A Low Molecular Weight Agonist Signals by Binding to the Transmembrane Domain of Thyroid-stimulating Hormone Receptor (TSHR) and Luteinizing Hormone/Chorionic Gonadotropin Receptor (LHCGR)*

Many cognate low molecular weight (LMW) agonists bind to seven transmembrane-spanning receptors within their transmembrane helices (TMHs). The thienopyrimidine org41841 was identified previously as an agonist for the luteinizing hormone/chorionic gonadotropin receptor (LHCGR) and suggested to bind within its TMHs because it did not compete for LH binding to the LHCGR ectodomain. Because of its high homology with LHCGR, we predicted that thyroid-stimulating hormone receptor (TSHR) might be activated by org41841 also. We show that org41841 is a partial agonist for TSHR but with lower potency than for LHCGR. Analysis of three-dimensional molecular models of TSHR and LHCGR predicted a binding pocket for org41841 in common clefts between TMHs 3, 4, 5, 6, and 7 and extracellular loop 2 in both receptors. Evidence for this binding pocket was obtained in signaling studies with chimeric receptors that exhibited improved responses to org41841. Furthermore, a key receptor-ligand interaction between the highly conserved negatively charged E3.37 and the amino group of org41841 predicted by docking of the ligand into the three-dimensional TSHR model was experimentally confirmed. These findings provide the first evidence that, in contrast to the ectodomain binding of cognate ligands, a LMW agonist can bind to and activate glycoprotein hormone receptors via interaction with their transmembrane domain.

The thienopyrimidine org41841 (Fig. 1A) was identified previously as the first low molecular weight (LMW) agonist for LHCGR (5). Interestingly, org41841 did not compete with 125 I-labeled LH for binding to LHCGR, indicating distinct binding sites for the two ligands (5). Similarly, a potent LMW antagonist for the FSHR did not displace 125 I-labeled FSH (6). These results suggest that org41841 or other LMW ligands may not bind to the extracellular leucine-rich domain like the glycoprotein hormones but rather within the transmembrane domain of these receptors (5,6). Although it is known that small ligands, such as retinal, catecholamines, and small peptides interact with the transmembrane cores of rhodopsin and other 7TMRs (7)(8)(9)(10), direct evidence for a binding site for LMW ligands in GPHRs has not been presented. It is noteworthy that for the glucagon-like peptide 1 receptor and for the metabotropic glutamate receptor determinants of binding of non-peptide small ligands were mapped to their extracellular amino termini (11,12). Therefore, it cannot be assumed that LMW ligands bind to the transmembrane domain of 7TMRs.
In this study, computer-based docking using three-dimensional models of LHCGR and TSHR identified a putative binding pocket for org41841 in a cleft between the transmembrane helices (TMHs) close to ECL2. Validation of this proposed binding site provides the first evidence that a LMW agonist can activate GPHRs by interaction with the transmembrane core.

Construction of Vectors and
Site-directed Mutagenesis of TSHR-cDNA for human TSHR was amplified by PCR from hTSHR-pSVL (13) and inserted into the pcDNA3.1(Ϫ)/hygromycin vector using restriction sites XhoI and BamHI. cDNA for human LHCGR was amplified by PCR from hLHR-pGS5 (14) and was inserted into the pcDNA3.1(ϩ)/hygromycin vector using restriction sites BamHI and XhoI. Constructs were confirmed by sequencing (MWG Biotech). Mutations were introduced into hTSHR-pcDNA3.1 via the QuikChange sitedirected mutagenesis kit (Stratagene). PCR products containing mutations were digested with Eco81I and Eco91I (MBI Fermentas) and used to replace the analogous Eco81I/Eco91I fragment in wild type hTSHR-pcDNA 3.1. All constructs were verified by sequencing (MWG Biotech).
Cell Culture and Transfection-HEK-EM 293 cells were grown in Dulbecco's modified Eagle's medium supplemented with 10% fetal bovine serum, 100 units/ml penicillin, and 10 g/ml streptomycin (Invitrogen) at 37°C in a humidified 5% CO 2 incubator. Cells were transiently transfected in 24-well plates (7.5 ϫ 10 4 cells per well) with 0.4 g of DNA/well using FuGENE TM 6 reagent (Roche Applied Science) according to the manufacturer's protocol.
Determination of Cell Surface Expression-After transfection, cells were cultured for 48 h, harvested using 1 mM EDTA/1 mM EGTA in phosphatebuffered saline, and transferred to Falcon 2058 tubes. Cells were washed once with phosphate-buffered saline containing 0.1% bovine serum albumin and 0.1% NaN 3 (binding buffer), incubated for 1 h with a 1:200 dilution of mouse anti-human TSHR antibody (Serotec) in binding buffer, washed twice, and incubated for 1 h in the dark with a 1:200 dilution of an Alexa Fluor 488-labeled F(abЈ) 2 fragment of goat anti-mouse IgG (Molecular Probes) in binding buffer. Before FACS analysis (FACS Calibur, BD Biosciences), cells were washed twice and fixed with 1% paraformaldehyde. Receptor expression was estimated by fluorescence intensity and transfection efficiency was estimated from the percentage of fluorescent cells.
Determination of Intracellular Cyclic AMP Accumulation-Transfected cells were cultured for 48 h before incubation for 1 h in serum-free Dulbecco's modified Eagle's medium containing 1 mM 3-isobutyl-1-methylxanthine (Sigma) and bovine TSH (0 -1.8 M) (Sigma) or human LH (0.1-1000 ng) or org41841 (0 -100 M) in a humidified 5% CO 2 incubator. Highly purified human LH was purchased from Dr. A. Parlow and the NIDDK National Hormone and Pituitary Program. Following aspiration of the medium, cells were lysed using lysis buffer 1 of the cAMP Biotrak Enzymeimmunoassay System * This work was supported by the Intramural Research Program of the NIDDK, National Institutes of Health. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. □ S The on-line version of this article (available at http://www.jbc.org) contains supplemental Table 1 and Fig. S1. 1 These authors contributed equally to this work. 2  (Amersham Biosciences). The cAMP content of the cell lysate was determined using the manufacturer's protocol. Data were analyzed using GraphPad Prism 4 for Windows.

