Identification of Relaxin-3/INSL7 as an Endogenous Ligand for the Orphan G-protein-coupled Receptor GPCR135*

GPCR135, publicly known as somatostatin- and angiotensin-like peptide receptor, is expressed in the central nervous system and its cognate ligand(s) has not been identified. We have found that both rat and porcine brain extracts stimulated 35S-labeled guanosine 5′-O-(3-thiotriphosphate) (GTPγS) incorporation in cells over-expressing GPCR135. Multiple rounds of extraction, purification, followed by N-terminal sequence analysis of the ligand from porcine brain revealed that the ligand is a product of the recently identified gene, relaxin-3 (aka insulin-7 or INSL7). Recombinant human relaxin-3 potently stimulates GTPγS binding and inhibits cAMP accumulation in GPCR135 overexpressing cells with EC50 values of 0.25 and 0.35 nM, respectively. 125I-Relaxin-3 binds GPCR135 at high affinity with a Kd value of 0.31 nM. Relaxin-3 is the only member of the insulin/relaxin superfamily that can activate GPCR135. In situ hybridization showed that relaxin-3 mRNA is predominantly expressed in the dorsomedial ventral tegmental nucleus of the brainstem (aka nucleus incertus), as well as in discrete cells in the lateral periaqueductal gray and in the central gray nucleus. GPCR135 is expressed abundantly in the hypothalamus with discrete expression in the paraventricular nucleus of the hypothalamus and supraoptic nucleus, as well as in the cortex, septal nucleus, and preoptical area. Relaxin-3 has previously been shown to bind and activate the LGR7 relaxin receptor. However, we believe that neuroanatomical colocalization of GPCR135 and relaxin-3, coupled with a clear high affinity interaction, suggest that GPCR135 is the receptor for relaxin-3. The identification of relaxin-3 as the ligand for GPCR135 provides the framework for the discovery of a new brainstem/hypothalamus circuitry.

Relaxin is a member of the insulin superfamily. The hallmark of this protein family is the presence of two peptide subunits that are arranged by three disulfide bonds (16 -19). Whereas insulin is known to play a major role in glucose metabolism and signals through the insulin receptor, a single transmembrane growth factor/tyrosine kinase-type receptor (20,21), relaxin is known as a hormone involved in growth and remodeling of reproductive and other tissues during pregnancy (22). Recently, two leucine-rich repeat-containing G-proteincoupled receptors (LGRs), LGR7 and LGR8 (23,24), have been identified as the receptors for relaxin (24). LGR7 and LGR8 belong to the typical hormone receptor family and share significant homologies with luteinizing hormone receptor (25) and thyroid stimulating hormone receptor (26). Activation of LGR7 or LGR8 by their ligands leads to intracellular cAMP accumulation. Recently, a new member of the insulin superfamily was discovered and named relaxin-3 (27). It was demonstrated to be an additional ligand for LGR7 (28), whereas the ligand INSL3 (29) has also been shown to be a ligand for LGR8 (30).
As part of a directed effort to identify novel receptors and ligands in the central nervous system, we cloned the full-length cDNAs for nearly all GPCR-like sequences from public and in-house data bases. One of these receptors, GPCR135, was used to fish out a ligand from brain extracts. GPCR135, publicly known as somatostatin-and angiotensin-like peptide receptor (SALPR) (31), shares significant homology to somatostatin receptor SSTR5 and angiotensin II receptor AT1 with 35 and 31% identity, respectively. GPCR135 mRNA is expressed in various regions in the brain, particularly in the substantia nigra and pituitary as determined by RT-PCR (31). GPCR135 mRNA can also be detected in the peripheral tissues, albeit at low levels (31). The predominant expression of GPCR135 mRNA in the brain suggests a role in central nervous system function. However, the endogenous ligand(s) for GPCR135 had not been identified. In the present studies, we report the puri-* 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.
The nucleotide sequence(s) reported in this paper has been submitted to the GenBank TM /EBI Data Bank with accession number(s) AY394501.
‡ To whom correspondence should be addressed. Tel.: 858-784-3059; Fax: 858-450-2040; E-mail: cliu9@prdus.jnj.com. 1 The abbreviations used are: GPCR, G-protein coupled receptor; HPLC, high performance liquid chromatography; RP-HPLC, reverse phase high performance liquid chromatography; LGR, leucine-rich repeat-containing G protein-coupled receptors; RT, reverse transcription; FLIPR, fluorescent imaging plate reader; GTP␥S, guanosine 5Ј-O- (3-fication, identification, and characterization of relaxin-3, a new member of the insulin/relaxin peptide superfamily (27), as the endogenous ligand for GPCR135. Relaxin-3 binds to and activates GPCR135 at high affinity. No other members of the insulin peptide family bind to or activate GPCR135. Because GPCR135 is expressed in the hypothalamus and relaxin-3 is expressed in the brainstem nuclei with known projections to the hypothalamus, we conclude that relaxin-3 is the endogenous ligand for GPCR135.
Molecular Cloning of GPCR135-The human GPCR135 coding region was identified from a human genome sequence (GenBank TM accession number NT_023085) based on its homology to somatostatin receptors. The complete coding region of human GPCR135 was PCR amplified from human genomic DNA (Clontech, Palo Alto, CA) using two primers with the forward primer: 5Ј-ACAGCTCGAGGCCACCAT-GCAGATGGCCGATGCAGCCACG-3Ј and the reverse primer: 5Ј-ACATCATCTAGATCAGTAGGCAGAGCTGCTGGGCAGCAG-3Ј. The PCR was performed at 94°C for 40 s, 65°C for 40 s, and 72°C for 3 min for 35 cycles. The PCR products were cloned into pCIneo and the insert region was sequenced to confirm the identity of the sequence.
