A Chicken Gonadotropin-releasing Hormone Receptor That Confers Agonist Activity to Mammalian Antagonists

Mammalian receptors for gonadotropin-releasing hormone (GnRH) have over 85% sequence homology and similar ligand selectivity. Biological studies indicated that the chicken GnRH receptor has a distinct pharmacology, and certain antagonists of mammalian GnRH receptors function as agonists. To explore the structural determinants of this, we have cloned a chicken pituitary GnRH receptor and demonstrated that it has marked differences in primary amino acid sequence (59% homology) and in its interactions with GnRH analogs. The chicken GnRH receptor had high affinity for mammalian GnRH (K i 4.1 ± 1.2 nm) , similar to the human receptor (K i 4.8 ± 1.2 nm). But, in contrast to the human receptor, it also had high affinity for chicken GnRH ([Gln8]GnRH) and GnRH II ([His5,Trp7,Tyr8]GnRH) (K i 5.3 ± 0.5 and 0.6 ± 0.01 nm). Three mammalian receptor antagonists were also pure antagonists in the chicken GnRH receptor. Another three, characterized by d-Lys6 ord-isopropyl-Lys6 moieties, functioned as pure antagonists in the human receptor but were full or partial agonists in the chicken receptor. This suggests that the Lys side chain interacts with functional groups of the chicken GnRH receptor to stabilize it in the active conformation and that these groups are not available in the activated human GnRH receptor. Substitution of the human receptor extracellular loop two with the chicken extracellular loop two identified this domain as capable of conferring agonist activity to mammalian antagonists. Although functioning of antagonists as agonists has been shown to be species-dependent for several GPCRs, the dependence of this on an extracellular domain has not been described.

Gonadotropin-releasing hormone (GnRH) 1 (Glu(P)-His-Trp-Ser-Tyr-Gly-Leu-Arg-Pro-Gly 10 NH 2 ) is synthesized in hypothalamic neurons and released into the hypothalamic-hypophysial portal system to regulate the synthesis and release of pituitary gonadotropins that in turn regulate the function of the testes and ovaries. Consequently, GnRH analogs are extensively employed as therapeutic agents in sex hormone-dependent diseases (1)(2)(3). The GnRH receptor is a rhodopsin-like G protein-coupled receptor (GPCR) (4). Progress has been made in defining the structure of mammalian GnRH receptors (5)(6)(7)(8) and in identifying amino acid residues that are important for ligand binding (9 -12) and coupling to G proteins (13,14). Some insight has also been obtained into the molecular entities involved in receptor activation. The interaction of Asn 2. 50(87) , in transmembrane domain 2 (TM2), with Asp 7.49(318) in TM7 of the mouse GnRH receptor appears to have a role in stabilizing the activated receptor conformation (7,8). In common with other GPCRs, the highly conserved Asp-Arg (DR) sequence at the intracellular boundary of TM3 has an integral role in activation of the GnRH receptor (15). Nevertheless, the structural features that determine whether a ligand will interact with the GnRH receptor as an agonist or antagonist, stabilizing the active or inactive conformation of the receptor, are poorly understood.
The high amino acid sequence homology of the mammalian GnRH receptors (over 85%) is paralleled by a similarity in pharmacological properties (4), which has made comparative sequence analysis of conserved and altered amino acids uninformative in identifying functionally important residues. In contrast, nonmammalian GnRH receptors have distinctly different pharmacology, both in ligand selectivity and G protein coupling (4, 16 -19). Biological assays indicate that the chicken GnRH receptor exhibits well defined differences in ligand se-lectivity compared with mammalian GnRH receptors (16). In particular, some antagonists of mammalian GnRH receptors act as agonists of the chicken GnRH receptor, stimulating luteinizing hormone (LH) release from chicken pituitary cells (20). Cloning and characterization of the chicken gonadotrope GnRH receptor would, therefore, potentially provide the means for identifying domains and residues involved in ligand selectivity and underlying receptor activation.