Synthesis of org41841 (N-tert-Butyl-5-amino-4-(3-methoxyphenyl)-2-(methylthio)thieno[2,3-D]pyrimidine-6-carboxamide)-
The synthesis of org41841 was performed as published (5). Analysis by C8 reversed phase liquid chromatography-mass spectrometry using a linear gradient of H 2 O with increasing amounts of CH 3 CN (0 -17 min, 30% 3 70% CH 3 CN at a flow rate of 1 ml/min: t R 13.5 min) found greater than 95% purity by peak integration. 1  The designation of the amino acids in the transmembrane domain was based on the Ballesteros-Weinstein nomenclature (17). In rhodopsin, interactions of the side chains of two consecutive threonines with the helical backbone of the preceding residues caused a structural bulge in TMH2. In TSHR and LHCGR, neither consecutive threonines nor prolines exist in TMH2 indicating a regular ␣-helix, which extends to residue position 2.71. At TMH5, a minor change of orientation (10 to 15 degrees twist) of the N-terminal half of TMH5 due to the lack of proline compared with rhodopsin was generated. Consequently residue position 5.42 was oriented toward the interior of the receptor. Loops were added by best fit and homology to fragments of other proteins (from Protein Data Bank). Gaps of missing residues in the loops of the template structure were closed by the "Loop Search" tool implemented in Sybyl6.8 (Tripos Inc., St. Louis, MO). The sheet-like fold of ECL2 and its general localization were kept as in rhodopsin based on rhodopsin structureconsistent results about different accessibility of two CC chemokine receptor 5 (CCR5) antibodies, each specific for the two different ␤-strand epitopes of ECL2 of CCR5 (18 -20). The stability of the ligand orientation in the binding site was studied by molecular dynamics runs in a water-vacuum-water box system (21) without any restraints (2 ns, periodic boundary box, charges neutralized by adding chlorine ions) using AMBER 7.0 (22). Quality and stability of the model were validated by checking the geometry by PROCHECK (23) and during the Molecular Dynamics run (overall backbone root mean square deviation 1.8 Å). Molecular Connolly surfaces were generated for the interior transmembrane residues using the MOLCAD module of the TRIPOS package.

