A large number of physiological events involve intercellular signaling through G protein-coupled receptors (GPCR). Furthermore, many drugs used in medicine today interact with such receptors. However, only a fraction of all potential receptors encoded in the genome have been assigned a cognate ligand. Therefore, the remaining orphan receptors constitute a vast reservoir of potential drug targets for therapeutic development. This makes the identification of ligand receptor pairs (deorphanization) one of the most important efforts in medicine today.

The human GPR92 gene is located on chromosome 12, region p 13.31, and contains an intronless open reading frame that encodes 372 amino acid residues. It belongs to the rhodopsin subfamily of GPCR and is structurally most closely related to GPR23/lysophosphatidic acid 4 (LPA₄), a recently identified receptor for LPA (Noguchi et al., 2003).

LPA, also known as monoacyl-sn-glycero-3-phosphate, is a lipid-signaling mediator with a wide spectrum of actions. Originally described as an intermediate in intracellular lipid biosynthesis, LPA is now accepted as a pluripotent extracellular mediator with mitogenic effects in almost every tissue (Moolenaar et al., 2004). At a physiological level, LPA influences such diverse biological processes as tissue remodeling, neurogenesis, myelination, and olfaction, as well as reproductive and immune functions (Mills and Moolenaar, 2003; Ishii et al., 2004).

From a pathophysiological point of view, increasing evidence indicates that LPA has an important role in wound healing by attracting macrophages and by stimulating migration and proliferation of mesenchymal and epithelial cells (Balazs et al., 2001; Sturm and Dignass, 2002). It also seems to participate in the development of atherosclerosis as an active constituent of oxidized low-density lipoprotein (Natarajan et al., 1995; Hayashi et al., 2001; Siess and Tigyi, 2004).

The diverse actions of LPA together with the fact that LPA is a major serum constituent at >1 μM (Goetzl and Lynch, 2000) prompted research toward the identification of high-affinity LPA receptors. To this date, four LPA receptors, all belonging to the genetic superfamily of G protein-coupled receptors, have been described: LPA₁ (Edg-2), LPA₂ (Edg-4), LPA₃ (Edg-7), and the structurally divergent GPR23/LPA₄ (Chun et al., 2002; Anliker and Chun, 2004a,b). Whereas LPA₁ is the receptor with the widest distribution, the expression of LPA₂ and LPA₃ is somewhat more restricted and GPR23/LPA₄ is expressed only in the ovary (Noguchi et al., 2003).

This study describes the identification of the previous orphan receptor GPR92 as a putative novel LPA receptor. This receptor is highly expressed in placenta, lung, brain, heart, spleen, and immune cells, particularly those associated with the gastrointestinal tract.

Materials and Methods

Chemicals. Lysophosphatidic acid, 1-oleoyl [oleoyl-9,10-³H(N)] (48.6 Ci/mmol), and inositol (myo-[2-³H(N)]) (18.5 Ci/mmol) were obtained from PerkinElmer Life Sciences, Inc. (Boston, MA). Sphingosine-1-phosphate (S1P) 18:1 and oleoyl LPA 18:1 were obtained from Larodan Fine Chemicals (Malmö, Sweden). Stearoyl LPA 18:0, palmitoyl LPA 16:0, myristoyl LPA 14:0, oleoyl lysophosphatidylcholine (LPC) 18:0, arachidoyl LPC 20:0, oleoyl lysophosphatidylethanolamine (LPE) 18:1, oleoyl lysophosphatidylserine (LPS) 18:1, oleoyl lysophosphatidylglycerol (LPG) 18:0, and diacylglycerol pyrophosphate (DGPP) 8:0 were from Avanti Polar Lipids (Alabaster, AL). All other chemicals were from Sigma-Aldrich (St. Louis, MO).

