G-protein-coupled Receptor Stimulation of the p42/p44 Mitogen-activated Protein Kinase Pathway Is Attenuated by Lipid Phosphate Phosphatases 1, 1a, and 2 in Human Embryonic Kidney 293 Cells*

Sphingosine 1-phosphate, lysophosphatidic acid, and phosphatidic acid bind to G-protein-coupled receptors to stimulate intracellular signaling in mammalian cells. Lipid phosphate phosphatases (1, 1a, 2, and 3) are a group of enzymes that catalyze de-phosphorylation of these lipid agonists. It has been proposed that the lipid phosphate phosphatases exhibit ecto activity that may function to limit bioavailability of these lipid agonists at their receptors. In this study, we show that the stimulation of the p42/p44 mitogen-activated protein kinase pathway by sphingosine 1-phosphate, lysophosphatidic acid, and phosphatidic acid, all of which bind to G i/o - coupled receptors, is substantially reduced in human embyronic kidney 293 cells transfected with lipid phosphate phosphatases 1, 1a, and 2 but not 3. This was correlated with reduced basal intracellular phosphatidic acid and not ecto lipid phosphate phosphatase activity. These findings were supported by results showing that lipid phosphate phosphatases 1, 1a, and 2 also abrogate the stimulation of p42/p44 mitogen-activated protein kinase by thrombin, a peptide G i/o -coupled re- ceptor

Phosphatidic acid phosphatase type 2 was originally identified as a plasma membrane enzyme that catalyzes the dephosphorylation of the putative second messenger, phosphatidic acid (PA) 1 to diacylglycerol (DG) (1). Subsequently, multiple isoforms of phosphatidic acid phosphatase type 2 were cloned (2)(3)(4)(5). It was found that these enzymes dephosphorylate a number of lipid phosphates in vitro other than PA, including lysophosphatidic acid (LPA), sphingosine 1-phosphate (S1P), ceramide 1-phosphate, and diacylglycerol pyrophosphate. Therefore, they have been renamed lipid phosphate phosphatases. Currently, there are four members of this family called LPP1, LPP1a, LPP2, and LPP3 (6). The LPP isoforms are integral membrane proteins with six transmembrane domains. These phosphatases contain a motif (KXXXXXXRP(X 12-54)PSGH(X 31-54 )SRXXXXXHXXXD), which is found in a super family of phosphatases (7). Recently, site-directed mutagenesis has been used to establish that the conserved residues K, R, P, S, G, H, R, and H in the catalytic domain of LPP1 are obligatory for activity (8).
There are also alternative routes through which S1P and LPA can be dephosphorylated in cells. This involves recently identified S1P-and LPA-specific phosphatases (9,10). The S1P-specific phosphatase has 8 -10 transmembrane domains and contains the same phosphatase motif as the LPP isoforms, whereas the LPA-specific phosphatase has an acid phosphatase motif (LXXVXXVXRHGXRXP).
LPA may also bind to G-protein-coupled receptors that are not members of the EDG family, such as PSP24␣/␤ and GPR45 (20). Furthermore, PA has been shown to stimulate cell migration via a novel receptor-mediated mechanism in breast cancer cells (21).