We report the cloning of a novel GnRH receptor from chicken pituitary that differs from the mammalian GnRH receptor in its primary structure, ligand selectivity, and in the agonistic behavior of certain mammalian GnRH receptor antagonists. Analysis of the functional properties of a range of antagonists of the mammalian GnRH receptors in the chicken GnRH receptor shows that agonism in the chicken receptor is conferred by a basic D-Lys or D-Ipr-Lys in position 6. Chicken-human chimeric receptors identified the receptor determinant of the agonist activity of the "antagonist" peptides as extracellular loop 2 (EC2) of the chicken GnRH receptor.

EXPERIMENTAL PROCEDURES
Reagents and Peptides-The sequences of GnRH analogs used in the study are shown in Table I. Agonists were synthesized by conventional solid-phase methodology and purified to more than 95% homogeneity by preparative C-18 reversed-phase chromatography (16 Receptor Amino Acid Residue Numbering-A numbering scheme, in which amino acids of GnRH receptors are numbered relative to the most conserved residues in the TM segments of the rhodopsin-like GPCRs, is used to facilitate comparison among different receptors (21). The amino acid identifier, which follows the name of the amino acid, consists of the TM number followed by the position of the amino acid relative to the most conserved residue in that TM, which is assigned the number 50, and the sequence number of the amino acid in its receptor, in parentheses. For example, the Asp residue that is located immediately amino-terminal to the most conserved residue, Pro 7.50 , in TM7 is designated Asp 7.49(319) in the human GnRH receptor and Asp 7.49(310) in the chicken GnRH receptor (see Fig. 2 for alignment of these receptor sequences).
Cloning of the Chicken GnRH Receptor Gene-A 120-base pair product encoding EC3 was amplified from 1 g of chicken genomic DNA using the degenerate primer pairs (JH5S/JH6A) to conserved regions in TM6 and TM7 of GnRH receptors (22) (Fig. 1). The 120-base pair product was cloned into the pMOS-blue vector (Amersham Pharmacia Biotech), labeled with [␣Ϫ 32 P]dCTP (Amersham Pharmacia Biotech) by random priming using the Megaprimer labeling system (Amersham Pharmacia Biotech), and used to screen a Charon 4A chicken genomic library (23). Three positive clones were isolated after tertiary screening.
DNA was purified from one of these clones (cGnRH-R.g-22). Southern blot analysis of a PstI digest of cGnRH-R.g-22 identified a 2.7-kb fragment that hybridized with the 120-base pair probe used to screen the library. The 2.7-kb fragment (pCH1) was cloned into pSKϩ (Bluescript, Strategene, La Jolla, CA) and partially sequenced. The pCH1 clone contained the entire 3Ј-coding region of the GnRH receptor gene, including an intron within IC3 (Fig. 1). Partial nucleotide sequence analysis of PstI fragments cloned into the pSKϩ vector from the GnRH-R.g-22 identified a clone (pCH2) encoding the region of the cGnRH-R from the 4th amino acid of the amino-terminal region to the middle of TM4. The pCH2 construct was verified to contain partial sequences of two introns at each end (Fig. 1).
Cloning of Chicken GnRH Receptor cDNA-Four antisense primers were designed from the pCH1 nucleotide sequence (Fig. 1). Two of these primers YS6, 5Ј-TCTAGTCTCCTTTTGGGTACATCTCTTC-3Ј, and YS5, 5Ј-TGGGTACATCTCTTCAGCACACCGT-3Ј, were designed to hybridize to sequences of the 3Ј-untranslated region. Primers YS4, 5Ј-GGTGCATGTGTGCAGCAAACC-3Ј, and YS3, 5Ј-GGGCATCCTCTG-GATCATGGC-3Ј, were designed on the basis of the EC3 sequences of the receptor. cDNA was synthesized from 2 g of total RNA isolated from the pituitaries of castrated chickens (Marathon cDNA synthesis kit, CLON-TECH, Palo Alto, CA). The YS6 primer was used to initiate first strand synthesis. Marathon cDNA adapters, which contain the AP1 and AP2 primers, were ligated to the ends of the cDNA. The 5Ј-end of the chicken GnRH receptor cDNA was amplified by three rounds of nested PCR with KlenTaq polymerase (CLONTECH), using the following combination of primers: round 1 (AP1/YS5), round 2 (AP2/YS4), and round 3 (AP2/YS3). Southern blot analysis of the PCR products identified three bands of 0.8, 0.9, and 1.2 kb that hybridized to ␣Ϫ 32 P-labeled pCH2. These bands were subcloned into the pMOS-blue vector. One clone, pCH3, was sequenced and shown to consist of the 5Ј-untranslated region of chicken GnRH receptor cDNA extending to the YS3 primer ( Fig. 1).
