LPA Is a Chemorepellent for B16 Melanoma Cells: Action through the cAMP-Elevating LPA5 Receptor

Lysophosphatidic acid (LPA), a lipid mediator enriched in serum, stimulates cell migration, proliferation and other functions in many cell types. LPA acts on six known G protein-coupled receptors, termed LPA1–6, showing both overlapping and distinct signaling properties. Here we show that, unexpectedly, LPA and serum almost completely inhibit the transwell migration of B16 melanoma cells, with alkyl-LPA(18∶1) being 10-fold more potent than acyl-LPA(18∶1). The anti-migratory response to LPA is highly polarized and dependent on protein kinase A (PKA) but not Rho kinase activity; it is associated with a rapid increase in intracellular cAMP levels and PIP3 depletion from the plasma membrane. B16 cells express LPA2, LPA5 and LPA6 receptors. We show that LPA-induced chemorepulsion is mediated specifically by the alkyl-LPA-preferring LPA5 receptor (GPR92), which raises intracellular cAMP via a noncanonical pathway. Our results define LPA5 as an anti-migratory receptor and they implicate the cAMP-PKA pathway, along with reduced PIP3 signaling, as an effector of chemorepulsion in B16 melanoma cells.


Introduction
Lysophosphatidic acid (LPA) is a multifunctional lipid mediator that stimulates migration, proliferation, survival and other functions in many different cell types, both normal and malignant. LPA acts through six known G protein-coupled receptors (GPCRs), termed LPA [1][2][3][4][5][6] , which show both overlapping and distinct signaling properties and tissue distributions [1,2]. The three classical and best studied LPA receptors, LPA 1-3 , belong to the so-called Edg subfamily of GPCRs. Three additional LPA receptors, termed LPA 4 (formerly P2Y9), LPA 5 (GPR92) and LPA 6 (P2Y5) stand apart from the Edg family and are more closely related to the purinergic receptor family, strongly suggesting that LPA receptors have evolved from distinct ancestor genes [1,2]. LPA receptors couple to multiple G protein-effector pathways, accounting for the multitude of cellular responses to LPA.
Cell migration plays a key role in embryogenesis, tissue renewal, immune responses and cancer metastasis. Most tumor cells, including carcinoma, lymphoma, glioma and melanoma cells, show an enhanced migratory response to LPA (positive chemotaxis) [12][13][14][15][16][17]. Where examined, LPA-induced migration is primarily mediated by Edg-family LPA 1 and LPA 2 receptors and involves both G i -and G 12/13 -mediated signaling pathways. One non-Edg-family receptor, notably LPA 4 , has been reported to suppress cell migration [18]. In general, cell migration is driven by pathways controlled by Rho GTPases and phosphatidylinositol 3kinase (PI3K), acting in a coordinate fashion to control the spatiotemporal dynamics of cytoskeletal components. However, soluble second messengers such as Ca 2+ [19,20] and cyclic AMP (cAMP) [21,22] also play crucial roles in directing cell migration and chemotaxis.
Here we show that, unexpectedly, LPA strongly impedes the basal and growth factor-induced migration of B16F10 melanoma cells. We show that the inhibitory effect of LPA is mediated by the LPA 5 receptor and that a rise in cAMP with consequent activation of protein kinase A (PKA) is an important effector of LPA 5mediated chemorepulsion, with a possible additional role for reduced phosphatidylinositol (3,4,5)-trisphosphate (PIP3) signaling. Our results identify LPA 5 as an anti-migratory receptor and they point to a mechanism of LPA-induced chemorepulsion likely to be relevant for tumor cells that predominantly express LPA 5 , acting to override positive chemotactic signals.

