LPA4/p2y9/GPR23 Mediates Rho-dependent Morphological Changes in a Rat Neuronal Cell Line*

Lysophosphatidic acid (LPA) is a potent lipid mediator that evokes a variety of biological responses in many cell types via its specific G protein-coupled receptors. In particular, LPA affects cell morphology, cell survival, and cell cycle progression in neuronal cells. Recently, we identified p2y9/GPR23 as a novel fourth LPA receptor, LPA4 (Noguchi, K., Ishii, S., and Shimizu, T. (2003) J. Biol. Chem. 278, 25600-25606). To assess the functions of LPA4 in neuronal cells, we used rat neuroblastoma B103 cells that lack endogenous responses to LPA. In B103 cells stably expressing LPA4, we observed Gq/11-dependent calcium mobilization, but LPA did not affect adenylyl cyclase activity. In LPA4 transfectants, LPA induced dramatic morphological changes, i.e. neurite retraction, cell aggregation, and cadherin-dependent cell adhesion, which involved Rho-mediated signaling pathways. Thus, our results demonstrated that LPA4 as well as LPA1 couple to Gq/11 and G12/13, whereas LPA4 differs from LPA1 in that it does not couple to Gi/o. Through neurite retraction and cell aggregation, LPA4 may play a role in neuronal development such as neurogenesis and neuronal migration.

Lysophosphatidic acid (LPA, 2 1-or 2-acyl-sn-glycero-3phosphate) is a naturally occurring bioactive lipid mediator that controls growth, motility, and differentiation (1). LPA plays important roles in many biological processes, such as brain development, oncogenesis, wound healing, and immune functions (2). The effects of LPA on target cells are mediated by activation of its specific G protein-coupled receptors (GPCRs). The LPA 1 (3), LPA 2 (4), and LPA 3 (5) receptors are the major members of the endothelial differentiation gene (EDG) family that interact with LPA. Pharmacological studies suggest that both LPA 1 and LPA 2 couple to at least three types of G proteins, G i/o , G q , and G 12/13 , whereas LPA 3 couples to G i/o and G q but not G 12/13 (6). Depending on the functional coupling of a given LPA receptor to G proteins, LPA activates diverse signaling cascades involving phosphoinositide 3-kinase, phospholipase C, mitogen-activated protein kinase, Rho family GTPase, and adenylyl cyclase (2,7).
LPA is present in the brain at relatively high levels compared with other organs (8,9). LPA influences the cell morphology of several neuronal cell lines, neural progenitors, and primary neurons (10). It has also been reported that LPA affects electrophysiology, cell survival, and cell cycle progression in neuronal cells (10,11). Targeted deletion of LPA 1 in mice produces olfactory deficits (12) and a behavioral abnormality (13). Furthermore, the use of LPA 1 knockouts revealed that LPA 1 is involved in the initiation of neuropathic pain (14). Exposure of the developing cerebral cortex to LPA produces dramatic changes in the folding of the brain, which do not occur in LPA 1 and LPA 2 double knockouts (15). However, the LPA receptor subtypes responsible for some neuronal effects have not been identified (16 -18).
Recently, we identified p2y 9 /GPR23 as a fourth LPA receptor (LPA 4 ) that is structurally distinct from the three LPA receptors of the EDG family (19). The expressed sequence tag cDNA encoding LPA 4 was originally isolated from human brain (20,21), and LPA 4 expression has been detected in rat embryonic hippocampal neurons (22) and immortalized hippocampal progenitor cells (18). These facts suggest that LPA 4 may have important roles in neurodevelopmental processes such as neurogenesis and neuronal migration. However, only very limited information is available regarding its physiological and biological functions. To assess the functional roles of LPA 4 in neuronal cells, we generated B103 cells stably expressing LPA 4 . This study demonstrates that treatment of the LPA 4 -expressing cells with LPA leads to morphological changes, including cell rounding and cadherin-dependent cell adhesion following cell aggregation, both of which are mediated by the Rho/Rho-associated kinase (ROCK) pathway. The effects of LPA 4 on the morphology of the neuronal cells were clearly distinct from those of LPA 1 , probably because LPA 4 does not couple to G i/o . Tokyo, Japan) in Dulbecco's modified Eagle's medium (DMEM) (Sigma) supplemented with 10% fetal bovine serum (BioWhittaker, Walkersville, MD) and 0.3 mg/ml G418 (Wako, Osaka, Japan). For some experiments, cells were pretreated with 100 ng/ml pertussis toxin (PTX) (List Biological Laboratories, Campbell, CA; from a 400 g/ml stock in 10 mM Tris-HCl (pH 7.4) and 2 M urea stored at 4°C) for 12 h, 5 M YM-254890 (a novel G q/11 inhibitor (23), a kind gift from Dr. J. Takasaki, Astellas Pharma Inc., Tokyo, Japan; from a 10 mM stock in dimethyl sulfoxide (Sigma) stored at Ϫ30°C) for 10 min, or 5 M Y-27632 (Calbiochem; from a 5 mM stock in water stored at Ϫ30°C) for 10 min. Pretreatment with vehicles of PTX and YM-254890 was used as a control.
