Deletions in the Third Intracellular Loop of the Thyrotropin Receptor A NEW MECHANISM FOR CONSTITUTIVE ACTIVATION*

Gain-of-function mutations of the thyrotropin receptor (TSHR) gene have been invoked as one of the major causes of toxic thyroid adenomas. In a toxic thyroid nodule, we recently identified a 9-amino acid deletion (amino acid positions 613–621) within the third intracellular (i3) loop of the TSHR resulting in constitutive receptor activity. This finding exemplifies a new mechanism of TSHR activation and raises new questions concerning the function of the i3 loop. Because the i3 loop is thought to be critical for receptor/G protein interaction in many receptors, we systematically reexam-ined the role of the TSHR’s i3 loop for G protein coupling. Thus, various deletion mutants were generated and functionally characterized. We identified an optimal deletion length responsible for constitutive activity. If the number of deleted amino acids was reduced, elevated basal cAMP accumulation was found to be con-comitantly diminished. Expansion of the deletion dra-matically impaired cell surface expression of the receptor. Shifting the deletion toward the N terminus of the i3 loop resulted in unaltered strong constitutive receptor activity. In contrast, translocation of the deletion toward the C terminus led to significantly reduced basal cAMP formation, most probably due to destruction of a conserved cluster of amino acids. In this study, we show for the first time that amino acid deletions

The glycoprotein hormone TSH 1 is the major regulator of growth and differentiation of the thyroid gland (1). TSH acts by binding to its receptor at the basolateral membrane of thyroid follicular cells. The thyrotropin receptor (TSHR) belongs to the large superfamily of heptahelical G protein-coupled receptors (GPCRs) (1)(2)(3). In the human thyroid, the ligand-bound TSHR leads to stimulation of adenylyl cyclase and phospholipase C by interacting with G s and G q/11 (4). The cAMP regulatory cascade controls growth and differentiated function (thyroid hormone secretion, iodide trapping), whereas Ca 2ϩ and diacylglycerol stimulate iodination and thyroid hormone synthesis (1).
Recent evidence suggests that agonist-dependent activation of GPCRs elicits a crucial conformational change in the receptor molecule resulting in a movement of transmembrane helices relative to one another and subsequent G protein activation (5,6). The reallocation of transmembrane domains (TMs) is brought about by breaking and formation of interhelical noncovalent bonds between specific amino acid residues (7)(8)(9)(10)(11). Interestingly, these intermolecular rearrangements can be mimicked by mutations that create constitutively active receptors. Gain-of-function mutations have been identified in many GPCRs (12). In the TSHR gene, gain-of-function mutations cause autosomal dominant nonautoimmune hyperthyroidism and toxic thyroid nodules (13,14). Constitutively activating mutations have been predominantly identified in TMs but also in connecting intra-and extracellular loops (15)(16)(17)(18)(19)(20)(21)(22). Within the overall receptor structure, the third intracellular loop (i3) and TM6 appear to be hot spots for gain-of-function mutations in different GPCRs. Several reports suggest that these domains may be particularly important for receptor/G protein interaction (12). In the adrenergic receptor system, cytoplasmic domains which are in juxtaposition to the plasma membrane, particularly near TMs 5, 6, and 7, have been implicated in G protein coupling (23). Site-directed mutagenesis and substitution of amino acid clusters with corresponding regions from ␣ 1and ␤ 2 -adrenergic receptors revealed selective effects of different regions in the i3 loop of the TSHR on inositol phosphate (IP) and cAMP signaling cascades. The N-and C-terminal i3 loop junctions appear to be involved in phosphoinositide hydrolysis, and sequence variation can result in loss-or gain-of-function (24 -26). Mutations of the conserved A623 in the C-terminal junction of the i3 loop to lysine or glutamic acid was reported to result in loss of TSH-stimulated IP formation leaving cAMP accumulation unaltered (27). TSH-induced cAMP accumulation, however, was affected only by mutations in the C-terminal region of the i3 loop (24 -26). Characterization of mutant receptors harboring deletions in the i3 loop confirmed the importance of this particular region for receptor/G protein interaction. In the glucagon (28), glucagon-like peptide-1 (29), and the muscarinic acetylcholine receptor (30), trimmed i3 loops resulted either in severely impaired or unaltered signaling abilities.