Recent studies by Xu et al. (22) demonstrate that the net association of LPA with EDG2 receptors is reduced in Rat2 fibroblasts transfected with LPP1. This leads to a decrease in the LPA-dependent stimulation of p42/p44 MAPK, inhibition of adenylate cyclase, calcium mobilization, phospholipase D activation, and DNA synthesis. Such findings support the possibility that LPP1 limits bioavailability of LPA at its receptor. In contrast, Hooks et al. (23) find that although ecto-LPP1 degrades ϳ90% of the LPA in the medium over a 24-h period in HEK 293 cells, this was insufficient to account for the decrease in LPA potency in mitogenic assays, suggesting an alternative unidentified mechanism. One possibility is that the LPP isoforms might dephosphorylate intracellular PA, which is a putative second messenger molecule that may have a role in growth factor and GPCR signaling. For instance, PA is involved in the membrane recruitment of Raf-1 (24), which contains a specific PA binding region (amino acids 390 -426) (25). Indeed, transfection of HIRcB fibroblasts with a PA binding region construct blocked subsequent insulin-stimulated p42/p44 MAPK and prevented Raf-1 translocation by sequestering PA (24). The LPP-catalyzed reduction in PA might also increase Ras-GTPase-activating protein (26), thereby reducing the halflife of GTP-bound Ras. Alternatively, the LPP isoforms may modulate the membrane dynamic governing GPCR signaling. For instance, PA and phosphatidylinositol 4,5-bisphosphate (the formation of which is in a positive feedback loop with PA (27)) play a role in membrane trafficking (28), where they initiate clathrin coat assembly and facilitate vesicle formation for endocytosis (29). In this regard, specialized areas of the plasma membrane, termed caveolae (30), provide platforms for pre-assembled clusters of signaling molecules, including G i , the localization of which is essential for GPCR signaling (31). Therefore, the LPP-catalyzed dephosphorylation of PA may have a role in the endocytic signaling process.
Recent studies show that intracellular S1P may also function as an intracellular second messenger. For instance, Rani et al. (32) show that platelet-derived growth factor stimulates an increase in intracellular S1P in fibroblasts and that this may activate the p42/p44 MAPK pathway. In this regard, the LPP isoforms may dephosphorylate intracellular S1P, thereby preventing its potential second messenger action.
In this study, we have shown that LPP1, LPP1a, and LPP2 but not LPP3 abrogate the activation of p42/p44 MAPK in response to LPA, S1P, PA, and thrombin. Significantly, thrombin is a peptide G i PCR agonist whose bioavailability at its receptor is not regulated by the LPP isoforms. We have also shown that the abrogation of p42/p44 MAPK activation by the LPP isoforms can be correlated with a reduction in basal intracellular PA and not ecto-LPP activity. Therefore, our findings suggest a model in which LPP1, LPP1a, and LPP2 may act on a basal intracellular PA pool to alter the membrane dynamic governing GPCR signaling to p42/p44 MAPK per se.
Cell Culture-HEK 293 cells and stable transfected cell lines generated from them were maintained in minimum essential medium (MEM) and 10% (v/v) fetal calf serum. Cells were maintained in MEM supplemented with 0.1% (v/v) fetal calf serum for 24 h before experimentation.
Transfection-Plasmid constructs were generated using pcDNA3.1 (ZeoϪ) containing either gpLPP1 or gpLPP1a (5). Plasmid constructs containing hLPP2 and hLPP3 were a generous gift from Dr. A. J. Morris (State University of New York, Stony Brook, NY). hLPP2 and hLPP3 were separately subcloned into pcDNA3.1 (ZeoϪ) at the BamHI and HindIII sites. HEK 293 cells were separately transfected with 1 g of either pcDNA3.1 or pcDNA3.1-LPP1, -LPP1a, -LPP2, and -LPP3 using LipofectAMINE Plus and stable transfectants selected using Zeocin. Multiple surviving colonies of each type were screened for enhanced membrane LPP activity in an in vitro mixed micellar assay. A single colony of each type was selected and propagated for subsequent experiments.
Membrane LPP Activity Assay-Membranes of the stable cell lines were prepared by their homogenization in ice-cold buffer (containing 50 mM Tris-maleate, 1 mM EDTA, 150 mM NaCl, and 10 mM mercaptoethanol) and centrifugation at 30,000 ϫ g at 4°C for 10 min. Pellets were resuspended in homogenization buffer (at 20 -200 g of protein/ml) and stored at Ϫ20°C. Membrane LPP activity was measured as the liberation of 32 P i from 32 P-labeled substrates (1000 dpm/pmol-625 dpm/ nmol) in the presence of Triton X-100 (fixed lipid:detergent ratio of 1:10), 37.5 mM Tris-maleate, 7.5 mM mercaptoethanol, and 0.2 mg/ml bovine albumin at 30°C for 5 min. Incubations were stopped by the addition of 5 volumes of chloroform/methanol/10 mM HCl (15:30:2, v/v). Organic and aqueous phases were resolved by the addition of 1.25 volumes each of chloroform and 0.1 M HCl. Liberated 32 P i was measured by counting radioactivity in the upper phase. 32 P i liberated from [ 32 P]S1P was measured by further processing the upper phase with an additional phase split with equal volumes of 2 M KCl and water-saturated butanol and counting radioactivity in the lower phase. Analysis of the reaction products by thin layer chromatography confirmed the specificity of the assay.
