This constraint could be released by a wide spectrum of aminoacid substitutions in this segment (59-62).The observation that aminoacid substitutions in a large number of residues scattered over the serpentine portion of the TSH receptor causes increase in its constitutive activity is fully compatible with the above model and provides arguments for its extension. The fact that mutations in residues distributed over most of the serpentine portion of the receptor are equally effective in activating it (which does not seem to be a general characteristic in all GPCRs) suggests that the unliganded TSH receptor might be less constrained than others. The readily measurable constitutive activity of the wild type receptor is compatible with this contention. Being already "noisy", the TSHr would be more prone to further destabilization by a variety of mutations. The sixth transmembrane segment and its continuation in the C-terminal portion of the third cytoplasmic loop is clearly a hot spot for activating mutations (Figure 16a-1), with a series of contiguous residues potentially implicated in keeping the receptor inactive. The observation in a drosophila homologue of the glycoprotein hormone receptors of a reciprocal aminoacid exchange between transmermbrane segments VI and VII suggested the existence of an interaction between Asp633 and Asn674 in the inactive TSH receptor (34). Since Asp633 is one of the residues most frequently affected by activating mutations, this led to the suggestion that an Asp633-Asn674 interaction was required to maintain the receptor in its inactive conformation. This hypothesis was supported by molecular dynamic simulations of the serpentine portion of the TSHr made on the template of the crystallographic structure of rhodopsin and by site-directed mutagenesis (34). Extension of such studies to the exceptionally wide panel of activating mutations of the TSHr has the potential to contribute to our general understanding of conformational changes implicated in the activation of rhodopsin-like GPCRs. Studies of this kind are still underway and have already yielded interesting results (35;63-65). The deletion of nine residues in the third cytoplasmic loop is believed to activate the receptor by facilitating binding of Gsa to portions of the transmembrane domains (39).The activating mutations identified in the extracellular loops and the aminoterminal extracellular domain are more difficult to interpret in the lights of a simple model based on constraint involving only the serpentine portion of the receptor. They are compatible with an extension of this model, in which the unliganded aminoterminal domain would contribute to the constraint keeping silent the serpentine portion (66). Experiments involving TSHr constructs with aminoterminal truncations have confirmed this hypothesis. Indeed, receptors devoid of the aminoterminal TSH-binding domain display higher constitutive activity than the wild type holoreceptor (66;67). However, full stimulation of the TSHr, whether by its normal agonist, TSAb, or activating mutations involving the extracellular domain, require more than release of the inhibitory constraint exerted by the ectodomain (66). Experiments with a series of site-directed mutagenesis constructs lead to the suggestion that, upon binding of TSH, the ectodomain would change its conformation, switching from a tethered inverse agonist (when unliganded) to a full agonist (66). According to this model, the true agonist of the serpentine portion of the TSHr would be the “activated” ectodomain (see figure 16a-2)

Figure 2. Putative model of the intramolecular interactions involved in the activation of the TSH receptor. The basal state of the receptor is characterized by an inhibitory interaction between the ectodomain and the serpentine domain. The ectodomain would function as a tethered inverse agonist. Removal of the ectodomain releases the serpentine domain from the inhibitory interaction, resulting in partial activation. Mutation of Ser281 into Leu switches the ectodomain from an inverse agonist into a full agonist of the serpentine domain. Binding of TSH or thyroid stimulating autoantibodies to the ectodomain are proposed to have a similar effect, converting it into a full agonist of the serpentine portion. It must be stressed that the scheme is purely illustrative. It emphasizes that, according to the model, activation does not require a direct interaction between the hormone and the serpentine domain. Such an interaction, however, is by no means excluded (adapted from )