We recently identified a 9-amino acid deletion (amino acid positions 613-621) within the i3 loop of the TSHR in a toxic thyroid nodule (31). In contrast to functional properties of all previously described deletion mutants in the corresponding region of other GPCRs, the TSHR deletion mutant displayed constitutive activity. This finding prompted us to further evaluate the role of the TSHR i3 loop for G protein coupling. Thus, we generated various TSHR deletion mutants within the i3 loop varying in length and location and characterized their functional profile.

EXPERIMENTAL PROCEDURES
Generation of Mutant TSHR Genes-To characterize functional properties of different deletions within the i3 loop of the TSHR (see Fig. 1), mutations were created by employing standard PCR mutagenesis techniques (32) using the human TSHR expression plasmid, TSHR-pSVl (33), as a template. PCR fragments containing the mutations were digested and used to replace the corresponding Eco81l/Eco91l fragment in the TSHR-pSVl vector. The identity of the various constructs and the correctness of all PCR-derived sequences were confirmed by restriction analysis and dideoxy sequencing with thermosequenase and dye-labeled terminator chemistry (Amersham Pharmacia Biotech).
Cell Culture and Transient Expression of Mutant TSHRs-COS-7 cells were grown in Dulbecco's modified Eagle's medium supplemented with 10% fetal bovine serum, 100 units/ml penicillin, and 100 g/ml streptomycin (Life Technologies, Inc.) at 37°C in a humidified 7% CO 2 incubator. For cAMP and radioligand binding assays, the cells were transfected using a DEAE-dextran method (34). In brief, 2 ϫ 10 6 COS-7 cells were seeded into 100-mm dishes and transfected with different TSHR deletion constructs (5 g/dish). Approximately 24 h after transfection, cells were split into 12-well plates at a density of 2 ϫ 10 5 cells/well. Functional assays were performed 48 h after transfection. For the measurement of agonist-induced phosphoinositide hydrolysis, COS-7 cells were transfected in 12-well plates (3 ϫ 10 5 cells/well) using LipofectAMINE (Life Technologies, Inc.) according the manufacturer's instructions. The cells were used for an IP accumulation assay 48 h after transfection.
Radioligand Binding Assays-Transfected COS-7 cells were washed once with Hank's solution without NaCl containing 280 mmol/liter sucrose, 0.2% bovine serum albumin (Sigma), and 2.5% low fat milk (19). Thereafter, cells were incubated in the same medium in the pres-ence of 140,000 -160,000 cpm of 125 I-bovine thyrotropin (bTSH) (25 Ci/g, 40 units/mg, Brahms Diagnostica) and the appropriate concentrations of nonlabeled bTSH (Sigma) at room temperature for 4 h. Cells were subsequently solubilized with 1 N NaOH after two washes with Hank's solution. Bound radioactivity was determined in a ␥-counter. bTSH, TSHR concentrations, and K D values are expressed as milliunits/ ml. Data were analyzed assuming a one-site binding model using the fitting module of SigmaPlot 2.0 for Windows (35).
cAMP Accumulation Assays-For cAMP assays cells were washed once in serum-free Dulbecco's modified Eagle's medium, followed by a preincubation with the same medium containing 1 mM 3-isobutyl-1methylxanthine (Sigma) for 20 min at 37°C in a humidified 7% CO 2 incubator. Subsequently, cells were stimulated with appropriate concentrations of bTSH for 1 h. Reactions were terminated by aspiration of the medium and addition of 1 ml 0.1 N HCl. Supernatants were collected and dried. cAMP content of the cell extracts was determined with a commercial kit (Amersham Pharmacia Biotech) according to the manufacturer's instructions.