Two sense primers were designed to the region 5Ј to the start codon of the chicken receptor as follows: YS1, 5Ј-GCTGAGCACTTGTGCT-GCCT-3Ј, and YS2, 5Ј-CACTTGTGCTGCCTGACTTGCTG-3Ј (Fig. 1). Two rounds of nested PCR with primer combinations (YS1/YS6 and YS2/YS5) yielded a ϳ1.2-kb band from the castrated chicken pituitary DNA. This band was subcloned into the pMOS vector. Two clones, pCH4 and pCH5, were isolated, and nucleotide sequencing confirmed that they contained the entire open reading frame of the chicken GnRH receptor. Comparison of the nucleotide sequences of these clones with those of the genomic clones identified a number of differences that might have arisen during the PCRs or represented polymorphisms. The pCH4 clone showed nucleotide substitutions resulting in amino acid substitutions of Gln for Lys 1.30(40) and Tyr for His 1.65(75) compared with the genomic sequences. The pCH5 clone contained Arg in place of Cys 5.23(189) of the genomic clone. Both clones showed a substitution of Arg for Gln 7.77(338) . As it is uncertain whether these differences represent polymorphism or PCR errors, three different DNA fragments from the pCH4, pCH5, and pCH1 were ligated to reconstruct a chicken GnRH receptor (cGnRH-R) encoding amino acid sequences identical to those of the genomic clone. The XbaI/SphI fragment from the pCH5 was ligated to the SphI/EcoRI fragment of the pCH4 and subcloned into the pSKϩ vector, eliminating the amino acid substitutions at positions 40, 75, and 189 (the XbaI and EcoRI sites were in the pMOS vector). To eliminate the substitution at position 338, the NotI/AvaI fragment from this chimeric clone was ligated to the AvaI/PstI fragment from pCH1 (NotI and PstI sites were in the pSKϩ vector) and subcloned into the pSK vector to yield the full-length cGnRH-R. The NotI/XhoI fragment of the cGnRH-R was subcloned into the mammalian expression vector pcDNA I/AMP (Invitrogen, Carlsbad, CA). Sequencing of the resulting chicken GnRH receptor clone verified that the encoded amino acid sequence was identical to that of the genomic clone.
Construction of Chimeric Chicken-Human GnRH Receptor-Silent restriction endonuclease sites for BsrGI, StuI, HpaI, and SnaBI were introduced at the extracellular ends of TM4 to TM7 of the human GnRH receptor by site-directed mutagenesis (24). The resulting receptor exhibited ligand binding and IP accumulation that were indistinguishable from those of the wild type human GnRH receptor. Chimeric GnRH receptors were constructed by excising the extracellular loops of the human receptor and replacing them with equivalent loops of the chicken GnRH receptor (Fig. 2), using appropriate restriction sites. EC2 and EC3 of the chicken GnRH receptor were generated by PCR amplification of the chicken GnRH receptor using Deep Vent DNA polymerase (New England Biolabs, Beverly, MA) and primers flanked by the appropriate restriction endonuclease recognition sequences. Mutations were confirmed by DNA sequencing.