Stable Expression of LPA 1 and LPA 4 -A DNA fragment containing the entire open reading frame of LPA 1 (NCBI accession number NM_001401) was first amplified from a cDNA prepared from human brain poly(A) ϩ RNA (Clontech) by PCR using Pfu turbo DNA polymerase (Stratagene, La Jolla, CA) and oligonucleotides (sense primer, 5Ј-AAGAAAATTTGTCTCC-CGTAGCTCT-3Ј and antisense primer, 5Ј-CATGAGTTGA-CTTTTCTCCTCTCTC-3Ј). The entire open reading frame of LPA 1 with an additional sequence encoding a hemagglutinin (HA) epitope (YPYDVPDYA) at the 5Ј-end was subsequently amplified from the resultant PCR products using KOD-Plus DNA polymerase (Toyobo, Osaka, Japan) and oligonucleotides (sense primer containing the KpnI and HA tag sequences, 5Ј-GGGGTACCGCCATGTACCCCTACGACG-TGCCCGACTACGCCGCTGCCATCTCTACTTCC-3Ј and antisense primer containing the SpeI sequence, 5Ј-GGAC-TAGTCTAAACCACAGAGTGGTCATT-3Ј). The resultant DNA fragment was digested with KpnI and SpeI and subsequently cloned into the mammalian expression vector pCXN2.1, a slightly modified version of pCXN2 (24) with multiple cloning sites, between the KpnI and SpeI sites. HA-tagged human LPA 4 cDNA was constructed and cloned into pCXN2.1 as described previously (19). B103 cells were transfected using the Lipofectamine 2000 reagent (Invitrogen). After 48 h, the transient expression of the HA epitope on the cell surface was confirmed by flow cytometric analysis (EPICS XL, Beckman Coulter, Fullerton, CA) with the 3F10 rat monoclonal anti-HA antibody (Roche Applied Science) and phycoerythrin-labeled anti-rat IgG (Beckman Coulter) as the secondary antibody. Stable transfectants were selected with 1 mg/ml G418 for 26 days. After staining the drug-resistant cells as described above, a group of HA-positive cells was sorted by flow cytometry (EPICS ALTRA, Beckman Coulter) and maintained with 0.3 mg/ml G418. Three weeks later, a second round of sorting was performed; the twice-immunopurified cells were used for experiments (termed B103-LPA 1 and B103-LPA 4

cells).
Binding Assay-Binding assay was done essentially as described previously (19), with minor modifications. Cells (4 ϫ 10 6 ) were seeded in collagen-coated 100-mm plates (Iwaki), followed by 24 h of serum starvation. The cells were washed with phosphate-buffered saline (PBS) 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 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 resultant pellet was homogenized in ice-cold binding buffer. Binding assays were performed in 96-well plates in triplicate. 20 g each of the membrane fractions from the twice-immunopurified cells was incubated in binding buffer containing 0.25% bovine serum albumin (BSA) (fatty acid-free, very low endotoxin grade; Serologicals Proteins, Kankakee, IL) with 2-fold serial dilutions (50 -3.125  cAMP Measurement-Cells (3.2 ϫ 10 4 ) were seeded in collagen-coated 96-well plates (Iwaki), followed by 24 h of serum starvation. To determine whether LPA receptors mediate the inhibition of adenylyl cyclase, an AlphaScreen cAMP assay kit (PerkinElmer Life Sciences) was used as recommended in the manufacturer's instructions. The cells were washed twice with buffer A (Hanks' balanced salt solution (HBSS) containing 25 mM HEPES-NaOH (pH 7.4) and 0.1% BSA (Serologicals Proteins)) and incubated in 100 l of buffer A containing 0.5 mM 3-isobutyl-1-methylxanthine (IBMX))(from a 20 mM stock in dimethyl sulfoxide stored at Ϫ30°C) (Sigma) for 15 min at room temperature. The reaction was initiated by adding 50 l of various concentrations of LPA in buffer A with 50 M forskolin (Wako; from a 10 mM stock in dimethyl sulfoxide stored at Ϫ30°C). After 30 min of incubation at room temperature, the reaction was terminated by adding 16.6 l of 10% Tween 20, followed by overnight storage at 4°C. After centrifugation at 800 ϫ g for 5 min, the cAMP concentration in the supernatant was measured in quadruplicate with a fusion system (PerkinElmer Life Sciences). To determine whether LPA receptors mediate the stimulation of adenylyl cyclase, the cAMP Biotrak EIA system (Amersham Biosciences) was used as recommended in the manufacturer's instructions. The cells were washed twice with HEPES-Tyrode's buffer (25 mM HEPES-NaOH (pH 7.4), 140 mM NaCl, 2.7 mM KCl, 1 mM CaCl 2 , 0.49 mM MgCl 2 , 12 mM NaHCO 3 , 0.37 mM NaH 2 PO 4 , and 5.6 mM D-glucose) containing 0.1% BSA (HEPES-Tyrode's BSA buffer) and incubated in 100 l of HEPES-Tyrode's BSA buffer con-taining 0.5 mM IBMX for 15 min at 37°C. The reaction was initiated by adding 100 l of various concentrations of LPA in HEPES-Tyrode's BSA buffer. After 30 min of incubation at 37°C, the reaction was terminated by adding 25 l of lysis buffer. Cell lysates in a volume of 100 l were used to determine the cAMP concentration using an enzyme immunoassay method.
