Ligand selectivity of gonadotropin receptors. Role of the beta-strands of extracellular leucine-rich repeats 3 and 6 of the human luteinizing hormone receptor.

The difference in hormone selectivity between the human follicle-stimulating hormone receptor (hFSH-R) and human luteinizing hormone/chorionic gonadotropin receptor (hLH-R) is determined by their approximately 350 amino acid-long N-terminal receptor exodomains that allow the mutually exclusive binding of human follicle-stimulating hormone (hFSH) and human luteinizing hormone (hLH) when these hormones are present in physiological concentrations. The exodomains of each of these receptors consist of a nine-leucine-rich repeat-containing subdomain (LRR subdomain) flanked by N- and C-terminal cysteine-rich subdomains. Chimeric receptors, in which the structural subdomains of the hFSH-R exodomain were substituted with those of the hLH-R, showed a similar high responsiveness to human chorionic gonadotropin (hCG) and hLH as long as they harbored the LRR subdomain of the hLH-R. In addition, these chimeric receptors showed no responsiveness to hFSH. The LRR subdomains of the gonadotropin receptor exodomains are predicted to adopt a horseshoe-like conformation, of which the hormone-binding concave surface is composed of nine parallel beta-strands. Receptors in which individual beta-strands of the hFSH-R were replaced with the corresponding hLH-R sequences revealed that hCG and hLH selectivity is predominantly determined by hLH-R beta-strands 3 and 6. A mutant receptor in which the hFSH-R beta-strands 3 and 6 were substituted simultaneously with their hLH-R counterparts displayed a responsiveness to hCG and hLH similar to that of the wild type hLH-R. Responsiveness to hFSH was not affected by most beta-strand substitutions, suggesting the involvement of multiple low-impact determinants for this hormone.

Follicle-stimulating hormone (FSH) 1 and luteinizing hormone (LH) stimulate the FSH receptor (FSH-R) and LH recep-tor (LH-R), respectively, which are expressed in different target cells (1). In some species, chorionic gonadotropin (CG) is also able to stimulate the LH-R. The coordinated interplay between the complementary and specific actions of FSH and LH(/CG) is required to guarantee successful reproduction. The interaction between these gonadotropins and their respective receptors is highly specific, and there is virtually no cross-reactivity between hormones and heterologous receptors (i.e. high receptor selectivity) except for LH and CG, which both act on the LH-R (2,3).
The FSH-R and LH-R, together with the thyroid-stimulating hormone receptor, represent the structurally unique glycoprotein hormone receptor (GpHR) subfamily of the G proteincoupled receptor superfamily. Characteristically, GpHRs are composed of two approximately equally sized but functionally distinct domains: an extracellular N-terminal half (exodomain), which is responsible for the selective recognition and high-affinity binding of its corresponding hormone, and a typical G protein-coupled receptor domain, consisting of seven transmembrane ␣-helices, which transduces the specific signal of hormone binding to the exodomain across the membrane to activate intracellular signaling pathways (4). The GpHR exodomain is subdivided into three structural subdomains: an Nterminal cysteine-rich subdomain (NCR subdomain) followed by a nine contiguous imperfect leucine-rich repeat-containing subdomain (LRR subdomain) and a C-terminal cysteine-rich subdomain (CCR subdomain; see Figs. 1 and 2A).
LLR motifs have been recognized in a large variety of proteins, with repeats composed of 20 -29 amino acids each (5). Crystal structure analyses of some of these LRR-containing proteins, such as the porcine ribonuclease inhibitor (6) and Listeria internalins (7), revealed that each LRR forms a righthanded structural unit that is composed of a short ␤-strand and a helical segment, which are positioned nearly antiparallel to each other and are linked by short loops. Tandem arrays of LRR units form a one-to three-quarter donut-like structure with the consecutive ␤-strands forming a parallel ␤-sheet at the concave surface, whereas the alternating helices are aligned next to each other to form the convex side of the structure. The concave surface of the ribonuclease inhibitor interacts with ribonucleases using multiple contact points (6).
