Orientation of Follicle-stimulating Hormone (FSH) Subunits Complexed with the FSH Receptor

Follicle-stimulating hormone (FSH) comprises an α subunit and a β subunit, whereas the FSH receptor consists of two halves with distinct functions: the N-terminal extracellular exodomain and C-terminal membrane-associated endodomain. FSH initially binds to exodomain, and the resulting FSH/exodomain complex modulates the endodomain and generates signal. However, it has been difficult to determine which subunit of FSH contacts the exodomain or endodomain and in what orientation FSH interacts with them. To address these crucial issues, the receptor was Ala-scanned and the hormone subunits were probed with photoaffinity labeling with receptor peptides corresponding to the N-terminal region of the exodomain and exoloop 3 of the endodomain. Our results show that both regions of the receptors are important for hormone binding and signal generation. In addition, the FSH β subunit is specifically labeled with the N-terminal peptide, whereas the α subunit is labeled with the exoloop 3 peptide. These contrasting results show that the FSH β subunit is close to the N-terminal region and that the α subunit is projected toward exoloop 3 in the endodomain. The results raise the fundamental question whether the α subunit, common among the glycoprotein hormones, plays a major role in generating the hormone signal common to all glycoprotein hormones.

Follicle-stimulating hormone (FSH) 1 consists of an ␣ subunit of 92 amino acids and ␤ subunit of 111 amino acids. Both of the subunits are necessary for hormone action (1,2). The crystal structure of human FSH (3) shows that the two subunits are tightly associated in a crescent with the C termini in the concave side and the N termini in the convex side (Fig. 1A). This is essentially the same as the human chorionic gonadotropin structure (4,5). An exception to this intermingled subunit structure is the two polarized tips of the crescent: the ␣ tip consisting of the ␣ loops 1 and 3 and the ␤ tip of the ␤ loops 1 and 3.
In contrast to the tightly held hormone structure, the FSH receptor (FSHR), a G protein-coupled receptor, has two distinct domains as shown in Fig. 1B. The extracellular N-terminal exodomain comprises ϳ350 amino acids, and the membraneassociated C-terminal endodomain with a similar number of amino acids consists of seven transmembrane helices, three exoloops, three cytoloops, and the C-terminal cytoplasmic tail (6 -8). The exodomain binds the hormone with high affinity (9 -16) and selectivity (17), whereas the hormone signal is generated in the endodomain (18 -22). FSH initially interacts with the exodomain, and the resulting FSH/exodomain complex modulates the endodomain to generate hormone signal. Important amino acids have been identified for the interaction of the hormone and receptor (23,24). However, the orientation of FSH ␣ and ␤ in the ternary complex of the hormone, exodomain, and endodomain has been a major enigma and difficult to determine.
The bulk of the exodomain comprises 8 -9 Leu-rich repeats (LRR) (7,(25)(26)(27)(28), which are flanked by the short upstream N-terminal region and the downstream hinge region. LRRs are thought to form a one-third of a doughnut structure (26 -28). FSH appears to interact with LRRs (23,29) and the N-terminal and hinge regions (19,30). However, it is unclear which subunit of FSH contacts the exodomain or endodomain and in what orientation FSH interacts with them, although the concave C-terminal side of FSH appears to interact with the receptor. For example, FSH may interact horizontally or vertically with LRRs ( Fig. 1, C and D) and in two directions. These are crucial pieces of information for understanding the interactions among the hormone, exodomain, and endodomain and the mechanistics of signal generation. In addition, the information will facilitate the design of agonists and antagonists and development of new therapeutics. Because of the importance, the interactions of glycoprotein hormones with their receptors have been modeled (3-5, 23, 28) based on the crystal structure of the LRRs of ribonuclease inhibitor complexed with its ligand (31). However, the evidence has been elusive.
In a step to resolve this issue, we set out to distinguish the interactions of the FSH subunits with the N-terminal region of the exodomain and exoloop 3 in the endodomain. Our results show the interaction of FSH␤ with the N-terminal region of the exodomain and the ␣ tip of FSH␣ with exoloop 3 (Fig. 1E).