Inositol Phosphate (IP) Assay-IP production was measured as described previously (26). Briefly, cells were labeled with myo-[ 3 H]inositol (1 Ci/ml, Amersham Pharmacia Biotech) in Medium 199, supplemented with 2% fetal bovine serum, penicillin, and streptomycin for 16 -22 h. Cells were washed twice with buffer A (140 mM NaCl, 4 mM KCl, 1 mM CaCl 2 , 1 mM MgCl 2 , 20 mM HEPES, 0.1% bovine serum albumin, and 8 mM D-glucose, pH 7.4) and incubated for 1 h at 37°C with 10 mM LiCl and appropriate concentrations of GnRH agonist or with antagonist in the presence and absence of 1 or 10 nM [Gln 8 ]GnRH (chicken GnRH receptor) or 1 nM mammalian GnRH (human GnRH receptor). Experiments were performed in duplicate and repeated at least three times.
Radioligand Binding Assay-Radioligand binding assays were performed on intact cells 48 h after transfection. Transfected cells in 12-well culture plates were washed and incubated for 3 h at 4°C with 125 I-GnRH-A (100,000 cpm) and various concentrations of unlabeled GnRH agonists or antagonists in buffer A as previously described (14,27). Nonspecific binding was determined in the presence of 1 M unla-beled antagonist 27. After incubation, the cells were washed three times and solubilized in 0.5 ml of 0.1 M NaOH, and the radioactivity was counted. All experiments were performed in triplicate and repeated at least twice.
Data Analysis-K i , ED 50 (peptide concentration required for halfmaximal IP formation), and IC 50 (antagonist concentration required to half-maximally inhibit IP formation) values were calculated by nonlinear regression analysis using the PRISM program (GraphPad Inc., San Diego, CA).

Cloning of Chicken GnRH
Receptor cDNA-Screening of chicken pituitary cDNA libraries failed to identify any GnRH receptor clones. As the chicken thyrotropin-releasing hormone receptor was cloned from the same libraries (27), it appears that GnRH receptor mRNA is expressed at low levels in chicken pituitaries, as in some mammalian species (28,29). The full-length chicken GnRH receptor cDNA was therefore isolated by the combined strategies of genomic library screening, 5Ј-rapid amplification of cDNA ends, and PCR of total RNA prepared from pituitaries of castrated chickens (see under "Experimental Procedures"). Sequencing of genomic clones revealed that the chicken GnRH receptor has introns located in TM4 and IC3 (Fig. 1) as in mammalian GnRH receptors (30,31). But, unlike the mammalian receptors, an additional intron is present in the amino-terminal domain (Fig. 1). Full-length chicken GnRH receptor cDNA clones obtained by PCR varied in the coding of the amino acids Lys 1.30 (40) , His 1.65(75) , Cys 5.23(189) , and Gln 7.77(338) (see "Experimental Procedures"). Since it was uncertain whether these were errors incorporated by PCR or polymorphisms, a "wild type" chicken GnRH receptor cDNA, which corresponded to the genomic sequence, was constructed ( Fig. 1) and used for all of the experiments described.
The cloned chicken GnRH receptor cDNA encodes a 375amino acid polypeptide with the seven hydrophobic putative TM domains, connected by three cytosolic and three extracellular loops, extracellular amino-terminal and cytosolic carboxyl-terminal domains that are characteristic of GPCRs (Fig. 2). The chicken GnRH receptor has low amino acid sequence identity (41%) and homology (59%) with the human and other mammalian GnRH receptors, excluding the highly variable amino-terminal domain and the carboxyl-terminal tail that is absent from mammalian GnRH receptors (4) but present in other nonmammalian GnRH receptors (17)(18)(19). The carboxylterminal domain has been shown to regulate GnRH receptor desensitization and internalization (32,33). Also similar to other nonmammalian GnRH receptors, the chicken GnRH receptor has Asp in both loci of the conserved helix 2/helix 7 functional microdomain. This microdomain consists of Asn 2.50(87) /Asp 7.49(319) in the human GnRH receptor and Asp 2.50 /Asn 7.49 in most other GPCRs, and it regulates GPCR coupling and expression (7,8). The presence of two Asp residues in the chicken GnRH receptor suggests that the nonmammalian GnRH receptors may represent an evolutionary intermediate between the Asp 2.50 /Asn 7.49 arrangement of the microdomain found in most GPCRs and the Asn 2.50 /Asp 7.49 of the mammalian GnRH receptors. The presence of Asp 2.50 / Asn 7.49 in the Drosophila melanogaster homolog of the GnRH receptor (34) supports this conclusion. Other residues that are important for coupling of mammalian GnRH receptors to cytosolic signal transduction, the Arg cage motif (DRXXX(I/V)) at the cytosolic end of TM3 (15) and Ala 5.29(261) in IC3 (14) are conserved in the chicken GnRH receptor. All of the residues previously shown to have a role in ligand binding of mammalian GnRH receptors, Asp 2.61(98) (12), Asn 2.65(102) (9), Lys 3.32(121) (11), and Glu 7.32(301) (10), are conserved in the chicken receptor (Fig. 2).