Results and Discussion
LPA, serum and autotaxin inhibit B16 cell migration in a highly polarized fashion When testing various tumor cell types for their chemotactic response to 1-oleoyl-LPA (LPA(18:1)) and serum (FCS) in transwell assays, we found that B16F10 melanoma cell migration is strongly inhibited by both FCS and LPA(18:1) ( Figure 1A,B). Under serumfree conditions, B16 cells show a high rate of basal, 'spontaneous' transwell migration during a 3 hr assay period. Addition of FCS or LPA to the lower transwell chamber inhibited the basal migration of B16 cells almost completely. Half-maximal inhibitory effects were observed at ,100 nM LPA and 1% FCS (containing up to 50 nM LPA; [3]) ( Figure 1A,B). Strikingly, the inhibitory effects were observed only when LPA and FCS were present in the bottom well of the transwell chamber. LPA or serum added to the top well had no detectable effect on cell migration, regardless of the concentration or the direction of the LPA gradient across the filter ( Figure 1C). In other words, the inhibitory response to LPA is highly polarized. In keeping with this notion, LPA had no detectable effect on the random motility of B16 cells in low-density culture, as evidenced by time-lapse video microscopy (not shown).
LPA is produced extracellularly from lysophosphatidylcholine (LPC) by autotaxin (ATX). Addition of LPC(18:1) alone to the lower chamber had no effect, but cell migration was significantly inhibited when LPC was added together with recombinant ATX ( Figure 1D). Again, inhibition of migration was only observed when ATX plus LPC were present in the lower chamber.
Even though the basal migration rate of B16 cells is already high, it could be further enhanced by hepatocyte growth factor (HGF). As shown in Figure 1E, LPA reduced the chemotactic activity of HGF by some 60%, but only when both LPA and HGF were present in the lower chamber. This result defines LPA as an anti-chemotactic agonist for B16 cells. Based on these findings, we conclude that LPA inhibits B16 cell migration in a highly polarized fashion, likely reflecting localized LPA receptor signaling in the extending pseudopodia.
We also tested the effect of LPA on B16 cell proliferation. Melanomas are often characterized by activating mutations in B-Raf, leading to constitutive activation of the mitogenic MAPK/ ERK1/2 pathway. B16F10 cells, however, express wild-type B-Raf [23]. Consistent with their normal B-Raf status, B16 cells showed very low basal MAPK/ERK1/2 activity, which was strongly enhanced upon LPA addition ( Figure 2A). Yet, LPA (up to 10 mM) did not significantly affect the rate of B16 cell proliferation, either in the presence or absence of limiting concentrations of serum ( Figure 2B).
Dissection of LPA signaling pathways: involvement of cAMP and protein kinase A We next set out to explore the signaling determinants of the chemorepulsive effect of LPA. Optimal transwell migration was observed with cells plated on fibronectin-coated membranes. When B16 cells were plated on poly-L-lysine, which does not interact with integrins, migration was similarly inhibited by LPA ( Figure 3A). Thus, the anti-migratory effect of LPA is independent of the nature of the adhesive substrate and, by inference, does not involve specific integrins. In general, GPCR-mediated cell migration depends on pertussis toxin (PTX)-sensitive G i proteins that couple to activation of Rasand Rac-mediated signaling pathways. Pretreatment of B16 cells with PTX (200 ng/ml) did not affect the basal cell migration rate nor its inhibition by LPA, which rules out the involvement of G ilinked pathways. In addition, we used the Rho-kinase (ROCK) inhibitor Y27632, the protein kinase C inhibitor Ro-31-8220 and two MEK inhibitors (PD598059 and U-0126), which were added to both chambers. However, none of these inhibitors influenced the anti-migratory effect of LPA during a 4 hrs time period (data summarized in Table S1). The lack of effect of Y27632 is somewhat unexpected, since the RhoA-ROCK pathway is known to mediate contractile and anti-migratory events and, furthermore, underlies LPA-induced chemorepulsion of neuronal growth cones; in the latter case, Y27632 can even convert chemorepulsion into chemoattraction [24]. We measured RhoA-GTP levels in B16 cells before and after LPA stimulation, but were unable to detect a significant effect of LPA on RhoA-GTP accumulation in these cells (data not shown).
Given the multifunctional role of cAMP in cytoskeletal regulation and cell migration, we tested the cAMP-raising agents forskolin and 8-Br-cAMP. Both stimuli partly mimicked LPA in inhibiting the basal migration rate of B16 cells ( Figure 3B and Table S1), indicating that cAMP is a candidate messenger for mediating LPA-induced chemorepulsion. cAMP acts through activation of protein kinase A (PKA), while at higher levels it exerts additional effects through activation of Epac [25]. Epac is an exchange factor for the Rap GTPase and is activated by direct binding of cAMP. The PKA inhibitor H-89 attenuated the inhibitory responses to LPA and forskolin without affecting the basal migration rate ( Figure 3C), whereas the Epac activator 8CPT-2Me-cAMP [26] had no detectable effect on cell migration (Table S1). We confirmed that H-89 inhibits PKA activity by using a FRET-based PKA sensor in forskolin-treated cells as described previously [27].
To monitor changes in cAMP, we used the FRET-based cAMP sensor CFP-Epac-YFP [27]. When expressed in mammalian cells, CFP-Epac-YFP shows a significant FRET signal that rapidly decreases upon a rise in cAMP [27]. In B16 cells expressing this biosensor, LPA induced an initial transient rise in cAMP lasting for about 5 min, followed by a prolonged phase of elevated cAMP levels ( Figure 4A,B). From these results we conclude that the cAMP-PKA pathway, rather than the cAMP-Epac-Rap pathway, contributes to LPA-induced chemorepulsion.