Cell Culture. The mouse cell line mICc12 derived from the small intestinal epithelium was obtained from Dr. A. Vandevalle (IN-SERM, Paris, France) and was grown as described by Bens et al. (1996). MODE-K, another mouse small intestinal cell line, was grown in DMEM supplemented with 10% FBS, and 1 mM sodium pyruvate (Invitrogen, Carlsbad, CA). The mouse T-cell line 2017 and the S194 mouse plasmacytoma cells (both from American Type Culture Collection, Manassas, VA) were grown in RPMI 1640 medium supplemented with 2% sodium pyruvate, 1% HEPES, and 10% FBS at 37°C in 5% CO₂. The reporter cell lines HF1 (Kotarsky et al., 2001) and HF1-GPR92 (stably expressing human GPR92) were grown in DMEM containing 10 and 3% FBS, respectively. CHO K1 cells were grown in DMEM containing 10% FBS. Stable CHO cell lines were transfected with the plasmid pIRESpuro-GPR92 (see below) or a plasmid coding for green fluorescent protein instead, using Metafectene (Biontex, Munich, Germany) and selected for 3 weeks by the addition of puromycin.

The following blood cell lines were studied: Nalm6, CCR-CEM, T45, U937, Molt-4, SKO007, TF1, TMβ1, and HL60 [kind gifts from Dr. Magnus Lindvall (Lund University, Lund, Sweden], all grown in RPMI 1640 medium containing 10% FBS. The HL60 cells were differentiated by 1% dimethyl sulfoxide for 7 days and the Nalm6 and CCR-CEM cells with 1 μM retinoic acid for 7 days.

Construction of Reporter Cells. The gene for the human GPR92 was isolated from genomic DNA by PCR using Pfx polymerase (Invitrogen) and the primers 5′-GCGAATTCACCATGTTAGCCAACAGCTCCTCA-3′ and 5′-GCGAATTCAGAGGGCGGAATCCTGGGGACA-3′. In brief, 100 ng of human gDNA (Promega, Madison, WI) was amplified with 20 μM of primers as described by the manufacturer; the product was subcloned in the pIRESpuro vector (BD Biosciences Clontech, Palo Alto, CA) and confirmed by sequencing. The reporter cell line HF1-GPR92 was constructed using the HF1 host reporter cell line and the pIRESpuro-GPR92 vector, essentially as described by Kotarsky et al. (2001).

Reporter Assay. HF1 and HF1-GPR92 reporter cells were seeded (1.6 × 10⁴ per well) in 96-well plates in 100 μl of medium. After 72 h, test substances diluted in phosphate-buffered saline (PBS) without Ca²⁺ or Mg²⁺ were added to the wells. After another 16 h, incubation was interrupted by the removal of the medium. Ten microliters per well of lysis buffer were added, and the plates were stored at -80°C until analysis. Luciferase activity was measured with a luciferase assay kit (BioThema, Handen, Sweden) according to the manufacturer's instructions. All samples were run in triplicate and repeated two to six times. The luminescence assays were done in a BMG Lumistar Galaxy luminometer (BMG Labtech, Offenburg, Germany) as described previously (Kotarsky et al., 2001, 2003b).

Phosphoinositide Hydrolysis. HF1 and HF1-GPR92 cells were assayed essentially as described previously (Kang and Leeb-Lundberg, 2002), with a few modifications. In brief, cells grown in 6-well dishes were incubated with 5 μCi/ml myo-[³H]inositol in DMEM, 5% heat-inactivated horse serum at 37°C for 24 h in 10% CO₂. Prior to experimentation, the cells were washed four times with 1 ml of Leibovitz's L-15 medium at room temperature and incubated in Leibovitz's L-15 medium, 1 mM LiCl for 30 min. Following replacement with 2 ml of the same medium, the cells were incubated with increasing concentrations of LPA at 37°C for 30 min. Inositol phosphates were then extracted and isolated using anion-exchange chromatography.

cAMP Production. HF1 and HF1-GPR92 cells were grown in 6-well dishes to approximately 70% confluency. After stimulating the cells for 20 min with LPA, the cell layer was washed and then extracted for 20 min with 300 μl of 0.1 M HCl per well. The cell extracts were analyzed using a cAMP enzyme-linked immunosorbent assay kit (Cayman Chemical, Ann Arbor, MI) according to the manufacturer's instructions.