EXPERIMENTAL PROCEDURES
Materials-Human FSH was purchased from the National Hormone and Pituitary Program. Denatured FSH was prepared by boiling the hormone in 8 M urea for 30 min. Rabbit anti-FSH␣ serum, rabbit anti-FSH␤ serum, and monoclonal anti-FSHR 106.105 antibody were kindly provided by Dr. James Dias. Anti-rabbit IgG conjugated with peroxidase was purchased from Pierce. Peptide mimics including wild type peptides corresponding to the Ser 9 -Lys 40 sequence (FSHR 9 -40 ) and exoloop 3 and a photoactivable peptide containing benzoyl phenylalanine (Bpa) in place of Phe 13 (FSHR 9 -40F13Bpa ) were synthesized by Genemed Synthesis (San Francisco, CA) and purified on a Vydac C 18 high pressure liquid chromatography column using solvent gradient from 100% of 0.1% trifluoroacetic acid in water to 20% of 0.1% trifluoroacetic acid in water and 80% 1-propanol.
Mutagenesis and Functional Expression of FSH Receptors-Mutant FSHR cDNAs were prepared in the pSELECT vector using the Altered Sites mutagenesis system (Promega), sequenced on a Beckman CEQ 2000XL capillary sequencer, subcloned into pcDNA3 (Invitrogen) as described previously (32), and sequenced again to verify mutation sequences. This procedure does not involve polymerase chain reaction and therefore does not have its infidelity problems. Wild type and mutant receptor constructs were transfected into HEK 293 cells by the calcium phosphate method as described previously (32). Stable cell lines were established in minimum essential medium containing 10% horse serum and 500 g/ml G418. These cells were used for hormone-binding cAMP production. All of the assays were carried out in duplicate and repeated 4 -5 times, and the means Ϯ S.D. were calculated. 125 I-FSH Binding and Intracellular cAMP Assay-Stable cells were assayed for 125 I-FSH binding in the presence of 100,000 cpm of 125 I-FSH (33) and increasing concentrations of unlabeled FSH. The K d values were determined by Scatchard plots. For intracellular cAMP assay, cells were washed twice with Dulbecco's modified Eagle's medium and incubated in the medium containing 0.1 g/ml isobutylmethylxanthine for 15 min. Increasing concentrations of FSH were then added, and incubation was continued for 45 min at 37°C. After removing the medium, the cells were rinsed once with fresh medium without isobutylmethylxanthine, lysed in 70% ethanol, freeze-thawed in liquid nitro-gen, and scraped. After pelleting cell debris at 16,000 ϫ g for 10 min at 4°C, the supernatant was collected, dried under vacuum, and resuspended in 10 l of cAMP assay buffer (Amersham Biosciences). cAMP concentrations were determined with an 125 I-cAMP assay kit (Amersham Biosciences) following the manufacturer's instructions and validated for use in our laboratory.
125 I-FSH Binding to Solubilized FSHR-Transfected cells were washed twice with ice-cold 150 mM NaCl, 20 mM HEPES, pH 7.4 (buffer A). Cells were scraped on ice, collected in buffer A containing protease inhibitors (1 mM phenylmethylsulfonyl fluoride, 5 mM Nethylmaleimide, and 10 mM EDTA), and pelleted by centrifugation at 1300 ϫ g for 10 min. Cells were resuspended in 0.6 ml of buffer A containing 1% Nonidet P-40, 20% glycerol, and the above protease inhibitors (buffer B), incubated on ice for 15 min, and diluted with 5.4 ml of buffer A containing 20% glycerol plus the protease inhibitors (buffer C). The mixture was centrifuged at 100,000 ϫ g for 60 min. The supernatant (500 l) was mixed with 100,000 cpm of 125 I-FSH and 6.5 l of 0.9% NaCl and 10 mM Na 2 HPO 4 at pH 7.4 containing increasing concentrations of unlabeled FSH. After incubation for 12 h  at 4°C, the solution was thoroughly mixed with 250 l of buffer A containing bovine ␥-globulin (5 g/ml) and 750 l of buffer A containing 20% polyethylene glycol 8000. After incubation for 10 min at 4°C, samples were pelleted at 1,300 ϫ g for 30 min and supernatants removed. Pellets were resuspended in 1.5 ml of buffer A containing 20% polyethylene glycol 8000, centrifuged, and counted for radioactivity. Monoclonal anti-FSHR antibodies were radioiodinated and used for binding to nonbinding mutant FSHRs expressed on the intact cell surface as described previously (34). introduced to terminate radioiodination. Radioiodinated peptides were mixed with 60 l of 16% sucrose solution in PBS and fractionated on Sephadex Superfine G-10 column (0.6 ϫ 15 cm) using PBS. Peptides were derivatized with 4-azidobenzoyl glycine (ABG) and radioiodinated as described previously (35).
