Identification of p2y9/GPR23 as a novel G protein-coupled receptor for lysophosphatidic acid, structurally distant from the Edg family.

Lysophosphatidic acid (LPA) is a bioactive lipid mediator with diverse physiological and pathological actions on many types of cells. LPA has been widely considered to elicit its biological functions through three types of G protein-coupled receptors, Edg-2 (endothelial cell differentiation gene-2)/LPA1/vzg-1 (ventricular zone gene-1), Edg-4/LPA2, and Edg-7/LPA3. We identified an orphan G protein-coupled receptor, p2y9/GPR23, as the fourth LPA receptor (LPA4). Membrane fractions of RH7777 cells transiently expressing p2y9/GPR23 displayed a specific binding for 1-oleoyl-LPA with a Kd value of around 45 nm. Competition binding and reporter gene assays showed that p2y9/GPR23 preferred structural analogs of LPA with a rank order of 1-oleoyl- > 1-stearoyl- > 1-palmitoyl- > 1-myristoyl- > 1-alkyl- > 1-alkenyl-LPA. In Chinese hamster ovary cells expressing p2y9/GPR23, 1-oleoyl-LPA induced an increase in intracellular Ca2+ concentration and stimulated adenylyl cyclase activity. Quantitative real-time PCR demonstrated that mRNA of p2y9/GPR23 was significantly abundant in ovary compared with other tissues. Interestingly, p2y9/GPR23 shares only 20-24% amino acid identities with Edg-2/LPA1, Edg-4/LPA2, and Edg-7/LPA3, and phylogenetic analysis also shows that p2y9/GPR23 is far distant from the Edg family. These facts suggest that p2y9/GPR23 has evolved from different ancestor sequences from the Edg family.

Lysophosphatidic acid (LPA, 1-or 2-acyl-sn-glycero-3-phosphate) 1 is a bioactive phospholipid with diverse physiological actions on many cell types (1,2). LPA induces mitogenic and/or morphological effects on the cells and has been proposed to be involved in biologically important processes, including neurogenesis, myelination, angiogenesis, wound healing, and cancer progression (1,3). LPA is present in serum at micromolar concentrations (4). LPA is generated mainly by two different pathways; 1) generation of lysophospholipids such as lysophosphatidylcholine (LPC), lysophosphatidylethanolamine (LPE), and lysophosphatidylserine (LPS) from membrane phospholipids by phospholipase A 2 (PLA 2 ) or phospholipase A 1 , followed by conversion of these lysophospholipids to LPA by lysophospholipase D (5) and 2) generation of phosphatidic acid (PA) from phosphatidylcholine (PC) by phospholipase D, followed by conversion of PA to LPA by specific classes of PLA 2 (6).
Construction of the Phylogenetic Tree-Peptide sequences of selected G protein-coupled receptors were obtained from GenBank TM and SwissProt. The phylogenetic tree was generated from peptide sequences of selected G protein-coupled receptors, using the all-against-all matching method (available at cbrg.inf.ethz.ch/Server/AllAll.html). The tree was constructed on the basis of point-accepted mutation distances between each pair of sequences estimated by the dynamic programming algorithm.
Cloning of p2y 9 /GPR23-The tBLASTn program was used to search the data base of GenBank TM for orphan G protein-coupled receptors sharing high identities with the human PAF receptor (14). A DNA fragment containing the entire open reading frame of p2y 9 /GPR23 (GenBank TM accession number NM_005296) was first amplified from human genomic DNA by PCR using KOD-Plus (Toyobo, Osaka, Japan) and oligonucleotides (sense primer, 5Ј-GTCCATAGTGTCAGAGTGGT-GAAC-3Ј; antisense primer, 5Ј-CATATCTGGACCTGAACACATTTC-3Ј). The entire open reading frame of p2y 9 /GPR23 with an additional sequence of hemagglutin (HA)-epitope at the 5Ј-end was subsequently amplified from the resultant PCR products using KOD-Plus and oligonucleotides (sense primer containing KpnI and HA tag sequences, 5Ј-GGGGTACCGCCATGTACCCCTACGACGTGCCCGACTACGCCGGT-GACAGAAGATTCATT-3Ј; antisense primer containing XbaI sequence, 5Ј-GCTCTAGACTAAAAGGTGGATTCTAG-3Ј). The resultant DNA fragment was digested with KpnI and XbaI and subsequently cloned into the mammalian expression vector pCXN2.1, a slightly modified version of pCXN2 (15) with multiple cloning sites, between KpnI and NheI sites.