Membrane Preparation for Radioligand Binding Studies. Membrane preparations were essentially done as described previously (Owman et al., 1997). In brief, HF1 and HF1-GPR92 cells grown on 10-cm Petri dishes were first washed twice with ice-cold PBS. The cells were then resuspended in buffer containing 50 mM Tris-HCl, pH 7.5, 1 mM EDTA, and 1 mM phenylmethylsulfonyl fluoride and homogenized using an Ultra-turrax tissue homogenizer (IKA Werke GmbH, Staufen, Germany) at 20,500 rpm for 10 to 15 s. Membranes were isolated by centrifugation, and the pellets were then resuspended in the above buffer and frozen at approximately 9 mg/ml.

Radioligand Binding. Membranes were diluted in PBS/0.1% BSA to a concentration of approximately 0.2 mg of protein/ml. Binding assays were then performed by incubating ∼80 μg of protein with various concentrations of [³H]LPA in a total volume of 0.5 ml. Nonspecific binding was determined in the presence 1 μM nonradioactive LPA. After incubation for 60 min at room temperature, assays were terminated by dilution with 4 ml of ice-cold PBS/0.3% BSA and rapid vacuum filtration on Whatman GF/C filters (Whatman Schleicher and Schuell, Keene, NH) previously soaked in 1% polyethyleneimine. The trapped membranes were washed with an additional 2 × 4 ml of ice-cold PBS/0.3% BSA. The filters were counted for radioactivity in a Beckman LS5000TD scintillation counter. No specific [³H]LPA binding was detected in mock-transfected cells. Binding constants were calculated using the Sigma plot program (SYSTAT Software, Point Richmond, CA). Association rate experiments were done by incubating membranes for various times after adding [³H]LPA, whereas dissociation rate experiments were done by incubating membranes with [³H]LPA for 30 min followed by the addition of 1 μM LPA and incubation for various times there-after.

Northern Blot. Total RNA was isolated from 10⁷ cells by the method of Chomczynski and Sacchi (1987). Twenty-five micrograms of total RNA was separated by electrophoresis and transferred to a nylon membrane according to standard procedures. The membrane and two pre-made tissue blots (BD Biosciences Clontech) were hybridized with the cloned human GPR92 open reading frame according to standard procedures (50% dimethyl formamide at 42°C) and subjected to autoradiography.

Quantitative Real-Time PCR. Real-time quantitative PCR was performed in a LightCycler system (Roche, Indianapolis, IN) using the SYBR Green I detection method. The reactions were performed in a total volume of 20 μl containing 5 μl of diluted cDNA (1:20), dH₂O (negative control) or purified gene-specific PCR product (for standard curves), 1× PCR buffer (Invitrogen), 3 mM MgCl₂, 7% dimethyl sulfoxide, 0.5 μM of each primer, 200 μM of each dNTP, 0.5 μg/μl BSA, 1:30,000 dilution of SYBR Green I, and 0.5 units of Platinum TAQ enzyme (Invitrogen). After denaturation at 94°C for 2.5 min, a total of 40 cycles were run (each for 10 s at 96°C, 10 s at 57°C, and 20 s at 72°C). Melting-curve profiles were analyzed, and the specificity of the bands further verified by electrophoresis on agarose gels. Primers used for quantitative (q)RT-PCR were constructed in accordance to Peixoto et al. (2004) and are listed in Table 1.

Isolation of Cells from Mouse Intestine and Lymphoid Organs. Epithelial cells and intraepithelial lymphocytes from the mouse intestinal mucosa were prepared as described by Svensson et al. (2002). In brief, dissected mouse small intestines were flushed with PBS, opened, and after the removal of the Peyer's patches, cut into small pieces. Epithelial cells and crypts were released by six consecutive EDTA treatments of 20 min each. The fractions were pooled, and the cells were counted in an inverted microscope. Intraepithelial lymphocytes were enriched by Percoll gradient centrifugation. Lamina propria lymphocytes were isolated by two consecutive treatments with collagenase in the presence of Ca²⁺ ions and Percoll gradient centrifugation (Svensson et al., 2002). Lymphocytes from mesenteric lymph nodes and spleen were isolated by passing the organs through a tissue strainer.