Photoaffinity Labeling of FSH-The following solutions were sequentially introduced to siliconized glass tubes: 20 l of 0.9% NaCl and 10 mM Na 2 HPO 4 , pH 7.4, in PBS, 10 l of FSH in PBS, and 10 l of 125 I-FSHR 9 -40F13Bpa in PBS. Competitive inhibition experiments were carried out as described for the photoaffinity-labeling experiments with the exception that 10 l instead of 20 l of PBS was introduced to each tube and the mixture was incubated with 10 l of increasing concentrations of nonradioactive receptor peptides. The mixtures were incubated at 37°C for 90 min in the dark, irradiated with a Mineralight R-52 UV lamp for 3 min as described previously (36), and solubilized in 2% SDS, 100 mM dithiothreitol, and 8 M urea. The samples were electrophoresed on 8 -12% polyacrylamide gradient gels. Gels were dried on filter paper and exposed to an imaging screen overnight, which was Deglycosylation-The FSH ␣ and ␤ subunits co-migrate on SDS-PAGE. To separate them on the gel, FSH was photoaffinity-labeled and deglycosylated with PNGase F. Enzymatic cleavage was done by incubation of the labeled FSH complex with 20 or 50 units of PNGase F (New England BioLabs) in 40 l for 18 h at 37°C. The samples were solubilized in SDS under the reducing condition and electrophoresed on 15% gel containing 9 M urea.

RESULTS
Activities of the N-terminal Region-In a first step to check the importance of the N-terminal region of human FSHR, each amino acid of the 8 SNRVFLCQESKVTEIPSDLPRNAIE 33 sequence was individually substituted with Ala. This sequence is diverse among the glycoprotein hormone receptors (Fig. 2), although these receptors share a high overall homology and structural similarity (38). In contrast, the FSHR sequence is highly conserved among species, implicating its importance. HEK 293 cells were stably transfected with mutant receptor plasmids and selected for stably expressing individual mutant receptors. These cells were assayed for 125 I-FSH binding and FSH-dependent cAMP induction. The results show that some of the Ala substitutions significantly impacted the hormone binding and/or cAMP induction.
Ala substitutions for Ser 9 , Asn 10 , Arg 11 , Val 12 , Phe 13 , and Leu 14 improved FSH binding (Fig. 3, A and B), FSH-dependent cAMP induction (Fig. 3C), or both. The K d values of FSHR N10A , FSHR R11A , and FSHR L14A were lower than the wild type value as were the EC 50 values of FSHR S9A , FSHR N10A, FSHR V12A , FSHR F13A , and FSHR L14A (see tables in Fig. 3). Ala substitutions for Gln 16 , Glu 17 , Lys 19 , and Val 20 did not impact the EC 50 values and maximal cAMP induction (Fig. 3F), and K d values of the mutants were similar to or somewhat higher than the wild type value (Fig. 3, D and E). In contrast, the S18A substitution resulted in a considerably lower EC 50 value despite a higher K d value. These results show an improved cAMP induction despite a lower hormone binding affinity and suggest an interesting and potentially crucial role of Ser 18 in modulating signal generation.
