Substitution of the Seat-belt Region of the Thyroid-stimulating Hormone (TSH) β-Subunit with the Corresponding Regions of Choriogonadotropin or Follitropin Confers Luteotropic but Not Follitropic Activity to Chimeric TSH*

The region between the 10th and 12th cysteine (Cys88-Cys105 in human thyroid-stimulating hormone β-subunit (hTSHβ)) of the glycoprotein hormone β-subunits corresponds to the disulfide-linked seat-belt region. It wraps around the common α-subunit and has been implicated in regulating specificity between human choriogonadotropin (hCG) and human follicle-stimulating hormone (hFSH), but determinants of hTSH specificity are unknown. To characterize the role of this region for hTSH, we constructed hTSH chimeras in which the entire seat-belt region Cys88-Cys105 or individual intercysteine segments Cys88-Cys95 and Cys95-Cys105 were replaced with the corresponding sequences of hCG and hFSH or alanine cassettes. Alanine cassette mutagenesis of hTSH showed that the Cys95-Cys105 segment of the seat-belt was more important for TSH receptor binding and signal transduction than the Cys88-Cys95 determinant loop region. Replacing the entire seat-belt of hTSHβ with the hCG sequence conferred full hCG receptor binding and activation to the hTSH chimera, whereas TSH receptor binding and activation were abolished. Conversely, introduction of the hTSHβ seat-belt sequence into hCGβ generated an hCG chimera that bound to and activated the TSH receptor but not the CG/lutropin (LH) receptor. In contrast, an hTSH chimera bearing hFSH seat-belt residues did not possess any follitropic activity, and its thyrotropic activity was only slightly reduced. This may in part be due to the fact that the net charge of the seat-belt is similar in hTSH and hFSH but different from hCG. However, exchanging other regions of charge heterogeneity between hTSHβ and hFSHβ did not confer follitropic activity to hTSH. Thus, exchanging the seat-belt region between hTSH and hCG switches hormonal specificity in a mutually exclusive fashion. In contrast, the seat-belt appears not to discriminate between the TSH and the FSH receptors, indicating for the first time that domains outside the seat-belt region contribute to glycoprotein hormone specificity.

Thyrotropin (thyroid-stimulating hormone (TSH)) 1 choriogonadotropin (CG), follitropin (follicle-stimulating hormone (FSH)), and lutropin (luteinizing hormone (LH)) are structurally related heterodimers that together form the glycoprotein hormone family (1). These hormones belong to the superfamily of cystine-knot growth factors (2,3) and activate specific Gprotein-coupled receptors notable for large extracellular domains containing multiple leucine-rich motifs (4). The primary structure of the ␣-subunit, which is encoded by a single gene, is identical in these hormones. The distinct ␤-subunits, despite conservation of all 12 cysteine residues and similar overall folding, are sufficiently different to confer specificity to each hormone (1)(2)(3).
The molecular mechanisms whereby glycoprotein hormones activate their receptors are largely unknown, but multiple contact points between ligand and receptor, perhaps in a stepwise fashion, appear necessary to induce conformational changes favoring receptor G-protein coupling and subsequent second messenger generation (5)(6)(7)(8)(9). Recently, we have described several ␣-subunit domains important for hTSH activity (10 -13), but there is little information on how the hTSH ␤-subunit contributes to receptor activation.
Previous studies have shown that the region between the 10th and 12th cysteine of the ␤-subunit is important not only for subunit association, receptor binding, as well as activation (14 -16), but also for specific receptor recognition (17)(18)(19)(20) of hCG and hFSH. In the crystal structure of hCG (2,3), this region corresponds to the "seat-belt" region (Cys 88 -Cys 105 in hTSH␤), so-called because it wraps around the ␣-subunit and orients it in the heterodimer while remaining covalently bonded to the ␤-subunit through disulfide linkages between Cys 9 -Cys 90 and Cys  . This seat-belt consists of two intercysteine segments, a surface-exposed hydrophilic loop between the 10th and 11th cysteine (Cys 88 -Cys 95 in hTSH␤) and a carboxyl-terminal segment between the 11th and 12th cysteine (Cys 95 -Cys 105 in hTSH␤), and is in close proximity to ␣-subunit domains important for the structural integrity and * Preliminary portions of these results were presented at the 10th International Congress of Endocrinology, San Francisco, CA (1996). 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.