Initial Identification and Characterization of the GPCR135 Ligand Activity from the Rat Brain-Five grams of frozen rat brain were homogenized in 50 ml of Ϫ30°C ethanol, 0.8 M HCl (3:1 ratio). The homogenate was extracted at 4°C for 2 h and then centrifuged at 4°C at 20,000 ϫ g for 30 min. The supernatant was loaded onto a 2-ml SP-Sephadex C-25 (Amersham Biosciences) column. The column was washed with 20 ml of 1 M acetic acid and eluted with 2 M pyridine and 1 M acetic acid. The eluted peptides were loaded onto a 500-mg C-18 Bond Elut column (Varian, Harbor City, CA), washed with 0.1% trifluoroacetic acid, and eluted with 60% acetonitrile and 0.1% trifluoroacetic acid. The eluted peptides were lyophilized, reconstituted in water, and tested for ligand activity in GPCR135-expressing cell membranes using GTP␥S binding assays. The reconstituted rat brain extract was also run on a HPLC gel filtration column (BioSep-SEC-S2000, Phenomenex, Torrance, CA) in 1 M acetic acid to characterize the molecular weight of the ligand. Briefly, fractions of 1 ml were collected, lyophilized, and tested for GPCR135 ligand activity using GTP␥S binding assays. In a parallel experiment, proteins or peptides with various molecular weights were run in the same conditions to serve as the molecular weight standards.
Purification of GPCR135 Ligand from Porcine Brain-Porcine brains (1.8 kg) were extracted with 18 liters of cold ethanol, 0.8 M HCl, mixed at 3:1. Portions of 300 g of frozen brains were homogenized in a blender (Waring, Winster, CT) for 1-2 min in 3 liters of the liquid cooled to Ϫ30°C. The homogenate was then stirred at 4°C for 2 h. The mixture was centrifuged at 5000 ϫ g for 20 min and filtered through Whatman 541 filter paper (Whatman, Maidstone, UK) and then through 0.22-m Millipore GP Express Plus filter (Millipore, Bedford, MA). Aliquots of 2.5 liters of the extract were mixed with 600 ml of Express-Ion TM Exchanger S (Whatman) equilibrated with 20 mM sodium acetate, 20% acetonitrile, pH 5.2, stirred for 20 min, and filtered. The ion exchanger was washed 3 times with 500 ml of the same buffer and then eluted with 500 ml of 20 mM sodium acetate, 0.5 M NaCl, 20% acetonitrile, pH 5.2, and the eluate was separated from the ion exchanger by filtration. The elution procedure was repeated twice, first with 500 ml and then with 400 ml of liquid, and the eluates were pooled. The remaining extract was processed identically. To the pooled eluate, an equal volume of 0.1% trifluoroacetic acid/water was added, and 10 liters of this liquid was mixed with 300 g of Amberlite XAD-16 HP (Supelco, Bellefonte, PA) and stirred for 20 min. Amberlite was washed on a filter with 0.1% triflu-oroacetic acid/water, and eluted by mixing with 700 ml of 50% acetonitrile, 0.1% trifluoroacetic acid/water. The eluate was separated from the Amberlite by filtration. Such elution from Amberlite was repeated twice, and the rest of the ion exchanger eluate was processed in the same manner. The eluates from Amberlite were pooled, and acetonitrile was removed with a rotary evaporator. The remaining liquid was lyophilized, and 0.52 g of dry material was obtained. The material was dissolved in 400 ml of 20 mM sodium acetate, 20% acetonitrile, pH 5.2, and filtered through a 0.22-m Millipore GP Express Plus filter.
The following chromatographies were carried out on Ä KTA Explorer™ and Purifier™ HPLC systems (Amersham Biosciences). The filtrate (78 ml per run) was pumped onto a Resource S (6-ml column, Amersham Biosciences) equilibrated with 20 mM sodium acetate, 20% acetonitrile, pH 5.2, and eluted with a 0 -20% linear gradient in 35 column volumes of the same buffer containing 2 M NaCl. The active fractions were pooled, diluted with 5 volumes of water, pH was adjusted to 2.5, and loaded onto the same column equilibrated with 20 mM sodium phosphate, 20% acetonitrile, pH 2.5. Peptides were eluted with a 0-25% linear gradient in 35 column volumes of this buffer containing 2 M NaCl. Active fractions were collected, mixed with an equal volume of 0.1% trifluoroacetic acid/water loaded onto a Source 5RPC (ST 4.6/ 150, Amersham Biosciences), and eluted with a 25-40% linear gradient in 12 column volumes of 0.1% trifluoroacetic acid, 95% acetonitrile/ water. The pooled active fractions were mixed with 750 l of 0.1% trifluoroacetic acid/water, and chromatographed on a RPC C2/C18 column (ST 4.6/100, Amersham Biosciences) using a 23-40% linear gradient in 15 column volumes of 0.1% trifluoroacetic acid, 95% acetonitrile/water. At this stage the active fractions appeared as a single peak and the material was subjected to structural analysis.
S-Carboxyamidomethylation-A part of the lyophilized peptide was dissolved in 200 l of reaction buffer consisting of 0.4 M Tris-HCl, pH 8.4, 6 M guanidinium hydrochloride (Sigma), 2 mM EDTA (Merck), and 4 l of 0.5 M dithiothreitol (Pierce) was added and kept at 40°C for 2 h. 12 l of 0.5 M iodoacetamide (Sigma) was added to the sample, and the sample was incubated at 40°C for 35 min followed by 30 min at room temperature in the dark. The sample was mixed with 2 volumes of 0.1% trifluoroacetic acid/water, pumped onto the RPC C2/C18 ST 4.6/100 column, and the peptides were separated with a linear gradient of 0.1% trifluoroacetic acid, 95% acetonitrile/water.