Ca 2ϩ Measurement-Cells serum-starved for 24 h were detached with PBS containing 2 mM EDTA, washed with HEPES-Tyrode's buffer, and then loaded with 3 M Fura-2 AM (Dojindo, Kumamoto, Japan) in HEPES-Tyrode's BSA buffer for 1 h at 37°C. The cells were washed twice and resuspended in HEPES-Tyrode's BSA buffer at a density of 1 ϫ 10 6 cells/ml. The cell suspension (0.5 ml) was applied to a CAF-100 spectrofluorometer (Jasco, Tokyo, Japan), and 5 l of 100 M LPA in HEPES-Tyrode's BSA buffer was added. The intracellular Ca 2ϩ concentration ([Ca 2ϩ ] i ) was measured as the ratio of emission fluorescence at 500 nm in response to excitation at 340 and 380 nm.
Cell Rounding Assay-Cells (1 ϫ 10 4 ) were seeded in poly-Dlysine-coated 12-well plates (BD Biosciences). After 24 h of incubation, the cells were washed three times with DMEM containing 0.1% BSA and serum-starved for 24 h. Three hours after a medium change, the cells were treated with 1 M LPA for 15 min. The cells were examined for a round cell morphology lacking any neurite extensions or filopodia. Extended neurites were defined as having a length greater than the cell body. The number of rounded cells was expressed as a percentage of the observed cells (Ͼ200 cells/well).
Rho Inhibition Study-Cells (5 ϫ 10 5 ) were seeded in poly-Llysine-coated 35-mm dishes (Iwaki) in DMEM supplemented with 10% fetal bovine serum. After 24 h, either the Clostridium botulinum C3 exoenzyme expression vector (pEF-C3) (25) (a kind gift from Dr. S. Narumiya, Kyoto University, Kyoto, Japan) or the corresponding control vector (pEF-BOS) (26) (a kind gift from Dr. S. Nagata, Osaka University, Osaka, Japan) was cotransfected with an enhanced green fluorescent protein (EGFP) expression vector (pEGFP-C1; Clontech) at a 4:1 weight ratio, with 3 g of total DNA, using the Lipofectamine 2000 reagent (Invitrogen). After 24 h, the cells were seeded in poly-D-lysine-coated 12-well plates and cultured for 24 h. The cells were then serum-starved for 12 h and treated with 1 M LPA for 15 min. Following fixation with 1% paraformaldehyde for 15 min at 4°C, EGFP images were obtained using a fluorescence microscope (Diaphoto, Nikon, Tokyo, Japan). EGFP-positive cells were examined for a round morphology without any neurite extensions or filopodia. At least 20 different fields were observed with a minimum of 100 EGFP-positive cells. The number of rounded cells was expressed as a percentage of the EGFP-positive cells.
Quantification of Cell Clustering-The degree of cell clustering was quantified by observing the distribution of the cell nuclei. Cells (1.5 ϫ 10 5 ) were seeded in poly-D-lysine-coated 24-well plates (BD Biosciences). After 24 h of incubation, the cells were washed three times with DMEM containing 0.1% BSA and serum-starved for 24 h. Three hours after a medium change, the cells were treated with 1 M LPA for 3 h, followed by fixation and staining with a Diff-Quik kit (Kokusai Shiyaku, Kobe, Japan). The distribution of the cells was mapped in rectangular areas (1710 ϫ 1290 m) by photographing the cultures (Cool Pix 990, Nikon). Each map was overlaid with grids at equal intervals (30 m) and divided into 2451 unit squares. The randomness in spatial distribution was tested by counting the number of unit squares containing at least one nucleus. The intensity of the cell clustering was expressed as the percentage of the unit squares without any nuclei.
Cell Dissociation Assay-The Ca 2ϩ sensitivity of cell-cell adhesion was estimated using trypsin treatment in the presence of either CaCl 2 (TC treatment) or EDTA (TE treatment) as described (27,28), with minor modifications. Briefly, 5 ϫ 10 5 cells were seeded in poly-D-lysine-coated 35-mm dishes (BD Biosciences) and cultured overnight. After 24 h of serum starvation, the cells were stimulated with 1 M LPA for 2 h and washed with HBSS containing either 2 mM CaCl 2 or 2 mM EDTA. The washed cells were treated with 0.01% trypsin for 30 min at 37°C and then dissociated by pipetting 10 times gently in 1 ml of HBSS with 0.01% trypsin. The number of cell clusters was counted with a particle counter (Beckman Coulter). The degree of cell-cell adhesion was expressed as the ratio of particles in the TC condition to particles in the TE condition (TC/ TE). Negative control experiments without LPA treatment were also performed.