In contrast to the work on the gonadotropins, the role of the seat-belt for hTSH is not known. A single study using a set of overlapping synthetic peptides spanning the entire hTSH subunit (21) showed that none of the peptides encompassing the seat-belt region, hTSH␤ 81-85 or hTSH␤ 91-105, inhibited TSH receptor binding, but a peptide containing the carboxyl terminus (␤101-112) possessed the highest TSH receptor binding activity. However, the role of the seat-belt region in the context of the intact hTSH heterodimer has not been investigated. Interestingly, recent studies on a naturally occurring hTSH␤ mutation from patients with secondary hypothyroidism have shown the importance of Cys 105 (corresponding to Cys 110 in hCG) for hTSH activity (22).
In the present study, using a chimeric mutagenesis approach, we demonstrate the importance of the seat-belt for hTSH action as well as specificity. Moreover, our findings reveal previously unrecognized differences in the regulation of specificity among the glycoprotein hormones.
Site-directed Mutagenesis-The chimeric hTSH were constructed with the PCR-based megaprimer method of site-directed mutagenesis (26), as described (11,13). Individual intercysteine segments C10-C11 (hTSH␤ Cys 88 -Cys 95 ) or C11-C12 (hTSH␤ Cys 95 -Cys 105 ) of the hTSH ␤-minigene were replaced with nucleotides coding for the respective sequence of hCG and hFSH or Ala cassettes. To replace the entire seat-belt (hTSH␤ Cys 88 -Cys 105 ), chimeras with individually mutated intercysteine segments were used as templates for subsequent PCR reactions. The hTSH␤ seat-belt was introduced into the hCG ␤-subunit in a single PCR reaction using a primer coding for the entire hTSH␤ seat-belt. Further hTSH/hFSH chimeras were constructed in which the carboxyl-terminal residues hTSH␤ 105-112 or amino acids hTSH␤ 44 -52 were replaced with the sequence of hFSH. In addition, Asp 94 of the determinant loop was replaced with Lys (TSH␤Lys 94 ) or Glu (TSH␤Glu 94 ). After subcloning into the expression vectors, the entire PCR products of all constructs were sequenced to verify the mutations and to rule out any undesired polymerase errors. Construction of the quadruple ␣-subunit mutant bearing Lys residues at positions ␣13, 14, 16, and 20 (␣4K) was described previously (13).
Transient Expression-CHO-K1 cells maintained as described (10) were transiently cotransfected with the various constructs using a transient transfection protocol based on a liposome formulation (Lipo-fectAMINE reagent, Life Technologies, Inc.) (10). After culture in CHO serum-free medium (CHO-SFM, Life Technologies, Inc.) for 48 h, conditioned media including control medium from mock transfections were harvested, concentrated with Centriprep 10 concentrators (Amicon, Beverly, MA), and stored at Ϫ70°C to prevent neuraminidase digestion.
Immunoassays-Wild type and mutant hTSH analogs were quantified with a panel of four different hTSH immunoassays, which were described in detail previously (12). hCG immunoreactivities were measured with two different specific third-generation immunoassays without crossreactivity to other glycoprotein hormones (Nichols Institute, San Juan Capistrano, CA; ICN, Costa Mesa, CA), and hFSH immunoreactivity was measured with an hFSH-specific third-generation immunoassay (Nichols Institute).
Hormone Binding Assays-The TSH receptor-binding activity of wild type and hTSH mutants was determined by their ability to displace 125 I-bTSH from a solubilized porcine thyroid membrane receptor preparation (Kronus, Dana Point, CA), as described previously (10). Binding to the CG/LH receptor was studied in MA-10 cells following a previously employed protocol (13,24), and FSH receptor binding was analyzed using a rat testis membrane radioreceptor assay as described in detail previously (20).