Some degree of stimulation of the thyroid gland by human chorionic gonadotrophin (hCG) is commonly observed during early pregnancy. It is usually responsible for decrease in serum thyrotropin with increases in free thyroxine concentrations that remain within the normal range (for references see (68;69)). When the concentrations of hCG are abnormally high, like in molar pregnancy, true hyperthyroidism may ensue. The pathophysiological mechanism is believed to be the promiscuous stimulation of the TSH receptor by excess hCG, as suggested by the rough correlation between serum hCG and free T4 concentrations (70;71). A convincing rationale is provided by the close structural relationships of the glycoprotein hormones and their receptors, respectively (72).A new syndrome has been described in 1998 in a family whith dominant transmission of hyperthyroidism limited to pregnancy (see fig.16a-3)(73). The proposita and her mother had severe thyrotoxicosis together with hyperemesis gravidarum during the course of each of their pregnancies. When non pregnant they were clinically and biologically euthyroid. Both patients were heterozygous for a K183R mutation in the extracellular aminoterminal domain of the TSH receptor gene. When tested by transient transfection in COS cells, the mutant receptor displayed normal characteristics towards TSH. However, providing a convincing explanation to the phenotype, it showed higher sensitivity to stimulation by hCG, when compared with wild type TSH receptor (73). The aminoacid substitution responsible for the promiscuous stimulation of the TSHr by hCG is surprisingly conservative. Also surprising is the observation that residue 183 is a lysine in both the TSH and LH/CG receptors. When placed on the available three dimensional model of the hormone-binding domain of the TSH receptor (74), residue 183 belongs to one of the beta sheets which constitute the putative surface of interaction with the hormones (75). In search for a structural explanation of the phenotype, Lys183 was mutated to a variety of aminoacids with different physicochemical properties (76). Unexpectedly, these experiments indicated that the gain of function of the Lys183Arg mutant toward hCG was not due to the presence of the Arg but to the loss of the Lys. Indeed, most mutants displayed a phenotype similar to that of Lys183Arg (76). A detailed functional analysis of site-directed mutagenesis constructs designed after molecular dynamic simulations led to the conclusion that the widening of specificity toward hCG was due to the release of the side chain of Glu157 (a neighbor of Lys183 in the tridimensional structure of the ectodomain) from an ionic bond with Lys183, making it available for interaction with hCG. This view of the molecular mechanism responsible for the broadening of specificity of the mutant has been fully confirmed when the three-dimensional structure of the ectodomain of the FSH receptor complexed with FSH became available (77). It is likely that an arginine in position 183 would confer a slight increase in stability to the illegitimate hCG/TSHr complex (78). This would be enough to cause signal transduction by the hCG concentrations achieved in pregnancy, but not by the LH concentrations observed after menopause. Indeed the mother of the proposita remained euthyroid after menopause. This is compatible with a relatively modest gain-of-function of the K183R mutant for stimulation by hCG.Contrary to other mammals, human and primates rely on chorionic gonadotropin for maintenance of corpus luteum in early pregnancy (79). The frequent partial suppression of TSH observed at peak hCG levels during normal pregnancy indicates that evolution has selected physiological mechanisms operating very close to the border of thyrotoxicosis. This may provide a rationale to the observation that, in comparison to other species, the glycoprotein hormones of primates display a lower biological activity due to positive selection by evolution of specific aminoacid substitutions in their alpha subunits (80). If this reasoning is correct, it is likely that further cases of hereditary gestational thyrotoxicosis will be identified with mutations in the alpha or beta subunits of hCG of the fetus.

A syndrome similar to gestational hyperthyroidism secondary to mutation in the TSH receptor has been identified in spontaneous ovarian hyperstimulation syndrome (81;82). In these cases, mutations in the FSH receptor renders it abnormally sensitive to hCG, illustrating the challenge hCG poses to the specificity of agonist recognition by glycoprotein hormone receptors during human pregnancy.

Figure 3. Pathophysiology of familial gestational hyperthyroidism secondary to mutation of the TSH receptor gene. Upper-left panel illustrates binding of TSH, or hCG to the ectodomain of the TSH receptor, according to crystallographic data from . The yellow dot indicates the position of the K183R mutation. Lower-left panel documents the thyrotropic activity of hCG during normal pregnancy (adapted from ). Upper right panel displays the pedigree, with the two ladies affected, together with a “snake plot “of the TSH receptor, with the mutation indicated. Lower right panel illustrates the increased sensitivity of the K183R mutant TSH receptor vis-à-vis hCG (adapted from )

Loss-of-function mutations in the TSHr gene are expected to cause a syndrome of "resistance to TSH". The expected phenotype is likely to resemble that of patients with mutations in TSH itself. These have been described earlier because of the prior availability of the information on TSH alpha and beta genes(83-85). A mouse model of resistance to TSH is available in the hyt/hyt line. Homozygous hyt/hyt mice are hypothyroid secondary to a developmental anomaly of their thyroid which remain hypoplastic (86). The cause has been traced to a mutation of the TSH receptor gene (Pro556Leu) (86;87). From this information one would expect patients with two mutated alleles to exhibit a degree of hypothyroidism in relation with the extent of the loss-of-function. Heterozygous carriers are expected to be normal or to display minimal increase in plasma TSH.

