Endothelial Differentiation Gene-2 Receptor Is Involved in Lysophosphatidic Acid-dependent Control of 3T3F442A Preadipocyte Proliferation and Spreading*

EDG-2, EDG-4,EDG-7, and PSP24 genes encode distinct lysophosphatidic acid (LPA) receptors. The aim of the present study was to determine which receptor subtype is involved in the biological responses generated by LPA in preadipocytes. Growing 3T3F442A preadipocytes express EDG-2 and EDG-4mRNAs, with no expression of EDG-7 or PSP24mRNAs. Quantitative reverse transcriptase-polymerase chain reaction revealed that EDG-2 transcripts were 10-fold more abundant than that of EDG-4. To determine the involvement of the EDG-2 receptor in the responses of growing preadipocytes to LPA, stable transfection of antisenseEDG-2 cDNA was performed in growing 3T3F442A preadipocytes. This procedure, led to a significant and specific reduction in EDG-2 mRNA and protein. This was associated with a significant alteration in the effect of LPA on both cell proliferation and cell spreading. Finally, the differentiation of growing preadipocytes into quiescent adipocytes led to a strong reduction in the level of EDG-2 transcripts. Results demonstrate the significant contribution of the EDG-2 receptor in the biological responses generated by LPA in 3T3F442A preadipocytes.

Lysophosphatidic acid (LPA 1 : 1-acyl-2-hydroxy-sn-glycero-3phosphate) is a bioactive phospholipid present in serum and other biological fluids (1,2). LPA controls a wide variety of cellular responses (mitogenesis, cytoskeletal rearrangements, cell adhesion, ion transport, apoptosis) through the activation of specific G-protein-coupled receptors (3,4). A first potential LPA gene receptor, called vzg-1 (5), was cloned in mouse and found homologous to the endothelial differentiation gene-1 (EDG-1) (6), a high affinity receptor for another bioactive phospholipid: sphingosine 1-phosphate (7). A human gene exhibiting 97% homology with vzg-1 was then identified and called EDG-2 (8). Two other human genes, called EDG-4 and EDG- 7, have also been cloned and proposed to be LPA receptors (9,10). At the beginning of the present work the cDNA sequences of EDG-4 and EDG-7 mouse orthologues were not yet available. Since then, the EDG-4 mouse orthologue was cloned and sequenced (11). A fourth gene receptor, called PSP24, was cloned in Xenopus (12) and mouse (13) and initially proposed to be a LPA-responsive receptor. However, the PSP24 receptor gene exhibits poor homology with the EDG receptor gene family and is actually related to the platelet-activating factor receptor gene.
Most evidence that EDG-2, EDG-4, EDG-7, and PSP24 are LPA receptors is deduced from their ability to increase or restore LPA activity following their overexpression in cells (5,9,10,12). However, the relative contribution of endogenously expressed LPA receptors in the biological activities of LPA remains poorly defined. Because pharmacological tools to study LPA receptors are very limited (no available antagonists, difficulties in performing receptor binding studies), one way to address the question is to alter LPA receptor expression by using antisense strategies or gene invalidation methods.
We recently observed that conditioned medium prepared from adipocytes exposed to an ␣ 2 -adrenergic stimulation increased proliferation and spreading (reflecting a reorganization of actin cytoskeleton) on an ␣ 2 -adrenergic-insensitive murine preadipose cell line: 3T3F442A. Analysis based on the use of a lysophospholipid-specific phospholipase (phospholipase B) and 32 P-phospholipid labeling, revealed the involvement of LPA in the trophic activities of adipocyte conditioned medium (14). Because of the intimate coexistence of adipocytes and preadipocytes within adipose tissue, LPA released by adipocytes could play an important role in paracrine/autocrine control of preadipocyte growth, a key event involved in adipose tissue development. Therefore, a better understanding of the cellular mechanisms of LPA action in preadipocytes could help to develop pharmacological and/or genetic strategies to control adipogenesis.