Ecto-LPP Activity Assay-Cells were rinsed twice in phosphate-buffered saline and incubated in MEM supplemented with 0.1% (w/v) fatty acid-free bovine albumin and 1 mM ␤-glycerophosphate at 37°C for 30 min. Subsequently, [ 32 P]S1P, [ 32 P]dioctanoyl-PA, or [ 32 P]oleoyl-LPA (final concentration 50 M, 10,000 -15,000 dpm/nmol) prepared in MEM, 0.1% fatty acid-free bovine albumin, 1 mM ␤-glycerophosphate was added, and the cells were incubated for an additional 30 min. 500 l of the medium was extracted using chloroform, methanol, 10 mM HCl (15:30:2). Organic and aqueous phases were resolved by the addition of chloroform and 0.1 M HCl. 32 P i liberated from [ 32 P]dioctanoyl-PA or [ 32 P]oleoyl-LPA was measured by counting radioactivity in the upper phase. 32 P i liberated from [ 32 P]S1P was measured after an additional phase split as described above. The inclusion of ␤-glycerophosphate was a precaution to minimize the possibility of phospholipase A-catalyzed degradation of 32 P-labeled substrates. Analysis of the reaction products by thin layer chromatography confirmed the specificity of the assay.
Measurement of Basal PA and DG-Lipid extracts prepared from cells labeled with [ 3 H]palmitate for 20 h were halved and analyzed by thin layer chromatography on silica G150 plates developed as above for PA or in hexane:diethyl ether:acetic acid (60:40:1, v/v) for DG. Radioactivity comigrating with unlabeled lipid standards was quantified by scintillation counting.
Phospholipase D Assay-Phospholipase D activity was measured using the transphosphatidylation assay.
Fluorescence Microscopy-HEK 293 cells stably transfected with empty vector or expressing recombinant LPP2 and LPP3 were grown on 6-well plates, rinsed three times with phosphate-buffered saline (PBS), and fixed using paraformaldehyde (3% w/v) in PBS. Subsequently, the cells were washed successively in PBS, 50 mM ammonium chloride in PBS, 0.1% (w/v) Triton-100 in PBS and incubated in blocking buffer (0.2% (v/v) fish gelatin and 0.1% (v/v) goat serum in PBS) before incubation with antibodies that specifically detect LPP2 or LPP3 (1:100 dilution of each antibody in PBS, 4 h). LPP expression was visualized using a fluorescein isothiocyanate-coupled anti-rabbit IgG (1:400 in PBS, 1 h) on a laser-scanning confocal imaging system and Laser Sharp software (Bio-Rad).

RESULTS
Lipid Phosphate Phosphatase Expression-We set out to evaluate whether members of the LPP family can modulate LPA-and S1P-dependent stimulation of the p42/p44 MAPK pathway either by limiting bioavailability of these lysophospholipid agonists at receptors or by regulating the intracellular level of the putative second messenger, PA. To investigate this, we generated four separate HEK 293 stable cell lines, each overexpressing either LPP1, LPP1a, LLP2, or LPP3.
Endogenous LPP activity of vector-transfected HEK 293 cells was very low, making these cells ideal for transfection. The characterization of recombinant LPP activities in the transfected cell lines is presented in Table I. LPP1, LPP1a, LLP2, or LPP3 activity versus endogenous LPP activity in vector-transfected cells was measured in cell membranes using 50 M [ 32 P]dioctanoyl-PA, [ 32 P]oleoyl-LPA, and [ 32 P]S1P. The increase in recombinant LPP activity was substantial in each transfected cell line. The fold increases were 74-, 56-, 21-, and 6-fold (dioctanoyl-PA), 82-, 172-, 74-, and 7-fold (oleoyl-LPA), and 128-, 41-, 271-, and 7-fold (S1P) for LPP1, LPP1a, LPP2, and LPP3, respectively. Similar rank order activities were measured using a lower substrate concentration (5 M) of each substrate. The fold increase in activities using LPA, PA, and S1P were substantially larger than any other previous report for the transfection of these enzymes into cells. One possibility for this larger difference is that we have used stable transfected HEK 293 cell lines, whereas other workers have used transient transfection where expression levels of the LPP isoforms are likely to be considerably lower.
