Protein Kinase A-mediated Phosphorylation of Serine 357 of the Mouse Prostacyclin Receptor Regulates Its Coupling to G s -, to G i -, and to G q -coupled Effector Signaling*

The prostacyclin receptor (IP) is primarily coupled to G (cid:1) s -dependent activation of adenylyl cyclase; however, a number of studies indicate that the IP may couple to other secondary effector systems perhaps in a species-specific manner. In the current study, we investigated the specificity of G protein:effector coupling by the mouse (m) IP overexpressed in human embryonic kidney 293 cells and endogenously expressed in murine erythroleukemia cells. The mIP exhibited efficient G (cid:1) s coupling and concentration-dependent increases in cAMP generation in response to the IP agonist cicaprost; however, mIP also coupled to G (cid:1) i decreasing the levels of cAMP in forskolin-treated cells. mIP coupling to G (cid:1) i was pertussis toxin-sensitive and was dependent on protein kinase (PK) A activation status. In addition, the mIP coupled to phospholipase C (PLC) activation in a pertussis toxin-insensitive, G (cid:1) i -, G (cid:2)(cid:3) -, and PKC-inde-pendent but in a G (cid:1) q - and PKA-dependent manner. Whole cell phosphorylation assays demonstrated that the mIP undergoes cicaprost-induced PKA phosphorylation. mIP S357A , a site-directed mutant of mIP, efficiently coupled to G (cid:1) pADVA (10 (cid:3) g/10-cm dish). The IP 3 levels produced were determined using the IP 3 -binding protein assay (29, 30). Levels of IP 3 produced by ligand-stimulated cells over basal stimulation in the presence of HBS were expressed in pmol of IP 3 /mg of cell protein (cid:5) S.E., and results are presented as -fold stimulation over basal (-fold increase (cid:5) S.E.). The data presented are representative of four independent experiments, each performed in duplicate. Measurement of Intracellular [Ca 2 (cid:1) ] Mobilization— Measurements of [Ca 2 (cid:1) ] i in Fura2/AM-preloaded cells were carried out essentially as described previously (26). To investigate the effect of PTx, cells were preincubated with PTx (50 ng/ml) at 37 °C in 5% CO 2 for 16 h prior to stimulation with cicaprost (1 (cid:3) M ). Where appropriate, the PKA (H-89, 10 (cid:3) M ) or PKC (GF 109203X, 50 n M ) kinase inhibitors were added 2 min prior to stimulation with cicaprost (1 (cid:3) M ). The phosphatidylinositol-specific phospholipase C (PI-PLC) (U73122, 1 (cid:3) M ) and phosphatidylcho- line-specific phospholipase C (D-609, 10 (cid:3) M ) inhibitors were added 5 min prior to stimulation of cells with cicaprost (1 (cid:3) M ). Drugs and inhibitors, with stock solutions dissolved in ethanol or dimethyl sulfox-ide, were diluted in HBSSHB (modified Ca 2 (cid:1) /Mg 2 (cid:1) -free Hanks’ buff- ered salt solution containing 20 m M HEPES, pH 7.67, 0.1% bovine serum albumin plus 1 m M CaCl 2 ) to the appropriate concentration such that the addition of 20 (cid:3) l of the diluted drug/inhibitor to 2 ml of cells resulted in the correct working concentration. In separate decreases in adenylyl cyclase activity and, in turn, decreases in cAMP generation. PKA-mediated mIP phosphorylation and coupling to PTx-sensitive G (cid:2) i may be inhibited by H-89. c , alternatively PKA-mediated phosphorylation of mIP at Ser 357 may pro-mote mIP:G (cid:2) q coupling leading to activation of PI-PLC and, in turn, leading to increases in IP 3 generation and mobilization of [Ca 2 (cid:1) ] i . PKA-mediated mIP phosphorylation and coupling to G (cid:2) q may be inhib- ited by H-89, and the PI-PLC inhibitor U73122 blocks mIP-mediated IP 3 generation and mobilization of [Ca 2 (cid:1) ] i .

