Activation of a high affinity Gi protein-coupled plasma membrane receptor by sphingosine-1-phosphate.

Sphingosine-1-phosphate (SPP) has attracted much attention as a possible second messenger controlling cell proliferation and motility and as an intracellular Ca(2+)-releasing agent. Here, we present evidence that SPP activates a G protein-coupled receptor in the plasma membrane of various cells, leading to increase in cytoplasmic Ca2+ concentration ([Ca2+]i), inhibition of adenylyl cyclase, and opening of G protein-regulated potassium channels. In human enbryonic kidney (HEK) cells, SPP potently (EC50, 2 nM) and rapidly increased [Ca2+]i in a pertussis toxin-sensitive manner. Pertussis toxin-sensitive increase in [Ca2+]i was also observed with sphingosylphosphorylcholine (EC50, 460 nM), whereas other sphingolipids, including ceramide-1-phosphate, N-palmitoyl-sphingosine, psychosine, and D-erythro-sphingosine at micromolar concentrations did not or only marginally increased [Ca2+]i. Furthermore, SPP inhibited forskolin-stimulated cAMP accumulation in HEK cells and increased binding of guanosine 5'3-O-(thio) triphosphate to HEK cell membranes. Rapid [Ca2+]i responses were also observed in human transitional bladder carcinoma (J82) cells, monkey COS-1 cells, mouse NIH 3T3 cells, Chinese hamster ovary (CHO-K1) cells, and rat C6 glioma cells, whereas human HL-60 leukemia cells and human erythroleukemia cells failed to respond to SPP. In guinea pig atrial myocytes, SPP activated Gi protein-regulated inwardly rectifying potassium channels. Activation of these channels occurred strictly when SPP was applied at the extracellular face of atrial myocyte plasma membrane as measured in cell-attached and inside-out patch clamp current recordings. We conclude that SPP, in addition to its proposed direct action on intracellular Ca2+ stores, interacts with a high affinity Gi protein-coupled receptor in the plasma membrane of apparently many different cell types.

Over the past years, sphingolipids have emerged as important second messengers of cellular signaling (1)(2)(3). Following stimulation of cells with nerve growth factor, tumor necrosis factor-␣, and interleukin-1␤, sphingomyelinases are activated leading to the generation of ceramide, which can be further metabolized to sphingosine and sphingosine-1-phosphate (SPP) 1 by the action of ceramidase and sphingosine kinase, respectively. Although ceramide and sphingosine have been the subject of extensive studies, recently, attention has also been focused on SPP. This sphingolipid has been demonstrated to be involved in a multitude of processes. Activation of sphingosine kinase and enhanced formation of SPP was shown to be induced by platelet-derived growth factor, with SPP being implicated as an important second messenger for the promotion of DNA synthesis in Swiss 3T3 fibroblasts (4). DNA synthesis and cell division of 3T3 cells could also be noted when SPP was added exogenously to intact 3T3 cells (5). At low concentrations, SPP is also able to inhibit efficiently motility and invasiveness of various tumor cells that cannot be mimicked by sphingosine or N-methylated sphingosines (6). Another important action of SPP has emerged from the observation in various cellular systems that SPP can cause release of Ca 2ϩ from internal stores by a non-inositol 1,4,5-trisphosphate receptormediated mechanism (7)(8)(9). On the other hand, a rapid increase in cytoplasmic Ca 2ϩ concentration ([Ca 2ϩ ] i ) could also be obtained by direct application of SPP to intact Swiss 3T3 cells (7).