Purification of CD8α⁺Intraepithelial Lymphocytes from Mouse Small Intestine. Intraepithelial lymphocytes purified by Percoll gradient centrifugation (see above) were incubated with blocking buffer [PBS, anti-FcRII/III antibody (2.4G2), and 2% FBS] for 20 min on ice followed by an incubation with biotin coupled anti-CD8α antibody and labeling with antibiotin-coupled paramagnetic beads (Miltenyi Biotec Inc., Auburn, CA). Cells were separated on a LS column (Miltenyi Biotech) according to the manufacturer's protocol. The purity of the cell preparation was determined to be 95 to 97% CD8α⁺ as verified by staining the cells with fluorescein isothiocyanate-labeled anti-CD8α antibody (BD Biosciences, San Jose, CA) and analysis in a flow cytometer (FACS Calibur; BD Biosciences).

RT-PCR Analysis. Total RNA was isolated from 2 × 10⁶ cells prepared from the mouse tissues as described above using the Absolutely RNA Mini kit (Stratagene, La Jolla, CA). Total RNA (4 μg of total RNA for cell lines and 1 μg of total RNA of tissue lymphocytes) was used to synthesize cDNA with Superscript III (Invitrogen). The cDNA preparations of CD8⁺ intraepithelial lymphocytes (Marsal et al., 2002) were a kind gift from Dr. Markus Svensson (Lund University, Lund, Sweden). Total RNA of K46, 40eI, and J558 B-cell lines was a kind gift of Dr. Mikael Sigvardsson (Lund University).

One microliter of the cDNA and a 10× dilution were used for PCR with 20 μM of the following PCR primers: mouse GPR92 forward, 5′-TCTGCGGCCGCGACATGATGTTTGCCAATTCTTCAG-3′; and mouse GPR92 reverse, 5′-AGGGAATTCTACGTTCACTTAGACTGCGTCCTCTCAG-3′. The following primers for mouse β-actin were used to assess the total amount of RNA present in the samples: β-actin forward, 5′-GGTGGGAATGGGTCAGAAGGACT-3′; and β-actin reverse, 5′-CCACGCTCGGTCAGGATCTTCAT-3′. The PCR-cycling conditions were as follows: 94°C for 2 min; 96°C for 30 s; 57°C for 30 s; 70°C for 1 min; 35 cycles and 72°C for 7 min.

Statistical Analysis. All data shown are calculated as means ± S.E.M. The level of significance between groups was assessed by unpaired Student's t test (95% confidence interval) using the Prism (GraphPad Software Inc., San Diego, CA) or InSTAT (GraphPad Software Inc.) software packages. Dose-response curves were plotted, and EC₅₀ and IC₅₀ values were calculated using Prism software.

Results

Sequence Analysis of the Orphan Receptor GPR92. Phylogenetically, the orphan receptor GPR92 maps to the “purinocluster” of GPCRs (Fredriksson et al., 2003) where one of the closest neighbors is GPR23/LPA₄, recently reported as an atypical LPA receptor (Noguchi et al., 2003). GPR92 displays 31% homology to GPR23/LPA₄ and between 21.3 and 22.6% homology with the classic LPA family of receptors (Fig. 1).

Identification of LPA as a Putative Ligand for GPR92. To search for the cognate ligand for GPR92, the GPR92 cDNA was stably introduced into the reporter cell line HF1. This cell line was chosen because it has been used earlier to identify ligands acting on orphan G protein-coupled receptors (Kotarsky et al., 2003a; Nilsson et al., 2003). The HF1 reporter cell line uses a synthetic enhancer region composed of six 12-O-tetradecanoylphorbol-13-acetate-response elements and a chimeric EGFP/luciferase reporter gene. HF1 cells are primarily activated by GPCRs coupling to G-proteins of the Gα_q and Gα_i classes (Kotarsky et al., 2001).