Based on sequence homology, the exodomains of GpHRs have been structurally modeled using the crystal structure of porcine ribonuclease inhibitor (8 -10). Consequently, the ␤-sheet at the inner circumference of the curved exodomain of GpHRs has been proposed to form the main hormone-binding site, with some additional hormone contact sites situated outside the LRR subdomain (11,12). Each of the LRR ␤-strands is composed of a highly conserved X 1 X 2 LX 3 LX 4 X 5 motif (Fig. 1), in which X indicates any amino acid and L refers to leucine, isoleucine, or other hydrophobic residues (5). The side chains of the conserved L residues are directed toward the helical segment and form the hydrophobic core of the LRR structure, whereas the side chains of the X residues are exposed to the surface of the presumed ligand-binding site (13)(14)(15).
Selective gonadotropin binding is determined exclusively by the exodomain of the LH-R and FSH-R. Studies using chimeric receptors revealed that the sequence NCR-LRR6 of the LH-R is important to confer LH/CG selectivity to the receptor, whereas the sequence NCR-LRR3 of the FSH-R, in conjunction with the FSH-R sequence LRR7-CCR, is important for FSH binding specificity (2,3). However, these results may be somewhat biased because the junctions of these chimeric proteins were introduced arbitrarily, depending mainly on the presence of common endonuclease restriction sites rather than considering the exact structural borders within the exodomain. Taking advantage of the current knowledge of the structural conformation of the exodomain of GpHRs, we examined in more detail the contribution of the NCR, LRR, and CCR subdomains of the human LH-R (hLH-R) exodomain in conferring human chorionic gonadotropin (hCG) and/or human LH (hLH) selectivity when placed in the context of a human FSH-R (hFSH-R) background. Furthermore, the contribution of individual (and combinations of) hLH-R ␤-strands in directing hCG and hLH (hCG/ hLH) selectivity to the receptor was studied systematically with mutant receptors generated by introducing hLH-R ␤-strands into the hFSH-R.
Construction of Mutant Receptor cDNAs-The cDNAs encoding the human FSH-R and LH-R, kindly provided by Dr. T. Minegishi (Gunma University School of Medicine, Maebashi, Japan) and Dr. E. Milgrom (Institut National de la Santé et de la Recherche Médicale, Paris, France), respectively, were subcloned into the pcDNA3.1/V5-His-TOPO expression vector (Invitrogen). To quantify receptor cell surface expression by enzyme-linked immunosorbent assay (ELISA), an HA epitope (Tyr-Pro-Tyr-Asp-Val-Pro-Asp-Tyr-Ala; derived from the influenza virus hemagglutinin) was introduced between the C termini of the signal peptide sequences and the N termini of the mature FSH-R (i.e. between Gly 17 and Cys 18 ) and the mature LH-R (i.e. between Leu 29 and Arg 30 ). HA epitope tagging did not significantly influence the ligand binding and signaling properties of the receptors compared with their wild type counterparts (data not shown).