Western Blotting-Cells (4 ϫ 10 6 ) were seeded in poly-L-lysine-coated 100-mm dishes. Following 24 h of serum starvation, the cells were treated with 1 M LPA for 3 h, washed twice with PBS, and harvested in buffer B (25 mM HEPES-NaOH (pH 7.4), 10 mM MgCl 2 , and 0.25 M sucrose). The cells were centrifuged at 800 ϫ g for 10 min at 4°C, suspended in ice-cold buffer B containing 20 M 4-amidinophenylmethylsulfonyl fluoride (Sigma) and a protease inhibitor mixture (Complete, Roche Applied Science), and sonicated three times for 30 s each at 4°C. The cell debris was removed by centrifugation at 800 ϫ g for 10 min at 4°C. The protein concentration of the homogenate was determined with a Bradford assay (Bio-Rad) using BSA as a standard. Five micrograms of protein sample containing 5% 2-mercaptoethanol was analyzed by 7.5% SDS-PAGE followed by transfer to a polyvinylidene difluoride membrane (Millipore Corp., Bedford, MA). The membrane was blocked with 5% skim milk (Difco) and probed with a mouse monoclonal antibody against N-cadherin or E-cadherin (BD Biosciences). The bands were visualized with an ECL chemiluminescence detection system (Amersham Biosciences) using horseradish peroxidaseconjugated anti-mouse IgG (Amersham Biosciences).
Immunofluorescence-Cells (3 ϫ 10 5 ) were seeded into poly-L-lysine-coated glass-bottomed 35-mm dishes (Matsunami, Tokyo, Japan) and serum-starved for 24 h. Following stimulation with 1 M LPA for 3 h at 37°C, the cells were fixed with 4% paraformaldehyde for 20 min at 4°C and rinsed twice with icecold PBS. Subsequently, the cells were incubated with a mouse monoclonal antibody against N-cadherin in PBS containing 1/4ϫ permeabilization reagent (Beckman Coulter) for 1 h at room temperature. The primary antibody staining was visualized with an Alexa 488-conjugated goat anti-mouse IgG (Invitrogen). Images were obtained using an LSM510 laserscanning confocal microscope (Carl Zeiss, Jena, Germany) equipped with an argon laser as the light source.
Statistical Analysis-All values in the figures are expressed as means Ϯ S.E. To determine statistical significance, the values were compared by analysis of variance followed by Tukey-Kramer test using Prism 4 software (GraphPad Software, San Diego, CA). The differences were considered significant if p values were less than 0.05.

Stable Expression of LPA 1 and LPA 4 in B103 Cells Results in Different Morphologies in Serum-containing Medium-To
address the functional roles of LPA 4 in neuronal cells, B103 rat neuroblastoma cells were stably transfected with the expression vector for either LPA 1 or LPA 4 . B103 cells were selected because they lack endogenous responses to LPA (29,30). Consistently, no specific binding was observed in B103-vector cells in the radioligand binding assays (see "Experimental Procedures"). To determine the intrinsic gene expression profiles of LPA receptors in B103 cells, we performed a reverse transcription-PCR analysis of total cellular RNA from the cells. Although LPA 4 mRNA expression was slightly detected, no mRNA expression of the other three receptors, LPA 1 , LPA 2 , and LPA 3 , was observed (data not shown). This finding is consistent with a recent report by Tsukahara et al. (31). The apparent discrepancy between the expression of LPA 4 and the lack of response to LPA might occur because the expression of LPA 4 is too low to respond to LPA. Alternatively, post-transcriptional/translational modifications (32) may produce discordance between mRNA and protein expression. Thus, we discounted the low expression of LPA 4 in B103 cells and took advantage of their unresponsiveness to LPA and their neuronal nature for the purpose of examining the functional roles of LPA 4 in neuronal cells.
For the construction of stably transfected cell lines, LPA 1 and LPA 4 were tagged with an HA epitope at the N terminus to enable us to determine the levels of expression on the cell surfaces. Fluorescence-activated cell sorting enriched a polyclonal population of the drug-resistant cells that expressed each LPA receptor. These populations of stable clones are free of any clonal deviation that could cause functional variations. Following two rounds of cell sorting, we observed that the fluorescence intensity of B103-LPA 1 cells was higher than that of B103-LPA 4 cells (Fig. 1A), although the B max value for B103-LPA 1 cells (0.8 pmol/mg of protein) was lower than that for B103-LPA 4 cells (6.0 pmol/mg of protein). The apparent discrepancy might be because of two possibilities as follows: the usage of organellar membranerich microsome fractions and the difference in HA antibody immunoreactivity to the HA epitope tagged to two receptors. To confirm that no expression of the other subtypes of LPA receptors was enhanced secondary to the transfection, reverse transcription-PCR was performed with specific primers for LPA 1 , LPA 2 , LPA 3 , and LPA 4 in B103-vector, B103-LPA 1 , and B103-LPA 4 cells. As in the parental B103 cells, we observed only a low expression of LPA 4 and virtually no expression of the other LPA receptors in all of the transfected cell lines (data not shown).