Thus, SPP appears to be an important component in the signaling system that is involved in Ca 2ϩ release and in the regulation of cell growth and motility. However, thus far, the molecular targets of SPP have been rather elusive. One possible molecular target may be an intracellular Ca 2ϩ -permeable channel located in the endoplasmic reticulum, which is gated by SPP (7)(8)(9). This channel may be reached by exogenously applied SPP. A unique feature of SPP response on intact cells, however, is the immediate and transitory rise in [Ca 2ϩ ] i (7). We therefore studied whether SPP may also activate a plasma membrane receptor rather than surpassing the plasma membrane to activate a putative intracellular Ca 2ϩ -permeable channel in the endoplasmic reticulum. In the present report, we provide evidence for this hypothesis and demonstrate that SPP at nanomolar concentrations activates pertussis toxin (PTX)-sensitive guanine nucleotide-binding proteins (G proteins) via a plasma membrane receptor apparently present in various cell types.

EXPERIMENTAL PROCEDURES
Materials-Psychosine, sphingosylphosphorylcholine, lysophosphatidic acid (LPA), and bovine serum albumin (lot 62H0154) were from Sigma. SPP and N-palmitoyl-sphingosine were obtained from Biomol. D-erythro-Sphingosine and C 8 -ceramide-1-phosphate were purchased from Calbiochem. Fura-2/AM was from Molecular Probes, PTX was from List Laboratories, and [ 35 S]GTP␥S (1322 Ci/mmol) was from Du-Pont NEN. All other reagents were from previously described sources (10 -14). Stock solutions of SPP (1 mM) were made up in 100% methanol and stored at Ϫ20°C. Prior to the experiments, SPP solutions were dried down, and SPP was dissolved in phosphate-buffered saline as a complex with bovine serum albumin (4 mg/ml) to a final concentration of 1 mM. D-erythro-Sphingosine was dissolved in absolute ethanol (100 mM) and diluted to 1 mM in 4 mg/ml bovine serum albumin. Sphingosylphosphorylcholine was dissolved as a complex in phosphate-buffered saline (1 mM) with bovine serum albumin (4 mg/ml). Dilutions of these stock solutions were in phosphate-buffered saline with 1 mg/ml bovine serum albumin. Bovine serum albumin by itself had no effect on [Ca 2ϩ ] i .
Cell Culture and PTX Treatment-Control human embryonic kidney (HEK) 293 cells and HEK 293 cells stably expressing the human m3 muscarinic acetylcholine receptor subtype were cultured as described in detail previously (10,11). In some experiments, cells were treated for 16 h with 100 ng/ml PTX to test the role of PTX-sensitive G proteins in SPP signaling. Atrial myocytes from hearts of adult guinea pigs were isolated and cultured in bicarbonate-buffered M199 medium as described previously (12). Atrial cells were plated at a low density (several hundred cells per 35-mm culture dish) and used experimentally for up to 8 days.
GTP␥S Binding Assay-Binding of GTP␥S to HEK 293 cell membranes prepared as described before (10) was performed in a reaction mixture (100 l) containing 0.4 nM [ 35 S]GTP␥S, 5 mM MgCl 2 , 1 mM EDTA, 150 mM NaCl, 1 mM dithiothreitol, 1 M GDP, 50 mM triethanolamine-HCl, pH 7.4, and the indicated additions. The reaction started by the addition of membranes (about 5-10 g of protein/tube) was conducted for 30 min at 30°C. Separation of membrane-bound and free radioactivity and determination of nonspecific binding were performed as described before (10).
Measurement of cAMP Accumulation-HEK 293 cells grown to 90 -100% confluency on 35-mm plates were preincubated for 20 min at 37°C with 5 mM theophylline to inhibit phosphodiesterases. Then, cells were stimulated for 5 min with 50 M forskolin plus 10 M SPP or vehicle. The reaction was stopped, and cAMP was measured as described previously (13). Protein was measured by the Lowry method.