LPA-induced PIP3 depletion from the plasma membrane
In many cases, directed cell migration is critically dependent on the lipid second messenger phosphatidylinositol (3,4,5)-trisphosphate (PIP3), which recruits PH-domain-containing proteins to regulate F-actin remodeling and other key signaling events. Local changes in PIP3 levels can be readily visualized by the  redistribution of green fluorescent protein (GFP)-tagged PH domains, such as the biosensor PH(GRP1)-GFP [28][29][30]. Most, if not all, chemoattractants activate phosphoinositide 3-kinase (PI3K) to induce rapid accumulation of PIP3 at the plasma membrane, as detected by the translocation of PIP3-sensitive probes from the cytosol to the plasma membrane [28][29][30]. Using PH-GRP1-GFP as a biosensor, we found that non-stimulated B16 cells show already relatively high basal PIP3 levels at the plasma membrane. LPA triggered a rapid depletion of PIP3 from the plasma membrane as inferred from the translocation of PH(GRP1)-GFP to the cytosol ( Figure 4C,D and Video S1). The PI3K inhibitor wortmannin induced a more sustained depletion of PIP3 from the plasma membrane ( Figure 4C,D) and inhibited the basal migration of B16 cells (Table S1). We also examined the PIP3-dependent phosphorylation state of Akt (protein kinase B) [31]. Consistent with the PIP3 imaging results, basal pAkt levels were high in non-stimulated B16 cells. LPA induced a modest fall in pAkt levels, while a PI3K inhibitor reduced basal pAkt levels to almost zero ( Figure S1).
It thus appears that LPA-induced chemorepulsion of B16 cells is associated with (i) an increase in intracellular cAMP and (ii) PIP3 depletion from the plasma membrane. PIP3 depletion is not secondary to a prior rise in cAMP levels, since the cAMP-raising agonist a-melanocyte-stimulating hormone (a-MSH) (see below) and forskolin failed to trigger PIP3 depletion ( Figure 4D and data not shown). Whether LPA-induced PIP3 depletion is due to inhibition of basal PI3K activity or enhanced PIP3 phosphatase activity, or both, remains to be determined.
Chemorepulsion is mediated by LPA 5 , an alkyl-LPA preferring receptor Which LPA receptor(s) mediate(s) the observed chemorepulsive effect? Quantitative PCR analysis revealed that B16 cells coexpress LPA 2 , LPA 5 and LPA 6 , with the relative expression levels of LPA 5 and LPA 6 being much higher than that of LPA 2 ( Figure 5A). We used siRNA to knock down LPA 5 and LPA 6 and assess the effects on migration. The inhibitory effect of LPA on cell migration was largely abolished after knockdown of LPA 5 , but not after knockdown of LPA 6 . LPA 5 knockdown did not significantly affect the basal cell migration rate, nor did it alter the expression levels of LPA 2 and LPA 6 ( Figure 5B,C). This strongly suggests that LPA 5 mediates the anti-migratory response of B16 cells.  [27]). LPA, 1 mM; forskolin (10 mM) was used to calibrate the response. C. Confocal images of LPA-and wortmannin-induced loss of PIP3 from the plasma membrane, as inferred from the translocation of PH(GRP1)-GFP to the cytosol. See also Video S1. D. Agonist-induced increase in cytosolic PH(GRP1)-GFP fluorescence, reflecting loss of PIP3 from the plasma membrane. Dynamic changes in fluorescence intensity (relative to non-stimulated control) were quantified using ImageJ software. LPA, 1 mM; wortmannin, 500 nM; a-MSH, 1 mM. doi:10.1371/journal.pone.0029260.g004 LPA 5 is unique in that it shows a clear preference for etherlinked LPA over the more common ester-linked LPA molecular species [32,33]. We therefore compared the efficacy of LPA(18:1) to that of the corresponding ether-linked analog 1-O-alkyl-LPA(18:1). As shown in Figure 5D, alkyl-LPA(18:1) was ,10-fold more potent than acyl-LPA(18:1) in inhibiting cell migration (IC 50 ,10 nM versus 100 nM). This difference in potency cannot be reconciled with Edg-family LPA receptors (LPA 1-3 ), which prefer the acyl forms of LPA. This result confirms critical involvement of LPA 5 in mediating the anti-migratory response to LPA.
To further establish the anti-migratory potential of LPA 5 , we evaluated LPA 5 in HeLa cells. These cells express LPA 1 , LPA 2 and LPA 5 at about equal mRNA levels ( Figure 6A), while their basal migration rate is very low ( Figure 6B). LPA induced a robust chemotactic response in HeLa cells. In LPA 5 -overexpressing HeLa cells, however, LPA-induced chemotaxis was significantly reduced ( Figure 6B). These results support the view that the relative expression pattern of LPA receptors in a given cell type determines migration outcome. That is, the chemorepulsive activity of LPA is observed in those cells where LPA 5 is the predominant receptor, overriding the chemoattractant activity of LPA 1 and LPA 2 .