Because the GPR92 gene displayed high expression in leukocyte cell lines (see below), the presence of a natural ligand in serum was investigated. It was found that serum was able to strongly activate the receptor in a concentration-dependent and -specific manner (Fig. 2A). Attempts to purify the active component from bovine serum enriched an active fraction containing serum albumin (data not shown). As lipid compounds can bind to serum albumin (Pages et al., 2001), different lipid mediators were checked for their ability to activate GPR92-expressing reporter cells. LPA and S1P were probably serum-derived candidate ligands. Exposure to LPA 18:0 caused a statistically significant (p = 0.0014) activation of the reporter cells expressing GPR92, whereas no response was observed in the parent cells (Fig. 2B). On the other hand, S1P produced no significant difference (p = 0.5922) compared with HF1-GPR92 reporter cells stimulated with PBS (Fig. 2C). A small stimulation by S1P was observed in the control HF1 cell line (Fig. 2C).

To further study the specificity of the GPR92 activation by LPA, HF1 reporter cells were stimulated with increasing concentrations of LPA 14:0 in the presence or absence of the LPA₁ /LPA₃ antagonist DGPP 8:0 (Sardar et al., 2002). The presence of 1 μM DGPP did not shift the dose-response curve of LPA 14:0 (Fig. 2D). This was further confirmed by the observation that increasing concentrations of DGPP had no effect on the LPA 14:0 activation of HF1pGPR92 reporter cells (Fig. 2E).

After transfection of GPR92 cDNA into the reporter cells, a significant amount of activity was observed in the presence of PBS only (Fig. 2A). This activity was not due to the presence of remaining serum, because it was also found after incubation in serum-free medium or after prolonged cultivation in medium supplemented with delipidated FBS (data not shown).

To exclude the possibility that introduction of the GPR92 cDNA into the reporter cells had changed the expression of another LPA receptor subtype, we analyzed the expression of all known LPA receptor subtypes in HF1 and HF1pGPR92 cells by qRT-PCR. Whereas LPA₁ and GPR23/LPA₄ transcripts could not be detected at all, transcripts for LPA₂ and LPA₃ were present at low levels. Importantly, overexpression of GPR92 did not alter the expression levels of any of the other LPA receptor subtypes (LPA₁, LPA₂, LPA₃, and GPR23/LPA₄) (Fig. 2F).

To investigate the specificity of the GPR92/LPA interaction, a number of compounds including lipid mediators and purinoceptor ligands were tested on HF1 and HF1pGPR92 reporter cells. Whereas the positive controls, LPA 14:0 and fetal bovine serum at 15%, activated GPR92-transfected reporter cells, none of the purinoceptor ligands tested resulted in a specific activation of GPR92-transfected cells (Fig. 3). In fact, the structurally related lipid mediator platelet-activating factor at 1 μM was the only mediator that caused a weak but specific activation of GPR92-transfected cells [10-15% LPA 14:0; 1 μM] (Fig. 3).

To investigate the pharmacological profile of GPR92, 11 different lipid analogs, all structurally related to LPA 18:1, were tested. It was found that mediators based on the LPA structure were by far the most effective lipids (Table 2). LPA with acyl chains comprising 14 or 16 carbon atoms displayed the highest potency (Table 2). Among ligands based on structures other than LPA, only LPS 18:1 and LPC 18:1 were able to activate GPR92 (Table 2), albeit at concentrations 10 to 100 times higher than those required for activation by LPA.

Activation of GPR92 by LPA: Phosphoinositide Hydrolysis. Stimulation of HeLa HF1 and HF1pGPR92 cells with increasing concentrations of LPA 18:1 for 30 min in the presence of 1 mM LiCl resulted in a concentration-dependent increase in phosphoinositide (PI) hydrolysis in the GPR92-transfected cells (Fig. 4A). However, no EC₅₀ value could be calculated, because the response was still increasing at the highest LPA concentration used, which was 30 μM. Nevertheless, the curve was similar to that observed using the reporter assay. No statistically significant decrease in the response was observed following treatment with 100 ng/ml pertussis toxin (165 ± 15% basal PI hydrolysis for pertussis toxin treated and 182 ± 8% basal PI hydrolysis for untreated controls).

To confirm the results obtained in HeLa cells, GPR92 was also stably expressed in CHO cells. Only the GPR92-transfected cells responded to LPA with an increase in PI hydrolysis (Fig. 4B). Pretreatment with pertussis toxin again had no effect on the response (Fig. 4B).