Although these stably transfected cell lines showed similar growth rates (data not shown), they showed distinctly different morphologies in serum-containing medium (Fig. 1B). As reported previously (33), B103-LPA 1 cells displayed a flattened and more migratory morphology compared with B103-vector cells. Interestingly, B103-LPA 4 cells had an epithelial like morphology and appeared to adhere more tightly to each other than B103-vector cells. These observations suggest that LPA 1 and LPA 4 have distinct signaling pathways that produce different cell morphologies.
LPA 4 Does Not Affect Adenylyl Cyclase Activity in B103 Cells-We examined whether LPA 4 mediates the inhibition of adenylyl cyclase activity in B103 cells, as the other three LPA receptors do (30) (Fig. 2A). In B103-LPA 1 cells, LPA caused a dose-dependent inhibition of adenylyl cyclase activity with IC 50 values below 10 nM (Fig. 2A). This inhibition was completely blocked by PTX treatment, indicating the primary role of G i/o proteins. However, LPA at concentrations up to 10 M did not blunt the forskolin-driven rises in cAMP accumulation in either B103-vector or B103-LPA 4 cells, suggesting that LPA 4 does not couple to G i/o proteins.
Previously, we reported that LPA induces cAMP accumulation in LPA 4 -expressing Chinese hamster ovary cells (19). However, LPA did not elevate basal cAMP levels in either B103-LPA 4 cells or B103-LPA 1 cells (Fig. 2B), suggesting that neither LPA 1 nor LPA 4 couples to G s in B103 cells.
LPA 1 and LPA 4 Mediate Ca 2ϩ Mobilization via Distinct Signaling Pathways-LPA has been shown to induce intracellular Ca 2ϩ mobilization in many cell types (34), and all EDG family LPA receptor subtypes mediate Ca 2ϩ mobilization when expressed in B103 cells (30). We therefore examined whether LPA 4 mediates Ca 2ϩ mobilization in B103 cells. Although B103-vector cells displayed no response to 1 M LPA, increases in [Ca 2ϩ ] i were observed both in B103-LPA 1 and B103-LPA 4 cells (Fig. 3). LPA induces phospholipase C-mediated Ca 2ϩ mobilization via the PTX-sensitive G i/o -and/or PTX-insensitive G q/11 -mediated pathways (34). To examine the signaling pathways leading to Ca 2ϩ mobilization in B103-LPA 1 and B103-LPA 4 cells, we treated the cells with a G q/11 -selective inhibitor, YM-254890 (23) (Fig. 3). ATP was used as a positive control, because ATP evokes Ca 2ϩ mobilization via P2Y receptors predominantly through G q/11 (35). The LPA-induced Ca 2ϩ response in B103-LPA 4 cells and the ATP-induced Ca 2ϩ response in both transfected cell lines were completely abolished by pretreatment with 5 M YM-254890 (Fig. 3). In B103-LPA 1 cells, YM-254890 only partially inhibited the LPA-induced response ( Fig. 3), but the combination of PTX and YM-254890 produced complete inhibition (data not shown). The degree of inhibition with YM-254890 in B103-LPA 1 cells was not altered at higher concentrations (up to 20 M; data not shown), indicating that 5 M YM-254890 was sufficient to inhibit the activation of G q/11 proteins. These results suggest that both G i/o and G q/11 proteins mediate Ca 2ϩ mobilization in B103-LPA 1 cells, whereas G q/11 is the dominant mediator of the response in B103-LPA 4 cells.

Both LPA 1 and LPA 4 Mediate Cell Rounding via Rho-dependent and G i/o -and G q/11 -independent
Pathways-LPA induces rapid growth cone collapse, neurite retraction, and neuronal cell rounding in several neuronal cell types  (10). Mouse LPA 1 and LPA 2 and human LPA 1 have been reported to mediate LPA-induced cell rounding in B103 cells (29,30,33); we examined whether human LPA 4 also mediates cell rounding in B103 cells by seeding cells at a low cell density (Fig. 4, A and B). Overexpression of LPA 1 and LPA 4 slightly increased the percentages of rounded cells even before LPA application. Within 15 min of LPA stimulation, about 80% of B103-LPA 4 cells became rounded and underwent neurite retraction (Fig. 4A). Cell rounding was observed in B103-LPA 1 cells as reported previously (33), but to a lesser degree than in B103-LPA 4 cells. LPA-induced cell rounding was not observed in B103-vector cells.
The role of Rho in LPA-induced cell rounding is now well established (36), and the G 12/13 types of heterotrimeric G proteins are known to be upstream activators of Rho proteins (1, 2, 7). On the other hand, there are reports that G q/11 activation induces cell rounding through Rho-dependent (37) and -independent (38) pathways. To determine which G proteins and signaling molecules are involved in LPA-induced cell rounding, we pretreated the cells with PTX, YM-254890, and a ROCK inhibitor, Y-27632 (Fig. 4B). In B103-LPA 4 cells, neither PTX nor YM-254890 inhibited LPA-induced cell rounding; in contrast, Y-27632 completely inhibited this morphological change. Y-27632 also hampered LPA-induced cell rounding in B103-LPA 1 cells, whereas YM-254890 did not affect the number of rounded cells. Interestingly, pretreatment with PTX increased the degree of LPA-induced cell rounding in B103-LPA 1 cells. To confirm the involvement of Rho, B103-LPA 4 cells were transfected with C3 exoenzyme, which inactivates Rho by ADP-ribosylation. The transfected cells were identified by cotransfection of an EGFP expression construct. C3 exoenzyme transfection blunted LPA-induced cell rounding in B103-LPA 4 cells, again indicating the involvement of Rho (Fig. 4C).