Stimulation of Inositol Phosphate Formation-Transfected COS-7 cells were incubated with 2 Ci/ml [myo-3 H]inositol (18.6 Ci/mmol, Amersham Pharmacia Biotech) for 18 h. Thereafter, cells were washed once with serum-free Dulbecco's modified Eagle's medium without antibiotics containing 10 mM LiCl. TSH-induced increases in intracellular inositol phosphate levels were determined by anion exchange chromatography as described previously (36).

RESULTS
A nine-amino acid deletion (⌬613-621) within the i3 loop of the TSHR (Fig. 1) resulted in constitutive activation of the cAMP signaling cascade reflected by a 5-fold increase in basal cAMP levels in conjunction with a markedly blunted maximal cAMP response (approximately 60% of wt receptor level) after stimulation with 100 milliunits/ml of bTSH (see Figs. 3 and 4). Taking this naturally occurring deletion mutant as a point of reference, we generated further mutants ( Fig. 1) to clarify the relative importance of distinct amino acids as well as of the extent and the location of the deletion for constitutive TSHR activity. All mutant receptors with deletions of more than one amino acid showed reduced cell surface expression (Table I). This was particularly prominent for the ⌬609 -621 and ⌬609 -624 deletion constructs (Table I), indicating that deletions in the i3 loop can severely interfere with correct folding and trafficking of the receptor to the plasma membrane. The TSHR has been described to be constitutively active, and elevated basal cAMP levels can be observed after transient transfection of COS cells (37). We confirmed this finding by comparing basal cAMP levels determined in COS-7 cells transfected with identical amount of either the human V2 vasopressin or the TSH receptor (see Fig. 3). Several TSHR deletion mutants (⌬609 -621, ⌬613-624, ⌬609 -624, and ⌬618 -624) showed decreased basal cAMP levels most likely due to decreased cell surface expression levels.
Deletion of Conserved and Nonconserved Regions within ⌬613-621-Sequence comparison within the family of glycoprotein hormone receptors reveals that the ⌬613-621 deletion can be subdivided into a highly conserved C-terminal and a less conserved N-terminal portion (Fig. 2). To test for differential functions of these two portions with regard to TSHR activation, we created the mutant receptors ⌬618 -621 and ⌬613-617 with deletions of 4 conserved or 5 nonconserved amino acids, respectively. Both mutant receptors remained constitutively active, although at levels lower than the initial ⌬613-621 mutant (Fig. 3). Receptor density (B max ), K D values, and cAMP levels after maximal bTSH stimulation were comparable between ⌬618 -621 and ⌬613-621 ( Fig. 4; Table I). When compared with the wt receptor, deletion of the less conserved Nterminal amino acids (⌬613-617) had only minor effects on maximal TSH-stimulated cAMP accumulation (Table I). In accord with the latter findings, the ⌬613-617 mutant showed a slightly increased cell surface expression compared with mutants ⌬613-621 and ⌬618 -621 (Table I).
N-and C-terminal Extension of the Deletion ⌬613-621-To investigate the functional importance of the size of the deletion, we generated three mutant receptors with expanded deletions. Only mutant ⌬613-624, the smallest deletion within this series of mutants, displayed expression levels of a similar order of magnitude as the initial mutant ⌬613-621 (Table I). TSH responsiveness in terms of cAMP accumulation, however, of this C-terminally extended mutant was markedly blunted (Fig.  3). On the contrary, mutant ⌬609 -621 located in the central and N-terminal portion of the i3 loop, responded to TSH chal-lenge with an increase in cAMP levels comparable to ⌬613-624 despite a profoundly diminished receptor density (Fig. 3, Table  I). Further extension of the deletion (⌬609 -624) led to an accompanying decrease in cell surface expression and hormonal responsiveness.