The prostanoid prostacyclin (prostaglandin I 2 ), mainly produced by the vascular endothelium, plays a key role in the local control of vascular hemostasis acting as a potent inhibitor of platelet aggregation and as a vasodilator (1,2). The actions of prostacyclin generally counteract those of thromboxane A 2 , and thus, the relative levels of these two prostanoids in the circulation are central to the maintenance of vascular hemostasis and vessel tone (3). Prostacyclin also exhibits proinflammatory and antiproliferative properties in vitro (4,5) and may confer a cytoprotective effect against tissue injury during acute myocardial ischemia or in response to hypoxia (6).
Prostacyclin signals through interaction with its signature G protein-coupled receptor (GPCR) 1 termed IP (7). The prostacyclin receptor (IP) is primarily coupled to activation of adenylyl cyclase via G s (3,8,9). Additional evidence indicates that the IP may couple to multiple G protein:effector systems including G q -dependent activation of PLC and to mobilization of intracellular calcium (10 -13). Iloprost, a stable carbacyclin analogue of prostacyclin, can stimulate the opening of ATP-sensitive K ϩ channels resulting in hyperpolarization and relaxation of the canine carotid artery (14). The IP agonists iloprost and cicaprost were shown to stimulate the GTPase activity of both G s and G i suggesting that the IP expressed in human erythroleukemia cells may potentially lead to activation and inhibition of adenylyl cyclase, respectively (15). On the other hand, the IP expressed in the rat medullary thick ascending limb coupled exclusively to G i inhibiting adenylyl cyclase as opposed to G s (16).
We have recently established that the IP may be unique among GPCRs in that it is isoprenylated (12). Whereas isoprenylation of mouse (m) IP is not required for ligand binding, it is absolutely required for its activation of adenylyl cyclase and for its efficient coupling to PLC (12,13). In addition, a number of studies have confirmed that the IP undergoes rapid agonistinduced receptor phosphorylation, internalization/sequestration, and down-regulation in human platelets and other cell types (10,(17)(18)(19)(20).
Whereas the primary manifestation of GPCR phosphorylation and sequestration was classically thought to be desensitization or dampening of the signal following agonist engagement, there is increasing evidence that this is not exclusively so (21). In fact, it is now widely held, for certain GPCRs at least, that "desensitization of the primary signaling event" may trig-ger the "activation of (an)other secondary signaling cascade(s)" (21). For example, for many GPCRs, receptor phosphorylation and subsequent internalization has been shown to be essential for GPCR activation of the mitogen-activated protein kinase cascades and for transactivation of growth factor receptors (22). Moreover, in a limited number of cases, agonist-mediated receptor phosphorylation has been reported to trigger molecular switching whereby a receptor alters its coupling from one G protein:effector system to a different G protein:effector system (23,24). Such molecular switching can facilitate GPCR coupling to more than one G protein:effector system (21).
The molecular mechanisms whereby a single IP can couple to more than one G protein:effector system has not been explored in detail. Thus, in the present study we investigated the G protein coupling characteristics of the mIP stably overexpressed in human embryonic kidney (HEK) 293 cells and in murine erythroleukemia (MEL) cells, which endogenously express mIP. Our studies confirmed that mIP, in addition to its coupling to G␣ s , may also couple G␣ i and G␣ q through a mechanism involving initial G␣ s signaling and cAMP-dependent protein kinase (PK) A phosphorylation of mIP at Ser 357 within its C-tail region; thereafter the phosphorylated mIP exhibited coupling to pertussis toxin (PTx)-sensitive G␣ i and inhibition of adenylyl cyclase and to G␣ q -mediated activation of PLC. These studies provide essential evidence demonstrating that agonist activation of the classical G s -coupled mIP can trigger its phosphorylation and, in turn, favor its signaling to G i -and G qcoupled signal transduction cascades.