Mass Spectrometry and Sequence Analysis-Positive-ion MALDI-TOF mass spectra were recorded on a Voyager DE Pro instrument (Applied Biosystems, Foster City, CA). For mass determination 1 l of sample from an HPLC fraction was mixed on a stainless steel MALDI target plate with 1 l of calibration solution and 1 l of ␣-cyano-4hydroxycinnamic acid solution and allowed to dry at room temperature. Carboxypeptidase Y digests were prepared by pipetting 1 l of samples from HPLC on the MALDI target, mixed with 1 l of carboxypeptidase Y solution and allowed to dry at room temperature. Calibration solution (1 l) and ␣-cyano-4-hydroxycinnamic acid solution (1 l) were added to the same spot and allowed to dry again. Acquired mass spectra were calibrated using monoisotopic peaks of singly or doubly charged ions of [Arg 8 ]vasopressin and human insulin (m/z 1084.4457 Th, m/z 2902.8266 Th, m/z 5804.6455 Th). On-target tryptic digests were prepared by pipetting 1 l each of sample, trypsin, and 0.2% ammonium bicarbonate solutions, and allowed to dry at room temperature. Finally, 1 l of matrix solution was added and allowed to dry again. The spectra were calibrated using monoisotopic peaks of fragments from self-digestion of trypsin (m/z 515.3306 Th, m/z 842.5100 Th, and m/z 2211.1046 Th).
Sequence analysis of tryptic fragments using tandem mass spectrometry was carried out on a Q-TOF Ultima instrument (Waters/Micromass, Manchester, UK) equipped with the standard Z-spray source. The samples were sprayed in 60% acetonitrile, 1% acetic acid/water from gold-coated borosilicate capillaries (Protana, Odense, Denmark) using a capillary voltage of 1.5 kV, and positive-ion spectra were recorded. Tryptic fragments were prepared from carboxyamidomethylated peptides by incubation with methylated trypsin in 0.2% ammonium bicarbonate at 37°C for 6 h, and the samples were lyophilized. Tandem MS/MS experiments were carried out at collision energies optimized within 25-35 eV, using 5-s scans during 5 min. Data were analyzed with a Biolynx peptide sequencing module of Masslynx 3.5. N-terminal sequence analysis was carried out on an Applied Biosystems Procise HT sequencer.
GTP␥S Binding Assays-The GPCR135 expression vector described above was transfected into CHO-K1 cells using LipofectAMINE (Invitrogen) according to the manufacturer's instructions. Two days after transfection, the cells were harvested and the cell membranes were prepared by homogenizing the cells in 50 mM Tris-HCl, 5 mM EDTA followed by centrifugation at 20,000 ϫ g at 4°C for 30 min. GTP␥S binding buffer (50 mM Tris-HCl, pH 7.4, 10 mM MgCl 2 , 10 M GDP, 1 mM EDTA, pH 8.0, and 100 mM NaCl) was added to the pellet and the pellet was homogenized using a Polytron tissue homogenizer. Protease inhibitors were added to the buffer at concentrations of 1 mM phenylmethylsulfonyl fluoride, 10 g/ml pepstain A, 10 g/ml leupeptin. Cell membranes and different concentrations of ligands were added to 96well plates and incubated at room temperature for 20 min. 35 S-GTP␥S (PerkinElmer Life Sciences) was then added to each well at a final concentration of 200 pM in a final volume of 200 l. The reactions were allowed to proceed at room temperature for 1 h, filtered though a 96-well GFC filter plate (Packard Instrument Co.), and washed with cold washing buffer: 50 mM Tris-HCl, pH 7.4, 10 mM MgCl 2 . 50 l of Microscint-40 (PerkinElmer Life Sciences) was added to each well and the plate was counted on a top counter (TopCount NTX, Packard).
Expression and Purification of Human Recombinant Relaxin-3-Human relaxin-3 complete coding region was PCR amplified from human brain cDNA (Clontech) using two primers with the forward primer: 5Ј-ACGATCGTCGACGCCACCATGGCCAGGTACATGCTGCTGCTG-CTC-3Ј and the reverse primer: 5Ј-ACGATAAAGCTTCTAGCAAAGG-CTACTGATTTCACTTTTGC-3Ј. The PCR products were cloned into a mammalian expression vector pCMV-sport1 (Invitrogen) between SalI and HindIII sites. The cloned cDNAs were sequenced to confirm the identities of human relaxin-3. The expression vector was transfected into COS-7 cells using LipofectAMINE (Invitrogen). Three days after the transfection, cell culture medium from the transfected cells was collected, adjusted to pH 3.0, and loaded onto a Sephadex C-25 cation exchange column. The column was washed with 1 M acetic acid and eluted with 2 M pyridine and 1 M acetic acid. The eluted proteins were loaded on a C-18 Bond Elut column (Varian), washed with 0.1% trifluoroacetic acid, and eluted with 60% acetonitrile and 0.1% trifluoroacetic acid. The eluted proteins were lyophilized, reconstituted in 50 mM Tris-HCl, pH 7.5, and used for GTP␥S binding assay.