HA-tagged receptor cDNAs were used as templates to generate hFSH-R and hLH-R exodomain chimeras using a fusion PCR-based FIG. 1. Amino acid sequence alignment of the exodomains of the hFSH-R and the hLH-R. Each exodomain was subdivided into three structural subdomains: the N-and C-terminal cysteine-rich subdomains (NCR subdomain and CCR subdomain, respectively), with conserved cysteine residues indicated by black boxes, and the LRR subdomain, consisting of nine consecutive LRR units. Conserved amino acid residues present in the LRRs and forming the hydrophobic core of the structural units are depicted in bold italic letters. The ␤-strand X 1 X 2 LX 3 LX 4 X 5 motifs (see text) that form the inner lining of the horseshoe-shaped LRR subdomain are indicated, and the ␤-strand amino acid residues are underlined. The X residues (see "Experimental Procedures"), which are directed to the surface of the protein and are thought to interact with the respective hormones, are indicated by gray boxes. Dashes indicate gaps introduced for optimal alignment. method (16). Briefly, 5Ј-and 3Ј-cDNA fragments were generated using overlapping primers (Invitrogen and Isogen) in combination with specific primers demarcating the cDNA insert. These PCR fragments were then fused in a self-primed PCR taking advantage of the introduced overlapping sequence. The fusion products were then PCR-amplified using the specific primers demarcating the cDNA insert. All PCRs were performed using the Advantage-HF PCR kit (Clontech). Four different exodomain chimeras ( Fig. 2A) were generated in which particular hFSH-R domains were substituted with either the entire exodomain of the hLH-R (generating the chimera LLL-hFSH-R), or with the LRR subdomain in combination with either the NCR subdomain or the CCR subdomain of hLH-R (LLF-hFSH-R and FLL-hLH-R), or with only the LRR subdomain of the hLH-R (FLF-hFSH-R). In a similar way, a set of mutant receptors was generated in which the residues indicated with X, and unique to the ␤-strand motifs (X 1 X 2 LX 3 LX 4 X 5 ) of the hLH-R, were introduced to replace residues on the homologous position in the hFSH-R (see Tables II, III, and V for the details of which residues were replaced). All cDNAs were subcloned into the pcDNA3.1/V5-His-TOPO expression vector and sequenced on automated ABI PRISM 310 and 377 DNA sequencers using Dye Terminator cycle sequencing chemistry in accordance with the manufacturer's instructions (Applied Biosystems).
Transient Expression of Wild Type, Chimeric, and Mutant Receptors-Human embryonic kidney (HEK-T 293) cells were cultured at 37°C under 5% CO 2 in culture medium (Dulbecco's modified Eagle's medium containing 2 mM glutamine, 10% fetal bovine serum, and 1ϫ antibiotic/antimycotic; all from Invitrogen). Transient transfections were performed in a 10-cm dish containing ϳ3.5 ϫ 10 6 cells, with 1 g of (wild type, chimeric, or mutant) HA-tagged receptor expression vector construct in combination with 10 g of a pCRE/␤-gal plasmid, using the modified bovine serum transfection method according to the instructions of the manufacturer (Stratagene) and as described previously (17). The pCRE/␤-gal plasmid consists of a ␤-galactosidase gene under the control of a human vasoactive intestinal peptide pro-moter containing five cAMP-response elements (18). "Empty" pcDNA3.1/V5-His vector was used for mock transfections. The next day, cells were collected and split into 96-well tissue culture plates (ϳ2 ϫ 10 5 cells/well) for ligand stimulation studies. In addition, duplicate aliquots (ϳ4.5 ϫ 10 5 cells/well) were transferred to poly-D-lysine (Sigma)-coated 24-well tissue culture plates for ELISA detection of receptor cell surface expression.
Detection of Ligand-induced cAMP Production-The receptor-mediated stimulation of cAMP-induced reporter gene activity was assayed according to Chen et al. (18) with minor modifications as described previously (17). Briefly, 2 days after transfection the cells were stimulated for 6 h with various concentrations of hFSH, hCG, and hLH in 25 l of Hepes-modified Dulbecco's modified Eagle's medium containing 0.1% bovine serum albumin and 0.1 mM 3-isobutyl-1-methylxanthine (all from Sigma). Ligand-induced changes in ␤-galactosidase activity (conversion of o-nitrophenyl-␤-D-galactopyranoside into o-nitrophenol) were measured at 405 nm and related to the forskolin-induced changes (10 M) in each 96-well plate. Therefore, the results are expressed as arbitrary units related to the forskolin-induced cAMP-mediated reporter gene activation. All experiments were repeated at least three times using cells from independent transfections, each performed in triplicate.