Materials and Methods
Site-directed Mutagenesis of the mIP-Conversion of Ser 357 of the mIP to Ala 357 , herein designated mIP S357A , was performed using the Stratagene Quick Change site-directed mutagenesis kit using pHM: mIP (12) as template and oligonucleotides 5Ј-CTT TCC AGA CCT GCA GCG GGG AGA AGA GAC C-3Ј (sense primer) and 5Ј-G GTC TCT TCT CCC CGC TGC AGG TCT GGA AAG-3Ј (antisense primer, the sequence complimentary to mutator Ser (TCG) to Ala (GCG) codon is in boldface italics). The plasmid was verified by double-stranded DNA sequencing using Sequenase Version 2.0 (United States Biochemical Corp.). The plasmid pRK5⅐␤ARK1-(495-689) encoding the carboxyl-terminal amino acid residues 495-689 of ␤ARK1 (25) was kindly provided by Prof. Robert Lefkowitz, Howard Hughes Medical Institute, Duke University Medical Center. The plasmid pCMV⅐G␣ i was generated by subcloning the full-length cDNA for rat G␣ i-1 into the EcoRI site of pCMV5. The plasmids pCMV⅐G␣ q and pCMV⅐G␣ s have been described previously (12,26).
Cell Culture and Transfections-MEL clone 707 cells and HEK 293 cells were obtained from the American Type Culture Collection and were maintained at 37°C in 5% CO 2 . MEL cells were routinely cultured in Dulbecco's modified Eagle's medium, 20% fetal bovine serum. HEK 293 cells were cultured in minimal essential medium with Earle's salts and 10% fetal bovine serum.
HEK 293 cells were transfected with 10 g of pADVA and 25 g of pCMV-or pHM-based vectors using the calcium phosphate/DNA coprecipitation procedure (26). For transient transfections, cells were harvested 48 h after transfection. HEK.mIP, HEK.HAmIP, HEK.mIP SSLC , and HEK.␤ 2 AR cell lines have been previously described (12). To create the HEK.HAmIP S357A stable cell line, HEK 293 cells were transfected with 10 g of ScaI-linearized pADVA plus 25 g of PvuI-linearized pHM:mIP S357A . Forty-eight hours post-transfection, G418 (0.8 mg/ml) selection was applied. After ϳ21 days, G418-resistant colonies were selected, and individual HEK.HAmIP S357A pure clonal stable cell lines or isolates were examined for IP expression by radioligand binding. All isolates used throughout the study were derived from pure clonal stable cell lines.
Measurement of cAMP-cAMP assays were carried out as described previously (12). Briefly, cells were harvested by scraping and were washed three times in ice-cold phosphate-buffered saline. Cells (ϳ1-2 ϫ 10 6 cells) were resuspended in 200 l of HEPES-buffered saline (HBS; 140 mM NaCl, 4.7 mM KCl, 2.2 mM CaCl 2 , 1.2 mM KH 2 PO 4 , 11 mM glucose, 15 mM HEPES-NaOH, pH 7.4) containing 1 mM 3-isobutyl-1methylxanthine and were preincubated at 37°C for 10 min. Thereafter ligands (50 l) were added, and cells were stimulated at 37°C for 10 min in the presence of the ligand (1 M cicaprost, 10 M forskolin, 1 M cicaprost plus 10 M forskolin, 10 M isoproterenol, or 10 M isoproterenol plus 10 M forskolin). For concentration-response studies, cells were stimulated with 1 pM-1 M cicaprost. As controls, cells were incubated in the presence of 50 l of HBS in the absence of ligand. To investigate the effect of PTx, cells were preincubated with PTx (50 ng/ml) for 16 h prior to stimulation with the respective ligand. To investigate the effect of kinase inhibitors on cAMP generation, cells were preincubated in the presence of GF 109203X (50 nM), H-89 (10 M), or vehicle (HBS) at 37°C for 10 min prior to stimulation with the respective ligand. In separate experiments, to examine the effect of co-transfection of G␣ s on cAMP generation, HEK 293, HEK.mIP, and HEK.mIP S357A cells were transiently co-transfected with pCMV⅐G␣ s (25 g/10-cm dish) plus pADVA (10 g/10-cm dish).