Kinetic estimates for K m and V max for LPP1, LPP1a, LPP2, and LPP3 using these substrates are shown in Table II. These data show that V max values for dioctanoyl-PA are considerably larger for LPP1 and LPP1a compared with LPP2 and LPP3. In contrast, LPP2 exhibits the highest V max for oleoyl-LPA. The V max values for S1P were considerably higher for LPP2 than LPP1 and LPP1a, which were in turn higher than LPP3.    LPP3 and ϳ 220 M for LPP1a and LPP2. We also found that LPP1, LPP2, and LPP3 exhibit low but significant ecto activity against [ 32 P]oleoyl-LPA (Table III). This is in line with other studies showing that LPP1 and LPP3 have ecto activity in transfected Rat2 fibroblasts and HEK 293 cells, respectively (34,35). However, our results are the first to provide evidence that LPP2 exhibits ecto activity against oleoyl-LPA. The exception is LPP1a, which does not display activity against this substrate. The increases in ecto activity in LPP1-, LPP1a-, LPP2-, and LPP3-transfected cells compared with vector-transfected cells were 3.6-, 0.8-, 7.7-, and 6.4-fold, respectively (Table III). These activities are similar to those reported for mouse LPP1 in Rat2 fibroblasts (34). The ecto-LPP activity is not an artifact of uptake of [ 32 P]oleoyl-LPA into the cells followed by intracellular metabolism and then release of phosphate into the extracellular medium. If this had been the case, then the relative ecto-LPP activities between the cell types would correlate exactly with relative activities measured in cell membranes. This is clearly not the case since the rank order of ecto-LPP activities (LPP2 ϭ LPP3 Ͼ LPP1 Ͼ ϾLPP1a ϭ vectortransfected cells) differs from that for LPP activities in membranes (LPP1a Ͼ LPP1 ϭ LPP2 Ͼ Ͼ LPP3 Ͼ Ͼ vector-transfected cells) (see Table I).
Additional evidence showing the expression of LPP2 and LPP3 was obtained using Western blot analysis with isoformspecific antibodies that recognize unique C-terminal regions in LPP2 and LPP3. Fig. 1a shows that a band with a molecular mass of 32 kDa was specifically immunostained in LPP2-transfected cell lysates using anti-LPP2 antibody and, significantly, was absent from vector-transfected cells. This band was not detected with anti-LPP3 antibodies. We can, therefore, attribute the protein to LPP2. A band with a molecular mass of 35 kDa corresponding to LPP3 was immunostained with anti-LPP3 antibodies in LPP3-transfected but not vector-transfected cell lysates. This band was not detected with anti-LPP2 antibodies. The molecular masses of these proteins determined on SDS-polyacrylamide gel electrophoresis are in agreement with the predicted values from the amino acid sequences of the recombinant proteins.
Both antibodies are highly specific and can be used for immunohistochemical staining of the respective transfected cells. Fig. 1b shows that the antibodies immunostained recombinant LPP2 and LPP3 at or close to the plasma membrane in transfected HEK 293 cells. We therefore propose that the LPP isoforms are located at or close to the inner membrane. This is based upon the fact that the ecto-LPP activities are very low and that the antibodies we have used are raised to C-terminal regions of the LPP isoforms that are predicted to be cytoplasmic-facing. In addition, LPP2 appears to be localized to an undefined intracellular compartment. As controls, we also show in Fig. 1b that the anti-LPP2 and anti-LPP3 antibodies do not immunostain vector-transfected cells nor do they cross- react with proteins in LPP3-and LPP2-transfected cells, respectively.