In a parallel experiment, we cloned the human relaxin-3 pro-peptide coding region into a modified pCMV-sport1 vector with the polycloning sites modified by replacing the sequence between PstI and EcoRI sites with the sequence: 5Ј-CTGCAGGCCGCCATGCTGACCGCAGCGTTG-CTGAGCTGTGCCCTGCTGCTGGCACTGCCTGCCACGCGAGGAGA-CTACAAGGACGACGATGACAAGGAATTC-3Ј, which codes an ␣ peptide signal sequence and a FLAG tag. The pro-peptide coding region of human relaxin-3 was PCR amplified using the forward primer: 5Ј-AC-GATAGAATTCGATGACGACGATAAGCGGGCAGCGCCTTACGGGG-TCAGGC-3Ј, and reverse primer: 5Ј-ACTATAGGATCCCTAGCAAAG-GCTACTGATTTCACTTTTGCTAC-3Ј, and human relaxin-3 cDNA as template. The PCR product was cloned into the modified pCMV-sport1 between EcoRI and BamHI sites and the insert region was sequenced to confirm the sequence identity. The construct encodes a secreted relaxin-3 fusion peptide with a FLAG tag at the N terminus. The FLAGrelaxin-3 fusion peptide expressing plasmid was transfected into COS-7 cells. Three days after transfection, the recombinant fusion peptide was affinity purified from the cell culture medium using an anti-FLAG affinity gel (Sigma). Briefly, the cultured medium was loaded onto the anti-FLAG affinity column. The column was washed with phosphatebuffered saline solution, and eluted with 0.1 M glycine-HCl, pH 2.8. The eluted protein was neutralized with 1 M Tris-HCl, pH 7.5. The Nterminal FLAG tag was cleaved from the fusion peptide by enterokinase (Novagen, Madison, WI) and the untagged relaxin-3 peptide was further purified by reverse phase HPLC using a C-18 column (Vydas, 4.6 ϫ 250 mm, CIY) and an acetonitrile, 0.1% trifluoroacetic acid gradient.
To increase the processing efficiency of the pro-relaxin-3 into mature relaxin-3 by pro-hormone convertase, we created a furin cleavage site with the amino acid sequence of RGRR in the pro-relaxin-3 at the junction of the C-chain and A-chain. This mutation was created by a two-step overlapping PCR. The 5Ј end of the cDNA construct was PCR amplified using two primers with the forward primer (P1): ACGATAC-TGCAGGCCGCCATGCTGACCGCAGCGTTGCTGA-3Ј and the reverse primer (P2): 5Ј-CAGCCAGGACATCTCGTCGGCCCCGAAGAACCCC-AGGGGTTCCTTG-3Ј and the FLAG-relaxin-3 cDNA as the template. The 3Ј end was PCR amplified using the forward primer (P3): 5Ј-GGT-TCTTCGGGGCCGACGAGATGTCCTGGCTGGCCTTTCCAGCAGC-3Ј, and the reverse primer (P4): 5Ј-ACTATAGGATCCCTAGCAAAGGCT-ACTGATTTCACTTTTGCTAC-3Ј and the FLAG-relaxin-3 cDNA as the template. The 5Ј end and the 3Ј end PCR products were purified and mixed together as the template for the second step PCR using the forward primer (P1) and the reverse primer (P4) as described above. The final PCR product was cloned into pCMV-sport1 between PstI and BamHI sites and the insert region was sequenced to confirm the identity of the relaxin-3 mutant. The resulting new relaxin-3 mutant ex-pression plasmid was co-transfected with a human pro-hormone convertase furin expression vector into COS-7 cells using LipofectAMINE as the transfection reagent. The human furin expressing vector was constructed by PCR amplifying human furin complete coding region from human brain cDNA (Clontech) using two primers with the forward primer, 5Ј-GACTAGAAGCTTGCCACCATGGAGCTGAGGCCCTGGT-TGCTATG-3Ј and the reverse primer, 5Ј-GACGATAGCGGCCGCAGT-GGGCTCATCAGAGGGCGCTCTG-3Ј, which were designed according to the published sequence (GenBank TM accession number NM_002569). The PCR product was cloned into pcDNA3.1 (Invitrogen) between the HindIII and NotI sites. The insert region of the furin expressing vector was sequenced to confirm the identity of furin. The secreted relaxin-3 was then purified using an anti-FLAG affinity column, cleaved by enterokinase, and further purified by RP-HPLC as described above. The final purified peptide was analyzed by mass spectrometry analysis and N-terminal Edman degradation (Molecular Structure Facilities, University of California, Davis).
Ca 2ϩ Mobilization Assays-Human embryonic kidney (HEK) 293 cells were transfected with GPCR135 and G qi5 (32) expressing plasmids using LipofectAMINE. Two days after transfection, cells were detached with phosphate-buffered saline plus 10 mM EDTA. The detached cells were washed with Dulbecco's modified Eagle's medium/F-12 medium (without phenol red, Invitrogen) and seeded in black poly-D-lysinecoated 96-well plates (BD Biosciences, San Jose, CA) at a cell density of 50,000 cells/well. Calcium dye Fluo-3 (AM) (TEFLABS, Austin, TX) was loaded into the cells at a final concentration of 4 M and ligandstimulated Ca 2ϩ mobilization was monitored using FLIPR (Molecular Devices).
Radioligand Receptor Binding Assays-Human relaxin-3 was labeled using chloramine T in the presence of Na 125 I (PerkinElmer Life Sciences) and the monoiodonated relaxin-3 was purified by RP-HPLC. The relaxin-3 peptide is labeled at Tyr 5 in the B-chain and the specific activity of the purified 125 I-relaxin-3 was ϳ2200 Ci/mmol. Cell membranes from COS-7 cells transiently expressing GPCR135 were incubated with 125 I-relaxin-3 either in the presence or absence of various concentrations of different competitors in 96-well plates at a final volume of 200 l in binding buffer: 50 mM Tris-HCl, pH 7.4, 2 mM EDTA plus 0.5% bovine serum albumin. The binding mixtures were incubated at room temperature for 1 h and then filtered through 96-well GFC plates (Packard) and washed with ice-cold washing buffer: 50 mM Tris-HCl, pH 7.4. Fifty microliters of Microscint-40 was added to each well and the plates were counted in a Topcounter NTX (Packard). The radioligand binding results were analyzed using GraphPad PRISM software.