Receptor Binding Assay-Competition ligand binding assays were carried out on purified cell membranes from HEK-T 293 cells expressing (mutant or chimeric) receptors. Two days after transfection, HEK-T 293 cells were rinsed with Dulbecco's phosphate-buffered saline (Sigma), subsequently harvested in ice-cold Tris buffer (10 mM Tris-HCl, 5 mM MgCl 2 ⅐6 H 2 O, pH 7.4), and centrifuged at 200 ϫ g at 4°C for 10 min. The pellet was resuspended in ice-cold Tris buffer containing 250 mM sucrose and homogenized by 40 strokes in a Dounce homogenizer at 4°C, and the cell suspension was centrifuged at 15,000 ϫ g at 4°C for 30 min. The pellet was resuspended in Tris buffer. Cell membranes (ϳ50 g of protein) were incubated for 18 -20 h in 300 l of Tris buffer  Table I. supplemented with 0.1% bovine serum albumin at room temperature with 10,000 cpm of [ 125 I]hFSH (i.e. NEX 173 with a specific activity of 163 Ci/g; purchased from PerkinElmer Life Sciences) in the presence of increasing concentrations of unlabeled hCG or hFSH. The reaction was terminated by adding 500 l of ice-cold Tris buffer supplemented with 0.1% bovine serum albumin and subsequently centrifuged at 15,000 ϫ g at room temperature for 5 min. The supernatant was aspirated, and the radioactivity in the membrane pellet was determined in an LKB ␥-counter (PerkinElmer Life Sciences). All binding studies were performed in triplicate in two independent experiments.
ELISA Detection of HA-tagged Receptors on the Cell Surface-Aliquots of the cells used for signal transduction experiments were used for cell surface receptor quantification by ELISA, essentially as described previously (19). Briefly, 2 days after transfection, cells were fixed using 4% paraformaldehyde in phosphate-buffered saline at room temperature for 30 min. Next, the samples were blocked with 1% nonfat dried milk in 0.1 M NaHCO 3 at room temperature for 4 h and subsequently incubated with anti-HA high-affinity antibodies (a 1:200 dilution in Tris-buffered saline containing 0.1% bovine serum albumin) overnight at 4°C. The next day, the samples were incubated with peroxidaseconjugated goat anti-rat IgG (1:500 dilution in 1% nonfat dried milk in 0.1 M NaHCO 3 ) at room temperature for 2 h. The peroxidase activity was then visualized using the 3,3Ј,5,5Ј-tetramethylbenzidine liquid substrate system (Sigma) for ϳ10 min. Absorbance values (at 450 nm) of mock transfected cells were subtracted, and mutant or chimeric receptor expression values were expressed as the percentage of the wild type hFSH-R expression. All experiments were repeated at least three times using cells from independent transfections, each performed in duplicate.
Data Analysis-The ligand concentrations that induce half-maximal stimulation (i.e. EC 50 values) were calculated by fitting the cAMP data to sigmoidal dose-response curves using GraphPad PRISM3 (GraphPad Software, Inc.). Ligand binding affinities (K i ) were calculated from one-site competition curves using GraphPad PRISM3. All data shown in Tables I, II, IV, and V are presented as the calculated mean Ϯ S.E. based on observations derived from at least three independent experiments. Statistical comparisons were performed on the log(EC 50 ) or log(K i ) values using one-way analysis of variance followed by the Bonferroni test using GraphPad PRISM3.

Hormone Specificity of Chimeric Gonadotropin
Receptors-To exactly determine which parts of the exodomains confer ligand selectivity to the gonadotropin receptors, we divided their exodomains into three structural subdomains (i.e. NCR, LRR, and CCR subdomains). Next, we substituted individual or combinations of the hFSH-R subdomains with their homologous hLH-R subdomains ( Fig. 2A). We then tested these chimeric receptors for their ability to mediate hCG-and hLHinduced cAMP production in transiently transfected HEK-T 293 cells. Substitution of the entire exodomain or the NCR subdomain in conjunction with the LRR domain of the hFSH-R with the corresponding hLH-R sequences (i.e. Fig. 2A, LLL-hFSH-R and LLF-hFSH-R, respectively) resulted in chimeric receptors that were readily expressed at the cell surface (130 and 148% of wild type hFSH-R cell surface expression, respectively; Fig. 2B). Moreover, these chimeric receptors were functionally similar to wild type hLH-R (Fig. 2, C and D, and Table  I) in their responsiveness to hFSH, as well as to hCG and hLH. An equal responsiveness to hCG and hLH is indicated by hCG/ hLH. Chimeric hFSH-Rs, in which only the hLH-R LRR subdomain or the combination of the hLH-R LRR and CCR subdomain was introduced (i.e. Fig. 2A, FLF-hFSH-R and FLL-hFSH-R, respectively), were severely hampered in cell surface expression (both Ͻ1% of the wild type hFSH-R cell surface expression; see Fig. 2B). Nevertheless, both FLF-hFSH-R and FLL-hFSH-R responded significantly better to hCG and hLH than the wild type hFSH-R (p Ͻ 0.001), although they were significantly less efficient than wild type hLH-R, LLL-hFSH-R, and LLF-hFSH-R ( Fig. 2C and Table I). Similar to the hLH-R, all chimeric hFSH-R constructs were devoid of responsiveness to hFSH ( Fig. 2D and Table I).