Hormone Activity Assays-The ability of the various chimeras to induce cAMP production was studied at the TSH receptor using JP09 cells (23), at the CG/LH receptor using MA-10 cells (24), and at the FSH receptor using FSH-R/293 cells (25). Briefly, confluent cells in 96-well tissue culture plates were incubated in a modified Krebs Ringer buffer for 2 h at 37°C, 5% CO 2 with serial dilutions of wild type and mutant hTSH, as well as control medium from mock transfections. The amount of cAMP released into the medium was assayed by radioimmunoassay (10). Progesterone production at the CG/LH or FSH receptor was determined using a commercially available progesterone radioimmunoassay kit (ICN) after incubation of the chimeric constructs with MA-10 cells or Y-1 cells, respectively, as detailed previously (20,24).  104 . The hCG/hTSH chimera 94 TSH 109 was obtained by introduction of the hTSH␤ seat-belt residues into the hCG ␤-subunit (Fig. 2). In addition, hTSH␤ Asp 94 , which is conserved in all known ␤-subunits and essential for hCG activity FIG. 1. Domains of hTSH important for activity. The schematic drawing of hTSH is based on a molecular homology model of hTSH (13) and built on a template of an hCG model derived from crystallographic coordinates (2). The ␣-subunit is shown in gray, and the ␤-subunit is black. The seat-belt region between the 10th (Cys 88 ) and 12th (Cys 105 ) Cys is depicted by an interrupted line; the N-terminal determinant loop (Cys 88 -Cys 95 in hTSH␤) corresponds to the dotted line (qqqq), and the carboxyl-terminal segment (Cys 95 -Cys 105 in hTSH␤) corresponds to the dashed line (----). Since the carboxyl terminus beyond hCG␤ 111 was not traceable in the original electron density map, hTSH␤ is only drawn to the corresponding residue 106. Domains of the ␣-subunit of known importance for hTSH activity are boxed. Because of hydrogen fluoride treatment of hCG prior to crystallization, the oligosaccharides, including that linked to ␣Asn 52 , are not shown. The ␣Asn 52 carbohydrate is predicted to project into the proposed receptor binding domain, which also includes the ␣40 -46 helix and the ␣-carboxyl terminus ␣88 -92 (2,3). In contrast, the ␣11-20 domain is not located in proximity to the ␤ seat-belt (see "Discussion"). (27), was replaced with Lys (hTSH␤Lys 94 ) or with Glu (hTSH␤Glu 94 ). Finally, sequences of the hTSH ␤-subunit outside the seat-belt were replaced with the corresponding sequence of hFSH to create 44 FSH 52 and 105 FSH 112 . These regions were chosen because they display the greatest charge heterogeneity among these ␤-subunits, based on the proposed role of variable charges for glycoprotein hormone specificity (17) (and see below). Receptor binding and biological properties of these analogs, described in detail below, are summarized in Table I. A comparison of the receptor specificity of glycoprotein hormone seat-belt chimeras from this and other studies (18 -20) is given in Table II.

Construction of Chimeric Glycoprotein Hormone Mutants-
Secretion of Chimeric Mutants-All chimeric hTSH heterodimers were secreted from the transfected CHO cells, and relative secretion, compared with hTSH-wt, ranged from 35.3 Ϯ 7.6% in the case of 89 Ala 94 to 100.2 Ϯ 14.7% in the case of 96 CG 104 . As evidence for accurate quantitation, hTSH immunoreactivity of each chimera was comparable in four different hTSH immunoassays, which recognize different epitopes of the hTSH molecule (10 -13) (data not shown). Similarly, secretion of 94 TSH 109 bearing the hTSH seat-belt sequence within the hCG ␤-subunit was 94.3 Ϯ 17.7% that of hCG-wt, as determined by two different hCG immunoassays. Therefore, in accord with previous studies on hCG and hFSH (18 -20), intercysteine loops appear to be interchangeable between the different ␤-subunits without major changes in subunit assembly and heterodimer secretion of hTSH or global conformational changes of the heterodimer. Secretion of 44 FSH 52 and 105 FSH 112 with replacements outside the seat-belt was less than 25% for both chimeras. The reasons for the reduced secretion of these mutants are unclear but could be related to decreased stability of messenger RNA, improper folding of the mutant subunit, or decreased efficiency of subunit assembly.
Ala Cassette hTSH Mutants-TSH receptor binding as well as thyrotropic activity of both 89 Ala 94 and 96 Ala 104 was substantially reduced (Fig. 3, A and B), showing that the native seat-belt sequence is important for hTSH activity. Interestingly, maximal cAMP stimulation of the 96 Ala 104 mutant was significantly lower than that of the 89 Ala 94 mutant (28.3 Ϯ 3.2 versus 63.0 Ϯ 4.6% of hTSH-wt, respectively), indicating that the carboxyl-terminal segment of the seat-belt is more important for hTSH activity than the determinant loop.
