H3 relaxin is a specific ligand for LGR7 and activates the receptor by interacting with both the ectodomain and the exoloop 2.

Leucine-rich repeat-containing, G protein-coupled receptors (LGRs) represent a unique subgroup of G protein-coupled receptors with a large ectodomain. Recent studies demonstrated that relaxin activates two orphan LGRs, LGR7 and LGR8, whereas INSL3/Leydig insulin-like peptide specifically activates LGR8. Human relaxin 3 (H3 relaxin) was recently discovered as a novel ligand for relaxin receptors. Here, we demonstrate that H3 relaxin activates LGR7 but not LGR8. Taking advantage of the overlapping specificity of these three ligands for the two related LGRs, chimeric receptors were generated to elucidate the mechanism of ligand activation of LGR7. Chimeric receptor LGR7/8 with the ectodomain from LGR7 but the transmembrane region from LGR8 maintains responsiveness to relaxin but was less responsive to H3 relaxin based on ligand stimulation of cAMP production. The decreased ligand signaling was accompanied by decreases in the ability of H3 relaxin to compete for (33)P-relaxin binding to the chimeric receptor. However, replacement of the exoloop 2, but not exoloop 1 or 3, of LGR7 to the chimeric LGR7/8 restored ligand binding and receptor-mediated cAMP production. These results suggested that activation of LGR7 by H3 relaxin involves specific binding of the ligand to both the ectodomain and the exoloop 2, thus providing a model with which to understand the molecular basis of ligand signaling for this unique subgroup of G protein-coupled receptors.

Relaxin and Leydig insulin-like peptide/relaxin-like factor (INSL3) 1 are peptide hormones with a two-chain structure similar to that of insulin (1,2). Relaxin is important for the function of reproductive tissues, heart, kidney, and brain (3), whereas INSL3 is essential for testis descent (4,5). We have recently demonstrated that two orphan leucine-rich repeatcontaining, G protein-coupled receptors (LGRs) with homology to gonadotropin and thyrotropin receptors, are capable of mediating the action of relaxin through a cAMP-dependent pathway (6). These two receptors, LGR7 and LGR8, share 50% sequence identity to each other, and contain a unique low density lipoprotein receptor-like cysteine-rich motif at the amino terminus. However, LGR7 and LGR8 do not have the consensus hinge region found in gonadotropin and thyrotropin receptors. In contrast to relaxin, INSL3 activates LGR8 but not LGR7; interactions between INSL3 and LGR8 were demonstrated by ligand-receptor cross-linking (7).
In addition to the two known human relaxin genes, H1 (8) and H2 (9), another related gene, designated H3 relaxin (H3), was identified recently. A synthetic peptide with a design based on this gene was found to possess relaxin activity in bioassays using the human monocyte cell line, THP-1 (10). Here, we demonstrate that H3 relaxin activates recombinant LGR7 but not LGR8. Taking advantage of the structural similarity of LGR7 and LGR8, and the differential specificity of relaxinrelated peptides to these receptors, we designed chimeric LGR7/LGR8 receptors to identify the domains in the receptor that are important for their ligand specificity. We demonstrate that both the ectodomain and exoloop 2 of LGR7 are important for ligand receptor binding and signaling.

MATERIALS AND METHODS
Hormones and Reagents-Porcine relaxin was purchased from the National Hormone and Peptide Program (Torrance, CA). Recombinant human H2 relaxin was a gift from Dr. Elaine Unemori (Connectics Co., Palo Alto, CA). H3 relaxin and human INSL3 were chemically synthesized and characterized as described previously (10,11). Anti-FLAG M1 monoclonal antibody and the FLAG peptide were purchased from Sigma Chemical Co. (St. Louis, MO). The soluble ectodomain of LGR7 was prepared as described previously (6). Briefly, cDNA for the ectodomain of human LGR7, named as 7BP, was fused in-frame with the prolactin signal peptide and the FLAG epitope at the 5Ј-end. At the 3Ј-end, the ectodomain was connected to the single transmembrane region of T cell surface antigen CD8 through a thrombin cleavage consensus region. Stable 293T cell lines expressing 7BP encoded in the pcDNA3.1-Zeo expression vector (Invitrogen Co., Carlsbad, CA) were selected using Zeocin (0.5 mg/ml). To release soluble 7BP anchored on the cell surface, cells were treated with thrombin (10 IU/ml) for 3 days under serum-free conditions. The cleaved 6-His-and FLAG epitopetagged recombinant 7BP in serum-free conditioned media was purified by sequential nickel and anti-FLAG affinity chromatography.