GnRH Agonist Interactions-The cloned chicken GnRH receptor exhibited high affinity binding to a series of GnRH agonists (Table II). It had high affinity for both mammalian GnRH and the native chicken ligand [Gln 8 ]GnRH (K i 4.1 Ϯ 1.2 and 5.3 Ϯ 0.5 nM, respectively), in contrast to the human GnRH receptor, which had similar high affinity for mammalian GnRH (K i 4.8 Ϯ 1.2 nM) but low affinity for [Gln 8 ]GnRH (K i 174 Ϯ 69 nM) (Table II). The chicken GnRH receptor had much higher affinity for GnRH II (0.60 Ϯ 0.01 nM), also contrasting with low affinity in the human GnRH receptor (K i 39 Ϯ 8.5 nM). The substitution of a D-amino acid for Gly 6 in GnRH is thought to constrain the peptide in the biologically active ␤-II-turn conformation and enhance binding affinity for mammalian GnRH receptors (see Ref. 4). This is reflected in the 3-fold enhancement of the affinities of [D-Arg 6 ]GnRH II and GnRH-A compared with unconstrained ligands in the human GnRH receptor (Table II). In contrast, there was little or no enhancement in binding affinity of the D-amino acid-containing ligands for the chicken GnRH receptor (Table II). These direct receptor binding studies confirm previous bioassay data and support the suggestion that incorporation of D-amino acids in position 6 of GnRH does not enhance binding affinity for the chicken GnRH receptor and that GnRH binds the chicken GnRH receptor differently from the mammalian receptor (16).
The cloned chicken GnRH receptor stimulated IP accumulation in response to GnRH agonists (Table II). The rank order of agonist ED 50 values was the same as the rank order of K i values determined in ligand binding assays (Table II) was not detected in cells transfected with the chicken GnRH receptor.
Agonism of Mammalian Receptor Antagonists-We have previously reported that some antagonists of mammalian GnRH receptors behave as agonists, stimulating LH release from cultured chicken pituitary cells (20). A series of mammalian receptor antagonists was used to define the structural basis of agonism in the cloned chicken pituitary. All of the analogs functioned as full antagonists with high binding affinity in the human GnRH receptor (Table II and Fig. 3). The analogs had much lower binding affinity for the cloned chicken GnRH receptor (Table II). Furthermore, three analogs exhibited distinct agonist activity, stimulating IP accumulation in cells expressing the cloned chicken GnRH receptor (Table II and Fig. 3), whereas the other three analogs functioned as pure antagonists (Table II). Antagonist 135-18 was a full agonist, and antagonists 135-25 and 26 were partial agonists (Fig. 3). All of the analogs that showed agonist activity contained a basic D-amino acid (D-Lys or D-Ipr-Lys) in position 6, and all peptides that were antagonists in the chicken GnRH receptor had uncharged side chains in this position (Table I).