[Ca 2ϩ ] i Measurements-[Ca 2ϩ ] i was determined with the fluorescent calcium indicator dye Fura-2 in a Perkin Elmer LS-5B spectrofluorimeter equipped with a fast-filter device as described before (14). Briefly, cells resuspended at approximately 0.5-1.0 ϫ 10 6 cells/ml were incubated with 1 M Fura-2/AM for 1 h at 37°C in a buffer containing 137 mM NaCl, 2.7 mM KCl, 0.9 mM CaCl 2 , 0.5 mM MgCl 2 , 6.5 mM Na 2 HPO 4 , 1.5 mM KH 2 PO 4 , 1 mg/ml bovine serum albumin, and 1 mg/ml D-glucose at pH 7.4. Thereafter, cells were washed twice, resuspended in fresh buffer, and used for fluorescence measurements within the next hour. To remove dye having leaked into the medium, aliquots were pelleted in a microcentrifuge, resuspended in fresh prewarmed medium without bovine serum albumin, and immediately transferred to a thermostatted cuvette (37°C) in the spectrofluorimeter. Excitation was alternating at 340 and 380 nm with emission being read at 495 nm. Fluorescence data were converted into Ca 2ϩ concentration with software supplied by the manufacturer. In some experiments, extracellular Ca 2ϩ was chelated by addition of 5 mM EGTA 30 s prior to agonist exposure (corresponding to 17 nM free Ca 2ϩ ).

Measurement of I K(ACh) Channels in Atrial
Myocytes-Activity of inwardly rectifying K ϩ channels (I K(ACh) ) in guinea pig atrial myocytes was measured using the cell-attached and the inside-out configuration of the patch clamp technique (15) as described in detail previously (12,16). In brief, pipettes fabricated from borosilicate glass with filament (Clark, Pangbourne, UK) were filled with a solution containing 150 mM KCl, 2 mM CaCl 2 , and 10 mM HEPES/KOH, pH 7.4. DC resistance of the pipettes was 2-4 megaohm. Current measurement was performed by means of a patch clamp amplifier (LM/EPC-7, List, Darmstadt, Germany). Current signals were passed through an analog filter (corner frequency, 3 kHz) and digitally stored on the hard disc of an AT computer. The computer was equipped with a hardware/software package (ISO2 by MFK, Frankfurt/Main, Germany) for voltage control, data acquisition, and data evaluation. Experiments were performed at ambient temperature (22-24°C). The bath in cell-attached measurements contained 120 mM NaCl, 20 mM KCl, 2 mM CaCl 2 , 1 mM MgCl 2 , and 10 mM HEPES/KOH, pH 7.4. For recordings from inside-out patches, this was replaced by HEPES-buffered 150 mM KCl solution supplemented with 1 mM EGTA and 1 mM ATP to prevent opening of ATP-dependent potassium channnels. Single channel activity was assessed by calculating from the current traces n*P o (n, number of channels; P o , open state probability).

Cellular Responses to SPP in HEK 293 Cells-Consistent
with studies in Swiss 3T3 cells (7), addition of SPP (1 M) induced a rapid increase in [Ca 2ϩ ] i in HEK 293 cells within 5 s of addition (Fig. 1). The increase in [Ca 2ϩ ] i was transient and within 40 -60 s returned to constant levels, which were slightly above initial basal levels. To study whether SPP mobilizes Ca 2ϩ from intracellular stores or induces influx of extracellular Ca 2ϩ , changes in [Ca 2ϩ ] i in response to SPP were measured following the chelation of extracellular Ca 2ϩ (Fig. 1, upper  panel). Inclusion of EGTA (5 mM) 30 s prior to stimulation with SPP (1 M) reduced [Ca 2ϩ ] i peak increases by 30 Ϯ 9% (mean Ϯ S.E., n ϭ 4) and eliminated the plateau phase. Thus, SPP predominantly increases [Ca 2ϩ ] i by releasing Ca 2ϩ from intracellular stores. To explore the potential involvement of PTXsensitive G proteins in SPP-induced [Ca 2ϩ ] i increase, we pretreated HEK 293 cells with PTX (100 ng/ml, 16 h). As shown in Fig. 1 (lower panel), pretreatment with PTX abolished to a large extent (70 Ϯ 12%, n ϭ 3) SPP-induced [Ca 2ϩ ] i elevation in HEK 293 cells.