The trophic action of LPA in growing 3T3F442A preadipocytes can be specifically desensitized by chronic exposure to a high concentration of LPA (14). Exposure of growing 3T3F442A preadipocytes to LPA leads to rapid and pertussis toxin-sensitive activation of the mitogen-activated kinases ERK1 and ERK2 (15). Whereas those observations suggest the involvement of a G-protein-coupled receptor(s) in the action of LPA in 3T3F442A preadipocytes, the identity of the discrete LPA receptor(s) responsible for these biological effects remains to be determined.
In the present work, we studied the expression of EDG-2, EDG-4, EDG-7, and PSP24 receptor genes and attempted to determine their relative contribution in the biological responses of 3T3F442A preadipocytes to LPA (proliferation and spreading). Based upon quantification of gene expression, and antisense cDNA transfection, the EDG-2 receptor was found to be predominantly involved in LPA-dependent control of 3T3F442A preadipocyte proliferation and spreading.

Cells
3T3F442A preadipocytes were grown in 10% donor calf serum-supplemented DMEM as reported previously (16). The medium was changed every 2 days. Conversion of 3T3F442A preadipocytes into adipocytes was obtained by cultivating confluent cells in DMEM supplemented with 10% fetal calf serum plus 50 nM insulin as described previously (14).

Nonquantitative RT-PCR Analysis
Total RNAs were extracted using RNeasy mini kit (Qiagen). One microgram of total RNA was treated with 1 unit of RNase-free DNase I (Life Technologies, Inc.) for 15 min at room temperature followed by further inactivation with 1 l of EDTA (25 mM) for 10 min at 65°C. Then, RNA was reverse-transcribed for 60 min at 37°C using Super-Script II (Life Technologies, Inc.) RNase H Ϫ RT and subjected to amplification. A minus RT reaction was performed in parallel to ensure the absence of genomic DNA contamination. PCR was carried out in a final volume of 50 l containing 3 l of cDNA, 1 l of dNTP (10 mM), 5 l of 10ϫ PCR buffer (10 mM Tris-HCl, pH 9, 50 mM KCl and 0.1% Triton X-100), 3 l of MgCl 2 (25 mM), 1.5 l of sense-and antisense-specific oligonucleotide primers (10 M), and 1.25 units of Taq DNA polymerase (Promega). Conditions for the PCR reaction were: initial denaturation step at 94°C for 2 min, followed by 35 cycles consisting in 1 min at 94°C, 1 min at 54°C (EDG-2), 57°C (EDG-4), 49°C (EDG-7), 57°C (PSP24), or 58°C (EDG-1), 72°C for 90 s. After a final extension at 72°C for 6 min, PCR products were separated on 1.5% agarose gel, and amplification products were visualized with ethidium bromide. In some experiments we analyzed the influence of the number of PCR cycles on the intensity of the amplification products.
EDG-4 primers were designed from the 570-bp fragment of mouse EDG-4 cDNA cloned in the present study (see below): sense, 5Ј-TGGC-CTACCTCTTCCTCATGTTCCA-3Ј and antisense, 5Ј-GGGTCCAGCA-CACCACAAATGCC-3Ј. Those primers are specific to mouse. EDG-7 primers were designed from a GenBank TM mouse expressed sequence tag (Clone ID 2192692, GenBank TM accession number AW107032), which exhibits 82% identity on nucleotide level and that we assumed to correspond to mouse EDG-7. This was confirmed later on after the identification of the mouse EDG-7 (GenBank TM accession number NM_012152): sense, 5Ј-AGTGTCACTATGACAAGC-3Ј and antisense, 5Ј-GAGATGTTGCAGAGGC-3Ј. These primers are 100% identical between mouse and human.