Collectively, the results clearly show that the individual LPP isoforms are stably expressed in the four transfected cell lines and are functionally active. Therefore, these cell lines provide an excellent model to address whether the LPP isoforms can modulate S1P and LPA signaling. S1P-and LPA-dependent Activation of p42/p44 MAPK-To investigate whether the LPP isoforms can modulate S1P and LPA signaling via G-protein-coupled receptors, we first had to establish that S1P and LPA stimulate p42/p44 MAPK via this mechanism in HEK 293 cells. This was necessary because, in the case of S1P, the stimulation of p42/p44 MAPK can also occur via an unidentified intracellular-dependent mechanism in certain cell types (32). S1P and LPA (0.5-10 M) stimulated a concentration-dependent activation of p42/p44 MAPK (data not shown). This was sustained for at least 30 min and declined toward basal thereafter (Fig. 2). Both lipid agonists were less effective than PMA, which short circuits G-protein-coupled and growth factor receptors to stimulate p42/p44 MAPK via a protein kinase C-dependent mechanism (Fig. 2). Most significant, in terms of establishing that these lipids are agonists at receptors, was the observation that the pre-treatment of cells with the bacterial toxin, pertussis toxin (0.1 g/ml for 18 h), which ADP-ribosylates the G-proteins, G i/o , and uncouples these G-proteins from their respective receptors, substantially reduced the S1P-and LPA-dependent activation of p42/p44 MAPK (Fig. 3a). In addition, we found that suramin, which can inhibit GPCR signaling blocked the S1P-and LPA-dependent activation of p42/p44 MAPK (Fig. 3b). These findings suggest that S1P and LPA exert their effects on p42/p44 MAPK via a G i PCR-dependent mechanism in HEK 293 cells.
In general we found that the extent of p42 MAPK phosphorylation in response to stimulation by S1P and LPA exceeded that of p44 MAPK, which on some Western blots was barely detectable. This is accounted for by the relative expression of p42 and p44 MAPK in these cells (data not shown).
The relatively high concentration requirement for LPA and S1P used in this study is in agreement with other reports in various cell types (18, 22, 36 -41). Indeed, Brindley and coworkers (22) show that decreasing extracellular calcium from 1.8 mM (concentration in MEM) to 10 M increased LPA binding by 20-fold, shifting the threshold for p42/p44 MAPK activation to the nanomolar range. Hence, the calcium dependence of the apparent K d values for ligand binding to receptors may explain the longstanding discrepancy of why micromolar LPA is often needed to activate p42/p44 MAPK at physiological calcium levels. These findings may explain the similar high concentration requirement for LPA and S1P seen in our experiments. However, this explanation remains only a possibility and further experiments are necessary to establish its validity.
In contrast with the effect on LPA and S1P, pertussis toxin and suramin had no effect on serum-or PMA-stimulated p42/p44 MAPK activation (Fig. 3, a and b). Lipid Phosphate Phosphatases Modulate S1P-and LPAstimulated Activation of p42/p44 MAPK-We next addressed the major question of this study as to whether the different LPP isoforms could affect the ability of LPA and S1P to stimulate p42/p44 MAPK. Fig. 4 shows that the activation of p42/ p44 MAPK by these lipid agonists was substantially reduced in LPP1-, LPP1a-, and LPP2-but not LPP3-transfected cells. Comparison with the ecto-LPP activities shows that there is no correlation with the abrogation of p42/p44 MAPK activation. For instance, LPP3 exhibits ecto activity against oleoyl-LPA and S1P that is larger than that of the other isoforms (Table  III). However, LPP3 does not abrogate the activation of p42/p44 MAPK by LPA or S1P (Fig. 4). There was a better correlation with membrane LPP activities against PA, LPA, and S1P measured in Table I. Finally, the expression level of p42 MAPK was not different between the vector-and LPP-transfected cell lines (Fig. 4).