Role of Asp 94 for hTSH Activity-A single mutation of the conserved Asp 94 to Lys (hTSH␤Lys 94 ) completely abolished TSH receptor binding and activation, whereas preserving the negative charge at this position by mutating Asp 94 to Glu (hTSH␤Glu 94 ) did not have a significant effect on TSH receptor binding or activation (Table I). This confirmed the importance of a negative charge in this particular position for glycoprotein hormone activity (27).
hTSH/hCG Chimeras-Replacement of the hTSH seat-belt segments with the respective sequences of hCG either substantially decreased ( 89 CG 94 ) or abolished measurable TSH receptor binding and activation ( 96 CG 104 , 89 CG 104 ) of the chimeras (Fig. 4, A and B). hTSH/hCG chimeras 89 CG 94 and 96 CG 104 showed only very little CG/LH receptor binding and activation, both with cAMP stimulation as well as progesterone production. Thus, at the highest doses possible within the limitations of the transient transfection system, between 100 -200 ng/ml, cAMP or progesterone production was only 10 -18% that of hCG-wt (Fig. 5, A-C). To more conclusively test whether the Cys 88 -Cys 95 determinant loop was important for differential hormonal activity, we constructed a hTSH mutant ␣4K/ 89 CG 94 . In ␣4K/ 89 CG 94 , residues ␣13, 14, 16, and 20 were substituted with Lys residues in addition to the replacement of Cys 88 -Cys 95 with the respective hCG residues. We had previously shown that introduction of positive charges into this ␣11-20 domain led to substantial increases of glycoprotein hormone receptor binding affinity (13). We therefore expected that increasing the binding affinity of 89 CG 94 (Fig. 5A) should accentuate its gonadotropic properties. Maximal cAMP and progesterone production with this ␣4K/ 89 CG 94 combination chimera increased to 35% and 50% of hCG-wt levels, respectively (Fig. 5, B and C). At the same time, the thyrotropic activity of ␣4K/ 89 CG 94 remained unchanged (data not shown).
Remarkably, replacement of the entire seat-belt of hTSH␤ with the hCG sequence ( 89 CG 104 ) resulted in a chimera with hCG receptor binding comparable to hCG-wt, suggesting that the two individual intercysteine segments confer hCG specificity in a synergistic fashion (Fig. 5A). Further, the 89 CG 104 chimera was able to induce biological responses in MA-10 cells expressing the CG/LH receptor (Fig. 5, B and C). Whereas potency and efficacy of progesterone production as well as efficacy of cAMP induction were similar to hCG-wt, the potency of 89 CG 104 for cAMP production was 10-fold less than that of hCG-wt. These differences may stem in part from differences in the cAMP and progesterone assay conditions in MA-10 cells (see "Experimental Procedures"). Moreover, such generation of full hormonal responses at submaximal cAMP levels, termed "the cAMP superfluity concept," has been well recognized in studies on structure-function relationships of glycoprotein hormones (8,28). Analogous findings for recombinant analogs with substantially higher progesterone-inducing than cAMP-inducing ability have been described by others (29). Interestingly, hTSH/hCG specificity appeared mutually exclusive since the 89 CG 104 chimera did not possess significant thyrotropic activity (Fig. 4, A and B).
hCG/hTSH Chimera-Conversely, the reciprocal chimera 94 TSH 109 , which bears the hTSH␤ seat-belt in the context of the hCG-␤ subunit, bound to the TSH receptor and was able to activate cAMP production in JP09 cells, with an EC 50 that was 26.7 Ϯ 4.7-fold higher than that of hTSH-wt (Fig. 6, A and B). Shown are the native amino acid sequences of the seat-belt region between the 10th and 12th Cys for hTSH, hCG, and hFSH ␤-subunits in the 1-letter code. For comparison, the Cys numbering is maintained according to its position in the respective subunit. Also depicted are the mutant seat-belt hTSH constructs and the hCG chimera prepared by site-directed mutagenesis, with the mutated segments underlined. In the Ala cassette constructs, either the determinant loop (between the 10th and 11th Cys, 89 TSH␤ 94 ) or the carboxyl-terminal segment (between the 11th and 12th Cys, 96 TSH␤ 104 ) were replaced with Ala residues. In the hTSH/hCG and hTSH/hFSH chimeras, individual intercysteine segments or the entire seat-belt of the hTSH ␤-subunit were replaced with the respective gonadotropin residues. In contrast, in 94 TSH 109 , the hTSH␤ seat-belt was introduced into the hCG ␤-subunit.