Construction of Chimeric Receptor cDNAs-Polymerase chain reaction-based, site-directed mutagenesis was performed to generate mutant receptor cDNAs (12) using cDNAs encoding human LGR7 (13) and human LGR8 (6). To predict the transmembrane helices in LGR7 and LGR8, a membrane protein topology prediction method based on a hidden Markov model (available at www.cbs.dtu.dk/services/TMHMM) was used (14). This method was also used to deduce the junctions of exoloops and transmembrane helices for the chimeric constructs. The junctional amino acid sequences for different chimeric receptors are * This work was supported in part by National Institutes of Health Grant HD23273 (to A. J. H.) and by Institute Block Grant 983001 to the Howard Florey Institute from the National Health and Medical Research Council (NHMRC) of Australia. 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.
listed in Table I. Polymerase chain reaction was performed using Vent DNA polymerase (New England BioLabs, Inc., Beverly, MA) in accordance with the manufacturer's instructions. All cDNAs were subcloned into the expression vector pcDNA3.1/Zeo (Invitrogen Co.). To allow efficient targeting of receptors to the cell surface, a lead cDNA sequence containing a prolactin signal peptide for secretion (MNIKGSP-WKGSLLLLLLVSNLLLCQSVAP) and an M1 FLAG (DYKDDDDK) epitope were added to the amino terminus of the mature region of all receptors (15). The expression constructs were purified using the Plasmid Maxi kit (Qiagen, Inc., Valencia, CA). Fidelity of the PCR products was confirmed by sequencing on both strands of the final constructs before use in expression studies.
Transfection of Cells and Analysis of Signal Transduction-Human 293T cells derived from human embryonic kidney fibroblast were maintained in Dulbecco's modified Eagle's medium/Ham's F-12 (Invitrogen Co.) supplemented with 10% fetal bovine serum, 100 g/ml penicillin, 100 g/ml streptomycin, and 2 mM L-glutamine. Before transfection, 2 ϫ 10 6 cells were seeded in 10-cm dishes (Becton Dickinson, Franklin Lakes, NJ). When cells were 70 -80% confluent, transient transfection was performed using 10 g of plasmid by the calcium phosphate precipitation method. Cells (2 ϫ 10 5 /ml) were placed on 24-well tissue culture plates (Corning, Corning, NY) and preincubated at 37°C for 30 min in the presence of 0.25 mM 3-isobutyl-1-methyl xanthine (Sigma Chemical Co.) before treatment with or without hormones for 16 h. At the end of incubation, cells and medium were frozen. After thawing, samples were heated to 95°C for 3 min to inactivate phosphodiesterase activity, and total cAMP was measured in triplicate by a specific radioimmunoassay as described previously (16). All experiments were repeated at least three times using cells from independent transfections. To monitor transfection efficiency, 0.5 g of ␤-galactosidase plasmid (17) was routinely included in the transfection mixture, and ␤-galactosidase activity in the cell lysate was measured as described previously (18). Statistical analysis was performed using Student's t test.
Determination of FLAG Epitope-tagged Receptors on the Cell Surface (M1 Binding)-The levels of cAMP production were normalized by correcting for varying expressions of receptors to monitor the levels of their tagged epitope. Transfected cells were washed twice with Dulbecco's PBS, and resuspended cells (2 ϫ 10 6 /tube) were incubated with the FLAG M1 antibody (50 mg/ml) (Sigma Chemical Co.) in Tris-buffered saline (pH 7.4) containing 5 mg/ml bovine serum albumin and 2 mM CaCl 2 (assay buffer) for 4 h at room temperature in siliconized tubes. Cells were then washed twice with 1 ml of assay buffer after centrifugation at 14,000 ϫ g for 15 s. The 125 I-labeled second antibody (antimouse IgG from sheep: ϳ400,000 cpm/tube) was added to the resuspended cell pellet and incubated for 1 h at room temperature. Cells were washed twice with 1 ml of assay buffer by repeated centrifugation before determination of radioactivity in cell pellets. Background binding was determined by adding excess amounts of the synthetic FLAG peptide (Sigma Chemical Co.) at a concentration of 100 g/ml.