Shifting ⌬613-621 within the i3 Loop-Having shown that the length of the deletion was a crucial parameter defining functional consequences, we addressed the issue as to whether the location of the deletion within the i3 loop would also be of importance. To this end, we created two mutant receptors lacking 7 and 9 amino acids, respectively, in regions adjacent to the initial ⌬613-621 deletion. Mutant ⌬618 -624 in which the deletion is shifted toward the C terminus of the i3 loop was characterized by a receptor density (Table I, Fig. 4) and a maximal cAMP response comparable to mutant ⌬613-621, yet did not display constitutive activity (Figs. 3 and 4). Like deletion mutant ⌬613-624, ⌬618 -624 was also devoid of the critical amino acids 622-624 in the conserved C-terminal part of the i3 loop. On the contrary, deletion mutant ⌬609 -617 located in the central and N-terminal portion of the i3 loop was effectively inserted into the plasma membrane and displayed prominent constitutive activity (Table I, Fig. 4). Maximal hormone-   (Figs. 3 and 4). Deletion of Single Amino Acids within ⌬613-621-To study the effects of single amino acid deletions within the i3 loop on TSHR function, we deleted individual amino acids in the conserved and nonconserved portions of ⌬613-621. When expressed in COS-7 cells, mutants ⌬615 and ⌬616 located within the less conserved portion of the i3 loop did not differ from the wt receptor in terms of receptor density, K D values, basal and TSH-stimulated cAMP levels (Table I, Fig. 3). A deletion of Lys 618 situated in the C-terminal highly conserved portion of ⌬613-621 did not have significant impact on the functional properties of the receptor (Table I, Fig. 3). The small decrease of K D values was most probably a consequence of a reduced receptor density (Table I). Deletion of Asp 619 resulted in profound constitutive activity of the mutant receptor. This result was unexpected because a mutant TSHR characterized by a deletion of Asp 619 in conjunction with a T620S substitution has previously been reported to be functionally indistinguishable from the wt receptor (38).
Effect of Deletion Mutants on Phosphoinositide Hydrolysis-In addition to the G s /adenylyl cyclase system the activated human TSHR is also known to stimulate phospholipase C activity (39). To address the question whether the various deletion mutations would affect the IP signaling pathway, we tested all receptor mutants for basal and hormone-induced IP production. None of the TSHR mutants constitutively stimulated the IP signaling pathway (data not shown). As shown in Table I, bTSH-induced IP accumulation closely correlated with the relative maximal increase in cAMP levels. Deletion mutations (⌬609 -621, ⌬613-624, and ⌬609 -624) that were characterized by low plasma membrane expression levels were unable to signal to phospholipase C. DISCUSSION A 9-amino acid deletion within the i3 loop of the TSHR (⌬613-621) leading to constitutive receptor activity prompted us to systematically analyze the effect of deletion mutations within the i3 loop on TSHR function. Our major finding is that various deletions in the i3 loop result in ligand-independent receptor activation as long as certain critical C-terminal amino acids are not affected. We show that maximal constitutive activity correlates with an optimal length of the deleted receptor portion. In contrast to previous assumptions, these data entertain the notion that several clusters of amino acid in the TSHR's i3 loop are not required for G-protein activation. However, the i3 loop is an important structural determinant to safeguard receptor folding, trafficking and activation by agonist.
First we asked the question whether the absence of distinct critical amino acids within the original deletion would be responsible for constitutive activity. Sequence comparison revealed that the four C-terminal amino acids within ⌬613-621 are highly conserved within the family of glycoprotein hormone receptors (Fig. 2). Therefore, we constructed two smaller deletions comprising either conserved C-terminal or non-conserved N-terminal amino acids (⌬618 -621 and ⌬613-617, respectively). Both mutant receptors retained constitutive activity, albeit at reduced levels compared with ⌬613-621. The observation that two separate portions of the original deletion each caused constitutive activity, let us put forward the proposal that a deletion per se was responsible for ligand-independent receptor activation. A reduction in the size of the deletion led to a decreased constitutive receptor activity irrespective of the exact position of the amino acids deleted.