Quantitative RT-PCR Analysis cDNA was synthesized from 2 g of total RNA in 20 l using random hexamers and murine Moloney leukemia virus reverse transcriptase (Life Technologies, Inc.). A minus RT reaction was performed in parallel to ensure the absence of genomic DNA contamination. Design of primers was done using the Primer Express software (Applied Biosystems). Real-time quantitative RT-PCR analyses were performed starting with 50 ng of reverse-transcribed total RNA with 200 nM concentration of both sense and antisense primers in a final volume of 25 l using the sybr green PCR core reagents in a ABI PRISM 7700 Sequence Detection System instrument (Applied Biosystems). Standard curves were determined after amplification of 5 ϫ 10 2 to 5 ϫ 10 6 copies of purified amplicons generated from 3T3-F442A cDNA by non quantitative RT-PCR. Quantification of EDG-2 and EDG-4 mRNA steady state copy numbers were performed using internal primers located within the amplicons' sequence. Internal sense and antisense primers and size of products, respectively, were for EDG-2 and EDG-4: 5Ј-CTGTGGTCAT-TGTGCTTGGTG-3Ј, 5Ј-CATTAGGGTTCTCGTTGCGC-3Ј, and 231 bp and 5Ј-GGCTGCACTGGGTCTGGG-3Ј, 5Ј-GCTGACGTGCTCCGCCAT-3Ј, and 214 bp.

Cloning of a Mouse EDG-4 cDNA Fragment
Mouse cDNAs were synthesized from total RNAs isolated from NIH3T3 cells with random primers and reverse transcriptase. PCR was done with degenerate primers 5Ј-CTiGCCiATCGCCGTiGAGCGiCA-3Ј and 5Ј-ACiACCTGiCCiGGiGTCCAGCA-3Ј corresponding to the third and the sixth transmembrane domains of the human EDG-4 receptor, respectively (9). The PCR conditions were 35 cycles of 94°C for 1 min, 60°C for 1 min, and 72°C for 2 min. A product of ϳ570 bp was obtained and cloned into pCR2.1-TOPO vector (Invitrogen). The sequence of this mouse cDNA fragment was 85 and 93% identical to human EDG-4 at the nucleotide and amino acid levels, respectively. Therefore, this cDNA fragment was assumed to correspond to the mouse EDG-4 gene. During the time of the present study, a full-length cDNA encoding mouse EDG-4 cDNA was cloned (11) (GenBank TM accession number AF218844). Our sequence was 100% homologous with this cDNA.

Stable Transfection with Antisense Vector
The coding region of human EDG-2 cDNAs (8) was subcloned in the antisense direction in pcDNA3.1 vector (Invitrogen). Construct was verified by restriction mapping and sequencing. Antisense EDG-2 cDNA vector or empty pcDNA3.1 vectors were transfected in exponentially growing 3T3F442A preadipocytes by calcium phosphate precipitation followed by G418 (neomycin) selection as described previously (16). Gene expression and functional analysis were performed on individual G418-resistant cell clones.

Western Blot Analysis
Preadipocyte proteins were solubilized in radioimmune precipitation buffer, and 50 g of protein were separated on 11% SDS-polyacrylamide gel electrophoresis and transferred on nitrocellulose as described previously (18). The blot was preincubated for 2 h at room temperature in TBST buffer (10 mM Tris-HCl, pH 8, 150 mM NaCl, 0.2% Tween 20) containing 5% dry milk (TBST-DM) and then overnight at 4°C in TBST-DM containing 1/5000 anti-Vzg-1 receptor antibody (5). After extensive washing with TBST-DM, the blot was incubated with peroxidase conjugate secondary anti-rabbit antibody 1/5000 e (Sigma) for 1 h and washed again. Immunostained proteins were visualized using the enhanced chemiluminescence detection system (ECL, Amersham Pharmacia Biotech).

Cell Proliferation and Spreading
Cell proliferation was determined as described previously (14). After 48-h culture in 10% donor calf serum-supplemented DMEM, cells were serum-deprived and exposed for an additional 48 h to various growth factors such as fetal calf serum, 1-oleoyl-LPA. Cell number was determined using Coulter counter. In some experiments LPA present in fetal calf serum was suppressed by overnight treatment at 37°C with 0.1 unit/ml phospholipase B (EC 3.1.1.5; Sigma).
Cell spreading was used as an index of actin cytoskeleton reorganization and was quantified as described previously (19). Briefly, precon-fluent cells were washed with phosphate-buffered saline and placed in serum-free DMEM for 30 -60 min to induce cell retraction characterized by a reduced cell area. Cell spreading was measured by the increase in the cell area generated after 20-min exposure to 1-oleoyl-LPA or sphingosine 1-phosphate. Cell area was measured under a microscope connected to a video camera and image analysis program (Visiolab).