To determine the potency and specificity of SPP's action, we performed concentration response experiments with SPP and a variety of related sphingolipids. SPP increased [Ca 2ϩ ] i in HEK 293 cells with an EC 50 value of 2.04 Ϯ 0.87 nM (mean Ϯ S.E.), and maximal increases in [Ca 2ϩ ] i (300 -400 nM) were observed at 1-10 M (Fig. 2). The slope of the concentration response curve was rather shallow, with a calculated Hill coefficient of 0.44 Ϯ 0.05. Out of the other sphingolipids studied, ceramide-1-phosphate, psychosine, and N-palmitoyl-sphingosine at 10 M did not alter [Ca 2ϩ ] i (data not shown), and D-erythro-sphingosine at 10 M only marginally elevated [Ca 2ϩ ] i . In contrast, sphingosylphosphorylcholine increased [Ca 2ϩ ] i in HEK 293 cells to the same maximal level as SPP. Responses to sphingosylphosphorylcholine were as fast as those to SPP but were found at much higher concentrations (EC 50 , 457 Ϯ 31 nM). Similar to the SPP response, [Ca 2ϩ ] i increases induced by sphingosylphosphorylcholine were largely blunted by PTX pretreatment (data not shown). LPA shares some structural similarity to SPP, i.e. a long hydrocarbon chain with a terminal phosphate group, and is known to exert many SPP-like effects (17). Therefore, we tested whether SPP acts by activating LPA receptors. For this, we investigated whether [Ca 2ϩ ] i transients induced by LPA were affected by a prior challenge with SPP. As shown in Fig. 3, this was not the case. Yet, the [Ca 2ϩ ] i response to SPP (1 M) was abolished following a first SPP challenge. In the reverse order of addition, the SPP response was not affected following a challenge with LPA (1 M), while prior stimulation with LPA abolished the effect of a second LPA challenge. From these results, it can be concluded that SPP does not increase [Ca 2ϩ ] i by activating the LPA receptor.
Since the [Ca 2ϩ ] i -elevating action of SPP was largely reduced by PTX treatment, we studied whether SPP by activating PTXsensitive G proteins inhibits adenylyl cyclase. As shown in Fig.  4, SPP (10 M) decreased forskolin-stimulated cAMP accumulation in HEK 293 cells by 30 -40%. This inhibitory response was completely blocked by pretreating cells with PTX (100 ng/ml, 16 h).
Finally, we determined whether SPP activates G proteins, by measuring binding of the labeled stable GTP analog GTP␥S to membranes of HEK 293 cells (Fig. 5). SPP (1 M) increased binding of [ 35 S]GTP␥S to membranes of HEK 293 cells by about 50%. An increase of similar magnitude (about 90%) was observed in response to carbachol (1 mM), which activates the m3 muscarinic acetylcholine receptor stably expressed in the same HEK 293 cells as described before (10). This HEK 293 cell clone showed identical [Ca 2ϩ ] i responses to SPP as untransfected HEK 293 cells.

SPP-induced [Ca 2ϩ ] i Responses Are Cell
Type-specific-The data thus far presented suggested that SPP activates in HEK 293 cells a plasma membrane receptor coupled to PTX-sensitive G proteins, although a direct G protein activation could not be excluded. To corroborate the receptor hypothesis, we therefore set out to test a variety of other cell types for their ability to generate a rapid, transient increase in [Ca 2ϩ ] i in response to SPP. In addition to HEK 293 cells, SPP (10 M) increased [Ca 2ϩ ] i in very divergent cell types, such as human transitional bladder carcinoma (J82) cells, monkey COS-1 cells, mouse NIH 3T3 cells, Chinese hamster ovary (CHO-K1) cells, and rat C6 glioma cells (Table I). Notably, however, differentiated human leukemia (HL-60) cells and human erythroleukemia cells failed to respond to 10 M SPP but were otherwise strongly responsive to known receptor agonists, e.g. N-formyl-methionylleucyl-phenylalanine and thrombin (data not shown).