Activation of GPR92 by LPA: cAMP Production. Stimulation of HeLa HF1pGPR92 cells with increasing concentrations of LPA 18:1 for 20 min in the presence of 1 mM 3-isobutyl-1-methylxanthine resulted in a concentration-dependent increase in the intracellular cAMP level (Fig. 4A). Again, LPA had no effect on control-transfected HeLa cells. The cAMP response curve was very similar to both the PI hydrolysis and the reporter cell-response curve.

Binding of LPA to GPR92. To directly determine whether LPA binds to GPR92, radioligand binding using [³H]LPA 18:1 was performed on a particulate preparation of HF1 and HF1pGPR92 cells. Figure 5A shows that [³H]LPA interacted with specific binding sites in the HF1pGPR92 membrane preparation. Analysis of equilibrium binding revealed that the radioligand bound with high affinity (K_D = 6.4 ± 0.9 nM) to an apparently single finite population of binding sites (B_max = 3.35 ± 0.36 pmol/mg protein, n = 3) (Fig. 5B). The [³H]LPA binding to GPR92 was completely reversible with a rapid association rate (k₁ = 1.25 ± 0.39 × 10⁷ M^-1 min^-1; n = 3) and a slow dissociation rate (k_-1 = 7.75 ± 0.72 × 10^-3 min^-1; n = 3), which corresponds to a K_D of 0.62 nM (Fig. 5C). Whereas the reason for the slightly lower K_D value calculated from the rate constants of the LPA binding compared with that obtained from equilibrium binding is currently not known, both K_D values indicate a high-affinity interaction in the nanomolar range. Analysis of the competition of [³H]LPA by several LPA analogs revealed that GPR92 has a strong preference for the LPA backbone regardless of the length of the acyl side chain (Table 3). Neither S1P (data not shown) nor the LPA₃/LPA₁ antagonist DGPP 8:0 (Sardar et al., 2002) competed for the LPA binding site present in GPR92-transfected cells. These results directly show that LPA is a specific, high-affinity ligand for GPR92.

Cell and Tissue Distribution of GPR92. Northern blot analysis of a set of human tissues showed that the GPR92 gene is primarily expressed in heart, placenta, brain, gut, and spleen (Fig. 6A). The results from spleen suggested a possible expression in blood cells. When a select number of leukocyte cell lines were subjected to Northern blot analysis, it was revealed that human TF1, Nalm6, and undifferentiated HL60 cells all strongly expressed GPR92 (Fig. 6B). Bands of different sizes were observed in both blots, which suggest that the GPR92 receptor can be expressed from different transcripts. Interestingly, the GenBank accession number BC033571 suggests a transcript length of 2352 base pairs for a GPR92 mRNA, which corresponds well to a predominant transcript detected. To obtain quantitative expression data for the GPR92 receptor gene, total RNA was isolated from a number of selected mouse tissues and the level of receptor gene expression determined by qRT-PCR. The highest expression was found in heart and small intestine. Lower levels were found in the other tissues examined (Fig. 6C).

Cellular Expression of Mouse GPR92 in Gut. Various approaches were used in attempts to identify the cell types expressing GPR92 in the gastrointestinal tract. First, PCR analysis was performed on cDNA prepared from two cell lines, MODE-K and mICc12, both derived from the mouse small intestinal epithelium. Neither cell line expressed the receptor (Fig. 7A). On the other hand, expression was verified in the murine B-cell plasmacytoma cell lines S194, 40eI, and J558 and in the mouse pro T-cell lymphocytoma cell line 2017 (Fig. 7A). Because the results indicated lymphocytes as the source of the receptor expression, we next analyzed different lymphocyte populations with semiquantitative PCR. The lymphocytes were isolated from mouse colon and small intestine and included intraepithelial lymphocytes (IEL), lamina propria lymphocytes (LPL), Peyers's patch lymphocytes, and lymphocytes from mesenteric lymph nodes. High levels of mouse GPR92 expression were observed in all populations. Interestingly however, the IEL compartment of the small intestine showed a stronger expression of GPR92 than the LPL compartment (Fig. 7B).