Hormone Selectivity of Mutant ␤-Strand hFSH Receptors-To identify which of the nine ␤-strands of the hLH-R confers hCG/hLH selectivity to its LRR subdomain, all X residues of individual hFSH-R ␤-strand motifs (with the consensus sequence X 1 X 2 LX 3 LX 4 X 5 ) were mutated to their corresponding hLH-R residues (see Table II). Next, each of the mutant ␤-strand hFSH-Rs was transiently expressed in HEK-T 293 cells and analyzed for cell surface expression as well as hCG-, hLH-, and hFSH-dependent cAMP production. Alternatively, this strategy may also lead to the identification of ␤-strands of the hFSH-R that usually are involved in repelling hCG and/or hLH from its LRR subdomain. Most of the mutant ␤-strand hFSH-Rs were expressed at the cell surface, with levels ranging from 21 to 147% of wild type hFSH-R expression (Fig. 3, A  and D). However, hFSH-R/L␤4 and hFSH-R/L␤5 were expressed at lower levels (4 and 15%, respectively), whereas hFSH-R/L␤1 and hFSH-R/L␤7 were undetectable in the anti-HA tag ELISA (Fig. 3, A and D).
The hFSH-R/L␤2, hFSH-R/L␤3, hFSH-R/L␤5, hFSH-R/L␤6, and hFSH-R/L␤9 constructs responded to hFSH with EC 50 values (ranging from 0.064 to 0.232 ng/ml; p Ͼ 0.05) similar to that of wild type hFSH-R (Fig. 3, C and F, and Table II). This finding indicates that these hFSH-R ␤-strands, when replaced by hLH-R ␤-strands, are not crucial in conferring hFSH selectivity to the hFSH-R exodomain. In contrast, the responsiveness to hFSH was decreased significantly in hFSH-R/L␤8 (5fold; p Ͻ 0.05), hFSH-R/L␤4 (14-fold; p Ͻ 0.001), hFSH-R/L␤7 (130-fold; p Ͻ 0.001), and hFSH-R/L␤1 (Ͼ 2600-fold; p Ͻ 0.001), TABLE I Summary of the ligand-induced intracellular cAMP production in HEK-T 293 cells transiently transfected with wild type hFSH-R, wild type hLH-R, and chimeric hFSH-R constructs Cyclic AMP production upon stimulation with hCG, hLH, and hFSH was measured in HEK-T 293 cells, transiently cotransfected with various receptor constructs and a plasmid (pCRE/␤-gal) containing a ␤-galactosidase gene under control of a promoter containing five cAMP-response elements. The identity of the exodomain subdomains present in the chimeric hFSH-Rs is indicated by capital F or L for each of the three subdomains if they were derived from the hFSH-R or the hLH-R, respectively (see text indicating that these ␤-strands contain important molecular determinants necessary for specific hFSH recognition and binding (Fig. 3, C and F, and Table II). However, the reduced ligand responsiveness of some of these receptor mutants may also be related to their impaired cell surface expression. However, the transfection of increasing amounts of these receptor constructs (5 or 10 g instead of 1 g) did not affect receptor cell surface expression or ligand responsiveness (data not shown).
Although both hFSH-R/L␤3 and hFSH-R/L␤6 approached the responsiveness of the LLL-hFSH-R to hCG and hLH, their EC 50 values were still significantly higher (p Ͻ 0.001; Table II).