Binding Assay-RH7777 cells and B103 cells were cultured on collagen-coated dishes in Dulbecco's modified Eagle's medium (DMEM, Sigma) supplemented with 10% fetal bovine serum (Cambrex Co., Walkersville, MD), 100 IU/ml penicillin, and 100 g/ml streptomycin (Roche Applied Science). Cells were transfected with p2y 9 /GPR23-pCXN2.1 or empty vector using LipofectAMINE 2000 reagent (Invitrogen). After 24 h of transfection, cells were washed with phosphatebuffered saline three times and serum-starved for 24 h in DMEM supplemented with 0.1% BSA. The cells were washed again with phosphate-buffered saline twice and scraped off. After further washing with binding buffer (25 mM HEPES-NaOH (pH 7.4), 10 mM MgCl 2 , and 0.25 M sucrose), the cells were suspended in the buffer with additional 20 M 4-amidinophenylmethylsulfonyl fluoride (Sigma) and a protease inhibitor mixture (Complete, Roche Applied Science), sonicated three times at 15 watts for 30 s, and centrifuged at 800 ϫ g for 10 min at 4°C. The supernatant was further centrifuged at 10 5 ϫ g for 60 min at 4°C, and the resultant pellet was homogenized in ice-cold binding buffer. Binding assays were performed in 96-well plates in triplicates. For Scatchard analysis, 40 g of the membrane fractions were incubated in binding buffer containing 0.25% BSA with various concentrations of [ 3 H]LPA for 60 min at 4°C. The bound [ 3 H]LPA was collected onto a Unifilter-96-GF/C (PerkinElmer Life Sciences) using a MicroMate 196 harvester (Packard, Wellesley, MA). The filter was then rinsed ten times with binding buffer containing 0.25% BSA and dried for 2 h at 50°C. 25 l of MicroScint-0 scintillation mixture (PerkinElmer Life Sciences) was added per well. The radioactivity that remained on the filter was measured with a TopCount microplate scintillation counter (Packard). Total and nonspecific bindings were evaluated in the absence and presence of 10 M unlabeled LPA, respectively. The specific binding value (dpm) was calculated by subtracting the nonspecific binding value (dpm) from the total binding value (dpm). For competition assay with related lipids, 20 g of the membrane fractions were incubated with 5 nM [ 3 H]LPA in the absence or presence of 1 M of unlabeled 18:1-LPA, 18:1-LPC, 18:1-LPE, 18:1-LPS, 18:1-LPG, 18:1-PA, PAF, S1P, or SPC. For competition assay with structural analogs of LPA, 10 g of the membrane fractions was incubated with increasing concentrations of unlabeled 18:1-, 18:0-, 1-alkyl-, and 1-alkenyl-LPA in the presence of 2.5 nM [ 3 H]LPA. Before conducting the binding assays, the cell surface expression of p2y 9 /GPR23 was confirmed by flow cytometric analysis (Epics XL, Beckman Coulter, Fullerton, CA) with anti-HA rat IgG (3F10, Roche Applied Science) and phycoerythrin-labeled anti-rat IgG (Beckman Coulter) as the second antibody.
cAMP Assay-CHO cell clones stably expressing p2y 9 /GPR23 were used to measure cAMP levels. After 24 h of serum starvation, cells were harvested and suspended in HBSS containing 0.1% BSA and 0.5 mM isobutylmethylxanthine. The density was 5 ϫ 10 6 or 5 ϫ 10 7 cells/ml for assay in the presence or absence of forskolin, respectively. 20 l of the cell suspension was applied to 96-well plates and incubated for 20 min at room temperature. The reaction was initiated by adding 10 l of ligand solution of 18:1-LPA in HBSS-BSA buffer with or without 5 M forskolin. After 30 min of incubation at room temperature, the reaction was terminated by adding 10 l of HBSS-BSA buffer containing 4% Tween 20. After centrifugation at 800 ϫ g for 5 min, cAMP contents in the supernatant were measured by a Fusion system (Packard) using an AlphaScreen cAMP assay kit (PerkinElmer Life Sciences). Pretreatment with pertussis toxin (List Biological Laboratories, Campbell, CA) was for 12 h at a concentration of 100 ng/ml. Results were expressed as -fold increases over respective controls.