Alternatively, we examined the functional consequences of N-and C-terminal extensions of the original deletion ⌬613-621. Expanded deletions led to a marked decrease in cell surface expression of the mutated receptors. These results indicate that the length of the loop connecting TM5 and TM6 cannot be reduced below a critical size, which would interfere with correct insertion of the receptor into the cell membrane. It is worth mentioning, however, that even large deletions did not completely abrogate hormone-stimulable cAMP formation and cell surface expression. Interestingly, mutants ⌬609 -621 and ⌬613-624 showed a similar cAMP response, although the ⌬613-624 construct was expressed at 5-6-fold higher levels indicating that the ⌬613-624 deletion targeted amino acids critically required for productive G s coupling. Indeed, amino acids 621-625 form a conserved B-X-X-B-B motif with B denoting basic and X nonbasic amino acids (Figs. 1 and 2). All glycoprotein hormone receptors share this amino acid cluster at corresponding positions within the i3 loop (40). The substitution of Ala 623 within this amino acid cluster with different amino acids leads to constitutive activation of the cAMP cascade (41) or to a selective loss of IP signaling (27) of the TSHR further supporting the involvement of this region in G protein activation. Moreover, similar motifs were identified as structural determinants for G i and G s coupling in other heptahelical receptors (30,42).
To assess the importance of the location of deletions within the i3 loop for TSHR function, the original deletion was shifted N-and C-terminally by three amino acids. The N-terminally shifted deletion mutant (⌬609 -617) showed strong constitutive activity and a maximal stimulation comparable to the wt receptor. A deletion within the i3 loop of the glucagon receptor comparable in length and location to ⌬609 -617, resulted in an attenuated glucagon-induced cAMP response (28). This N-terminal deletion mutant showed a very low cell surface expression and consequently, low agonist-induced cAMP accumulation. These results are at odds with our results obtained with deletion mutant ⌬609 -617 of the TSHR. However, receptor regions involved in G protein coupling vary in location and sequence between different GPCRs (3), and therefore, similar structural modifications at corresponding locations do not necessarily have to yield identical functional effects in different GPCRs.
In contrast to the latter results, a C-terminally shifted deletion mutant (⌬618 -624) showed no constitutive activity. Basal values for cAMP accumulation were lower than for wt TSHR, and the maximal response was strongly attenuated. Interestingly, this effect is in good agreement with the functional properties of a comparable deletion mutant generated in the glucagon receptor (28). In the case of the TSHR, the decreased cAMP response was not caused by a decrease in cell surface expression because expression levels were comparable to the original deletion ⌬613-621 and also to the shorter deletion mutants ⌬613-617 and ⌬618 -621 which all displayed constitutive activity. A conspicuous difference between these four deletion mutants is the destruction of the B-X-X-B-B motif in ⌬618 -624 indicating the necessity of this amino acid cluster for effective G s coupling. It should not go unnoticed, however, that although the destruction of the conserved motif attenuated constitutive activity, the ability of the receptor to couple to G s was not utterly precluded as shown by reproducible increases in intracellular cAMP levels after TSH stimulation.
As point mutations within the original deletion ⌬613-621 have been reported (18,41) to result in constitutive activity, we analyzed the influence of distinct single amino acid deletions within the i3 loop on TSHR function. For this purpose, we generated four mutant receptors with deletion of Pro 615 , Gly 616 , Lys 618 , and Asp 619 , respectively. Only the ⌬619 mutant showed constitutive activity. A D619G substitution also leads to constitutive activity of the TSHR (18) indicating an important role of this amino acid for receptor activation. Decreasing the length of the i3 loop by one amino acid does not by itself lead to constitutive activity as exemplified by the deletion mutants Pro 615 , Gly 616 , and Lys 618 . Therefore, deletions of critical amino acids may have an effect similar to activating substitutions of these residues in that critical interhelical, intrahelical, or intraloop bonds are disrupted, thereby releasing a structural constraint in the receptor and exposing critical activating receptor domains for interaction with G proteins. Interestingly, the deletion of Asp 619 in conjunction with a T620S substitution FIG. 5. Proposed mechanism of TSHR activation by deletions in the third intracellular loop. A, a polypeptide derived from the lower portion of TM6 of the LHR is capable of activating G s (43). In the nonstimulated TSHR this region is prevented from interacting with G s because it is concealed in the cell membrane. B, insertions of additional amino acids in the middle portion of TM6 of the m2 muscarinic receptor resulted in a constitutive activation presumable by a relative movement of TM6 toward the cytoplasm (44). C, deletions in the i3 loop of the TSHR possibly lead to a relative movement of transmembrane regions exposing the lower part of TM6 toward the cytoplasm thus enabling G s to couple. has been reported not to lead to constitutive activity (38).