Ala substitution for Thr 21 , Glu 22 , or Ser 25 did not significantly impact hormone binding or cAMP induction (Fig. 4). On the other hand, the I23A substitution partially impaired the cAMP induction with a 23-fold higher EC 50 value and 2.6-fold lower maximal cAMP level. The P24A, D26A, and L27A substitutions completely abrogated hormone binding and therefore cAMP induction, suggesting the importance of these residues and this region. The P28A, R29A, N30A, A31G, I32A, and E33A substitutions did not dramatically impact the K d and EC 50 values or the maximal cAMP production (Fig. 4). The nonbinding mutants were either incapable of binding the hormone or trapped in cells, and these possibilities have successfully been tested by assaying hormone binding after solubilization of cells in nonionic detergents (34,39,40). The binding assay for receptors solubilized in Nonidet P-40 showed that FSH did not bind to any of the C15A, P24A, D26A, and L27A mutants (Fig.  5A), indicating that they are incapable of hormone binding. However, the result does not necessarily prove the surface expression of the nonbinding mutants and nonbinding on the surface. Therefore, the cells expressing them were probed with 125 I-anti-FSHR antibody. The antibody bound to the cells (Fig.  5B), indicating the surface expression of the mutants and their inability to bind the hormone on the cell surface. These results, taken together, show several distinct effects of Ala substitutions as shown in the summary bar graph (Fig. 6). C15A, P24A, D26A, and L27A abolished hormone binding. In contrast to these nonbinding mutations, N10A, R11A, and L14A improved hormone binding. On the other hand, I23A impaired cAMP induction by dramatically increasing the EC 50  ably, S9A, V12A, F13A, S18A, and I32A reduced the EC 50 value by 2-3-fold while maintaining or slightly enhancing the maximum cAMP induction level. These results suggested the importance of this region of the receptor in hormone binding and cAMP induction and raised a question as to whether this region directly interacts with the hormone or indirectly impacts the global structure of the receptor.
Photoaffinity Labeling of FSH-To examine the two general possibilities, a peptide mimic corresponding to the receptor sequence of 8 SNRVFLCQESKVTEIPSDLPRNAIELRFVLTK 40 was synthesized, FSHR 9 -40 (Fig. 7A). A Tyr residue was attached to the N terminus for radioiodination, and the N terminus was acetylated while the C terminus amidated. Phe 13 was substituted with Bpa for photoaffinity labeling (41). The ketone moiety of the Bpa group can be activated with UV at Ͼ350 nm and is capable of reacting with unreactive ␣-CH bonds of amino acids (41,42,44). To determine whether the resulting peptide 125 I-FSHR 9 -40F13Bpa could bind and photoaffinity-label FSH, it was incubated with FSH and irradiated with UV for increasing time periods. Samples were solubilized in SDS under reducing conditions and then electrophoresed. The autoradiographic phosphorimage of the gel (Fig. 7B) revealed labeling of the FSH band. The autoradiograph suggests that the two subunits of human FSH comigrated. The band was not labeled when the sample was not irradiated with UV, indicating the requirement for UV irradiation. The extent of the labeling was dependent on the irradiation time, reaching maximum labeling after 30-s irradiation. The result shows that the labeling is saturable. The hormone was labeled next with increasing concentrations of 125 I-FSHR 9 -40F13Bpa while maintaining FSH at a constant concentration (Fig. 7C). Conversely, increasing concentrations of FSH were labeled with a constant concentration of 125 I-FSHR 9 -40F13Bpa (Fig. 7D). If the labeling was specific, the concentrations should reach a plateau under both conditions. Indeed, the labeling plateaued under both conditions, indicating saturable and specific labeling of a substantial portion (Ͼ50%) of FSH. Furthermore, the labeling should be inhibited by nonradioactive peptide and unmodified wild type peptide. FIG. 7. Photoaffinity labeling of FSH with photoactivable FSHR 9 -40F13Bpa . The FSH receptor peptide corresponding to the sequence Ser 9 -Lys 40 (FSHR 9 -40 ) was synthesized with a Tyr at the N terminus for radioiodination and Bpa at the position of Phe 13 for photoaffinity labeling (A). The peptide was radioiodinated, and the resulting 125 I-FSHR 9 -40F13Bpa was incubated with FSH and irradiated with UV. B, the sample was irradiated with UV for increasing time periods from 0 to 150 s, solubilized in SDS under the reducing condition, and electrophoresed on polyacrylamide gel. After drying gels, they were exposed to a phosphorimaging screen and scanned on a PhosphorImager. The peptide appeared as the lower band, and the FSH ␣ and FSH ␤ subunits comigrated and appeared in the upper band. The intensity of each band in a gel lane was measured, and the percentage of the labeled FSH band in a gel lane was calculated based on the total intensity of a gel lane and presented in the bar graph above the autoradiograph. C, increasing amounts of 125 I-FSHR 9 -40F13Bpa from 0 to 3.7 M were incubated with a constant amount (0.1 M) of FSH and photolyzed for 60 s. The samples were processed as described above. D, increasing amounts of FSH from 0 to 0.2 M were incubated with a constant amount (3.1 M) of 125 I-FSHR 9 -40F13Bpa . E, FSH (80 nM) was incubated with 3 M 125 I-FSHR 9 -40F13Bpa in the presence of increasing concentrations of unlabeled FSHR peptides, FSHR 9 -40 , FSHR 9 -40F13Bpa , exoloop 1 peptide (FSHR exo1 ), exoloop 2 peptide (FSHR exo2 ), and exoloop 3 peptide (FSHR exo3 ). The samples were irradiated and processed as described in B.