Quantitative Real-time PCR-Human first strand cDNAs from 16 tissues were purchased from Clontech (Human MTC Panel I and II), whose concentrations were normalized to the mRNA expression levels of four different housekeeping genes (␣-tubulin, ␤-actin, glyceraldehyde-3-phosphate dehydrogenase (GAPDH), and PLA 2 ). The cDNA levels of GAPDH and p2y 9 /GPR23 were quantified using a LightCycler apparatus (Roche Applied Science). The PCR reactions were set up in microcapillary tubes in a volume of 20 l, consisting of 2 l of cDNA solution, 1ϫ FastStart DNA Master SYBR Green I (Roche Applied Science), 0.5 M each sense and antisense primers and 3 mM MgCl 2 . The PCR program for GAPDH was as follows; denaturation at 95°C for 3 min and 50 cycles of amplification consisting of denaturation at 95°C for 0 s, annealing at 60°C for 5 s, and extension at 72°C for 40 s. The PCR program for p2y 9 /GPR23 was as follows: denaturation at 95°C for 5 min and 50 cycles of amplification consisting of denaturation at 95°C for 15 s, annealing at 60°C for 5 s, and extension at 72°C for 6 s. For GAPDH, a human GAPDH control Amplimer Set (Clontech), designed to amplify a 983-bp fragment was used as primers; sense primer: 5Ј-TGAAGGTCGGAGTCAACGGATTTGGT-3Ј and antisense primer: 5Ј-CATGTGGGCCATGAGGTCCACCAC-3Ј. The primers for p2y 9 / GPR23 were designed to amplify a 139-bp fragment; sense primer: 5Ј-AAAGATCATGTACCCAATCACCTT-3Ј and antisense primer: 5Ј-CTTAAACAGGGACTCCATTCTGAT-3Ј. The PCR products were detected by measuring the fluorescence of SYBR Green I, which selectively bound to double-stranded DNA and emitted the greatly enhanced fluorescence. The cDNA level of each sample was quantified by Fit Points Method in LightCycler analysis software. The control cDNA contained in Clontech human MTC Panels and a linearized p2y 9 / GPR23-pCXN2.1 were used as standards for GAPDH and p2y 9 /GPR23, respectively. In both cases, the quality of PCR products was assessed by monitoring a fusion step.
Northern Hybridization-Human poly(A) ϩ RNA samples from kidney and skeletal muscle were purchased from Clontech. Total RNA of human megakaryoblastic MEG-01 cells was extracted with Absolutely RNA RT-PCR Miniprep kit (Stratagene, La Jolla, CA). Poly(A) ϩ RNA was isolated from 200 g of the total RNA using MACS mRNA isolation kit (Miltenyi Biotec, Bergisch Gladbach, Germany). For Northern analysis, 2.5 g of poly(A) ϩ RNA was hybridized with [ 32 P]dCTP-labeled probes of human p2y 9 /GPR23 and human ␤-actin, as described previously (16). The washed membrane was subjected to autoradiography for 4 days (p2y 9 /GPR23) or 3 h (␤-actin).
Effect on Intracellular Ca 2ϩ Mobilization-To characterize the intracellular signals of p2y 9 /GPR23, the effect of LPA on intracellular Ca 2ϩ mobilization was examined in detail using a CAF-100 spectrofluorometer. We established four independent clones of CHO cells stably expressing p2y 9 /GPR23 and a polyclonal population of mock-transfected CHO cells. Cell surface expression of p2y 9 /GPR23 was confirmed by flow cytometric analysis (Fig. 4A). The other three clones showed similar expression patterns. Either in p2y 9 /GPR23-expressing or mocktransfected CHO cells, 18:1-LPA induced an increase in [Ca 2ϩ ] i in a dose-dependent manner (Fig. 4B). Although mock-transfected CHO cells displayed a significant increase in [Ca 2ϩ ] i , the increase in [Ca 2ϩ ] i was enhanced ϳ2-fold by the stable expression of p2y 9 /GPR23. Similar results were obtained in all four different clones. The effects of 18:1-LPA were reproducibly Effect on cAMP Formation-18:1-LPA induced an increase in cAMP levels in p2y 9 /GPR23-expressing CHO cells either in the absence or presence of 5 M forskolin (Fig. 5, A and B), and pretreatment of the cells with pertussis toxin further increased the cAMP levels (Fig. 5, A and B). In mock-transfected CHO cells, LPA induced no change or a decrease in cAMP levels in the absence or presence of forskolin, respectively (Fig. 5, A and  B), and pretreatment of the cells with pertussis toxin attenuated an LPA-induced decrease in cAMP levels (Fig. 5B).