GPCRs are assumed to exist in equilibrium between inactive and active states, and only the active state effectively interacts with G proteins. At the structural level, the molecular events accompanying the functional transition from inactive to active states are largely unknown. Recent studies with spin-labeled rhodopsin (6) or rhodopsin molecules carrying engineered metal-ion-binding sites (5) emphasized the importance of rigid body movements of helices relative to one another during the activation process. Furthermore, a model of the LHR suggested possible structural and functional effects of constitutively activating mutations (43). A tightly packed hydrophobic cluster between the intracellular halves of TM5 and TM6 is postulated to be essential for receptor quiescence. According to the model, activating mutations would then disrupt the hydrophobic packing and disturb the relative positioning of TM5 and TM6 in the plasma membrane (43). Our results with various deletion mutants of the TSHR are fully compatible with such models and may provide further insight into the molecular mechanism of GPCR activation.
Based on our results we propose a novel model of TSHR activation. We show that shortening of the TSHR's i3 loop rather than deletion of distinct amino acids is responsible for constitutive activity. This assumption is based on the fact that several deletion mutations which do not overlap in the deleted sequence display constitutive activity. Therefore, it is very likely that a similar mechanism accounts for shifting the receptor into the active conformation. Additionally, the extent of constitutive activity appears to be dependent on the length of the deletion. It can not be excluded, however, that a loss of one or more distinct amino acids within a TSHR deletion mutant is responsible for constitutive activation.
The i3 loop has previously been thought to directly mediate receptor/G protein interaction. Surprisingly, none of the deletion mutants examined showed a complete abolishment of G s coupling. It is likely that a decrease in the length of the i3 loop will affect the conformation of the adjacent transmembrane domains. The N-terminal part of TM6 appears to be most important for LHR activation. There is evidence that a peptide consisting of the wt sequence of the lower portion of TM6 of the LHR can activate adenylyl cyclase (44). Within this receptor region there is more than 90% amino acid identity among glycoprotein hormone receptors. Thus, the N-terminal portion of TM6 is capable of activating G s , if freed from all conformational constraints imposed by neighboring receptor sequences. Agonist binding (Fig. 5A), activating point mutations, and a shortening of the i3 loop (Fig. 5C) may all lead to a relative movement of TM5 to TM6, most likely allowing a relative movement of TM6 toward the cytoplasm, thus enabling critical transmembrane sequences to interact with G s . A similar hypothesis has been derived from alanine-insertion studies (Fig.  5B) with the m2 muscarinic receptor (45). The mechanism of TSHR activation proposed in Fig. 5 is, however, limited to the mutations made and tested in this study. Since at present no data are available whether insertion mutations within TM6 of a glycoprotein hormone receptor may display constitutive activity as observed with the m2 muscarinic receptor a direct comparison of receptor activation between these two GPCRs remains speculative.
In conclusion, we performed a thorough analysis of a naturally occurring mutant of the human TSHR and developed a novel model describing TSHR activation at the molecular level. Additional studies with the TSHR will have to show whether this model is correct and whether it can be extended to all glycoprotein hormone receptors or even to other GPCRs more distantly related.