LPA 5 mediates an increase in cAMP
Thus far, the signaling properties of LPA 5 /GPR92 have been inferred from heterologous expression studies [34][35][36][37]. LPA 5 / GPR92-transfected B103 neuroblastoma cells undergo cytoskeletal contraction (Rho kinase-dependent), Ca 2+ mobilization and an increase in cAMP after LPA addition, whereas control B103 cells do not [35]. The rise in cAMP is not mediated by the canonical Ga s pathway, but most likely involves bc subunits; there is no evidence for LPA 5 coupling to G i pathways [35].
To assess whether the increase in cAMP in LPA-treated B16 cells ( Figure 4B) is mediated by LPA 5 , we monitored cAMP levels in LPA 5 -depleted B16 cells. While the LPA-induced rise in cAMP was not affected in cells treated with non-targeting siRNA, it was absent in LPA 5 knockdown cells ( Figure 7A,B), indicating that LPA 5 mediates the cAMP signal. We next expressed LPA 5 in HEK293 cells, as these cells do not normally show a rise in cAMP upon LPA stimulation ( Figure 7C). In LPA 5 -transfected HEK293 cells, LPA evoked a significant increase in cAMP with kinetics similar to those observed in B16 cells ( Figure 7D).
Taken together, these results demonstrate that LPA 5 is a cAMPelevating receptor, both in its native cellular context and in transfected cells, and they reinforce the notion that the anti-   migratory activity of LPA 5 can be attributed, at least in part, to a rise in cAMP with subsequent activation of PKA.