Intracellular cAMP Accumulation and Measurement-CHO-K1 cells stably expressing human GPCR135 were established by transfection of GPCR135 expressing plasmids into CHO-K1 cells and culturing the transfected cells under G418 selection (400 mg/liter). The receptor expressing cells were seeded in 96-well plates at a cell density of 30,000 cells/well. 24 h later, the cell culture medium was replaced with Dulbecco's modified Eagle's medium/F-12 (Invitrogen) plus 2 mM isobutylmethylxanthine (Sigma). Different test compounds for cAMP stimulation were added to each well. The reactions were incubated at room temperature for 25 min. Twenty microliters of 0.5 N HCl was added to each well to stop the reaction and the extract accumulated cAMP. The cAMP concentrations in the extracted media were measured using cAMP Flash Plates (PerkinElmer Life Sciences) as described by the manufacturer.
In Situ Hybridization-Two male Sprague-Dawley rats were perfused with 4% paraformaldehyde and 30-m thick coronal sections of the whole brain were cut on a sliding microtome and mounted on glass slides. The tissue was digested with proteinase K, dehydrated, and hybridized overnight with either the relaxin-3 or GPCR135, antisense or sense probes, at concentrations of about 10 7 cpm/ml. For post-hybridization, the tissue was treated with RNase A, washed, dehydrated, and exposed to a Fuji Film imaging plate (BAS-SR 2025) for 2 days. The image plates were scanned using a Fuji Bio-Imaging Analyzer System (BAS 5000). The slides were then defatted, dipped in NBT2 nuclear emulsion (Kodak), and developed after 17 days. Arbitrarily, positively labeled cells were defined as any accumulation of silver grains within a cell-sized area that was 3-5 times above background levels.
Antisense and sense RNA probes for rat relaxin-3 or GPCR135 were synthesized from cloned respective cDNA fragments in pBluescript (Stratagene) using either T7 or T3 RNA polymerases. Probes were labeled with 35 S-UTP (PerkinElmer Life Sciences). The labeled sense strands served as controls and did not show any specific labeling of cellular localization. Specific activities of 35 S-UTP probes were ϳ2-3 ϫ 10 6 cpm/g.

RESULTS
Identification and Sequence Analysis of GPCR135-Using human somatostatin receptor SSTR5 as the query, we searched the human genome draft sequence data base in the Gen-Bank TM using NCBI Blast program and found a contig that encoded a putative GPCR with 35% similarity to SSTR5, which we designated as GPCR135. The complete coding region of GPCR135 was cloned from human genomic DNA and the open reading frame was confirmed by DNA sequencing. The complete coding region of GPCR135 has been submitted to Gen-Bank TM (GenBank TM accession number AY394501). A recent search of GenBank TM indicated that GPCR135 is identical to a published orphan GPCR named the somatostatin-and angiotensin-like peptide receptor (31). The sequence homology of GPCR135 to known G-protein-coupled receptors, such as somatostatin receptor and angiotensin receptor, suggested that GPCR135 might be a peptide receptor.
Identification and Characterization of GPCR135 Ligand from Rat Brain and Porcine Brain Extracts-Using an ethanol/ HCl extraction method, we extracted peptides from different rat tissues and tested the extracts for GPCR135 ligand activity using a GTP␥S binding assay. Our results showed that extracts from rat brain stimulated the GTP␥S binding for GPCR135 expressing cell membranes, whereas similarly derived extracts from other tissues showed no activity (Fig. 1A). Mock transfected cell membranes did not respond to any extracts (data not shown). HPLC gel filtration analysis of the rat brain extract indicated that GPCR135 ligand activity eluted with a predicted molecular mass of 4 to 6 kDa (Fig. 1B). To increase the scale of FIG. 1. Identification and characterization of GPCR135 ligand activity in rat brain extracts. A, GPCR135 ligand activity in different rat tissues. Extracted rat tissues at the dilutions shown were used as ligands in the GTP␥S binding assay using human GPCR135 expressing cell membrane. B, molecular weight characterization of GPCR135 ligand from rat brain extract. Crude rat brain ethanol/HCl extract was run through a HPLC gel filtration column. Fractions were collected and assayed for GPCR135 ligand activity in GTP␥S binding assays using human GPCR135 expressing cell membranes. In a parallel experiment, peptides and nucleotides with known molecular weights were run using the same conditions to serve as the molecular mass standards. purification, we switched to porcine brain, which afforded a greater mass of tissue and was shown to retain the GPCR135 ligand activity in ethanol/HCl extract (data not shown).
Isolation and Structural Analysis of the Porcine GPCR135 Ligand-The porcine GPCR135 ligand was purified as described under "Experimental Procedures" and GTP␥S binding assays were used to monitor the GPCR135 ligand activities through the entire purification process. After the final HPLC purification, the GPCR135 ligand activity appeared as a single peak (Fig. 2, A and B). Initial N-terminal sequence analysis revealed the presence of two peptide chains with sequence similarities to human and rat relaxin-3. To establish the structure of the porcine peptide, the protein was reduced with dithiothreitol and carboxyamidomethylated. The two peptides were separated by RP-HPLC and analyzed separately. N-terminal sequence analysis results were DVLAGLSSNCCKWGC-SKSEISSLC for one of the peptides, and RASPYGVKLCGRE-FIRAVIFTCGGSRW for the other peptide. Analysis by MALDI-TOF MS gave m/z 2718.32 Th (calculated value m/z 2718.20 Th) for the monoisotopic singly charged ion of the first peptide, and m/z 3143.58 Th (calculated value m/z 3143.61 Th) for that of the second peptide, both in carboxyamidomethylated forms. To further confirm the N-terminal sequencing result, the peptides were digested with trypsin and carboxypeptidase Y. The digested peptide fragments were analyzed by MALDI-TOF MS, which gave fragment ion mass values consistent with the sequences. Sequence analysis of the tryptic fragments by tandem mass spectrometry also fully confirmed the N-terminal sequence analysis. Measurements of the intact protein by MALDI-TOF MS indicated that porcine relaxin-3 has a monoisotopic molecular mass 5511.56 Da, which is in good agreement with the monoisotopic mass value calculated from the disulfide-linked amino acid sequences, 5511.62 Da. The measured value is in agreement with the three disulfide bridges in the protein. Thus the structural analysis established that porcine relaxin-3 consists of two peptide chains with the amino acid sequences shown above, linked together with disulfide bridges in an arrangement characteristic to insulin-relaxin family of proteins. The extractions yielded ϳ2-3 pmol of relaxin-3 per kg of brain. The sequences of those two porcine peptides show high homology (Ͼ90%) to human, rat, and mouse relaxin-3 A-chain and B-chain, respectively (Fig. 2C), indicating that the isolated peptide is porcine relaxin-3.