The agonist behavior of three antagonists in the chicken receptor may arise from interactions of these peptides with amino acid residues that are unique to the chicken receptor. We attempted to identify the domains of the chicken receptor that are involved using chimeric receptors in which EC domains of the human receptor were substituted with EC domains of the chicken receptor. Since EC1 is highly conserved (Fig. 2), exchanges were confined to EC2 and EC3. The chimeric receptor containing EC3 of the chicken GnRH receptor did not bind GnRH or stimulate IP accumulation in response to GnRH, suggesting that the EC3 chimera was poorly expressed or uncoupled from activation of phospholipase C (data not shown). However, the chimera containing the chicken receptor EC2 substituted in the human receptor exhibited high affinity binding and IP accumulation. The EC2-containing chimera bound antagonists 26, 135-25, and 135-18 with high affinities (K i 7.8 Ϯ 2.9; 12 Ϯ 4.7; and 18 Ϯ 5.3 nM, respectively), which were similar to their affinities for the wild type human receptor and higher than those for the chicken receptor (Table II), indicating that the NH 2 -terminal domain, EC1, EC3, and superficial regions of the TM domains of the human receptor are major contributors to high affinity building of the antagonists (Tables  II). Antagonists 26 and 132-25, which were partial agonists in the chicken GnRH receptor, behaved as antagonists in the chimera (Fig. 4), similar to the wild type human receptor. Antagonist 135-18, which was a full antagonist in the human GnRH receptor and a full agonist in the chicken GnRH receptor, exhibited partial agonist behavior in the chimera (Fig. 4). As was found for the chicken GnRH receptor, no constitutive activity was detectable in the chimeric receptor. DISCUSSION The definitive molecular delineation of ligand binding, signal propagation, and G protein coupling of the human GnRH receptor and the development of GnRH analogs is a major goal in reproductive medicine. Progress in this regard has been made through the cloning of GnRH receptors and a combination of molecular modeling and mutagenesis studies (4). Since the various mammalian receptors have close sequence homology and similar ligand selectivity (4), information on the primary sequence of a related, but pharmacologically distinct, nonmammalian receptor would potentially contribute in these endeavors. The chicken GnRH receptor that we have cloned exhibits marked pharmacological differences in its interaction with GnRH agonist and antagonist analogs and has sequence differences from the mammalian receptors that may be used to identify functional residues.
Agonist-binding Site Differs in Chicken and Mammalian GnRH Receptors-The conservation of the Asp 2.61(98) , Asn 2.65(102) , and Lys 3.32(121) residues in the chicken GnRH receptor is expected, as these residues are believed to interact with the amino-(Glu(P) 1 -His 2 ) and carboxyl-terminal (Gly 10 -NH 2 ) residues of the GnRH ligands, which are conserved in the native mammalian and chicken forms of GnRH (9,11,12). As expected from biological assays, the chicken GnRH receptor does not distinguish between its cognate native ligand, [Gln 8 ]GnRH (K i 5.3 Ϯ 0.5 nM), and mammalian GnRH (K i 4.1 Ϯ 1.2 nM). Surprisingly, the affinity of the chicken receptor for both ligands was as high as the affinity of the human GnRH Amino acid numbers are indicated for the chicken GnRH receptor above the sequence and at the left side for both receptors. Sequences that are conserved in both receptors are shaded. The putative TM segments, EC and IC loops are assigned according to those predicted for the human receptor (4) and TM segments are indicated by double bars above the chicken GnRH receptor sequence; period, space. Human GnRH receptor sequences that were substituted with chicken GnRH receptor sequences in the chimeric receptors are indicated by bars below the human receptor sequence. receptor which is selective for mammalian GnRH (K i 4.8 Ϯ 1.2 nM) and binds [Gln 8 ]GnRH with low affinity (K i 174 Ϯ 69 nM). The high affinity of the chicken receptor is unexpected because high affinity binding of mammalian GnRH depends on an interaction of Arg 8 with an acidic residue in the EC3 domain (Glu 3.32(301) in the mouse and Asp 3.32(302) in the human) (10). The equally high affinity of mammalian GnRH and [Gln 8 ]GnRH binding to the chicken receptor suggests, therefore, that the Arg 8 -Glu 3.32(301) /Asp 3.32(302) interaction does not occur but is compensated by alternative interaction(s) in the chicken receptor. Thus, it appears that, although there are interactions that are common to the mammalian and chicken receptors, the binding sites differ in unique interactions.