LPA 4 Mediates ROCK-dependent Cell Aggregation-As described earlier, B103-LPA 4 cells appeared to form aggregates in serum-containing medium to a greater extent than B103vector cells (Fig. 1B). To determine whether the binding of LPA to LPA 4 mediates the induction of cell-cell adhesion, B103-LPA 4 cells at a medium cell density were stimulated with 1 M LPA after 24 h of serum starvation. Although the rapid cell rounding after LPA application was difficult to evaluate at this cell density because of the formation of cell-cell contacts, LPA caused a slow but dramatic aggregation in B103-LPA 4 cells (Fig.  5A, panel f ). The morphological change observed in B103-LPA 4 cells was transient, reaching a maximum 2-3 h after the treatment and then returning to the base line 24 h after the treatment (data not shown).
To investigate the signaling pathways downstream of LPA 4 that are involved in the cell aggregation, we treated B103-LPA 4 cells with several inhibitors. The LPA-induced morphological changes in B103-LPA 4 cells were completely prevented by Y-27632 (Fig. 5A, panel o). In contrast, neither PTX nor YM-254890 inhibited the cell aggregation (Fig. 5A, panels i and  l). We quantified the degree of cell aggregation by examining the randomness in the spatial distribution of the cells (see the "Experimental Procedures" ; Fig. 5B). These results suggest that Rho mediates LPA-induced cell aggregation in B103-LPA 4 cells in a G i/o -and G q/11 -independent manner. Rho regulates the reorganization of the actin cytoskeleton, which can modify the intensity of adhesion (28,39). To examine whether the reorganization of the actin cytoskeleton was involved in this effect, B103-LPA 4 cells were pretreated with cytochalasin D (an inhibitor of actin polymerization). In these cells, morphological changes were not observed after LPA stimulation, indicating that actin reorganization is involved in the LPA-induced cell aggregation (data not shown). Like B103-LPA 4 cells, PTX- treated B103-LPA 1 cells became aggregated after LPA stimulation (Fig. 5A, panel h).
LPA 4 Mediates N-cadherin-dependent Cell-Cell Adherence-Through LPA-induced cell aggregation, B103-LPA 4 cells formed tightly compact aggregates (Fig. 5A, panel f), which dissociated very little after pipetting (data not shown). Cell-cell adhesion mechanisms can be Ca 2ϩ -dependent and Ca 2ϩ -independent, and cadherins are the major components of the Ca 2ϩdependent system. Cadherin-dependent adhesion was originally defined as being trypsin-resistant in the presence of Ca 2ϩ and trypsin-sensitive in the absence of Ca 2ϩ (40). We examined whether LPA-induced cell adhesion was mediated by cadherins using a cell dissociation assay, one of the adhesion assays for the evaluation of the cadherin activity (27,28). We defined the TC/TE index as a ratio of the cell particle number after trypsin treatment in the presence of Ca 2ϩ (TC) to the number after trypsin treatment with EDTA (TE). Cadherin-dependent adhesion remains after trypsin-Ca 2ϩ treatment, whereas trypsin-EDTA treatment disrupts cell adhesion nearly completely. In either treatment, an increase in particles would occur when a large aggregate breaks into small particles by pipetting; the higher the number of particles, the lower the aggregation (adhesion). Thus, the TC/TE index is negatively correlated with cadherin-mediated adhesion. The aggregation level of B103-LPA 4 cells increased after LPA stimulation ( Fig. 6A; note that the index inversely reflects cadherin activity). LPA treatment did not significantly affect the TC/TE index in either B103vector cells or B103-LPA 1 cells. These results suggest that LPA increased the cadherin-mediated adhesive activity in B103-LPA 4 cells. Even LPA-untreated B103-LPA 4 cells had significantly more cadherin-dependent adhesion activity, i.e. a lower TC/TE index, than LPA-untreated B103-vector and B103-LPA 1 cells (Fig. 6A).
The cadherins constitute a large superfamily of molecules that includes the classic cadherins, the desmosomal cadherins, the protocadherins, and the cadherin-like signaling receptors (41). The levels of the two classic cadherins most commonly expressed in the nervous system, N-and E-cadherin, were determined by Western blotting of LPA-treated or -untreated B103 cells. Consistent with a previous report (42), these cells abundantly expressed N-cadherin (Fig. 6B), whereas E-cadherin was undetectable (data not shown). The expression level of N-cadherin was not up-regulated by LPA treatment in any of the transfected cell lines (Fig. 6B), and N-cadherin was intact in the cells undergoing TC treatment. In contrast, TE treatment resulted in complete digestion of N-cadherin (data not shown), as reported previously (43). We next examined whether LPA increases N-cadherin-mediated cell-cell contacts. LPA promoted the assembly of N-cadherin in the form of a thick, bright band at the cell-cell contact area in B103-LPA 4 cells but not in B103-vector cells (Fig. 6C).