At the same time, 94 TSH 109 did not bind to the CG/LH receptor, nor did it stimulate cAMP or progesterone production in MA-10 cells at concentrations up to 1000 ng/ml (Table I).
hTSH/hFSH Chimeras-Analogous replacement of individual intercysteine segments of hTSH␤ with the corresponding hFSH residues only slightly reduced TSH receptor binding or activation of these hTSH/hFSH chimeras (Fig. 7, A and B). Further, in contrast to 89 CG 104 , the 89 FSH 104 construct bearing the entire hFSH seat-belt sequence was able to significantly bind to the TSH receptor and induce 50% of maximal hTSH-wt cAMP production in JP09 cells. Interestingly, none of the three hTSH/hFSH chimeras showed significant follitropic activity. Unlike hFSH-wt, the chimeras did not stimulate cAMP production at the hFSH receptor expressed in 293 cells (Fig. 8). Further, they did not show significant binding in a rat testis FSH radioreceptor assay and did not stimulate progesterone production in Y-1 cells expressing the hFSH receptor (Table I).
Since charge heterogeneity could play a role for hTSH/hFSH specificity (17), we replaced candidate regions hTSH␤ 44 -52 and the carboxyl-terminal residues 105-112 with hFSH sequences. Thus, hTSH␤ 44 KYALSQDVC 52 has a net charge of 0, and the corresponding hFSH␤ sequence ARPKIQKTC has a FIG. 3. Thyrotropic activity of Ala cassette hTSH mutants. A, TSH receptor binding. Increasing doses of wild type or mutant hTSH were incubated with porcine membranes in the presence of a constant amount of 125 I-bTSH. 125 I-bTSH bound to membranes was precipitated and quantitated in a ␥ counter, and radioactivity precipitated in the presence of concentrated medium from mock transfections was defined as 100%. Values are shown as mean Ϯ S.E. of triplicate determinations. Experiments were repeated twice. B, cAMP induction at the TSH receptor. cAMP induction by the Ala cassette mutants was assessed in CHO cells expressing the rhTSH receptor (JP09). Increasing concentrations of wild type or mutant hTSH were incubated with JP09 cells, and the cAMP concentration in the resulting supernatants was assayed by radioimmunoassay. Basal cAMP concentrations ranged from 8 -19 pmol/ml. The amount of cAMP released from the cells in the presence of concentrated medium from mock transfected cells was not different from base-line levels (buffer only). A representative experiment, repeated at least twice, is shown. Values from triplicate determinations are depicted as mean Ϯ S.E. In this and the following figures, S.E. values are shown for all data points. In certain cases, the S.E. values are smaller than the size of the symbol.

FIG. 4. Thyrotropic activity of hTSH/hCG seat-belt chimeras.
A, TSH receptor binding. TSH receptor binding for the hTSH/hCG chimeras was performed as described in Fig. 3B. B, cAMP induction at the TSH receptor. The ability of the chimeras to induce cAMP production at the hTSH receptor was assessed in JP09 cells as described in Fig. 3B. Consistent with previous observations (11), hCG-wt did not bind to or activate the TSH receptor in the concentration range employed. A representative experiment, repeated at least twice, is shown. charge of ϩ3. hTSH␤ 105 CTKPQKSY 112 has a net charge of ϩ2, and the corresponding hFSH␤ sequence CSFGEMKE has a charge of Ϫ1. The hTSH␤ 44 -52 region corresponds to the carboxyl-terminal part of the long ␤2 loop identified by Keutmann et al. (30), which forms a wedge-shaped and partly surface-exposed extrusion in proximity to the determinant loop (2). However, none of the resulting hTSH/hFSH chimeras showed any follitropic activity in the three different assay systems (Table I). Interestingly, the thyrotropic activity of 105 FSH 112 , but not that of 44 FSH 52 was significantly reduced (Fig. 9), in accord with previous studies on hTSH structure-function using a synthetic peptide approach (21). DISCUSSION Previous studies on glycoprotein hormone specificity had focused on analogs that bound either to the CG/LH or the FSH receptor. These studies had suggested that domains within the seat-belt region of the ␤-subunit are involved in directing gonadotropin specificity (18 -20). It is unknown, however, how hTSH specificity is achieved and whether the seat-belt is critical for interaction with the TSH receptor. We directly compared the effects of systematically replacing the hTSH␤ seatbelt and its individual intercysteine segments with the corresponding regions of two different hormones, hCG and hFSH in parallel. Conversely, the hTSH␤ seat-belt residues were introduced into hCG. This strategy allowed us to characterize the role of the seat-belt for hTSH activity as well as specificity and to reveal divergent principles of specificity regulation among the members of the glycoprotein hormone family (see Table II).