This suggestion that [Gln 8 ]GnRH utilizes a different binding site in the chicken GnRH receptor is not unexpected.
[Gln 8 ]GnRH is not configured in the folded ␤ϪII-turn conformation characteristic of mammalian GnRH (36,37), and sub-stitution of a D-amino acid for Gly 6 , which enhances the folded conformation, does not enhance binding affinity for the chicken receptor as it does in the human receptor (Table II). The aminoand carboxyl-terminal residues, Glu(P) 1 , His 2 , and Gly 10 NH 2 , of the ligand and their cognate mammalian receptor binding residues, Asp 2.61(98) , Asn 2.65(102) , and Lys 3.32(121) , are all conserved in species of fish, amphibians, birds, and mammals (4,(17)(18)(19). We therefore propose that the interactions between these residues are also conserved. Since the receptors have different requirements for ligand conformation, we further propose that the spatial arrangement of the receptor-binding sites differs between the chicken and mammalian receptors. This allows the accommodation of binding of the configured ligand in mammalian receptors and the nonconfigured ligands in the chicken GnRH receptor, through the same interactions. This proposal suggesting that the receptors are configured differently is supported by the presence of two Asp residues (Asp 2.50(87) and Asp 7.49(310) ) in the helix 2/helix 7 functional microdomain of the wild type chicken GnRH receptor, an arrangement that is not tolerated in the mammalian GnRH receptor (7,8).
Mammalian Antagonists Function as Agonists in the Chicken GnRH Receptor-The six mammalian GnRH receptor antagonists studied had decreased affinities for the chicken GnRH receptor compared with the human receptor, emphasizing the differences in the ligand-binding site. Three of the selected mammalian GnRH receptor antagonists were also pure antagonists in their interaction with the chicken GnRH receptor. Another three pure mammalian antagonists, 26, 135-25, and 135-18, behaved as partial or full agonists in their interaction with the chicken GnRH receptor. There was no correlation of agonist/antagonist behavior and binding affinities at the chicken GnRH receptor (Table II).
Analysis of the sequences of the antagonist analogs revealed that a single feature is unique to those with agonistic activity at the chicken receptor. All three analogs with agonist activity have a basic D-amino acid substitution (D-Lys or D-Ipr-Lys) for Gly 6 ( Table I). Antagonists 135-18 and 135-25, which are full and partial agonists, respectively, differ only in the presence of Ile and 1-MePal in position 5. This suggests that the larger aromatic side chain (1-MePal) prevents 135-25 from acting as a full agonist in the chicken receptor. This conclusion is supported by the partial agonism of antagonist 26 that also has a large aromatic side chain (Tyr) in position 5. It appears that the large aromatic side chain changes the orientation of the adjacent D-Lys moiety such that its ability to interact with a cognate receptor amino acid is impaired, thus decreasing agonist activity.
EC2 of Chicken GnRH Receptor Is a Determinant of Agonist Behavior of Antagonists-Insertion of the chicken GnRH receptor EC2 domain into the human receptor conferred partial agonism to antagonist 135-18, which was a full agonist in the chicken receptor. Antagonists 135-25 and 26, which were partial agonists in the chicken receptor, had no agonistic activity in the chimera. Thus, although the chicken EC2 domain is a determinant of agonist activity of antagonist 135-18, it is insufficient to confer the same degree of agonist activity as is found in the complete chicken receptor. This suggests that a combination of appropriate domains is required or that the molecular dynamics of the transition between active and inactive states differ in the human and chicken receptors such that the interaction of antagonist 135-18 with the EC2 chickenhuman chimera cannot adequately stabilize the active conformation.