DISCUSSION
A number of studies have shown that LPA mediates morphological changes in neuronal cells through the Rho-ROCK pathway (10,11,44). It has been proposed that these effects are mediated by LPA 1 and/or LPA 2 (10,11,44). Recently, we identified p2y 9 /GPR23 as a fourth LPA receptor (LPA 4 ) that is structurally distinct from the EDG family of LPA receptors (19). The expression of LPA 4 in neuronal cells implies a significant role for this receptor in the nervous system (18). The results in this study demonstrate that LPA 4 caused morphological changes in B103 neuronal cells, including cell rounding and N-cadherin-associated cell aggregation, both of which were mediated by the Rho-ROCK pathway.
Ca 2ϩ mobilization and adenylyl cyclase inhibition are the major cellular responses to LPA (45). When expressed in B103  FEBRUARY 23, 2007 • VOLUME 282 • NUMBER 8 JOURNAL OF BIOLOGICAL CHEMISTRY 5821 neuronal cells, each of the EDG family LPA receptors, LPA 1 , LPA 2 , and LPA 3 , mediates both of these reactions (30). LPA 1 is likely to mediate the Ca 2ϩ response through G q/11 proteins (Fig.  3). The PTX-sensitive inhibition of adenylyl cyclase ( Fig. 2A) suggests that LPA 1 also couples to G i/o . We showed that LPA 4 mediates the Ca 2ϩ response in B103 cells (Fig. 3). This is consistent with our previous report (19) that the stable expression of LPA 4 in Chinese hamster ovary cells significantly enhanced the LPA-induced Ca 2ϩ mobilization. From our results using YM-254890, LPA 4 probably mediates Ca 2ϩ mobilization through G q/11 proteins (Fig. 3). In contrast, LPA did not inhibit adenylyl cyclase in B103-LPA 4 cells (Fig. 2A). These results indicate that unlike the other LPA receptor subtypes, LPA 4 does not couple to G i/o proteins.

LPA 4 Changes the Morphology of Neuronal Cells
Neurite retraction and neurite formation play a role in the remodeling of neurons for guidance and synaptic plasticity (46). Neurite retraction in neuronal cells is induced by lysophospholipids, including LPA and sphingosine 1-phosphate (S1P), in addition to semaphorins, netrins, and ephrins (10,11,47). LPA induces neurite retraction through LPA 1 or LPA 2 when expressed in B103 cells (30). In this study, we showed that 1 M LPA induced cell rounding in B103-LPA 4 cells (Fig. 4A). Sugiura et al. (48) reported that rat brain contains 3.73 nmol of LPA/g of tissue. These results suggest a role for LPA 4 in LPAinduced neurite retraction. Neurite initiation and formation involve actin cytoskeletal changes, and as a regulator of actin reorganization, Rho GTPase has a profound effect on neuritogenesis (47). For example, S1P induces Rho-dependent neurite retraction through the S1P 2 (49,50), S1P 3 (49), and S1P 5 receptors (50,51). Several studies have also revealed a critical role for Rho and ROCK in LPA-induced neurite retraction (36,52,53), although some studies have reported Rhoindependent neurite retraction (38,54). Judging from its complete inhibition by C3 exoenzyme and Y-27632, the cell rounding induced by LPA 4 depended on Rho and ROCK in B103-LPA 4 cells (Fig. 4, B and C).
In general, activation of G 12/13 proteins leads to an increase in RhoA guanine nucleotide exchange, activation of ROCK, and actin polymerization (55,56), although some studies have implied that G q/11 proteins can also activate Rho (57,58). Based on our results, it is conceivable that the G 12/13 proteins are upstream regulators of Rho in B103-LPA 1 cells and B103-LPA 4 cells, because YM-254890 inhibited cell rounding in both cell types (Fig. 4B). In many studies reporting the induction of neurite retraction by LPA (10), the LPA receptor subtypes responsible have not been identified, or LPA 1 and LPA 2 were suggested as candidate subtypes based on mRNA expression in the neuronal cells examined (59 -62). This study suggests the possibility that LPA 4 was involved in the neurite retraction in some of these studies.