Tissue Distribution-To explore the physiological function of p2y 9 /GPR23 in vivo, it is important to know the tissue distribution of the receptor. By using cDNAs prepared from 16 human tissues as templates, quantitative real-time PCR was performed to estimate the mRNA expression levels. In a set of samples, ovary showed the highest expression of p2y 9 /GPR23 mRNA, whereas other tissues showed only weak expressions (Fig. 6A). Northern hybridization of human poly(A) ϩ RNA from kidney, skeletal muscle, and megakaryoblastic MEG-01 cells (20) detected a transcript of about 4.4 kb (Fig. 6C). DISCUSSION LPA is a lipid mediator with diverse physiological activities (1,3). Many structural analogs of LPA have been identified in mammalian cells and tissues. Most are 1-acyl-LPAs with unsaturated fatty acyl-chains (oleoyl, linoleoyl, and arachidonoyl), and smaller amounts are with saturated fatty acylchains (palmitoyl and stearoyl) (21). Recently, 1-alkyl-, 1-alkenyl-, and 2-acyl-LPAs were also found (22)(23)(24). LPA has been widely considered to elicit its physiological functions through three types of G protein-coupled receptors, Edg-2/ LPA 1 , Edg-4/LPA 2 , and Edg-7/LPA 3 (2,3). However, there are some reports implying the existence of an additional LPA receptor(s). First, in the study of Edg-2/LPA 1 (Ϫ/Ϫ) Edg-4/ LPA 2 (Ϫ/Ϫ) double knockout mice, some LPA-induced responses, such as inositol phosphate production, adenylyl cyclase inhibition, and stress fiber formation, were absent or severely reduced but still remained at high LPA concentrations in embry- onic fibroblast cells (10). They reported that Edg-7/LPA 3 was not detected by Northern blotting or reverse transcriptase-PCR in these cells. Second, RH7777 cells that do not express Edg-2/LPA 1 , Edg-4/LPA 2 , or Edg-7/LPA 3 have a mitogenic response to LPA and LPA analogs (12). Third, LPA-induced platelet aggregation showed different ligand specificities with Edg receptor-mediated response; the platelet response lacks the stereoselectivity (11), requires micromolar concentrations of LPA (11), and displays a distinct ligand selectivity with a preference to 1-alkyl-LPAs (13). Here we identified p2y 9 /GPR23 as the fourth LPA receptor (LPA 4 ). p2y 9 /GPR23 was identified as a novel G protein-coupled receptor from an analysis of the expressed sequence tag (EST) data base, and the complete clone was isolated from human genomic DNA (25,26). The p2y 9 /GPR23 gene is located on chromosome X, region q13-q21.1, and contains an intronless open reading frame of 1113 bp encoding 370 amino acids (25,26). However, information is limited regarding the specific ligands, tissue distribution, and biological functions of this orphan receptor.
Mock-transfected CHO cells displayed an increase in [Ca 2ϩ ] i (Fig. 4, B and C), possibly due to the presence of endogenous LPA receptors. Despite this background response of CHO cells, the stable expression of p2y 9 /GPR23 significantly enhanced the LPA-induced Ca 2ϩ response by ϳ2-fold (Fig. 4B) in four independent clones. These results strongly suggest that p2y 9 / GPR23 could elicit an intracellular Ca 2ϩ mobilization, a well documented cellular effect of LPA (27).
In mock-transfected CHO cells, LPA induced a decrease in cAMP levels in the presence of forskolin, which was inhibited by pretreatment of the cells with pertussis toxin (Fig. 5B), suggesting the existence of endogenous LPA receptors coupling with pertussis toxin-sensitive G protein (G i/o ). By contrast, LPA induced an increase in cAMP levels in p2y 9 /GPR23-expressing CHO cells, and pretreatment of the cells with pertussis toxin further potentiated the LPA-induced cAMP accumulation (Fig.  5, A and B). It is, therefore, possible that p2y 9 /GPR23 is coupled with G s , and that the effect of LPA on p2y 9 /GPR23 is unmasked by blocking the pertussis toxin-sensitive signals from endogenous LPA receptors in CHO cells. Muscarinic M 2 and somatostatin sst 5 receptors are coupled with G i , inhibiting adenylyl cyclase in CHO cells. However, at higher agonist concentrations, these receptors can also mediate activation of adenylyl cyclase by a mechanism involving G s activation (30,31). Conversely, at lower concentrations of LPA, p2y 9 /GPR23 might inhibit the production of cAMP via G i , like Edg-2/LPA 1 , Edg-4/LPA 2 , and Edg-7/LPA 3 (2, 3).