To confirm that the ligand responsiveness of a receptor represents its affinity for that ligand, displacement studies were performed for a selected number of mutant receptors that displayed a responsiveness to hFSH similar to that of wild type hFSH-R. K i values similar to those obtained for the wild type hFSH-R were found for hFSH-R/L␤3, hFSH-R/L␤6, and hFSH-R/L␤3,L␤6 ( Fig. 4B and Table III). The mutant receptors hFSH-R/L␤3 and hFSH-R/L␤6 displayed a significantly increased affinity for hCG compared with the wild type hFSH-R (p Ͻ 0.001), which is in accordance with the functional assay results. Likewise, the simultaneous introduction of hLH-R ␤-strands 3 and 6 into hFSH-R (hFSH-R/L␤3,L␤6) resulted in a further synergistic increase of the receptor affinity for hCG (p Ͻ 0.001; Fig. 4A and Table III).
Cys Residues in hLH-R ␤-Strands 4 and 5 Are Important for hLH-R Cell Surface Expression-In contrast to the situation in the hFSH-R, Cys residues are present on position X 4 of ␤-strands 4 and 5 of the hLH-R (Fig. 1). These Cys residues form a putative disulfide bridge (10) and have been shown to be involved in LH-R cell surface expression (20). Because the "unpaired" Cys residues in hFSH-R/L␤4 and hFSH-R/L␤5 may have caused reduced receptor cell surface expression (Fig. 3A) by inappropriate receptor folding and/or intracellular trafficking, each individual Cys was substituted with a Ser residue (hFSH-R/L␤4-C133S and hFSH-R/L␤5-C158S, respectively; Table II). In addition, the adjacent Cys pair was restored by substituting hFSH-R ␤4and ␤5-strands in conjunction with their corresponding hLH-R ␤-strands (Table II, hFSH-R/ L␤4,L␤5). The hFSH-R/L␤4-C133S and hFSH-R/L␤5-C158S receptors were equally well expressed at the cell surface as the wild type hFSH-R. The hFSH-R/L␤4,L␤5 receptor was expressed at ϳ56% of wild type hFSH-R levels ( Fig. 3G) but at higher expression levels than hFSH-R/L␤4 or hFSH-R/L␤5 (Fig. 3A). All three mutant receptor constructs (hFSH-R/L␤4-C133S, hFSH-R/L␤5-C158S, and hFSH-R/L␤4,L␤5) displayed a similar responsiveness to hFSH, hCG, and hLH as the wild type hFSH-R (Fig. 3, H and I, and Table II). Table I), and mutant hFSH-R constructs Cyclic AMP production upon stimulation with hCG, hLH, and hFSH was measured in HEK-T 293 cells, transiently cotransfected with various receptor constructs and a plasmid containing a ␤-galactosidase gene under control of a promoter containing five cAMP-response elements (pCRE/␤-gal). Residues that were introduced into the hFSH-R ␤-strands are underlined. Cell Surface Expression of the hFSH-R Is Sensitive to Mutations in ␤-Strands 1 and 7-The mutant hFSH-R/L␤1 and hFSH-R/L␤7 receptors displayed completely impaired cell surface expression (Fig. 3, A and D), together with a 2600-and 130-fold decreased hFSH responsiveness, respectively (Fig 3, C and F, and Table II), and no response to hCG or hLH (Fig. 3, B and E, and Table II). To test whether the observed reduction in responsiveness is related directly to reduced cell surface receptor expression, HEK-T 293 cells were transfected with 10, 100, and 1000 ng of wild type HA-tagged hFSH-R construct. Although these amounts of transfected construct appeared to be expressed at different levels (3 and 23% and set at 100%, respectively) on the cell surface, the production of cAMP in response to hFSH, hCG, and hLH stimulation appeared to be only slightly, although significantly, affected by the different receptor numbers (Fig. 5 and Table IV); a 33-fold reduction in cell surface receptor expression was accompanied by a less than 5-fold reduction in ligand responsiveness. To determine whether replacing hFSH-R ␤-strands 1 or 7 with the corresponding hLH-R ␤-strands either affected hFSH binding directly or severely disrupted the exodomain conformation, the hLH-R-specific residues in hFSH-R/L␤1 and hFSH-R/L␤7 were  Table II. ND, not detectable.  