LPA has been shown to stimulate cell motility and to modulate tumor cell invasion, both of which are mediated mainly by LPA 1 and G i/o proteins (33,63,64). In the presence of serum, B103-LPA 1 cells exhibited a flattened morphology and were widely dispersed throughout the dish (Fig. 1B) (33). This morphological phenotype was probably evoked by LPA through the LPA 1 -G i/o -Rac signaling axis (33). In sharp contrast to B103-LPA 1 cells, B103-LPA 4 cells formed cell aggregates in serum-containing medium (Fig. 1B), apparently through activation of the G 12/13 -Rho-ROCK signaling axis (Fig. 5). This cell-cell adhesion involved N-cadherin without de novo synthesis (Fig.  6). Because Rho affects cadherin-dependent adhesion through actin cytoskeleton reorganization (28), we presume that LPAinduced cytoskeletal changes affect the subcellular distribution of N-cadherin, as shown in Fig. 6C, leading to strong cell-cell adhesion in B103-LPA 4 cells. This is supported by the current results that treatment with Y-27632 and cytochalasin D abolished the LPA-induced cell aggregation (Fig. 5 and data not shown). N-cadherin is widely expressed in the nervous system and has critical roles in neural development and functions, including synapse formation and myelination. Weiner et al. (65) previously reported that LPA induced cell-cell junctions containing N-cadherin in rat Schwann cells. Furthermore, LPA was reported to induce cell clustering in neural progenitor cells prepared from embryonic rat hippocampus (66) and in mouse postmitotic cortical neurons (16). Our results raise the possibility that in addition to LPA 1 and LPA 2 , LPA 4 might also be involved in these effects in neural cells and have critical roles in the development and function of the nervous system. In contrast to LPA 1 and LPA 2 , which activate Rac through G i/o (17,64), LPA 4 is unlikely to activate the G i/o -Rac pathway because LPA 4 did not inhibit adenylyl cyclase activity (Fig. 3A). Therefore, LPA 4 might have a unique role in keeping a proper balance between Rho and Rac activation, which is important for neuronal development and function (47).
We observed that PTX significantly enhanced the intensity of LPA 1 -mediated cell rounding (Fig. 4B). This "permissive effect" is consistent with a previous report that PTX enables LPA-induced cell rounding in 1321N1 astrocytoma cells (61). It is known that Rho activity is inhibited by Rac activation through G i/o proteins (67). Indeed, Rac activation functionally antagonizes Rho-mediated neurite retraction in 1321N1 astrocytoma cells (61). LPA 1 was shown to couple to G i/o and activate Rac strongly in B103 cells (33) and other cells, including mouse embryonic meningeal fibroblast (MEMF) and mouse skin fibroblast (MSF) cells (17,64). Taken together, we suggest that PTX treatment of B103-LPA 1 cells suppresses G i/o proteins and subsequently suppresses Rac activation by LPA, which in turn permits Rho-mediated cell rounding. This mechanism probably also accounts for the LPA-induced aggregation of PTX-treated B103-LPA 1 cells (Fig. 5, A, panel h, and B).
We showed here that LPA 4 has Rho-dependent morphological effects. It has been reported that LPA-induced Rho activation is mediated by LPA 1 and/or LPA 2 . However, pathways independent of LPA 1 and LPA 2 have also been proposed. Contos et al. (17) showed that MEMF cells from LPA 1 and LPA 2 double knockouts remained capable of forming stress fibers in response to LPA. This study proposed the presence of unknown LPA receptors in MEMF cells because of the absence of LPA 3 mRNA. Consistent with this, Hama et al. (64) reported that LPA activates Rho in MSF cells from LPA 1 and LPA 2 double knockouts. Our results, together with the abundant expression of LPA 4 in MSF cells (64), suggest that LPA 4 may also be involved in LPA-induced Rho activation in these cells. Furthermore, Hama et al. (64) observed that Rac activation was totally dependent on LPA 1 and LPA 2 , supporting our hypothesis that LPA 4 does not activate the G i/o -Rac signaling axis.
We also observed that the stable expression of LPA 1 or LPA 4 slightly increased the population of rounded cells even before LPA application (Fig. 4B). Furthermore, serum-starved B103-LPA 4 cells adhered to one another more strongly than B103vector or B103-LPA 1 cells did (Fig. 6A). These morphological effects might be because of the constitutive activation of these LPA receptors. Indeed, there are many reports showing that the constitutive expression of GPCR for lipid mediators, including lysophospholipids and prostanoids, has morphological effects (30,49,68,69). For example, cell rounding and cadherin-dependent adhesion in the absence of ligand occur in human embryonic kidney 293 cells transfected with FP B receptor, an isoform of prostanoid GPCR (69,70). The involvement of phosphatidylinositol 3-kinase and ␤-catenin was proposed for these constitutive activities (70). Another conceivable explanation for the morphological effects observed in serum-starved B103-LPA 1 and B103-LPA 4 cells involves autocrine ligand secretion and subsequent receptor activation (33). Both hypotheses are consistent with our results that the treatment of these cells with Y-27632 decreased the percentage of rounded cells (Fig. 4B).
In summary, as shown in Fig. 7, we demonstrate for the first time that the novel LPA receptor subtype LPA 4 is coupled to the activation of Rho in a rat neuronal cell line. The activation of Rho through LPA 4 leads to morphological changes, including cell rounding and cell aggregation. LPA is well known to induce neurite retraction and cell clustering in neural cells. The identification of Rho as an effector of LPA 4 will give insight into some of the physiological and morphological effects of LPA that could not be explained by the EDG family LPA receptors. A full understanding of the potential roles of the endogenous LPA 4 receptor in the development and function of the nervous system awaits future studies.