PA Stimulation of p42/p44 MAPK-PA has been shown to stimulate cell migration via a receptor-mediated mechanism in breast cancer cells (22). Given that the LPP isoforms also dephosphorylate PA, we tested first whether PA could stimulate p42/p44 MAPK via a receptor-mediated mechanism and, second, whether the LPP isoforms might limit PA bioavailability at this receptor. Fig. 5a shows that PA (30 nM) stimulated p42/p44 MAPK activation via a pertussis toxin-and suraminsensitive mechanism, suggesting that this response is probably mediated via a G i/o -coupled receptor. This receptor is clearly different from those that bind LPA and S1P, as the concentration requirement for PA was in the nM range. The response to PA is unlikely to be mediated via EDG2 and EDG4, as PA has a 50-fold lower binding affinity for these receptors compared with LPA (24). Fig. 5b shows that the PA-dependent activation of p42/p44 MAPK activation was substantially reduced in LPP1-, LPP1a-, and LPP2-but not LPP3-transfected cells. Comparison with the ecto-LPP activities against PA shows that there is no correlation with the abrogation of p42/p44 MAPK activation. For instance, LPP2 does not exhibit significant ecto activity against PA (Table III) but does abrogate the activation of p42/p44 MAPK (Fig. 5b).
GPCR-mediated Stimulation of p42/p44 MAPK-We next tested whether the specificity of action by the LPP isoforms was restricted to the lipid agonists used in this study. We evaluated whether the LPP isoforms can act on other G i PCR agonists whose bioavailability at receptors is not subject to regulation by these enzymes. For this purpose we used the peptide agonist, thrombin, which stimulates p42/p44 MAPK via a pertussis toxin-sensitive mechanism (Fig. 6a), thereby suggesting that it binds to a G i/o -coupled receptor. Fig. 6b shows that the stimulation of p42/p44 MAPK by thrombin was abrogated in LPP1-, LPP1a-, and LPP2-but not LPP3-transfected cells. This correlated exactly with the effect of the LPP isoforms on LPA, S1P, and PA stimulation of p42/ p44 MAPK. Thrombin does not cause release of S1P, PA, nor LPA from HEK 293 cells (data not shown). In addition we have done control experiments to confirm that thrombin does not use intracellular S1P as a second messenger to stimulate the p42/ p44 MAPK pathway. This is based upon data shown in Fig. 6c, where pre-treatment of cells with the sphingosine kinase inhibitor DL-threo-dihydrosphingosine (used at a concentration that has been shown to completely inhibit platelet-derived growth factor-stimulated sphingosine kinase activity and to block platelet-derived growth factor-stimulated p42/p44 MAPK activation in Swiss 3T3 fibroblasts (32)) did not block the thrombin-dependent activation of p42/p44 MAPK (Fig. 6c). These data appear to exclude the dephosphorylation of intracellular S1P as a possible mechanism by which LPP1, LPP1a, and LPP2 abrogate stimulation of p42/p44 MAPK by thrombin.
We also looked at other agents that do not use G-proteins to signal to p42/p44 MAPK, such as epidermal growth factor, serum factors, and PMA. Fig. 4 shows that none of the LPP isoforms altered p42/p44 MAPK activation in response to serum or PMA. The response to epidermal growth factor was only partially reduced (Ͻ20%) by LPP1, LPP1a, and LPP2.
PA Metabolism-We next investigated whether the LPP isoforms alter the intracellular levels of PA and DG, both of which have been implicated as second messengers for LPA, S1P, and thrombin. Our experiments show that there was a correlation between the ability of the LPP isoforms to abrogate activation of p42/p44 MAPK and changes in basal intracellular PA and DG levels. Table IV shows that basal intracellular PA levels were decreased, whereas DG levels were increased in LPP1-, LPP1a-, and LPP2-transfected but not LPP3-transfected cells.
Importantly, the correlation was specific with changes in basal PA/DG levels and not agonist stimulation of PA formation. This was based upon two pieces of evidence. First, neither S1P nor LPA increased intracellular PA (Fig. 7). Second, although PMA stimulates phospholipase D to generate PA in these cells (Fig. 7), the ability of this agent to activate p42/p44 MAPK was not altered at all by expression of LPP1, LPP1a, LPP2, and LPP3 ( Fig. 4 and Fig. 5b).