Table III. mutated. An Ala substitution of all X residues in ␤-strands 1 or 7 (i.e. hFSH-R/Ala␤1 and hFSH-R/Ala␤7 in Table V) impaired receptor surface expression as well as ligand responsiveness in a similar way as found for hFSH-R/L␤1 and hFSH-R/L␤7 ( Fig.  6 and Table V). In contrast, an Ala substitution of individual X residues in ␤-strands 1 and 7 did not affect the receptor responsiveness to hFSH, hCG, and hLH ( Fig. 6 and Table V), whereas their cell surface expression levels varied from 2 to 140% of wild type hFSH-R. DISCUSSION Mammalian glycoprotein hormones are bound with high affinity and high selectivity by the exodomains, which are highly related and are thought to share a similar structural conformation of their corresponding receptors. Nevertheless, sequences in the exodomains of these receptors have diverged sufficiently to generate the above mentioned selectivity toward their respective glycoprotein hormones. Taking advantage of their structural similarity, the hormone selectivity of the hLH-R was studied in the present study by substituting hFSH-R-specific sequences with their corresponding hLH-R se-quences. In contrast to previous studies (2,3,21), chimeric junctions were designed to coincide with the presumed structural borders of the extracellular subdomains (i.e. the NCR, LRR, and CCR subdomains, as well as the ␤-strands within the LRR subdomain).

TABLE II Summary of the ligand-induced intracellular cAMP production in HEK-T 293 cells transiently transfected with wild type hFSH-R, the chimeric LLL-hFSH-R (see
Receptor chimeras in which the extracellular NCR, LRR, and/or CCR subdomains of the hFSH-R were substituted by their homologous hLH-R counterparts indicated that the determinants involved in ligand selectivity are confined mainly to the LRR subdomain (cf. LLL-hFSH-R, LLF-hFSH-R, and hLH-R). Similar to other extracellular LRR-containing proteins, it is thought that the hydrophobic core of the LRR subdomain of gonadotropin receptors is protected from the solvent at both ends by its flanking NCR and CCR subdomains (5). Despite the conservation of four almost equally spaced Cys residues, which are presumably disulfide-bonded and essential for correct folding (10,20), the NCR subdomain of the hFSH-R was not compatible with the LRR subdomain of hLH-R as revealed by the severely impaired cell surface expression of the FLL-hFSH-R and FLF-hFSH-R chimeras.   Table IV.

Summary of the ligand-induced intracellular cAMP production in HEK-T 293 cells transiently transfected with different quantities of
wild type hFSH-R construct Cyclic AMP production upon stimulation with hCG, hLH, and hFSH was measured in HEK-T 293 cells, transiently cotransfected with various amounts of hFSH-R construct and a plasmid containing a ␤-galactosidase gene under control of a promoter containing five cAMP-response elements (pCRE/␤-gal). The EC 50 values presented are the calculated mean Ϯ S.E. of EC 50 values derived from at least three independent experiments. Human LH and hCG yielded a similar efficacy for all constructs as determined for every construct in at least two independent assays; for clarity, only the hCG data are shown. EC 50 wt/red values were calculated by dividing the EC 50 of 1 g of transfected hFSH-R with the EC 50 values of the 0.1 and 0.01 g of hFSH-R transfections. red, reduced amount. The current knowledge of the putative spatial arrangement of the LRR subdomain of GpHRs (13,14) was used to systematically exchange only those residues that were expected to have their side chains directed toward the presumed hormonebinding site (i.e. the X amino acid residues of the nine ␤-strand motifs arranged in parallel, each with the consensus sequence X 1 X 2 LX 3 LX 4 X 5 ). Human FSH, hCG, and hLH binding selectivity is very likely associated with the ϳ73% divergence of these X residues between hFSH-R and hLH-R. Identical X residues (e.g. the highly conserved Asp at position X 5 in ␤-strand 5) may be involved in common hormone-receptor contacts (10). Substi-tution