FIG. 8. Photoaffinity labeling of denatured FSH and other glycoproteins.
A, increasing concentrations of 125 I-FSHR 9 -40F13Bpa were incubated with 80 nM denatured FSH, irradiated with UV, and processed as described in the legend to Fig. 7. FSH was denatured by boiling in 8 M urea for 30 min. B, a constant amount of 125 I-FSHR 9 -40F13Bpa was incubated with increasing concentrations of denatured FSH, treated with UV, and processed. C, a constant amount of 125 I-FSHR 9 -40F13Bpa was incubated with 5 nM each of FSH, phospholipase A (PLA), urokinase, growth hormone (GH), LH, or TSH, treated with UV, and processed as described in the legend to Fig. 7. D, cells stably expressing FSHR were incubated with 125 I-FSH and washed as described in Fig. 3 but in the presence of increasing concentrations of FSHR 9 -40 or FSHR 9 -40F13Bpa . Cells were washed three times and counted for the bound radioactivity. Therefore, FSH was incubated with 125 I-FSHR 9 -40F13Bpa in the presence of increasing concentrations of wild type peptide and nonradioactive FSHR 9 -40F13Bpa (Fig. 7E). The peptides inhibited the photoaffinity labeling in a dose-dependent manner and eventually blocked the labeling. Peptides corresponding to exoloops 1-3, FSHR exo1 , FSHR exo2 , and FSHR exo3 , were also tested. FSHR exo2 inhibited the labeling as FSHR 9 -40 did. In contrast, FSHR exo1 inhibited the labeling with a ϳ30-fold less potency. On the other hand, FSHR exo3 failed to block the labeling.
Labeling Specificity-Although the photoaffinity labeling was specific for FSH, our data do not show the biological specificity of the affinity labeling. To address this concern, a constant amount of denatured FSH was incubated with increasing concentrations of 125 I-FSHR 9 -40F13Bpa and treated with UV. Denatured FSH was not labeled at all despite high concentrations of the peptide (Fig. 8A). Denatured FSH was not labeled when increasing concentrations of denatured FSH were incubated with a constant amount of 125 I-FSHR 9 -40F13Bpa and treated with UV (Fig. 8B). When FSH was denatured by boiling in 8 M urea for 30 min, it did not bind to FSHR and induce cAMP production. To test whether the denatured FSH remained in solution, the mixture of radioactively labeled FSH and unlabeled FSH was denatured and varying volumes of the mixture were transferred to other tubes and the radioactivity was counted. The transfer was quantitative with 99 -100% efficiency, indicating that denatured FSH was present in the photoaffinity-labeling tube. These results indicate the specificity of the affinity labeling for biologically active FSH. To determine the labeling specificity, luteinizing hormone (LH), thyroid-stimulating hormone (TSH), growth hormone, phospholipase A, and urokinase were subjected to photoaffinity labeling with 125 I-FSHR 9 -40F13Bpa (Fig. 8C). None of them was labeled. If 125 I-FSHR 9 -40Bpa13 specifically binds to FSH and labels it as indicated by the results, the peptide is expected to inhibit the in vivo binding of FSH to the receptor on intact cells. Indeed, FSHR 9 -40Bpa13 and FSHR 9 -40 inhibited 125 I-FSH binding to the receptor (Fig. 8D).