We also found that forskolin facilitated LPA-induced cAMP accumulation in p2y 9 /GPR23-expressing CHO cells (Fig. 5, A  and B). At least ten types of mammalian adenylyl cyclase are at present identified (32). All types of adenylyl cyclase are activated by G s and by forskolin, and some types of adenylyl cyclase (type II, IV, V, and VI) are synergistically activated in the presence of both G s and forskolin (32). One possible explanation of our results is that the latter types of adenylyl cyclase might be involved in the LPA-induced cAMP accumulation in p2y 9 /GPR23-expressing CHO cells. Indeed, type VI and type VII adenylyl cyclases are expressed in CHO cells (33).
The mRNA levels of p2y 9 /GPR23 were significantly high in ovary (Fig. 6). Various species of LPA such as linoleic, arachidonic, and docosahexaenoic acids were detected from ascites of ovarian cancer patients (24), and they had many effects on the ovarian cancer progression such as cell proliferation, prevention of apoptosis, resistance to cisplatin, and production of vascular endothelial growth factor (34). Consistently, a prominent expression of Edg-4/LPA 2 has been shown in primary cultures and established lines of ovarian cancer cells (35). LPA was also found at relatively high concentrations in human ovarian follicular fluid from healthy subjects (36), suggesting the relevance of LPA for normal ovarian functions as well. Tokumura (37) recently described in his review article that LPA increased the intracellular cAMP level in mouse cumulus cells. This phenomenon is consistent with our findings that the activation of p2y 9 /GPR23 evoked cAMP accumulation in CHO cells. Thus, p2y 9 /GPR23 might explain some of the pathological and physiological roles of LPA in ovary. It remains to be determined whether the expression of p2y 9 /GPR23 is modulated in ovarian cancer cells. Although the EST cDNA encoding p2y 9 /GPR23 was originally isolated from human brain (25,26), high expression of p2y 9 /GPR23 was not detected in brain in our study. It is possible that specific types of cells in re- stricted areas express p2y 9 /GPR23, which will be examined by in situ hybridization in the near future.
Interestingly, p2y 9 /GPR23 shares only 20 -24% amino acid identities with Edg-2/LPA 1 , Edg-4/LPA 2 , and Edg-7/LPA 3 . Phylogenetic analysis also shows that p2y 9 /GPR23 is far distant from the Edg family (Fig. 1). These facts suggest that p2y 9 / GPR23 has evolved from ancestor sequences that are different from those of the Edg family. There are several examples of structurally unrelated receptors recognizing the same ligand. Prostaglandin D 2 binds to DP and CRTH2 (38) and histamine has four structurally distant receptors, H 1 -H 4 (39). In addition, some neurotransmitters and nucleotides have both metabotropic (G protein-coupled) and ionotropic (ion channel) receptors. These examples show a limitation of the ligand search strategy utilizing a structural similarity of receptor.
As described above, there are some reports implying the existence of additional receptors for LPA. It is possible that LPA-induced responses in embryonic fibroblast cells of Edg-2/ LPA 1 (Ϫ/Ϫ) Edg-4/LPA 2 (Ϫ/Ϫ) double knockout mice (10) might be mediated by p2y 9 /GPR23, although we do not have any direct evidence. However, a mitogenic response to LPA in RH7777 cells (12) might be due to the activity of intracellular receptors (40), rather than G protein-coupled receptor, because mocktransfected RH7777 cells exhibited no significant binding to [ 3 H]LPA ( Fig. 2A). As to the putative LPA receptor in platelets (11,13), there may be a receptor other than p2y 9 /GPR23, because the ligand preference of platelets to 1-alkyl-LPAs is not consistent with that of p2y 9 /GPR23. Existence of further unidentified LPA receptors is, therefore, expected.
In conclusion, we report here the identification of p2y 9 / GPR23 as a novel fourth LPA receptor (LPA 4 ). Cells expressing p2y 9 /GPR23 displayed intracellular Ca 2ϩ mobilization, cAMP accumulation, and luciferase activation. The K d value of p2y 9 / GPR23 (45 nM) was equivalent to those of Edg-4/LPA 2 (73.6 nM) and Edg-7/LPA 3 (206 nM) (9). Although p2y 9 /GPR23 mRNA was significantly detected in ovary, its biological functions in vivo remain to be determined. Nevertheless, the present findings introduce a further complexity for LPA and its receptors.
In addition, our study shows a limitation of the "de-orphaning" strategy based on the receptor structure.