To study whether the PTX sensitivity of the [Ca 2ϩ ] i response was cell type-dependent, we chose to test in NIH 3T3 fibroblasts whether PTX-sensitive G proteins are involved in SPPinduced [Ca 2ϩ ] i increases as well. While in untreated NIH 3T3 cells SPP (10 M) elevated [Ca 2ϩ ] i by 130 nM, in PTX-pretreated NIH 3T3 cells SPP-induced [Ca 2ϩ ] i increase was completely abolished (data not shown).
SPP Activates I K(ACh) in Atrial Myocytes-After finding activation of PTX-sensitive cellular responses, we tested whether SPP activates muscarinic acetylcholine receptor-regulated, inwardly rectifying potassium channels (I K(ACh) ) in atrial myo- cytes. Receptor regulation of these channels is mediated by PTX-sensitive G i proteins (18 -20). SPP is a potent and efficient activator of I K(ACh) in guinea pig atrial myocytes, as studied in whole cell recording configurations, and its effect is fully PTXsensitive (21). This experimental system was, therefore, used to address the question of whether the action of SPP is brought about strictly from the extracellular face of the plasma membrane, where the interaction of SPP with its putative receptor should take place. In the cell-attached mode, a membrane channel under the mouth of the pipette is accessible to signals from outside this area only if these can transverse the plasma membrane (15,16). Thus, if SPP were producing its effect by direct interaction with G proteins, it should cause activation of I K(ACh) channels if applied to the cell outside the membrane patch isolated by the pipette, whereas a strictly extracellular action should fail to produce activation of I K(ACh) channels under the pipette. As illustrated in Fig. 6, which shows a slow speed recording of basal I K(ACh) activity, superfusion of the cell with a solution containing 1 M SPP for about 15 s had no effect on channel activity. The mean activity, expressed as n*P o , was 0.25 Ϯ 0.12 before and 0.27 Ϯ 0.19 (n ϭ 4) during superfusion of the cells with 1 M SPP. This result strongly argues against  a direct interaction with G proteins reached after transversing the plasma membrane.
Further evidence for a receptor-mediated action of SPP was obtained from single channel recordings in the inside-outside mode. When the pipette solution, facing the original outer face of the plasma membrane, was supplemented with 1 M SPP, exposure of the inside of the membrane to GTP (50 M) resulted in a dramatic increase in channel activity (Fig. 7, trace A). In the absence of SPP, exposure to GTP alone caused a slight increase in channel activity (trace B), which is due to the fact that also basal activity of this channel is G protein-regulated (22). Notably, the same low degree of channel activity was obtained when GTP (50 M) and SPP (1 M) were applied simultaneously to the intracellular face of the isolated patch (trace C). Peak values (n ϭ 4) for n*P o upon exposure of the patches to GTP were 4.6 Ϯ 3.5 (A), 0.31 Ϯ 0.15 (B), and 0.26 Ϯ 0.19 (C). In summary, the patch clamp data clearly demonstrate that SPP acts only from the extracellular face of the plasma membrane and requires GTP for activation of I K(ACh) channels. DISCUSSION In the present study, we tested the hypothesis that exogenous SPP acts on intact cells via activation of a plasma membrane receptor. We demonstrate that SPP rapidly increases [Ca 2ϩ ] i in various cells, inhibits forskolin-stimulated adenylyl cyclase in HEK 293 cells, and activates I K(ACh) channels in atrial myocytes. These responses were either to a large extent or completely PTX-sensitive, indicating activation of G i -type G proteins. Since SPP is a lipid that may be readily taken up by cells, it was important to determine whether SPP activates G proteins directly or requires a plasma membrane receptor to activate G proteins. We could demonstrate in atrial myocytes that SPP only acts from the extracellular face of the plasma membrane and is not able to activate G i -regulated I K(ACh) channels when applied to the intracellular face of the plasma membrane. We therefore conclude that SPP acts by binding to a plasma membrane receptor that couples predominantly to PTX-sensitive G proteins. Consistent with this conclusion is the observation that the calcium responses to SPP varied strongly from one cell type to another, and in some cell types no distinct calcium response to SPP was noted at all.