Labeling of FSH ␤ Subunit-Because the two subunits of purified human FSH appeared to comigrate on SDS-PAGE, it was unclear which of the subunits was labeled. To resolve the subunits, FSH was deglycosylated with PNGase F and electrophoresed (Fig. 9A, lane 2). The two subunits were clearly separated into two distinct bands. Because the ␤ subunit is larger than the ␣ subunit, the upper band is probably the ␤ subunit. To conclusively determine the identity of the upper band, deglycosylated FSH was electrophoresed, the gel was blotted on nitrocellulose membrane, and the membrane was probed with anti-FSH␣ and anti-FSH␤ antibodies. Anti-FSH␣ antibody conspicuously labeled the lower band, whereas the anti-FSH␤ antibody recognized primarily the upper band and faintly the lower band (Fig. 9, A and B). These results show that the lower band represents the FSH␣ subunit, whereas the upper band is the FSH␤ subunit, indicating that FSH␤ was labeled. To compare this labeling of FSH␤ with the N-terminal peptide, the exoloop 3 peptide, FSHR exo3 , was used for labeling FSH. The peptide was derivatized with a UV-activable reagent, ABG, and radioiodinated. The resulting 125 I-ABG-FSHR exo3 was incubated with FSH, irradiated with UV, deglycosylated, and electrophoresed. The labeled FSH subunit appeared in the lower band, indicating the labeling of the FSH␣ subunit (Fig. 9C) as reported previously (35).
The labeling of FSH␤ by 125 I-FSHR 9 -40F13Bpa contrasts the labeling of the FSH␣ subunit by the FSHR exoloop 3 peptide. If this contrasting labeling is specific and reflects the true interaction between FSH and the receptor, some of the peptides representing parts of the ␣ subunit sequence might inhibit the labeling. Particularly, if some of the ␣ peptides block the labeling of the ␣ subunit but not the ␤ subunit, the result would support the selective labeling results and show labeling specificity. Four ␣ peptides, ␣ 1-15 , ␣ 26 -46 , ␣ 61-75 , and ␣ 81-92 were tested, and none of them inhibited the labeling of the FSH␤ subunit by 125 I-FSHR 9 -40F13Bpa (Fig. 10A). In contrast, the labeling of the FSH␣ subunit by 125 I-ABG-FSHR exo3 was blocked by ␣ 26 -46 and somewhat by ␣ 61-75 (Fig.  10B). ␣ 1-15 and ␣ 81-92 Peptides failed to block the labeling. These results support the differential labeling and its selectivity and validity.

DISCUSSION
Our Ala-scanning results indicate that the Ser 9 -Glu 33 sequence of the FSH receptor is important for surface expression, hormone binding, and signal generation. The photoaffinitylabeling results show that FSHR 9 -40F13Bpa photoaffinity labels FSH but not LH, TSH, growth hormone, phospholipase A, and urokinase. The labeling is saturable and dependent on the concentrations of FSH and derivatized 125 I-FSHR 9 -40 , UV activation, and UV exposure time. 125 I-FSHR 9 -40F13Bpa photoaffinity labels bioactive FSH but not denatured hormone, and the labeling is blocked by nonderivatized wild type peptide and nonradioactive FSHR  . The labeling specificity is further underscored by the fact that it labeled the ␤ subunit but not the ␣ subunit. These results suggest that the N-terminal region of the FSH receptor is in close proximity to FSH, probably interacting with the hormone. This conclusion is consistent with the previous reports that some residues of the region FIG. 9. Immunoblot of FSH ␣ and ␤ subunit bands. A, FSH was treated with PNGase F, solubilized in SDS under the reducing condition, and electrophoresed along with nondeglycosylated FSH. Gel lanes were either stained with Coomassie Brilliant Blue (CBB) or blotted and stained with anti-FSH, anti-FSH␣, or anti-FSH␤ antibodies. B, FSH was incubated with increasing concentrations of 125 I-FSHR 9 -40F13Bpa , treated with UV, deglycosylated with PNGase F, solubilized, and electrophoresed along with 125 I-FSH. C, the peptide corresponding to exoloop 3 (FSHR exo3 ) was derivatized with a UV-activable reagent, ABG, and radioiodinated. The resulting 125 I-ABG-FSHR exo3 was incubated with FSH, treated with UV, deglycosylated with PNGase F, solubilized, and electrophoresed (E3). E3 was compared with FSH labeled with 125 I-FSHR 9 -40F13Bpa (N).