The relative proportion of EDG-2 versus EDG-4 mRNAs was quantified using real time RT-PCR. As shown in Table I, EDG-2 mRNAs were found to be 10-fold more abundant than EDG-4 mRNAs (Table I). By using Northern blot analysis on total RNA, EDG-2 (as well as EDG-1) mRNAs were easily detectable after 24-h autoradiography (Fig. 2), whereas EDG-4 mRNAs remained undetectable after 7 days exposure (not shown). Results showed that although EDG-2 and EDG-4 transcripts could be detected in growing 3T3F442A preadipocytes, EDG-2 transcripts were predominantly expressed.
Stable Transfection of Antisense EDG-2 cDNA in Growing 3T3F442A Preadipocytes-To determine the contribution of EDG-2 receptor in the bioactivity of LPA, an EDG-2 antisense cDNA was stably transfected into 3T3F442A preadipocytes. Thirteen G418-resistant cell clones were isolated, and the presence of antisense mRNAs was determined by RT-PCR. For that, specific primers designed for antisense EDG-2 mRNA detection (see "Material and Methods") were used. Six cell clones were found to express antisense EDG-2 mRNA (Fig. 3A). Based upon Northern blot analysis, the six clones exhibited a lower expression of endogenous EDG-2 mRNAs as compared with 3T3F442A preadipocytes transfected with the empty vector (Fig. 3B). Clone 7 and clone 24 were those expressing the lowest amount of endogenous EDG-2 mRNA (Fig. 3B). It was then tested whether down-regulation of endogenous EDG-2 mRNAs observed in clone 7 and clone 24 was accompanied by a reduction in EDG-2 protein level. Thus, Western blot analysis was performed with a polyclonal antibody raised against mouse-EDG-2 receptor (Vzg-1) (5). As described previously (5), the receptor was detected as a protein with a molecular mass between 31 and 45 kDa (Fig. 3C). In clone 7 and clone 24, the amount of this protein was lower when compared with 3T3F442A preadipocytes transfected with the empty vector (Fig. 3C). No detectable reduction in the amount of EDG-2 receptor protein was observed in clones 9, 12, and 22 (not shown). Results showed that stable transfection with EDG-2 antisense cDNA allowed us to isolate 3T3F442A preadipocyte clones with decreased expression of endogenous EDG-2 receptor.
In parallel, we tested whether stable expression of EDG-2 antisense cDNA would affect EDG-4 receptor expression. EDG-4 transcripts were thus quantified in the six clones depicted in Fig. 3B. Because EDG-4 mRNAs could not be detected by Northern blot, real time RT-PCR was used for their quantification. When compared with 3T3F442A preadipocytes transfected with the empty vector, the EDG-4 mRNA level was found to be increased in clone 7, clone 19, and clone 22 and significantly decreased in clone 12 (Table II). Therefore, downregulation of EDG-2 transcripts was not systematically accompanied by a compensatory up-regulation of EDG-4 transcripts.

TABLE I Expression of EDG-2 and EDG-4 transcripts in growing 3T3F442A
preadipocytes EDG-2 and EDG-4 mRNAs were quantified from growing 3T3F442A preadipocytes using real time RT-PCR (see "Materials and Methods"). In each experiment EDG-2 and EDG-4 transcripts were quantified from the same cDNAs preparation. Results were obtained from four experiments and expressed in copies of mRNA per 50 ng of total RNA.  2. Northern blot detection of EDG-2 and EDG-1 mRNA in 3T3F442A preadipocytes. Twenty g of total RNA extracted from growing 3T3F442A preadipocytes or mouse brain were analyzed by Northern blot using specific 32 P-labeled probes directed against EDG-2 (A) and EDG-1 mRNAs (B). An 18 S ribosomal RNA probe was used to ensure equal well loading. Data presented are representative of at least three separate experiments. Fig. 3C). It was also noticeable that clone 7 exhibited an alteration of preadipocyte responsiveness to LPA despite the existence of the high level of expression of EDG-4 transcripts (Table  II). Clone 12, which exhibited no alteration of EDG-2 transcripts but a significant reduction of EDG-4 transcripts (Table  II), revealed no significant alteration in the proliferative response to LPA. Results showed that reduction of EDG-2 receptor expression was accompanied by an alteration in the proliferative response of 3T3F442A preadipocytes to LPA. In parallel, a poor contribution of EDG-4 receptor in this response was suggested.