Ala cassette mutations showed that the primary sequence of I-hCG (70,000 -100,000 counts/well) for 20 h at room temperature. After repeated washing, cells were dissolved in 1 N NaOH and counted in a ␥ counter. B, cAMP production at the CG/LH receptor. cAMP induction by the hTSH/hCG chimeras was determined in MA-10 cells according to the protocol described in Fig. 4B and under "Experimental Procedures." Basal cAMP levels were Ͻ 1.5 pmol/ml. C, progesterone production in MA-10 cells. Cells were incubated with hCG-wt or hTSH/hCG constructs for 6 h at 37°C, 5% CO 2 . Progesterone concentrations were determined in the supernatant using a commercially available progesterone radioimmunoassay kit (ICN). A representative experiment, repeated at least twice, is shown. the seat-belt is essential for hTSH receptor binding and activation. Of central importance was the negatively charged Asp 94 of the determinant loop since a single mutation of this residue to Lys, but not to Glu, abolished hTSH receptor binding and activity. The critical role of the negative charge of Asp 94 , which is conserved in all known glycoprotein hormone ␤-subunits and forms a non-bonded interaction with ␣Thr 54 (2, 3), was first identified in hCG by Chen et al. (27), suggesting that this residue is universally important for the members of the glycoprotein hormone family.
Our chimeric studies demonstrated that the seat-belt region of the hTSH ␤-subunit, if placed into the context of the hCG ␤-subunit, confers thyrotropic activity although the seat-belt alone was not sufficient for full thyrotropic activity. This suggests that additional hTSH␤ domains beside the seat-belt may contribute to hTSH specificity or that the hCG ␤-subunit con-tains segments that restrict interaction with the TSH receptor. In this respect, it had been shown that removal of the Cterminal extension peptide of hCG (31) as well as of the Nlinked carbohydrate side chain at ␣Asn 52 increased the weak inherent thyrotropic activity of hCG (11). This thyrotropic activity of native hCG however, unlike the chimera described here, requires 1000-fold higher concentrations than TSH itself in most systems (11,31). In contrast to the results with the hCG/hTSH chimera, the hCG␤ seat-belt, in the context of the hTSH ␤-subunit, was sufficient for full CG/LH receptor binding and secretory response. This reciprocal exchange of hCG/hTSH receptor specificity was mutually exclusive as both chimeras possessed no significant residual activity at their native receptor.
Remarkably, introduction of the hFSH seat-belt into the hTSH ␤-subunit did not result in FSH receptor binding or follitropic activity, and the hTSH/hFSH chimera retained most FIG. 6. Thyrotropic activity of the hCG/hTSH seat-belt chimera 94 TSH 109 . This chimera bears the hTSH␤ seat-belt introduced into the hCG ␤-subunit. A, TSH receptor binding. TSH receptor binding experiments were carried out as described in Fig. 3B. B, cAMP induction at the TSH receptor. cAMP production at the hTSH receptor was determined in JP09 cells as described in Fig. 3B (see legend to Fig. 3). 94 TSH 109 did not possess significant gonadotropic properties (Table I). A representative experiment, repeated at least twice, is shown.

FIG. 7. Thyrotropic activity of hTSH/hFSH seat-belt chimeras.