Recombinant Expression of Functional Relaxin-3 in Mammalian Cells-To confirm that relaxin-3 is indeed a ligand for GPCR135, we expressed human relaxin-3 full-length cDNA in COS-7 cells. The ligand in the cell culture medium, extracted by ion exchange and C-18 Bond Elut columns, was tested for GPCR135 ligand activity in the GTP␥S binding assay. Our results indicated that cultured medium from the relaxin-3 cDNA-transfected cells specifically stimulates GTP␥S binding in GPCR135 expressing cell membranes but not in mock transfected cells. To facilitate the purification process of the recombinant relaxin-3 peptide, we engineered an expression construct by replacing the signal peptide coding region of relaxin-3 with an ␣ peptide signal sequence followed by a FLAG sequence. The resulting construct encodes a secreted fusion relaxin-3 peptide with a FLAG tag at the N terminus. We expressed the construct in COS-7 cells and the recombinant fusion relaxin-3 peptide was purified from the conditioned medium using an anti-FLAG affinity column. Following enterokinase cleavage, the untagged relaxin-3 peptide was further purified by C-18 reverse phase HPLC and tested for its ligand activity. The purified peptide specifically stimulates GTP␥S binding in GPCR135 expressing cell membranes. However, SDS-PAGE analysis indicated that the purified peptide is heterogeneous in molecular mass ranging from 5 kDa (predicted molecular mass for relaxin-3 mature peptide) to 13 kDa (pre-dicted molecular mass for relaxin-3 pro-peptide). Reverse phase HPLC indicated that the purified peptide was not uniform in column retention time (data not shown). Those results strongly suggested that the pro-hormone process that removes the relaxin-3 C-chain was not complete. To increase the efficiency of C-chain removal and obtain the homogeneous mature form relaxin-3, we created a furin cleavage site (33,34) with the amino acid sequence of RGRR at the junction of C-chain and A-chain in the relaxin-3 pro-peptide, which is designated as relaxin-3-RR (Fig. 3A). In addition, we cloned a human furin cDNA and co-transfected furin cDNA with relaxin-3-RR into COS-7 cells. After purification through an affinity column, the peptide is cleaved with enterokinase to remove the tag and further purified by RP-HPLC. The purified peptide was evaluated by SDS-PAGE as well as by an analytical RP-HPLC for purity and was tested for ligand activity in GTP␥S binding assays. Our results indicated that the purified relaxin-3 peptide from relaxin-3-RR vector is uniform in SDS-PAGE analysis (with a molecular mass of 5 kDa) as well as in retention time in HPLC (Fig. 3B) and has potent ligand activity in GTP␥S binding assays (Fig. 3C). Mass spectrometry analysis indicated that purified relaxin-3 has a molecular mass of 5501.1, which is in good agreement with the predicted molecular mass (5500.4). N-terminal sequencing results for the purified peptide revealed two sequences (DVLAG . . . and RAAPY . . . ) with an almost 1:1 ratio, which match the predicted N termini of the A-chain and B-chain of human relaxin-3.
Pharmacological Characterization of GPCR135-With the purified relaxin-3 available, we performed radioligand binding and functional studies of relaxin-3 for its ability to stimulate GTP␥S binding, stimulate or inhibit cAMP accumulation, and stimulate Ca 2ϩ mobilization in GPCR135 overexpressing cells.
Functional Characterization of GPCR135 by GTP␥S Binding and cAMP Accumulation Inhibition Assays-We have shown that relaxin-3 stimulated GTP␥S binding in GPCR135 overexpressing cell membranes. To further characterize the function of GPCR135, we performed GTP␥S binding assays using GPCR135 expressing cell membranes with different doses of relaxin-3 and other members of the relaxin/insulin family. Our results indicated that relaxin-3 potently stimulates GTP␥S binding in GPCR135 overexpressing cell membranes. The EC 50 value observed in GTP␥S binding assays is 0.25 nM for GPCR135 (Fig. 5A). Relaxin-3 B-chain showed weak agonist activity with an EC 50 of ϳ100 nM. Other members of the relaxin/insulin family showed no agonist activity. We characterized GPCR135 to see if it is coupled to cAMP stimulation or inhibition. Our results showed that relaxin-3 does not stimulate cAMP accumulation (Fig. 5B) but instead dose-dependently inhibits forskolin-stimulated cAMP accumulation in GPCR135 overexpressing cells with an EC 50 of 0.35 nM (Fig. 5C), which is similar to that from GTP␥S binding assays.

Characterization of GPCR135 by Ca 2ϩ Mobilization
Assays-G qi5 has been reported to be able to shift the signal transduction of GPCRs from cAMP inhibition to calcium mobilization (32). We co-expressed GPCR135 and G qi5 in 293 cells. Stimulation of the transfected cells with relaxin-3 evoked calcium mobilization in the transfected cells co-expressing GPCR135 and G qi5 but not in mocked transfected 293 cells and 293 cells expressing only GPCR135 or G qi5 as monitored by FLIPR. The EC 50 of relaxin-3 to GPCR135 observed in the calcium mobilization assays is ϳ5 nM (Fig. 5D).