LPA is abundant (0.5-2.5 M) in serum (20) and contributes to its biological activity (21)(22)(23)(24). In 3T3F442A preadipocytes transfected with the empty vector, 10% serum led to a large increase in cell number, which was significantly reduced (about 30%) by pretreatment of the serum with phospholipase B (Fig.  5). Phospholipase B is a lysophospholipase previously shown to hydrolyze LPA and suppress its bioactivity (14,21,25). Therefore, LPA significantly contributed to the proliferative response of 3T3F442A preadipocytes to serum. In clone 7, the response to serum was significantly lower as compared with 3T3F442A   4. Influence of LPA on the proliferation of G418-resistant cell clones transfected with EDG-2 antisense cDNA. Each cell clone was seeded and grown in 10% donor calf serum-supplemented DMEM. After 48 h, serum was removed, and each cell clone was grown for an additional 48 h in the presence or absence of 1 M 1-oleoyl-LPA. Cell number obtained in each clone was determined as described under "Material and Methods" and compared with that obtained with empty pcDNA3.1 transfected cells. Each column represents the mean Ϯ S.E. of three to five independent experiments, depending on cell clone. Statistical analysis was performed using the Student's t test: *, p Ͻ 0.05 when comparing LPA activity in each clone to that measured in empty pcDNA3.1 transfected cells.
preadipocytes transfected with the empty vector. In addition, it was not significantly modified by phospholipase B treatment (Fig. 5). Similar results were obtained with clone 24 (not shown). Results showed that clone 7 exhibited a strong reduc-tion in its proliferative response to the LPA present in serum. Finally, it was noticeable that the phospholipase B-insensitive proliferative response to serum was not significantly different between clone 7 and empty vector transfected cells (Fig. 5). This showed that clone 7 exhibited no alteration in the proliferative response to phospholipase B-insensitive growth factors.
Influence of EDG-2 Antisense cDNA Transfection on LPA-dependent Spreading-In several cell types, including 3T3F442A preadipocytes (14,25), 1-oleoyl-LPA (LPA) and sphingosine 1-phosphate (S1P) (another bioactive phospholipid acting via a distinct receptor than LPA) induce a rapid and powerful reorganization of actin cytoskeleton, leading to a rapid spreading of the cells previously retracted by serum deprivation (Cont in Fig. 6C). In 3T3F442A preadipocytes transfected with empty vector (pcDNA3.1 in Fig. 6C), LPA (left panel in Fig. 6C) and S1P (right panel in Fig. 6C) induced a dose-dependent increase in cell spreading. This effect was quantified by measurement of cell surface (Fig. 6, A and B). The detectable spreading effect was observed with a 10 nM concentration of both LPA and S1P (Fig. 6, A-C). In clone 7, the dose-response curve generated by LPA was significantly shifted to the right, with a detectable spreading effect observed only at 100 nM (Fig. 6, A and C, left  panel). On the contrary, the dose-response curve generated by S1P was not significantly different between 3T3F442A preadipocytes transfected with empty vector and clone 7 (Fig. 6, B and C, right panel). A weak but not significant reduction in maximal response of sphingosine 1-phosphate was observed in clone 7. Results showed that clone 7 exhibited a significant and specific reduction in its spreading response to LPA.