A, TSH receptor binding. TSH receptor binding experiments for the hTSH/hFSH chimeras were carried out as described in Fig. 3B. B, cAMP induction at the TSH receptor. The ability of the chimeras to induce cAMP production at the hTSH receptor was assessed in JP09 cells as described in Fig. 3B. A representative experiment, repeated at least twice, is shown. of its thyrotropic activity. In contrast to this finding, hCG could be converted to hFSH by placing the hFSH␤ seat-belt into the hCG ␤-subunit (18, 19), and hFSH adopted partial CG/LH receptor binding after exchange of its determinant loop with the corresponding hCG sequence (20) (see Table II). Hence, the role of the seat-belt in conferring specificity appears to depend on the particular subunit into which it is introduced. These findings are best reconciled by considering the concept of "negative specificity," which proposes that specificity of glycoprotein hormones evolved independently from signal transduction by the introduction of domains that block inappropriate ligandreceptor interactions (9,19). In this respect, it is interesting to note that the net charge of the determinant loop, the N-terminal part of the seat-belt region, is similar in hTSH (Ϫ2) and hFSH (Ϫ3) but different from that in hCG (ϩ1). Thus, it is conceivable that a net positive charge of the determinant loop, in conjunction with the carboxyl-terminal segment of the seatbelt, interferes with hormone binding to the TSH as well as FSH receptor; whereas a net negative charge, again in conjunction with carboxyl-terminal residues reduces interaction with the CG/LH receptor. From an evolutionary standpoint, it is justifiable to assume that diversification and ligand selectivity did not evolve by development of new mechanisms of receptor activation but rather by the emergence of inhibitory domains that impose steric hindrances, thus allowing only the intended ligand to interact with the common activation domain. This concept of negative specificity is not without precedent among cystine-knot growth factors. Thus, binding specificity among members of the neurotrophin family is achieved by the cooperation of distinct active and inhibitory binding determinants that restrict ligand-receptor interactions, enabling the creation of analogs with multiple specificities (32).
Our findings extend the original observation of Moore et al. (17), who proposed that the variable charge of this loop may act as a determinant of hormone specificity. However, our data show that the carboxyl-terminal segment of the seat-belt is of similar importance for specificity, and charge differences of the determinant loop per se are, therefore, not sufficient for the switch of hormonal activity. Indeed, conversion of hTSH to a full hCG analog required concomitant replacement of the determinant loop as well as the carboxyl-terminal segment of the seat-belt, which displays a high degree of sequence but not charge heterogeneity among the glycoprotein hormones. Interestingly, the luteotropic activity of the hTSH/hCG determinant loop chimera could be increased by concomitant introduction of a cluster of Lys residues into the spatially unrelated ␣11-20 domain, previously shown to increase receptor binding of hTSH as well as hCG (13).
In an attempt to identify domains determining hTSH/hFSH specificity outside the seat-belt, we focused on regions hTSH␤ 44 -52, which correspond to the carboxyl-terminal part of the long ␤2 loop identified by Keutmann et al. (30), and the ␤ carboxyl terminus 105-112. These domains were chosen because they display the greatest degree of charge heterogeneity between their ␤-subunits. The decrease of thyrotropic activity of the 105 FSH 112 chimera confirmed the importance of the hTSH␤ carboxyl terminus, which was identified with a synthetic peptide approach (21), for hTSH activity. However, introduction of hFSH residues into these regions of the hTSH ␤-subunit did not confer FSH receptor binding or follitropic activity to any of these chimeras. This was in agreement with the findings of Campbell et al., who showed that the long ␤2 loop was not important for hCG versus hFSH specificity (18). Thus, charged residues appear to play a lesser role in determining hTSH/hFSH specificity. It is possible that hTSH/hFSH specificity is not located within distinct segments of the ␤-subunit but is mediated by a combination of topically related domains although the present study was not designed to systematically test this possibility.
Our study cannot define the molecular mechanisms whereby the seat-belt determines glycoprotein hormone specificity as this will require complete elucidation of the structure of hormone-receptor complexes. In this respect, the seat-belt could either directly contact the receptor or influence the conformation of functionally important but unrelated portions of the hormone. An indirect effect of the seat-belt on the conformation of the ␣-subunit would be consistent with antibody binding studies showing that the conformation of the ␣-subunit could change depending upon with which ␤-subunit it associated (33), as well as with a recent model of glycoprotein hormonereceptor interaction predicting that the seat-belt does not directly contact the receptor (6). It could also explain the lack of FIG. 8. Follitropic activity of hTSH/hFSH seat-belt chimeras. hFSH-wt or hTSH/hFSH chimeras were incubated with FSH-R/293 cells following the protocol described for JP09 cells in Fig. 3B. Basal cAMP levels were Ͻ 2 pmol/ml. The hTSH/hFSH chimeras did not show specific binding in a rat testis radioreceptor assay. Further, there was no significant progesterone production in Y-1 cells (Table I). A representative experiment, repeated at least twice, is shown.