GPCR135 and Relaxin-3 mRNA Tissue Expression Profile-Using RT-PCR, we systematically studied the mRNA tissue expression patterns of GPCR135 and relaxin-3 in 19 different human tissues. Our results showed that GPCR135 mRNA was detected in brain, testis, thymus, and adrenal gland. Relaxin-3 mRNA was only found in brain and testis (Fig. 6A). Using in situ hybridization, we further studied the mRNA expression of GPCR135 and relaxin-3 mRNA distribution in the rat brain (Fig. 6B). Our results showed that relaxin-3 and its receptor are discretely expressed in different areas of the central nervous system. GPCR135 mRNA was distributed throughout the brain, with particularly strong expression in the paraventricular nucleus of the hypothalamus and the supraoptic nucleus. No signal was detected in tissues hybridized with the sense probe (data not shown). The distribution of the ligand was in more discrete and limited areas of the brain, namely the lateral periaqueductal gray, nucleus incertus, and central gray regions in the brainstem (Fig. 6B). DISCUSSION Hundreds of orphan G-protein-coupled receptors have been found by searching the human genome data base, yet agonists for those receptors are critical to validation of their activity and elucidation of their functions. Ligands for orphan receptors have most often been identified by testing known modulators (5-7, 9 -11) or by purifying novel ligands from extracts (1,3,4,12). In this report, we demonstrated that relaxin-3, a member of the insulin superfamily, is the endogenous ligand for an orphan receptor (GPCR135) predominantly expressed in the brain through peptide purification approaches.
Initial purification of porcine GPCR135 ligand was achieved from 200 g of porcine tissue (2 pig brains). However, this scale only provided enough material to obtain mass data. Therefore the extraction and purification were repeated with 1.8 kg, and finally with 20 kg of porcine brain, which yielded 10 -15 ng of relaxin-3 per kg of brain. The overall small amount is consistent with the expression of the peptide in a very limited area of the brain.
The two-subunit structure of relaxin-3 has been predicted based on the known structures of relaxin/insulin superfamily members (27). The predicted cleavage sites, which are involved in processing of the signal peptide and C-chain removal for pre-prorelaxin-3, are based on assumptions as well. The purification and characterization of the natural endogenous porcine relaxin-3 have verified the predicted structure of relaxin-3.
We explored the possibility of producing functional human relaxin-3 by recombinant approaches. Initial attempts to express relaxin-3 in COS-7 cells using the natural relaxin-3 cDNA sequence resulted in very little mature peptide with the majority of pro-peptide unprocessed or incompletely processed. In the relaxin-3 pro-peptide sequence, there exists a natural furin cleavage site (RWRR) (35,36) between the B-chain and C-chain. Although COS-7 cells express furin protease (37), the fact that the cleavage between relaxin-3 B-chain and C-chain is not completely processed in COS-7 cells suggests that the overexpressed relaxin-3 in COS-7 cells exceeds the processing capacity of the endogenous furin. Co-expression of additional furin leads to complete cleavage between the B-chain and Cchain, which agrees with our hypothesis. At the junction of the C-chain and A-chain of relaxin-3, there is another putative pro-hormone convertase cleavage site (RGSR). Our results indicated that this site is not well cleaved by endogenous COS-7 cell protease nor by overexpressed furin and pro-hormone convertase-1 (PC-1) (38). A mutation that changes this native FIG. 5. Functional characterization of GPCR135 using relaxin-3 and related peptides as ligands. A, relaxin-3 stimulates 35 S-GTP␥S binding in GPCR135 expressing cells. Different peptides were added at various concentrations to the human GPCR135 expressing cell membranes to stimulate GTP␥S incorporation. The specific 35 S-GTP␥S incorporation was obtained by subtracting counts without ligand from the counts with ligand. B, inhibition of forskolin-stimulated cAMP production by relaxin-3. Chinese hamster ovary cells stably expressing GPCR135 and control Chinese hamster ovary cells were stimulated with buffer, 200 nM relaxin-3, 5 M forskolin, or 5 M forskolin plus 200 nM relaxin-3. cAMP from the treated cells was extracted and measured using cAMP flash plates (PerkinElmer Life Sciences). C, dose response of relaxin-3 inhibition of cAMP production in GPCR135 expressing cells. Chinese hamster ovary cells stably expressing GPCR135 were stimulated with different concentrations of peptides, including relaxin-3, at various concentrations. Forskolin was then added to all samples at a final concentration of 5 M. cAMP from the stimulated cells was extracted and measured using cAMP flash plates. D, relaxin-3 stimulates Ca 2ϩ mobilization in HEK293 cells co-expressing GPCR135 and G qi5 . HEK293 cells, either mock transfected (293), transfected with G qi5 (293/Gqi5), human GPCR135 (GPCR135), or co-transfected with human GPCR135 and G qi5 (GPCR135/Gqi5), were used for Ca 2ϩ mobilization assays. Relaxin-3 stimulated intracellular Ca 2ϩ mobilization was monitored by FLIPR. cleavage site into a preferred furin site (RGRR) resulted in complete processing at this site when co-expressed with furin. The method we developed for expression and purification of relaxin-3 could be used for expression of other members of the relaxin/insulin family.