Binding studies were performed with 100 l of 33 P-labeled H2 relaxin, labeled as previously described (20), and 100 l of competitor or blank in binding buffer at 25°C for 60 min. Saturation binding was performed using increasing concentrations of 33 P-labeled H2 relaxin, whereas competition experiments were performed with 100 pM 33 Plabeled H2 relaxin in the absence or presence of increasing concentrations of unlabeled peptides. Nonspecific binding was determined by an excess of H2 relaxin (1 M). After incubation, the cells were washed with PBS followed by their recovery from the plates using 500 l of 1 M NaOH before transfer to scintillation vials. Liquid scintillation mixture (Ultima Gold, Packard, Meriden, CT) was added to these vials for counting in a liquid scintillation analyzer (Packard 1900 TR). Data are expressed as mean Ϯ S.E. of the percentage of specific binding of triplicate determinations performed at least three times. Furthermore, data were analyzed using the LIGAND program (21). All Scatchard plots were linear, and the best fit to the data, given by LIGAND, was obtained with a one-binding site model. IC 50 values, determined from the competition curves, were analyzed by one-way analysis of variance followed by a Newman-Keuls multiple comparison test.

RESULTS
LGR7 Is Activated by Porcine Relaxin, H2 Relaxin, and H3 Relaxin, Whereas LGR8 Is Activated by Porcine Relaxin, H2 Relaxin, and INSL3-Earlier studies indicated that porcine relaxin activates both LGR7 and LGR8, whereas different INSL3 preparations activate only LGR8 (6,7). Furthermore, H3 relaxin, like H2 relaxin, stimulates cAMP production in the THP-1 cells and competes for 33 P-relaxin binding to its receptors (10). Based on these results, we tested the ability of these peptides in the activation of human LGR7 and LGR8. 293T cells were transiently transfected with LGR7 and LGR8 plasmids and ligand signaling was estimated based on total cAMP production ( Fig. 1). To correct for varying receptor expression, all data for this and subsequent experiments were normalized based on cell surface M1 antibody binding to the tagged FLAG epitope of the recombinant receptors (Table II). The EC 50 values and maximal levels of cAMP production for different doseresponse curves are shown in Table III. In cells expressing LGR7, treatment with porcine relaxin and H2 relaxin led to dose-dependent increases in cAMP production. Although with lower efficacy, treatment with H3 relaxin also stimulated a dose-dependent increase in cAMP levels, whereas treatment with INSL3 was ineffective (Fig. 1A). In contrast, cells expressing LGR8 responded to treatment with INSL3, porcine relaxin, and H2 relaxin with increases in cAMP production (Fig. 1B). Moreover, treatment with H3 relaxin was ineffective in activating LGR8.
To demonstrate the receptor binding of relaxin-related peptides, cells expressing LGR7 and LGR8 were incubated with 33 P-labeled H2 relaxin with or without increasing competing ligands. As shown in Fig. 2A, Scatchard plot analyses of saturation binding studies indicated that 33 P-labeled H2 relaxin shows a higher affinity for LGR7 (K d : 0.209 Ϯ 0.025, n ϭ 4) than LGR8 (1.062 Ϯ 0.127; n ϭ 4) (p Ͻ 0.05). Consistent with these results and the observed potencies in stimulating cAMP production, H2 relaxin was most potent in competing for LGR7 binding by 33 P-labeled H2 relaxin, whereas H3 relaxin has lower binding affinity than H2 relaxin. In contrast, INSL3 showed minimal affinity (Fig. 2B). For LGR8, the displacement data showed that INSL3 and H2 relaxin, but not H3 relaxin, could compete for 33 P-labeled H2 relaxin binding to this receptor (Fig. 2C). The IC 50 values for different peptides are shown  Table III. These results demonstrate that H3 relaxin binds to LGR7, but not LGR8, and stimulates cAMP production mediated by LGR7. Thus, H3 relaxin is a specific ligand for LGR7, but not LGR8. In contrast, INSL3 binds to LGR8 and activates cAMP production, acting as a specific ligand for LGR8.
We have used an anchored receptor approach to generate soluble ectodomains of the gonadotropin and thyrotropin receptors (15), as well as the ectodomain of LGR7. The soluble ectodomain of LGR7, designated as 7BP, was able to block relaxin actions in vitro and in vivo (6). To test whether the ectodomain of LGR7 is capable of interacting with H3 relaxin, we treated 293T cells expressing LGR7 with H3 relaxin and H2 relaxin together with 7BP. As shown in Fig. 2D, co-treatment with 7BP completely blocked the stimulatory effects of H3 and H2 relaxin in a dose-dependent manner, thus demonstrating the ability of the ectodomain of LGR7 to bind these ligands.