DISCUSSION
The first major finding of this study is that the stimulation of p42/p44 MAPK by LPA, S1P, and PA is substantially reduced in LPP1-, LPP1a-, and LPP2-but not LPP3-transfected HEK 293 cells. To date, only LPP1 has been shown to abrogate LPA-dependent stimulation of p42/p44 MAPK and mitogenesis in Rat2 fibroblasts and HEK 293 cells, respectively (22,23). Cell lysates were taken for Western blotting with antibodies that react with the phosphorylated/activated forms of p42/p44 MAPK. Blots were stripped and re-probed with antibodies that react with p42 MAPK to ensure equal protein loading. These are representative results of an experiment performed three times.
Therefore, the current study is not only the first to extend this to other LPP isoforms but, significantly, is also the first to report that the LPP isoforms can attenuate the stimulation of p42/p44 MAPK by S1P and PA. The inability of LPP3 to abrogate activation of p42/p44 MAPK by the lipids may be due to the fact that its expression in HEK 293 cells is considerably less than the other LPP isoforms.
Some of the LPP isoforms exhibited V max values that were considerably larger than any other previously reported for these enzymes in transfected cells (2,3,5). These values simply reflect particularly good expression of the enzymes in the stable-transfected HEK 293 cells. The estimated K m values for dioctanoyl-PA were ϳ200 -290 M for all the enzymes, whereas those for oleoyl-LPA were 14 -60 M for LPP1/LPP1a and ϳ190 -250 M for LPP2/LPP3. These values are very similar to those obtained by Hooks et al. (43), who determined K m values for dioleoyl-PA of 98, 150, and 100 M and for oleoyl-LPA of 170, 340, and 110 M for LPP1, LPP2, and LPP3, respectively, in transfected HEK 293 cells. In addition, we report for the first time that LPP2 has both intracellular and extracellular activity and have provided evidence of its localization in transfected HEK 293 cells.
The second major finding of this study is that the abrogation of p42/p44 MAPK activation by LPP1, LPP1a, and LPP2 is not due to limitation of the bioavailability of the lipid agonists at their respective receptors. This was based upon two lines of evidence. First, there was no correlation between ecto-LPP activity against LPA, S1P, and PA and the abrogation of p42/p44 MAPK activation. Second, LPP1, LPP1a, and LPP2 also abrogated the stimulation of p42/p44 MAPK by thrombin, a peptide G i/o -coupled receptor agonist whose bioavailability at its receptor is not subject to regulation by the LPP isoforms.
These findings contrast with studies by Xu et al. (22), where ecto activity has been suggested to account for the effect of LPP1 on LPA-stimulated signaling in Rat2 fibroblasts. However, it is important to note that the study by Xu et al. (22) did not investigate the effect of LPP1 on other GPCR agonists and so the specificity of its action, and therefore the mechanism was not fully defined. Our findings do agree to some extent with those of Hooks et al. (23). These authors found that although ecto-LPP1 degrades ϳ90% of the LPA in the medium over a 24-h period in HEK 293 cells, this was insufficient to account for the decrease in LPA potency in mitogenic assays, suggesting an alternative unidentified mechanism. Moreover, these authors reported, using degradation-resistant phosphonate analogs of LPA and stereoselective agonists of the EDG receptors, that the mitogenic response of LPA was independent of EDG2, -4, and -7. These authors therefore suggest that the responses to LPA may be mediated via a distinct GPCR, possibly PSP24, an LPA receptor that is not a member of the EDG family (20). In our studies, we found that LPA activates p42/p44 MAPK via a GPCR that is sensitive to pertussis toxin and suramin, although we do not know as yet whether this receptor is a . c, vector-transfected HEK 293 cells were pre-treated with vehicle (Me 2 SO (DMSO)) or DLthreo dihydrosphingosine (DHS, 1, 5, 10 M, 15 min) before thrombin (0.3 units/ml, 10 min). Cell lysates were taken for Western blotting with antibodies that react with the phosphorylated/activated forms of p42/ p44 MAPK. Blots were stripped and re-probed with antibodies that react with p42 MAPK to ensure equal protein loading. These are representative results of an experiment performed three times.  member of the EDG family. We have also found that exogenous PA stimulates p42/p44 MAPK via a G i/o -coupled receptor that appears to exhibit a much higher affinity for PA compared with LPA at its receptor. Therefore, it is possible that PA binds to a distinct receptor from that which binds LPA.