RESULTS AND DISCUSSION
Activation of TSHR and LHCGR signaling was determined by measuring accumulation of intracellular cAMP. org41841 (Fig. 1A) acts as a partial agonist of LHCGR with an EC 50 of 220 nM (95% confidence interval: 140 -360 nM) and a maximum response 34 Ϯ 2.1% of that stimulated by LH. Due to the high sequence homology between TSHR and LHCGR within the transmembrane core, we tested org41841 and found that it is also a partial agonist for TSHR with a maximum response 23% of TSH but with lower potency (EC 50 : 7700 nM) than for LHCGR (Table 1). Amino acid alignments between TSHR and LHCGR (Fig. 1B) and computer-based docking using three-dimensional models identified a putative binding pocket for org41841 in a cleft between TMHs 3, 4, 5, 6, and 7 close to ECL2 in both receptors (Fig. 1C). The binding cleft is smaller for TSHR than for LHCGR and two transmembrane helical residues, F5.42 (LHCGR: T5.42) and Y6.54 (LHCGR: F6.54), and residue L570 in ECL2 (LHCGR: F515) differ in the central cores of the predicted binding pockets (Fig.  1C). Furthermore, the binding pocket of TSHR is characterized by strong hydrophobic and bulky residues at the junctions between TMH4/ECL2, ECL2/ TMH5, and TMH6/ECL3 that form an extracellular cover over the binding cleft. In contrast, generally less hydrophobic and/or bulky residues are involved at LHCGR. Taken together, these findings might explain the higher potency of org41841 for LHCGR.
The putative binding pocket is covered by ECL2 and adjacent residues of the three junctions of TMH4/ECL2, TMH5/ECL2, and TMH6/ECL3, which contain six different residues between TSHR (I560 in ECL2, P5.34, A5.36, L5.37, A5.38, and I6.59) and LHCGR (V505 in ECL2, T5.34, S5.36, Q5.37, V5.38, and A6.59) (Fig. 1B). Therefore, a construct (M9) was engineered with six additional mutations replacing TSHR residues with the corresponding residues of LHCGR from the region that covers the binding cleft. Thus, M9 contained nine substitutions: I560V (LHCGR: V505), L570F (LHCGR: F515), P5.34T, A5.36S, L5.37Q, A5.38V, F5.42T, Y6.54F, and I5.59A. Mutant M9 was expressed on the cell surface at near 100% of TSHR. Activation of this construct by TSH revealed a small reduction in potency (EC 50 ϭ 9.3 nM), but the maximal response to the native ligand was not affected (supplemental Table 1, Fig. 1D). Importantly, M9 responded to org41841 with an improved EC 50 of 2700 nM and a greatly improved efficacy for signaling to 99% of the maximal value observed for TSH stimulation of TSHR (Table 1, Fig. 1D). Thus, org41841, which is only a partial agonist with 23% of TSH activity at TSHR, acts as a full agonist for M9. This provides further support for the proposed binding pocket. In M9, all but one of the 6 residues that were changed in the region that covers the binding cleft were either less hydrophobic and/or less bulky than in TSHR and M9 lacks a proline (P5.34T) known for its impact on backbone conformation. Consequently, the three TMH/ECL junctions in M9 are differently packed than in TSHR and provide more space for the ligand and less hydrophobic properties at the extracellular cover of the small ligand-bind- Initial docking studies of org41841 into the three-dimensional model of TSHR by a constrained simulated annealing process allowing C␣ atoms only small conformational changes without loss of the overall geometry of the receptor delineated the conformational space and orientations of the ligand in the pocket. To find optimal ligand-receptor interaction, a further restraint was used with respect to the experimentally determined contact point E3.37 (see below) and the potential complementary NH 2 group of the ligand. This docking of org41841 followed by molecular dynamics simulations allowed two preferred binding orientations within the binding cleft (supplemental Fig. S1). Common to both is the pivotal and experimentally supported role of the anchor-point E3.37 for hydrogen-bonding interaction with the ligand. In docking version A, the t-butyl group is oriented toward the transmembrane core between residues M6.48 and Y7.42 and the aryl-meta-methoxy group points toward F5.42. In docking version B, the t-butyl and aryl-meta-methoxy groups are oriented in opposite directions from their positions in A.
The predicted hydrogen bonding interaction between the amino group of the ligand and residue E3.37 in the receptor was tested. A mutant receptor with an alanine substitution at the highly conserved position E3.37 exhibited cell surface expression (93%) and response to TSH (Fig. 2, left) similar to TSHR. However, mutant E3.37A was unable to form the predicted hydrogen bond with and was not activated by org41841 (Fig. 2, right). The finding that loss of the hydrogen bond acceptor in mutant E3.37A disrupts receptor activation strongly supports the critical role of the predicted hydrogen bond between org41841 and E3.37 and thereby adds critical validation of the model.
In conclusion, a bioinformatic comparison of the related receptors for TSH and CG/LH allowed the successful prediction that a LMW ligand for one receptor would also activate the other. Utilizing the sequence differences and the fact that org41841 was less effective at TSHR than at LHCGR, this comparative approach provided experimental evidence for the proposed binding pockets within the TMHs, since changing TSHR residues to the corresponding residues of LHCGR in the proposed pocket consistently improved the response to org41841. Our results provide the first report of a LMW ligand for TSHR and direct evidence that this ligand binds within the transmembrane core to activate GPHRs.
Indeed, org41841 activates the TSHR with too low a potency to be clinically useful. Further experiment-based refinements of both the binding pocket and the orientation of the ligand within the pocket may guide the synthesis of new compounds by structure-based rational design and might lead to identification of highly potent LMW ligands for glycoprotein hormone receptors. LMW ligands of LHCGR and FSHR have the potential to become therapeutics for infertility treatment or oral contraception. It is noteworthy that in vivo efficacy of org41841 for LHCGR was demonstrated in an ovulation induction model (5) supporting the potential pharmacological utility of such synthetic ligands. Similarly, LMW antagonists of TSHR may have therapeutic potential for TSHR-mediated hyperthyroidism (24,25), while agonists might replace injected recombinant TSH in diagnostic screening for thyroid cancer (26).