The next experiment was aimed to determine the major cellular expression site of mouse GPR92 in the small intestinal epithelium. Because the IEL compartment mainly consisted of CD8⁺ T cells, we subjected magnetic associated cell sorting-purified CD8α⁺ IEL, Peyer's patch lymphocytes, and an epithelial cell preparation to qRT-PCR. The highest expression level was observed in CD8α⁺ IEL followed by Peyers's patch lymphocytes and epithelial cells (Fig. 7C).

The CD8⁺ cells present in the small intestinal epithelium consist of different subpopulations defined by their expression of different CD8 and TCR chains. Three subpopulations of CD8⁺ T cells present in the epithelium, CD8αα⁺ TCRαβ, CD8αβ⁺ TCRαβ, and CD8αα⁺ TCRγδ, were investigated for their expression of GPR92 by semiquantitative PCR. Indeed, all were found to be positive for GPR92 expression (Fig. 8A). To strengthen the relevance of the finding that GPR92 is expressed on the CD8⁺ cells, we analyzed the relative expression of all known LPA receptor subtypes on these cells. The results showed that the GPR92 transcript is by far the most abundant transcript. Transcripts for LPA₃ and GPR23/LPA₄ could not be detected at all, and LPA₂ and LPA₁ transcripts were present at lower levels (Fig. 8B).

Discussion

LPA is the simplest glycerophospholipid and mediates a wide range of cellular responses, including platelet aggregation, cell proliferation, modulation of chemotaxis, smooth muscle cell contraction, protection from apoptosis, and transcellular migration (van Leeuwen et al., 2003; Moolenaar et al., 2004). LPA has been shown to interact with at least four GPCRs known as LPA₁, LPA₂, LPA₃, and GPR23/LPA₄ (Chun et al., 2002; Noguchi et al., 2003). In the present study, we identify the orphan receptor GPR92 as a putative additional LPA receptor.

The LPA receptor family can be divided into two subfamilies where one is composed of the first three members, LPA₁, LPA₂, and LPA₃, whereas the other subfamily consists of GPR23/LPA₄ and GPR92. The fact that the latter subfamily of LPA receptors is structurally more related to the purinoreceptor cluster of GPCRs than to the original LPA receptors indicates different ancestor genes. All of the human LPA receptor genes are localized to different chromosomes, but the ones forming the original subfamily belong to the same paralogon group (Fredriksson et al., 2003), which suggests that they have a common ancestral gene. Although LPA₄ and GPR92 share a structural similarity, they are part of different paralogons.

LPA bound to GPR92 with an affinity in the low nanomolar range. This affinity is higher than that of the previously described GPR23/LPA₄ (Noguchi et al., 2003). On the other hand, the affinities of LPA binding to the classic LPA receptors have not been reported. The pharmacological binding profile revealed that GPR92 was unable to discriminate between different LPA analogs varying in the length of their acyl groups. The LPA inhibitor DGPP failed to displace radiolabeled LPA from GPR92 membrane preparations. DGPP at 1 μM has been shown to inhibit signaling on the classic LPA receptors LPA₃ and LPA₁ (Sardar et al., 2002; Ohta et al., 2003).