FIG. 9. Thyrotropic activity of hTSH/hFSH chimeras 44 FSH 52 and 105 FSH 112 . The ability of the chimeras to induce cAMP production at the hTSH receptor was assessed in JP09 cells as described in Fig. 3B. In contrast, these chimeras did not posses follitropic activity (Table I). A representative experiment, repeated at least twice, is shown. TSH receptor binding of hTSH␤ peptides spanning the seatbelt region (21), as well as observations that mutations of identical ␣-subunit residues have hormone-dependent effects on activity (10 -13, 25, 34). In this respect, we have shown that identical mutations in the ␣33-44 domain, which includes a positively charged ␣-helix at ␣40 -46, truncation of the ␣-carboxyl terminus, and deletion of the carbohydrate consensus sequence at ␣Asn 52 , all affected hTSH subunit association or activity differently than in the analogous hCG and hFSH mutants (10 -12). Intriguingly, these ␣-subunit domains are in close proximity to the seat-belt in the crystal structure of hCG and an hTSH homology model (5,13). It has thus been proposed that they may form a composite receptor-binding domain (2,3). In contrast, a peripheral ␣-subunit receptor binding domain located at the tip of the ␣1 loop, ␣11-20, appears to be important for all glycoprotein hormones (13). It is tempting to spec-ulate that the lack of specificity of the ␣11-20 domain is related to its distance from the seat-belt region. On the other hand, the ␣-subunit domains in close proximity to the ␤-subunit seat-belt may be spatially oriented by the seat-belt to contact the appropriate receptor in a hormone-dependent fashion. In this respect, a direct cooperation between the complementary charged residues Lys 91 of the ␣-subunit and Asp 397 of the CG/LH receptor has recently been demonstrated (35).
Thus our findings suggest that, during the evolutionary divergence of the glycoprotein hormones from a common ancestor gene (36), determinants of ligand specificity have evolved independently and in different ways. The seat-belt region appears to be critical to direct glycoprotein hormone binding to either the CG/LH receptor or to the TSH and FSH receptor. Determinants mediating discrimination between the TSH and FSH receptor remain to be elucidated.

TABLE I
Receptor binding and biological activity of chimeric analogs at the TSH, CG, and FSH receptor Thyrotropic, luteotropic, or follitropic activities of the chimeric analogs constructed by PCR-based mutagenesis and obtained by expression in CHO cells were tested in a variety of bioassays as described under "Experimental Procedures." Due to limitations of the amounts provided by transient transfection and due to the considerably reduced activity of certain analogs, it was not possible to calculate the half-maximum binding inhibition or stimulation in all cases. Therefore, the relative values of this table do not compare K d or EC 50 . Rather, for the radioreceptor assays (RRA), values represent relative binding inhibition of an analog at the concentration of the appropriate wild type hormone resulting in half-maximal inhibition. Similarly, for cAMP and progesterone assays, the relative stimulation by hTSH mutants at the EC 50 of the wild type hormone is given. *, no significant relative receptor binding or biological activity was observed at half-maximal concentration of wild-type hormone. Shown are mean values from individual experiments performed in triplicate and repeated at least twice. The S.E. were usually Ͻ10% (see also Figs. 3-9 for representative experiments). Prog., progesterone.

TABLE II Glycoprotein hormone seat-belt chimeras and their receptor specificity
Results are from this paper unless specified otherwise. Receptor specificity was determined either by receptor binding and/or activity assays (see text for details). For most chimeras, binding affinity and activation correlated closely. hTSH bearing the entire hCG seat-belt was 10-fold more potent for progesterone compared with cAMP stimulation (see text). The region between the 10th and the 11th Cys (C10-C11) corresponds to the determinant loop, and the region between the 11th and the 12th Cys (C11-C12) to the carboxyl-terminal segment of the seat-belt. ϩϩϩ, specificity equivalent to native hormone (80 -100%); ϩϩ, 20 -80%; ϩ, 5-20%; Ϫ, Ͻ5% of native hormone. The known cross-reactivity of natural glycoprotein hormones, such as thyrotropic activity of hCG, by comparison, occurs with much higher amounts and requires a Ͼ1000-fold molar excess relative to the native hormone in the assays employed (11,31