The LPP isoforms have no effect on the stimulation of p42/ p44 MAPK by other agents that do not use G-proteins to signal, such as serum factors and phorbol ester. Taken together, these findings represent a major advance as they clearly show that LPP1, LPP1a, and LPP2 may function to abrogate GPCR signaling per se.
Basal intracellular PA levels were decreased, whereas DG levels were increased in LPP1-, LPP1a-, and LPP2-transfected but not LPP3-transfected cells. This is in line with studies by Leung et al. (4) using HEK 293 cells transfected with LPP1 or LPP1a. Agonist-stimulated PA has been implicated in the activation of Raf and the inhibition of Ras-GTPase-activating protein (25,26), both of which are up-stream components in the p42/p44 MAPK pathway. However, neither S1P nor LPA increase PA in HEK 293 cells. We propose that the LPP isoforms act on a basal pool of PA, which may have a distinct function from agonist-stimulated PA. One possibility is that basal PA contributes to the structural integrity of lipid micro domains within the plasma membrane, in which signal complexes are assembled. Alternatively, PA may regulate the formation of endocytic vesicles required for GPCR signaling.
The magnitude of the increase in DG levels in LPP1-, LPP1a-, and LPP2-transfected cells did not equate with the reduction in PA. Identical reciprocal changes in the amounts of DG and PA might be expected if DG production was exclusively via the action of the LPP isoforms. This suggests additional routes of DG synthesis in these cells. However, it is unlikely that there is a role for increased basal intracellular DG in abrogating p42/p44 MAPK activation, even though the chronic increase in intracellular DG levels does raise the possibility of down-regulation of protein kinase C, an enzyme that has been shown to regulate p42/p44 MAPK activation. This would have the effect of attenuating the action of agonists that use protein kinase C to stimulate p42/p44 MAPK activation, such as LPA and S1P. However, we suggest that functional changes in protein kinase C expression are unlikely given that the activation of p42/p44 MAPK by acute treatment of cells with PMA, which stimulates the same DG-sensitive protein kinase C isoforms that would be down-regulated by chronic DG, was unaffected by the LPP isoforms.
We speculate here that the LPP isoforms may act on a small pool of basal intracellular PA that specifically regulates the membrane dynamic for signaling via G-protein-coupled receptors. There are good reasons to believe that this may be the case. First, the attenuation of p42/p44 MAPK activation by the LPP isoforms is common to several GPCR agonists used in this study, including thrombin. Second, LPP isoforms may localize to caveolin-enriched lipid rafts, where the enzyme could function to maintain structural integrity by precisely regulating the PA concentration. This is important because G i ␣ is also localized in lipid rafts containing caveolin (31), which is required for GPCR signaling. Third, the LPP isoforms may dephosphorylate a pool of PA that plays a role in clathrin coat assembly and vesicle formation for endocytosis (29). This is important because the endocytosis of G-protein-coupled receptor signal complexes is required for efficient phosphorylation of MEK-1 (mitogen-activated protein kinase/extracellular signal-regulated kinase kinase 1) by Ras-Raf (45)(46)(47). Furthermore, the ratio of LPA/PA at the inner leaflet of the plasma membrane is important for vesicular budding (48). This involves endophilin I, an enzyme that exhibits lysophosphatidic acyltransferase activity and generates PA, which is required for endocytic vesicle formation. Furthermore, endophilin interacts with dynamin II (44) and GPCRs (42).
In conclusion, the overexpression of LPP1, LPP1a, and LPP2 in cells might deplete the intracellular pool of PA sufficiently to disrupt (i) the recruitment of specific signaling proteins to the caveolae and/or (ii) the endophilin I-mediated endocytic process required for GPCR-dependent stimulation of p42/p44 MAPK. This proposal is currently under investigation in our laboratory.