Comparing LPA to a-MSH and sphingosine 1-phosphate
To confirm that cAMP serves as an anti-migratory signal we used a-MSH, an agonist of Ga s -coupled melanocortin receptors. It has been reported that a-MSH suppresses the metastatic potential of B16 cells but does not affect their proliferation [38]. a-MSH induced a very rapid and prominent increase in cAMP ( Figure 8A,B) and partly inhibited B16 cell migration, regardless of whether a-MSH was present in the lower or upper chamber ( Figure 8C); the latter result clearly differs from LPA action. The anti-migratory effect of a-MSH was less pronounced than observed with LPA, while the rise in cAMP is more rapid and more sustained, similar to what is observed with forskolin. The marked difference between the a-MSHand LPA-induced cAMP signals must reflect a fundamental distinction between the canonical (a s ) and the noncanonical (bc) pathway, respectively, by which GPCRs can stimulate cAMP production. Together, these results support the view that cAMP is an anti-migratory signal for B16 cells. However, a-MSH and forskolin appear to act globally, whereas LPA signals in a polarized manner. Furthermore, LPA triggers PIP3 depletion from the plasma membrane, whereas a-MSH does not ( Figure 4D).
Inhibition of B16 cell migration is also induced by sphingosine 1-phosphate (S1P), acting through the S1P 2 receptor, which has been associated with altered Rac/Rho-kinase signaling [39]. Unlike LPA, however, S1P inhibited B16 cell migration when the agonist was present alone in the upper chamber; furthermore, S1P failed to affect basal cAMP levels in B16 cells ( Figure S2). This suggests that S1P 2 and LPA 5 differ fundamentally in their antimigratory signaling mechanisms.

LPA 5 subcellular localization
Next, we asked whether the polarized response of B16 cells to LPA is attributable to an asymmetrical distribution of LPA 5 . Certain GPCRs show a polarized subcellular localization as exemplified by the P2Y receptor family [40]. In MDCK epithelial cells, a subset of P2Y receptors resides at the basolateral membrane, while other P2Y members are only found at the apical plasma membrane [40]. We examined the distribution of HA-tagged LPA 5 in MDCK, HeLa and B16 cells. However, none of these cell types showed a clear asymmetric distribution of LPA 5 ( Figure S3). We therefore assume that the polarized response of migrating B16 cells to LPA is a reflection of localized effector pathways in the pseudopodia, rather than of localized LPA 5 expression.

Conclusions
How does the cAMP-PKA pathway, with a possible additional role for reduced PIP3 signaling, contribute to the chemorepulsive effect of LPA 5 ? PKA can exert both stimulatory and inhibitory effects on cytoskeletal dynamics and cell migration [41], but its action is very complex since PKA acts on numerous target proteins depending on agonist, cell type and subcellular localization of PKA [42]. In epithelial cells, the PKA catalytic subunit (encoded by PRKCA) emerged as an anti-migratory gene from RNAi screens [43,44]. In fibroblasts and carcinoma cells, the cAMP-PKA pathway can inhibit cell migration by interfering with lamellipodia formation [45]. Other studies show that PKA is enriched in protrusive pseudopodia [46,47], but the specific PKA substrates remain to be identified. Inhibition of cell migration by LPA 5 most likely occurs through PKA-mediated and PIP3-dependent modulation of components in the Rho/Rac and cytoskeletal assembly pathways. Localized PKA activation, along with PIP3 depletion, would then lead to destabilization and retraction of extending pseudopodia.
Chemoattractants and repellents cooperate during tumor cell migration and metastasis, and hence the outcome of cell migration is dictated by the relative expression levels of individual receptors. While the precise mechanisms underlying LPA 5 -mediated che- morepulsion remain to be determined, the present findings warrant further studies on the expression patterns and clinical outcome of LPA 5 in melanoma and other cancers. By inhibiting cell migration, LPA 5 -selective agonists might have therapeutic potential for those tumors in which LPA 5 is a predominant receptor. Caution will be needed, however, since LPA 5 signaling may also impact on platelet and mast cell functions [32,33,48].