H3 Activates Chimeric Receptor LGR7/8, Whereas INSL3 Activates Chimeric Receptor LGR8/7-To further confirm the importance of the ectodomain of LGR7 and LGR8 for ligand binding, we constructed chimeric receptors with their extracellular regions switched. LGR7/8 is comprised of the extracellular region from LGR7 and the transmembrane region to the carboxyl terminus from LGR8. In contrast, LGR8/7 has the extracellular region from LGR8 and the transmembrane region and C-tail from LGR7. As shown in Fig. 3, H2 relaxin stimulated cAMP production in transfected 293T cells expressing LGR7/8 (Fig. 3A) or LGR8/7 (Fig. 3C), whereas treatment with INSL3 resulted in a dose-dependent cAMP increase only in cells expressing LGR8/7 (Fig. 3C). Even though H3 relaxin stimulates cells expressing LGR7/8 to produce cAMP in a dosedependent manner (Fig. 3A), the efficacy of cAMP production is lower compared with the H3 relaxin stimulation of wild type LGR7 (Fig. 1A). H3 was 30-fold less potent than H2 in stimulating wild type LGR7 (Table III). In contrast, H3 was 60-fold less potent than H2 in activating the chimeric LGR7/8. Receptor binding analyses also showed that H3 relaxin exhibited a decreased ability to compete for 33 P-H2 relaxin binding to the chimeric receptor LGR7/8 (Fig. 3B) as compared with the H3 relaxin competition for 33 P-H2 relaxin binding to the wild type LGR7 (Fig. 2B). Although the ability of H3 relaxin to bind and stimulate the LGR7 receptor could be blocked by the soluble ectodomain of LGR7, these results suggested that additional regions of LGR7 might participate in receptor signaling by the H3 relaxin. In contrast, the chimeric receptor LGR8/7 responded to INSL3 and H2 relaxin but was not stimulated by H3 relaxin. Furthermore, INSL3 and H2 relaxin competed effectively for 33 P-H2 relaxin binding to LGR8/7 (Fig. 3D).

Replacement of Exoloop 2, but Not Exoloop 1 or 3, of LGR7 in the Chimeric Receptor LGR7/8 Restores Ligand Binding and Signaling by H3 Relaxin-We hypothesized that exoloops in
LGR7, in addition to the ectodomain, are important for interaction with H3 relaxin and designed additional chimeric constructs by replacing the individual exoloop of LGR7 into the chimeric receptor LGR7/8. The chimeric receptor LGR7/8(EL2) with exoloop 2 and the ectodomain from LGR7 responded to H3 relaxin treatment with an EC 50 value comparable to the H3 stimulation of wild type LGR7 (Figs. 1A versus 4B and Table  III). In contrast, LGR7/8(EL1) and LGR7/8(EL2), chimeric receptors with the other exoloops replaced, responded to H3 relaxin treatment with an ED 50 similar to the H3 stimulation of LGR7/8 (Figs. 4A and 4C versus Fig. 3A and Table III). Likewise, replacement of exoloop 2, but not exoloop 1 or 3, in the chimeric receptor LGR7/8 restores receptor binding by H3 relaxin (Fig. 5).
Taking advantage of the structural similarity between LGR7 and LGR8, and because leucine-rich repeats in the ectodomain have been postulated to be important for proteinprotein interactions (22)(23)(24), we designed chimeric constructs to investigate the importance of the ectodomain of these receptors in ligand signaling. Similar to wild type LGR7, LGR7/8 responded to porcine relaxin and H2 relaxin stimulation, whereas neither LGR7 nor LGR7/8 responded to treatment with INSL3. Although treatment with H3 relaxin in cells expressing LGR7/8 led to dose-dependent increases in cAMP production, this chimeric receptor was less responsive to H3 relaxin as compared with the wild type LGR7. Further analyses of competition data based on ligand-receptor binding also indicated that the chimeric receptor LGR7/8 showed lower affinity for H3 relaxin. These results suggest that, in addition to the ectodomain, the transmembrane region of LGR7 plays an important role for optimal H3 relaxin binding and activation of LGR7. We further demonstrated that replacement of exoloop 2 of LGR7 in the chimeric receptor LGR7/8 completely restored receptor binding and cAMP production. In contrast, LGR7/8(EL1) or LGR7/8(EL3) showed similar EC 50 and IC 50 values for LGR7/8. These results indicate that exoloop 2, but not exoloop 1 or 3, is responsible for H3 relaxin binding to the receptor and optimal signal transduction.