are important for hormone binding and receptor trafficking (45)(46)(47) and that the similar region of the LH receptor interacts with human chorionic gonadotropin (34,48).
In contrast to the labeling of the ␤ subunit by the N-terminal peptide, the exoloop 3 peptide labeled the ␣ subunit. The significance of these contrasting results is 2-fold. First, it supports the validity and specificity of the photoaffinity labeling. Second, it provides the crucial information on the overall arrangement of the ternary complex involving the exodomain, FSH, and endodomain. The results suggest that the ␤ subunit is near the N-terminal region of the exodomain, whereas the ␣ subunit is close to the exoloops of the endodomain. These are consistent with the previous reports that only the ␣␤ dimer is capable of high affinity binding to receptors and inducing biological responses (49,50). Interestingly, all glycoprotein hormones utilize similar if not identical signal pathways consisting of adenylyl cyclase and phospholipase C␤. Therefore, the ␣ subunit has been implicated in the signal generation (49).
Based on these results, it is now possible to project the hormone interacting with both the exodomain and endodomain. Furthermore, they suggest that the hormone is probably in a vertically tilted position with respect to LRRs of the exodomain and the endodomain. To help visualize the arrangement and facilitate modeling of the ternary complex, one of several possible models is presented in Fig. 1E. The model suggests that parts of the ␣ subunit might interact with LRRs and, conversely, that some parts of the ␤ subunit may be close to the exoloops. Such interactions could be probed by strategically attaching a photoactivable group at appropriate positions of the hormone subunits and/or using a reagent that can reach farther than ABG and Bpa. These two reagents can reach 10 and 7 Å, respectively. Other additional information will also be necessary to more precisely define the ternary structure. For example, the crystal structure of FSH shown in the model does not likely represent its structure in the ternary complex, because the gonadotropin undergoes conformational changes, particularly the interaction between the two subunits upon the initial interaction with the receptor (51). These conclusions are consistent with observations that the original quaternary structure of unbound hormone dimers is not essential for hormone action (1, 52).
The results that Ala substitution for some N-terminal residues improved hormone binding, cAMP induction or both sug-gest an interesting possibility that this region is involved in modulating not only hormone binding but also signal generation. The most dramatic improvement is seen in the S18A substitution, which improved the EC 50 value of cAMP induction by 3-fold as compared with the wild type value. Additionally, the maximum level of cAMP production only slightly increased. These observations indicate that the affinity and maximum level of cAMP induction are distinctly regulated. They suggest that FSH activates FSHR S18A more effectively than the wild type receptor does, which in turn results in better activation of the G protein. It will be interesting to see whether the number of activated G protein molecules is the limiting factor. The improved EC 50 is not related to the hormone binding affinity because the binding affinity of the mutant is somewhat less than the wild type affinity. These novel observations suggest an intriguing possibility that FSHR S18A is more sensitive to hormone binding and is capable of activating the G protein with higher affinity without significantly impacting the level of activation. Because the exodomain is likely to modulate the endodomain to generate hormone signals at the exoloops (18 -20, 38, 43), a simple possibility is that the affinity of the modulation at the interface between the exodomain and exoloops is improved in FSHR S18A . Several other Ala substitutions, S9A, V12A, and F13A, also showed similar yet less dramatic results.
In conclusion, the evidence is presented that in the ternary exodomain/FSH/endodomain complex, FSH is vertically oriented with the ␤ subunit close to the N-terminal region and the ␣ tip projecting toward the exoloop 3.