With the high quality relaxin-3 peptide available, we were able to perform a pharmacological characterization of GPCR135. We employed GTP␥S binding assays, cAMP accumulation/inhibition assays, radioligand binding assays, and Ca 2ϩ mobilization assays to characterize the in vitro pharmacology of GPCR135. The EC 50 values of relaxin-3 for stimulation of GTP␥S binding and cAMP accumulation inhibition are consistent with the K d and K i values derived from radioligand binding assays, which are ϳ200 -300 pM for GPCR135. However, relaxin-3 stimulates Ca 2ϩ mobilization in 293 cells coexpressing GPCR135 and G qi5 at a much higher EC 50 value (5 nM) compared with that from other assays. One difference between the Ca 2ϩ mobilization assay and the other three as-says is that Ca 2ϩ mobilization is not measured at equilibrium. Another potential cause for this discrepancy in EC 50 values is that G qi5 , an artificial chimeric G-protein, may not couple well to all G i -linked receptors. In addition, the above difference could also be because of levels of receptor expression varying in the different expression conditions.
Recently, two closely related LGRs, LGR7 and LGR8, have been shown to be receptors for relaxin (24), a member of the insulin superfamily that plays an important role in reproductive tissues. More recently INSL3 has been shown to be a selective ligand for LGR8 (30) and relaxin-3 has been shown to be an additional ligand for LGR7 (28). Because crossover activity has been demonstrated for relaxin receptors, we tested whether other members of relaxin/insulin family are also ligands for GPCR135. Our results indicated that except for relaxin-3, none of the other relaxin/insulin family members, including relaxin and INSL3, activate GPCR135. In our assays, porcine relaxin stimulates cAMP accumulation in LGR7 and LGR8 transfected 293 cells, whereas INSL3 only stimulates cAMP production in LGR8-transfected cells (data not shown), which are consistent with previous reports (24,28,30). However, both relaxin and INSL3 showed no ligand activity or binding affinity for GPCR135 in any assay tested. We also transiently expressed human relaxin 1, relaxin 2, and INSL5 in COS-7 cells. The conditioned media from both relaxin 1 and relaxin 2 cDNA-transfected cells stimulated cAMP production in LGR7 and LGR8 expressing 293 cells, however, both of them showed no activity for GPCR135 either in the functional assays or in the radioligand binding assays. Conditioned medium from INSL5 expressing cells showed no ligand activity for LGR7, LGR8, or GPCR135 in any assay tested, leaving relaxin-3 as the only known ligand for GPCR135.
LGR7 and LGR8 belong to the GPCR hormone receptor family with significant homology to thyroid stimulatory hormone receptor (26,39) and luteinizing hormone receptor (25,40). These hormone receptor GPCRs typically have long N-terminal extracellular domains (Ͼ300 amino acids) and are known to be involved in cAMP stimulation (24,39,40,41). GPCR135 is essentially not homologous to LGR7 and LGR8. GPCR135, which has a short N-terminal extracellular domain (Ͻ100 amino acids) and is coupled to cAMP inhibition, is a typical neuropeptide-like receptor with significant homology to somatostatin receptors.
Four subtypes of receptors/proteins have now been identified to be putative receptors or binding proteins for members of the relaxin/insulin family of peptide. The first are insulin and IGF receptor as the single transmembrane cytokine/growth factortype receptors (21,42). The second are the IGF-binding proteins (43,44), which are secreted soluble binding proteins for IGF1 and IGF2. The third are LGR7 and LGR8 as the hormone-type receptors for relaxin, relaxin-3, and INSL3. The fourth is the typical type I GPCR receptor that we describe here as GPCR135. There is currently no evidence that we know of to support that other relaxin/insulin family members also activate type I GPCRs.
The mRNA expression for GPCR135 and relaxin-3 has been previously reported (27, 31 45). GPCR135 (somatostatin-and angiotensin-like peptide receptor) mRNA was shown by RT-PCR to be expressed predominantly in the brain, whereas relaxin-3 mRNA has been shown to be almost exclusively expressed in the brainstem. With our identification of relaxin-3 as the ligand for GPCR135, we attempted to systematically map and compare the mRNA expression patterns of GPCR135 and relaxin-3, particularly in the brain. Our results show that mRNA transcripts for GPCR135 and relaxin-3 are present in the brain as detected by RT-PCR and in situ hybridization.
Several lines of evidence suggest that GPCR135 is the receptor for relaxin-3. First, although relaxin-3 can bind and activate LGR7, it is the only member of the ligand family to activate GPCR135. Second, LGR7 is expressed in the uterus and relaxin is expressed in the ovary, which is consistent with its reproductive role. Third, relaxin-3 is expressed in the nucleus incertus, which has afferent projects to the hypothalamus and forebrain (46) where GPCR135 is abundantly expressed. This hypothesis does not preclude the possibility that relaxin-3 is also an endogenous ligand for LGR7 or other yet to be identified receptors. Definitive assignment will require additional experiments including co-localization and possibly knockout animals.
The dominant brain expression profiles of relaxin-3 and GPCR135 suggest that this ligand/receptor pair plays a role in the central nervous system. Given the fact that GPCR135 and LGR7 share a ligand but have vastly different tissue expression patterns and signal via opposite transduction pathways (G i versus G s ), we speculate that the two receptors exert different physiological functions, but, perhaps, in a coordinated manner orchestrated by their common ligand, relaxin-3. For example, because LGR7 is known to be involved in tissue remodeling during pregnancy, perhaps the physiological role of GPCR135 may involve feeding, energy expenditure, metabolism, or other related central functions acting in concert to meet the reproductive needs of the body. It is also equally as likely that GPCR135 and LGR7 mediate unrelated physiological functions but share only relaxin-3 as a ligand. Additional studies will be required to truly understand the interplay of these two receptor systems.
In summary, we have purified and identified relaxin-3 as a ligand for GPCR135 from porcine brain. Recombinant human relaxin-3 binds and activates GPCR135 with high affinity. Pharmacological studies of GPCR135 indicated that relaxin-3 is the only member of the relaxin/insulin family that activates GPCR135. The central nervous system expression patterns of relaxin-3, GPCR135 coupled with the high affinity interaction of relaxin-3 to GPCR135 strongly suggest that relaxin-3 is an endogenous ligand for GPCR135.