In the human genome, there are seven peptide hormones belonging to the relaxin family, all with the putative two-chain, three cysteine-bounded structure. The homology between the A-and B-chain of H1 and H2 relaxin is 62 and 85%, respectively. These two hormones show similar biological activity. H2 relaxin is expressed in the corpus luteum and is the major circulating form (25), whereas the expression of H1 relaxin is restricted to decidua, trophoblasts, and prostate (26). The distinctive RXXXRXX(I/V) motif in the B-chain of H1 and H2 relaxin is believed to be the contact site for receptor binding (27). Although the paralogous INSL3 produced by the testis and ovary (4, 5) is a specific ligand for LGR8, another human paralog relaxin 3, designated as H3 relaxin, has the RXXXRXXI motif in the B-chain and stimulates cAMP production by THP-1 cells expressing the relaxin receptors (10). The distribution of H3 relaxin in human tissues is unknown; however, the predominant site of relaxin 3 expression in rodents is the brain (10,28). H3 relaxin is believed to be a neuropeptide that activates its receptor in neuronal synapses (28), thus being consistent with its lower efficacy in activating recombinant LGR7 as compared with the more potent endocrine hormone H2 relaxin.
The known LGRs from vertebrates and invertebrates can be divided into three distinct subgroups based on phylogenetic analysis. The first subgroup contains the mammalian gonadotropin and thyroid-stimulating hormone receptors, fly LGR1, and LGRs from sea anemone and Caenorhabditis elegans, whereas the second subgroup consists of mammalian orphan receptors LGR4, LGR5, and LGR6, as well as fly LGR2 (29,30). The third group of LGRs, including mammalian LGR7 and LGR8 as well as snail LGR (13), is distinct from the other groups in that it has the unique low density lipoprotein receptor-like cysteinerich motifs in the amino terminus but is missing the typical hinge region known to be important for gonadotropin and thyroidstimulating hormone receptor activation (31).
At least three steps are involved in the ligand signaling of glycoprotein hormone receptors, each probably requiring unique but overlapping domains (31)(32)(33)(34). First, the heterodimeric ligands interact with the ectodomain of the receptor, consisting of leucine-rich repeats that could form a 1/3donut structure important for ligand-interaction. Second, ligand binding leads to the disruption of the constraint on the transmembrane region exerted by the interactions between the ectodomain (likely the hinge region) and exoloop 2. Third, the relaxed transmembrane region, as the result of ligand binding, interacts with the Gs protein to activate the adenyl cyclase. In this model, it is likely that the common ␣-subunit of the glycoprotein hormones interacts with the leucine-rich repeats of the receptor ectodomains, whereas the unique ␤-subunits of these ligands stabilize the ligand-receptor complex by binding to the exoloops (33). Due to the lack of a hinge region in LGR7 and LGR8 comparable to those found in glycoprotein hormone receptors, it is unclear whether the ectodomains of these relaxin receptors are capable of constraining their transmembrane region similar to glycoprotein hormone receptors.
Although the present studies using chimeric receptors and the soluble ectodomain of LGR7 suggest an important role for the ectodomain in ligand-receptor binding and signal transduction, our data demonstrate that H3 relaxin binds to both the ectodomain and exoloop 2 of LGR7 to induce maximal signal transduction. We propose a model for the activation of LGR7 by H3 relaxin (Fig. 7). First, H3 relaxin binds to the ectodomain of LGR7 through the putative contact motif RXXXRXX(I/V) (Fig.  7A). This interaction could be blocked by the soluble ectodomain of LGR7. Subsequently, H3 relaxin also binds to exoloop 2 of LGR7 to stabilize the ligand-receptor complexes (Fig. 7B). Binding of H3 relaxin to both regions of LGR7 evokes efficient receptor activation by interacting with the Gs protein and stimulating cAMP production (Fig. 7C). Because LGR7 and LGR8 show 59% homology in exoloop 2, H2 relaxin could interact with the consensus sequence of these receptors and the present model could apply to the H2 relaxin.
In conclusion, we demonstrate that H3 relaxin is a specific ligand for LGR7, and using chimeric receptors, H3 relaxin is shown to bind both the ectodomain and the exoloop 2 for the activation of its receptor. Further studies using chimeric LGR7 and LGR8 receptors could provide useful information regarding the mechanism of receptor activation by H2 relaxin and INSL3 and aid in the understanding the structural-functional relationship between ligands and receptors for this unique group of G protein-coupled receptors.