During the course of this study, the groups of Spiegel and Sturgill (23,24) reported that exogenous SPP decreases cellular cAMP levels, stimulates inositol phosphate formation, increases [Ca 2ϩ ] i , activates mitogen-activated protein kinase, and stimulates DNA synthesis in Swiss 3T3 fibroblasts and that these cellular actions of SPP are largely or completely abolished by PTX pretreatment. These observations led the authors to conclude that SPP may selectively activate PTXsensitive G proteins in a receptor-independent fashion or alternatively activate a specific cell surface receptor that is coupled to these G proteins. Although the data reported in Swiss 3T3 cells are, at least partially, in agreement with those reported herein, there are also important differences. First, the effects of  6. Lack of an effect of SPP on I K(ACh) channel activity in atrial myocyte-attached membrane patch. Channel current was measured using an approximately symmetrical K ϩ distribution across the patch, whereas the bath solution contained 20 mM K ϩ to "clamp" the resting potential (E R ) at around Ϫ50 mV. Membrane potential across the patch was E R Ϫ60 mV, resulting in a unitary inward current through open K ϩ channels of approximately 4 pA. This experimental condition was used because of the strongly inward-rectifying properties of I K(ACh) channels, which pass hardly any current in the outward direction. Several observations suggest that SPP binds to a specific SPP receptor in the plasma membrane. SPP was effective in the nanomolar range (EC 50 of 2 nM), whereas various other sphingolipids, including ceramide-1-phosphate, D-erythrosphingosine, psychosine, and N-palmitoyl-sphingosine, did not or only marginally increase [Ca 2ϩ ] i in HEK 293 cells, even at micromolar concentrations. Likewise, the structurally related phospholipid LPA, which effectively elevated [Ca 2ϩ ] i at nanomolar concentrations, did not desensitize the SPP-induced [Ca 2ϩ ] i response in HEK 293 cells at micromolar concentrations. This lack of cross-desensitization for induced [Ca 2ϩ ] i transients in HEK 293 cells is in agreement with a similar study in A431 cells described recently (25). On the other hand, sphingosylphosphorylcholine fully mimicked the SPP response, although about 200-fold higher concentrations of sphingosylphosphorylcholine than of SPP were required for increasing [Ca 2ϩ ] i in HEK 293 cells. This sphingolipid also mimics SPP's actions on Ca 2ϩ release from endoplasmic reticulum, however apparently at similar concentrations as of SPP (8,9). The shallow concentration response curve of SPP compared to the rather steep one of sphingosylphosphorylcholine for increasing [Ca 2ϩ ] i in HEK 293 cells may even suggest that distinct types of receptors are involved in SPP's actions.
The synthesis of SPP from sphingosine is catalyzed by the enzyme sphingosine kinase. However, present information about this enzyme is rather scarce. Sphingosine kinase appears to be a cytosolic enzyme in platelets while being membraneassociated in rat brain and other tissues (26). Since our study indicates that exogenous SPP acts at a plasma membranelocated receptor, it will be of interest to study how SPP synthesis is regulated by hormones and growth factors, how SPP is released from cells, and what physiological functions SPP plays in different tissues.
In conclusion, this study demonstrates that SPP can regulate intracellular second messengers and membrane channels through activation of specific receptors apparently present in many different cell types and coupled to PTX-sensitive G i -type G proteins. SPP has also been implicated as an intracellular second messenger releasing Ca 2ϩ from internal stores. Thus, the present study suggests that SPP has at least two molecular targets of action, i.e. the proposed sphingolipid-gated Ca 2ϩpermeable channel in the endoplasmic reticulum as well as a high affinity G i protein-coupled receptor in the plasma membrane.