GPR92-mediated LPA stimulation leads to increases in PI hydrolysis and cAMP production, suggesting that this receptor is coupled at least to the G_q and possibly also to the G_s class of G proteins in HeLa cells. This signaling pattern is similar to that of GPR23/LPA₄, but differs from the classic LPA receptors, which preferentially couple to G_i, G_12/13 and/or G_q. The potencies of LPA to stimulate both of these cellular responses as well as the reporter assay were considerably lower than the affinity of LPA for the receptor measured in particulate membranes. The reason for this difference is currently unknown. LPA at higher concentrations acts as an agonist through this receptor, since it stimulates both PI hydrolysis and cAMP production in a receptor-dependent manner. Whereas our results clearly show that LPA is a ligand for GPR92, the details of the LPA signal behavior for this receptor clearly requires further investigation. The fact that the LPA₁/LPA₃ antagonist DGPP failed to inhibit LPA stimulation of GPR92 is consistent with both its inability to displace radiolabeled LPA from GPR92 and to the structural difference between LPA₁/LPA₃ and GPR92. The order of potencies for LPA analogs with different chain length to stimulate the expression of the reporter gene showed that GPR92 was most efficiently activated with LPA analogs with an acyl chain shorter than 18 carbon atoms. This is in apparent contrast to the previously identified GPR23/LPA₄ (Noguchi et al., 2003) and the classic LPA receptors (Bandoh et al., 2000), which are more efficiently activated by long acyl chain LPA analogs. However, it must be noted that these results originate from different experimental assay systems. Nevertheless, it has been reported that a substantial amount of LPA consists of LPA 16:0 both in plasma (Baker et al., 2001) and brain tissue (Sugiura et al., 1999). A physiological difference between different LPA analogs has so far not been established. However, there is a discrepancy between the binding results, which indicate that chain length is unimportant for the displacing ability of the LPA analogs, and the activation results, showing a preference for shorter chain lengths. A possible interpretation of this observation could be the presence of two binding states for the GPR92 receptor, with only the lower affinity state coupled to activation of the receptor. In summary, GPR92 seems to be most related to GPR23/LPA₄, because they share structural similarity, a similar signaling pattern, and a comparable affinity for LPA.

The LPA receptors differ with respect to their distribution in various tissues (see Anliker and Chun, 2004a; Ishii et al., 2004 for review). Although LPA₁ is broadly expressed throughout the body, the expression of LPA₂ and LPA₃ is more restricted. GPR23/LPA₄, the first member of the purinoreceptor-related subfamily, is expressed almost entirely in the ovaries (Noguchi et al., 2003). Expression of the classic LPA receptors (LPA_1-3) has been studied in human leukocytes (Goetzl et al., 2000; Zheng et al., 2000). Whereas the LPA₃ receptor could not be detected on any of the cell populations studied, the LPA₂ receptor was mapped to CD19⁺ B cells and CD4⁺ T cells, and the LPA₁ receptor was expressed by CD4⁺ T cells and CD14⁺ monocytes (Goetzl et al., 2000; Graler and Goetzl, 2002). Interestingly, none of the LPA receptor subtypes studied was found in CD8⁺ T cells isolated from human blood (Goetzl et al., 2000; Zheng et al., 2000).

With regard to GPR92, although we also detected a high expression in the heart, we decided to focus on its expression in lymphocytes, especially of lymphocytes in the gastrointestinal tract. The intestine can be divided in an epithelial layer and the underlying lamina propria. The major cell types present within the epithelium includes the epithelial cells and intraepithelial lymphocytes that consist mainly of CD8⁺ T cells. Therefore, we purified CD8⁺ lymphocytes and epithelial cells and quantified GPR92 mRNA abundance. Highly purified CD8⁺ IEL derived from the small intestinal epithelium expressed significantly higher levels of GPR92 than the Percoll gradient-purified epithelial cells. The fact that GPR92 could not be found in two cell lines derived from the mouse small intestinal epithelium substantiates the idea that GPR92 is predominantly expressed in the lymphocyte compartment.

The CD8⁺ T-cell preparation was further tested for the presence of other LPA receptor subtypes, and it was found that GPR92 was the most abundant, whereas LPA₂ and LPA₁ were found at lower levels. Although this is the first report showing that CD8⁺ T cells express LPA receptors, an exact mapping of the receptor expression in the different cell populations would be facilitated by specific antibodies.

In conclusion, we have identified the former orphan receptor GPR92 as a putative LPA receptor, highly expressed in the heart and gastrointestinal tract. GPR92 is in many respects similar to the previously identified GPR23/LPA₄. It binds LPA with high affinity and activates PI hydrolysis and cAMP production. However, its expression pattern markedly differs from GPR23/LPA₄. GPR92 is highly expressed in lymphocytes, especially in the gastrointestinal tract. It is coexpressed with two classic LPA receptors in intraepithelial CD8⁺ T cell present in the small intestinal mucosa.