A few patients with convincing resistance to TSH have been described before molecular genetics allowed for identification of the mutations (88;89). Another family has been described more recently, but no mutation was found in the receptor gene (90). The first cases described in molecular terms were euthyroid sibs with elevated TSH (91). Sequencing of the TSH receptor gene identified a different mutation in each allele of the affected individuals, making them compound heterozygotes. The substitutions were in the extracellular amino terminal portion of the receptor (maternal allele: Pro162Ala; paternal allele: Ile167Asn). The functional characteristics of the mutant receptors were studied by transient expression in COS cells: the paternal allele was virtually completely non-functional, while the maternal allele displayed increase in EC50 for stimulation of cAMP production by TSH. Recent additional experiments have shown that the paternal allele is expressed in normal amounts in COS cells, but that it remains trapped intracellularly and does not reach the cell surface. However, even when assayed on cell membranes, the paternal allele does not bind TSH, which suggests that the mutations has profound structural consequences affecting both the routing of the receptor to the plasma membrane and its ability to bind TSH. When both mutations are displayed on a tentative model of the extracellular domain, their location is compatible with the observed phenotype: the Pro162Ala mutation affects a residue predicted to be at the surface of the molecule, which may explain its interference with effects of TSH. The Ile167Asn mutation affects a residue protruding within the hydrophobic tunnel between the alpha helices and the beta sheets of the doughnut-shaped model (74). It is expected that a polar residue would be incompatible with such position and results in severe misfolding of the whole extracellular domain. Coexpression in COS cells of the wild type and mutated receptors did not show evidence for dominant negative effects of the mutants. A series of other cases have been published (92;93) and a compilation of loss-of-function mutations is shown in figure 16a-4 (for a complete list of mutations, see http://gris.ulb.ac.be/).

Figure 4. Schematic representation of the TSH receptor with the position of loss-of-function mutations indicated. For references and a complete list of mutations see text and http://gris.ulb.ac.be/.

Further familial cases with loss-of-function mutations of the TSH receptor have been identified in the frame of screening programs for congenital hypothyroidism (for a complete list of mutations see (94-96) and “Glycoprotein-hormone receptor information system database”, GRIS, at http://gris.ulb.ac.be/ (17). The patients displayed the usual criteria for congenital hypothyroidism, including high TSH, low free T4 and undetectable trapping of 99Tc. The thyroids were small and normally located at ultrasonography. Surprisingly, in some of the cases, plasma thyroglobulin levels were normal or high. As may be expected for loss-of-functions, mutations were diverse and patients were compound heterozygotes except when consanguinity was present. In one case from Brussels, the patients were sibs born from consanguinous parents and were homozygous for a mutation in the transmembrane segment IV(Ala553Thr), close to the hyt mutation of the mouse (see figure 16a-3). When transiently expressed in COS cells, the mutants were barely expressed at the cell surface. However, the residual expression was compatible with some TSH binding and stimulation of cAMP production by TSH (94).

Studies in mice with complete invalidation of the TSH receptor gene demonstrate convincingly that, at least in mice, a functional TSH receptor is not required for the development of an orthotopic gland (97). The absence of iodide trapping by the glands from patients harboring a non-functional TSH receptor is explained by the complete dependence of NIS activity on activation of the cAMP regulatory pathway (98).

Patients heterozygous for complete loss-of-function mutations of the TSH receptor may display mild increase of plasma TSH. However, the patients are euthyroid and the penetrance is not complete (99). There is indication that in some cases, the mutant allele might exert a dominant negative effect on expression of the wild type allele, by trapping misfolded dimers in the endoplasmatic reticulum (100).

Families with autosomal dominant transmission of apparent resistance to TSH have been described where the phenotype does not cosegregate with the TSH receptor gene (101;102). In some of these families linkage has been demonstrated with a locus on chromosme 15q25.3-26.1, but the gene responsible for the affection has not been identified yet (103).