Expression of EDG-2 mRNAs during the Conversion of Growing 3T3F442A Preadipocytes into Growth-arrested Adipocytes-When cultured in fetal calf serum and insulin (see "Material and Methods"), confluent 3T3F442A preadipocytes can be converted into growth-arrested adipocytes (26). Conversion into adipocytes is also characterized by a increased expression of mRNAs encoding adipocyte-specific proteins. Among them is the adipocyte-lipid-binding protein encoded by the aP2 gene (27). The influence of adipose conversion of 3T3F442A preadipocytes was tested on the expression of EDG-2 mRNAs. By using Northern blot analysis, it was observed that adipocyte conversion (characterized by a rapid increase in the aP2 mRNA level) was accompanied by a coordinate and strong reduction in the EDG-2 mRNA level (Fig. 7). Results suggested that the EDG-2 receptor likely played a more important role in growing preadipocytes than in growth-arrested adipocytes. DISCUSSION The results of the present study show evidence of a predominant contribution of the EDG-2 receptor in the responses of 3T3F442A preadipocytes to LPA. Among the four potential LPA receptor genes (EDG-2, EDG-4, EDG-7, and PSP24), only EDG-2 and EDG-4 transcripts were found in 3T3F442A preadipocytes. Quantitative analysis of transcript abundance revealed predominance of EDG-2 gene receptor over EDG-4 gene receptor. This predominance was also found in another preadipose cell line: 3T3L1. 2 Therefore the EDG-2 receptor is likely primarily involved in the action of LPA in preadipocytes. This assessment is supported by experiments showing that antisense-directed reduction in EDG-2 receptor expression significantly altered the responses of 3T3F442A preadipocytes to LPA. It was indeed possible to isolate 3T3F442A-derived cell clones exhibiting significant reduction of endogenous EDG-2 mRNA and protein level associated with significant reduction of the cellular responses to LPA: proliferation and cytoskeleton reorganization (Figs. 4 -6). This reduction was specific to LPA, since, in parallel, responses to sphingosine 1-phosphate (Fig. 6) or to other phospholipase B-insensitive growth factors (Fig. 5) were not significantly altered.
Nevertheless, this antisense strategy did not completely block the action of LPA. This very likely resulted from a partial blockade of EDG-2 receptor expression following antisense cDNA transfection, since a substantial decrease, but not total disappearance, of EDG-2 expression is elicited by EDG-2 antisense stable expression. In addition, one cannot completely exclude the possible contribution of another receptor in the residual responses generated by LPA. Besides EDG-2 receptors, 3T3F442A preadipocytes also express EDG-4 receptor. However, the data of Table II reveal that variations of EDG-4 transcript expression in EDG-2 antisense-expressing clones cannot be correlated with modifications of the proliferative response to LPA. Although we are aware that these data should be confirmed at the protein level, they strongly suggest the poor contribution of the EDG-4 receptor in the responses of 3T3F442A preadipocytes to LPA.
Finally, EDG-2 transcripts were predominantly expressed in growing preadipocytes and were strongly reduced in growtharrested adipocytes. The precise mechanisms involved in downregulation remain unclear and are currently under investigation. Whatsoever, this observation strongly supports the role of EDG-2 receptors in the proliferative response to LPA in growing preadipocytes.
Although several studies have shown that overexpression of EDG-2 cDNA restores or increases LPA sensitivity in mammalian cells (5,8), the specific contribution of the endogenously expressed EDG-2 receptor remained poorly documented. Goetzl et al. (28) have shown that the antiapoptotic response of a human T-lymphocyte cell line to LPA could significantly be reduced by transfection with EDG-2 plus EDG-4 antisense cDNA. The specific contribution of the EDG-2 receptor remained to be determined. The present study brings evidence for a specific contribution of the EDG-2 receptor endogenously expressed in preadipocytes.
Previous work from our laboratory (14) revealed that LPA can be produced by adipocytes and plays an important role in paracrine/autocrine control of proximal preadipocytes. Because of the involvement of the EDG-2 receptor in the control of the proliferation of preadipocytes, this receptor appears to be an interesting target to control preadipocyte proliferation, one of the key events of adipose tissue development. FIG. 7. Expression of EDG-2 and aP2 mRNAs during conversion of 3T3F442A preadipocytes into adipocytes. 3T3F442A preadipocytes were grown in donor calf serum (SVD)-supplemented DMEM until confluence. At confluence, the medium was replaced by fetal calf serum-supplemented DMEM plus insulin (SVFϩinsulin) to induce adipose conversion (see "Material and Methods"). Total RNAs were extracted at different time during the course of preadipocytes conversion into adipocytes, and mRNAs were detected by Northern blot. Data are representative of at least three separate experiments.