of individual ␤-strands of the hFSH-R with the corresponding hLH-R ␤-strands allowed the identification of ␤-strands 3 and 6 as modules containing hCG-and hLH-selective determinants, as these mutant receptors mediate a 60 (␤-strand 3)-to 100-fold (␤-strand 6), respectively, enhanced responsiveness to hCG as well as to hLH. Moreover, both ␤-strands appeared to act synergistically, as a mutant receptor (i.e. hFSH-R/L␤3,L␤6) displayed a responsiveness to hCG and hLH similar to that of wild type hLH-R. Hence, hLH-R selectivity toward hCG and hLH can be confined mainly to just these two ␤-strands. In fact, none of the other ␤-strands ap-  B and E, human CG-stimulated, cAMP-mediated reporter gene activity in HEK-T 293 cells transiently transfected with wild type or mutant ␤-strand hFSH-R constructs. C and F, human FSH-stimulated, cAMP-mediated reporter gene activity in HEK-T 293 cells transiently transfected with wild type or mutant ␤-strand hFSH-R constructs. Human LH and hCG stimulated all constructs with a similar efficacy; for clarity, only the hCG-induced cAMP-mediated reporter gene activity is shown. Results shown are the mean Ϯ S.E. of triplicate observations from a single representative experiment. Mean EC 50 values are presented in Table V. ND, not detectable. peared to confer additional hCG/hLH selectivity to a major extent. These results are consistent with previous studies (2,3) in which hCG-selective determinants were predicted to be localized within the NCR-LRR6 sequence of the LH-R.
Most of the mutant hFSH-R constructs that harbored hLH-R ␤-strands did not exhibit severely impaired hFSH responsiveness, indicating that no essential hFSH-selective determinants are present in the corresponding hFSH-R ␤-strands. Moreover, this means that mutations in these ␤-strands did not alter the ligand-stimulated capacity of the receptors to induce cAMP production. Hence, the observed efficacy of hFSH, hCG, and hLH to stimulate receptor-mediated signaling correlates with the respective hormone binding affinities of these receptors as confirmed by ligand binding studies. Although some mutant receptors that were expressed at low levels (e.g. hFSH-R/ L␤3,L␤6 and hFSH-R/L␤5) exhibited efficient responsiveness to ligand stimulation, reduced responsiveness to hFSH stimulation was often related to severely impaired surface expression of mutant ␤-strand hFSH-Rs (e.g. hFSH-R/L␤1, hFSH-R/ L␤4, and hFSH-R/L␤7). This was likely because of disrupted protein folding/conformation, because reduced numbers (i.e. transfecting different concentrations of wild type hFSH-R constructs) of correctly folded receptors on the cell surface had only minor effects (Ͻ5-fold) on the level of ligand responsiveness. Also hFSH-R/Ala␤1 and hFSH-R/Ala␤7 exhibited disrupted cell surface expression and, as a consequence, reduced ligand responsiveness. However, Ala substitution of the individual residues in ␤-strands 1 and 7 revealed that these ␤-strands are not involved the binding of hFSH, hCG, or hLH.
In conclusion, the primary objective of this study was to identify hCG/hLH selective determinants in the extracellular N terminus of the hLH-R by generating potential gain-of-function mutant receptors in the context of an hFSH-R background. Using this strategy, hCG/hLH selectivity was tracked down to hLH-R ␤-strands 3 and 6. Alternatively, the results observed may also be explained by the loss of hCG/hLH repulsion in these mutant hFSH-Rs. Surprisingly, hFSH signaling (and binding) was hardly affected by the introduction of most hLH-R ␤-strands, suggesting that selective hFSH responsiveness, in contrast to the responsiveness to hCG and hLH, appears to be determined by numerous hFSH-R ␤-strands, each, however, with a relatively low contribution. The identification of the amino acid residues in ␤-strands 3 and 6 that are critically involved in the molecular mechanism determining hCG/hLH selectivity is currently under investigation.