Proposed Mechanism of Agonist Activity of Antagonists-Contemporary thinking proposes that agonist analogs bind and stabilize the receptor in the active conformation, antagonists bind both the active and inactive states of the receptor, and inverse agonists bind and stabilize the inactive form (38). According to this hypothesis, antagonist analogs, in binding both inactive and active conformations of the receptor, would not disturb the equilibrium between both forms; agonists would drive the equilibrium in favor of the active conformation, and inverse agonists would drive the equilibrium in favor of the inactive conformation. The agonist activity of mammalian GnRH receptor antagonists at the chicken receptor is therefore interpreted as resulting from a structural alteration that confers a preference for binding to the active receptor conformation. D-Lys 6 or D-Ipr-Lys 6 of these antagonists appears to interact with a functional group, which is available only in the chicken receptor in the active conformation. These analogs can therefore stabilize the active conformation of the chicken receptor but not the active form of the mammalian receptor. Since a D-Lys or D-Ipr-Lys residue in position 6 of the mammalian GnRH receptor antagonists is the unique feature associated with agonist activity, it presumably interacts with a receptor amino acid side chain that is accessible in the active conformation of the chicken receptor and is absent from, or inaccessible in, the active conformation of the human GnRH receptor. A candidate for a strong interaction with D-Lys 6 would be an acidic residue in the receptor. Since substitution of the chicken EC2 in the human receptor conferred partial agonism to antagonist 135-18, the Glu 5. 34(200) and Glu 5.35(201) residues in the chicken EC2 (His 5.34(207) and Gln 5.35(208) in mammalian receptors) are candidates. Glu 5.35(201) is also present in the Xenopus GnRH receptor in which antagonist 135-18 also behaves as an agonist. 2 An alternative proposal is that ligand binding induces the active state of the receptor (39). In the ␤-adrenergic (40) and 5HT2A (41) receptors, a different positioning of certain ligands alters their ability to induce the active state of the receptor. It may be conceived that the interaction of D-Lys 6 or D-Ipr-Lys 6 with extracellular EC2 of the chicken receptor changes positioning of these ligands to allow interaction with receptoractivating sites.
Mutations in a number of GPCRs have been described that confer agonism to antagonists. Mutations in TM3 of the AT1 angiotensin receptor (39,42), rhodopsin (43), and the ␤-adrenergic receptor (40), in TM4 ofand ␦-opioid receptors (44), in TM6 of the -opioid receptor (45), in IC3 of the ␣ 2A -adrenergic receptor (46), and in IC2 of the V2 vasopressin receptor (47) conferred agonistic activity to antagonist ligands. The present study is the first description of a mutation in an extracellular domain that confers agonism to an antagonist. Receptor activation is believed to involve movement (e.g. rotation) of the TM domains and that this is propagated into structural changes in the connecting IC loop domains (reviewed in Refs. 48 and 49). Our demonstration that mutation of EC2 allows the interaction with a single residue in an antagonist to confer agonism indicates that a distinct relationship exists between EC and TM domains and that ligand interaction with EC domains can stabilize the receptor in the active conformation. Indirect evidence that molecular interactions with EC domains can lead to receptor activation has been obtained from antibody studies. Antibodies against EC2 of M1 and M2 muscarinic, AT1 angiotensin, and ␤ 1 -and ␤ 2 -adrenergic receptors caused receptor activation (50 -53), pointing to a role for this domain in stabilizing receptor conformation.
In conclusion, our findings here identified a single amino acid side chain (D-Lys 6 or D-Ipr-Lys 6 ) in mammalian GnRH receptor antagonists that confers agonist activities to these antagonists when interacting with the chicken GnRH receptor. This phenomenon is conferred to the human receptor with incorporation of EC2 of the chicken GnRH receptor. This suggests that an interaction of the D-Lys 6 or D-Ipr-Lys 6 side chain with a residue in EC2 of the chicken GnRH receptor stabilizes the active conformation of the receptor. Identification of other contact sites of this ligand will assist in delineating molecular distances of TM domains of the GnRH receptor in the active and inactive conformations and shed light on the molecular mechanism of ligand-mediated receptor activation.