Materials and Methods
Cell culture and materials

RNA isolation and real-time PCR
Total RNA was extracted using RNeasy Micro Kit (Qiagen). cDNA was synthesized by reverse transcription from 2 mg RNA using oligodT 15 primer and SSII RT enzyme (InVitrogen). RT-qPCR was performed on a 7500 Fast System (Applied Biosystems) as follows: 95uC for 2 min., 95uC for 10 min., 40 cycles at 95uC for 15 sec. followed by 60uC for 1 min. for annealing and extension. The final reaction mixtures (20 ml) consisted of diluted cDNA, 16SYBR Green Supermix (Applied Biosystems), 200 nM forward primer and 200 nM reverse primer. Reactions were performed in 96-well plates in duplicate. Primer sequences are listed in Table S2.

RNA interference and transfections
siRNA pools specific for LPA 5 and LPA 6 and non-targeting siRNAs were purchased from Dharmacon, Inc. B16F10 cells were transfected with 2 mM siRNA using Dharmafect-2 according to the manufacturer's protocol. Cells were used 3 days after transfection. Transfections of LPA 5 and CFP-Epac-YFP plasmids were done using Fugene according to the manufacturer's protocol, using 10 mg of DNA on 25 ml of Fugene for a 10-cm dish with 3610 6 cells in suspension. Transfection efficiency was determined by side-by-side transfection with GFP and was typically around 75%. PH(GRP1)-GFP was transfected into B16F10 cells using Fugene according to the manufacturer's protocol, using 1 mg of DNA on 3 ml of Fugene6 per well on a 6-well plate. Efficiency was typically around 30%.

Cell migration assays
Cell migration was measured using 48-well chemotaxis chambers (Neuro Probe, Inc.) equipped with 8 mm-pore polycarbonate membranes, which were coated with either fibronectin or poly-L-lysine (20 mg/ml). Cells (1610 6 /ml) were added to the upper chamber. Fatty acid-free BSA (1 mg/ml) was used as a lysophospholipid carrier. Migration was allowed for 3 hrs for B16F10 and 24 hrs for HeLa cells at 37uC in humidified air containing 5% CO 2 . Migrated cells were fixed in Diff-Quik Fix and stained using Diff-Quik II. Migration was quantified by color intensity measurements using TINA software.

Western blotting
Cells were washed in ice-cold PBS (phosphate-buffered saline containing 2 mM Ca 2+ and Mg 2+ ), lysed in RIPA buffer with protease inhibitors and spun down. 15 ml of lysate was loaded onto a 4-12% NuPageH Bis-Tris Gel. Electrophoresis was carried out at 200 V for 40 min in NuPageH MOPS running buffer. Protein was transferred to a nitrocellulose membrane using the iBlotH Dry Blotting System. The membrane was blocked for 1 hour at roomtemperature in 50 mg/ml non-fat milk powder in TBST. Incubation with antibodies was done overnight at 4uC. Proteins were visualized using chemiluminescence.

Monitoring cAMP
Intracellular cAMP levels were monitored in real-time using the FRET-based CFP-Epac-YFP sensor, as described [27]. Transfected cells on coverslips were placed on an inverted NIKON microscope and excited at 425 nm. Emission of CFP and YFP was detected simultaneously through 470620 and 530625 nm bandpass filters. Data were digitized and FRET was expressed as the ratio of CFP to YFP signals, the value of which was set to 1.0 at the onset of the experiments. Changes were measured as percentage deviation from this initial value.

PIP3 imaging
Agonist-induced changes in the levels of phosphatidylinositol (3,4,5)-trisphosphate (PIP3) in the plasma membrane were monitored using a biosensor consisting of the PIP3-specific PH domain of GRP1 (general receptor for phosphoinositides-1) fused to GFP, PH(GRP1)-GFP [28]. PH(GRP1)-GFP-transfected B16F10 cells were grown on glass coverslips and serum starved overnight. Live-cell imaging was done on a Leica TCS SP2 confocal microscope equipped with an Ar 488 laser system. Images were taken using a 636 NA 1.32 objective. Coverslips were mounted in a metal ring system and buffer was applied (140 mM NaCl, 5 mM KCl, 2 mM MgCl 2 , 1 mM CaCl 2 , 23 mM NaHCO 3 , 10 mM HEPES, 10 mM glucose). Cells were selected randomly. Agonists were added while images were collected at appropriate time intervals (2-7.5 sec).