In each case, cAMP reactions were terminated by heat inactivation at 100°C for 5 min, and the level of cAMP produced was quantified using the cAMP-binding protein assay (12). Levels of cAMP produced by ligand-treated cells over basal stimulation, determined in the presence of HBS, were expressed as pmol of cAMP/mg of cell protein Ϯ S.E. Results are expressed as -fold stimulation relative to basal (-fold increase Ϯ S.E.). Data were analyzed using the unpaired Student's t test. p values of less than or equal to 0.05 were considered to indicate a statistically significant difference.
Measurement of IP 3 Levels-Intracellular IP 3 levels were measured as described previously (29,30). Briefly, cells were harvested, washed twice in ice-cold phosphate-buffered saline and were then resuspended at ϳ5 ϫ 10 6 cells/ml in HBS containing 10 mM LiCl. To investigate the effect of PTx, cells were preincubated with PTx (50 ng/ml) for 16 h prior to harvesting. To investigate the effect of the kinase inhibitors, H-89 (10 M) or GF 109203X (50 nM) was added, and the cells were incubated for 5 min at 37°C in 5% CO 2 prior to harvesting. Cells (200 l) were then preincubated at 37°C for 10 min. Thereafter cells were stimulated for 2 min at 37°C in the presence of cicaprost (1 M), or for concentrationresponse studies, cells were stimulated with 1 nM-10 M cicaprost. To determine basal IP 3 levels, cells were incubated in the presence of an equivalent volume (50 l) of the vehicle HBS. To examine the effect of co-transfection of G␣ q on IP 3 generation, HEK.mIP cells were transiently co-transfected with pCMV⅐G␣ q (12) (25 g/10-cm dish) and pADVA (10 g/10-cm dish). The IP 3 levels produced were determined using the IP 3 -binding protein assay (29,30). Levels of IP 3 produced by ligand-stimulated cells over basal stimulation in the presence of HBS were expressed in pmol of IP 3 /mg of cell protein Ϯ S.E., and results are presented as -fold stimulation over basal (-fold increase Ϯ S.E.). The data presented are representative of four independent experiments, each performed in duplicate.
Measurement of Agonist-mediated IP Phosphorylation in Whole Cells-Agonist-mediated IP phosphorylations in whole HEK.HAmIP (isolate 2 cells) and HEK.HAmIP S357A cells were carried out essentially as described previously (29). Briefly, cells were washed once in phosphate-free Dulbecco's modified Eagle's medium containing 10% dialyzed fetal bovine serum and were metabolically labeled for 1 h in the same media (1.5 ml/10-cm dish) containing 100 Ci/ml [ 32 P]orthophosphate (8,000 -9,000 Ci/mmol) at 37°C in 5% CO 2 . Where appropriate, H-89 (10 M), GF 109203X (50 nM), or the vehicle HBS was added for the duration of the labeling period. Thereafter 1 M cicaprost or the vehicle HBS was added for 10 min at 37°C in 5% CO 2 . Metabolic labeling of cells was terminated by transferring the dishes to ice and aspiration of the medium. Thereafter cells were quickly washed once in ice-cold phosphate-buffered saline (2 ml/dish) and were lysed with 0.6 ml of Radioimmune Precipitation Buffer (50 mM Tris-HCl, pH 8.0, 150 mM NaCl, 1 mM EDTA, 1% Nonidet P-40 (v/v), 0.5% sodium deoxycholate (w/v), 0.1% SDS (w/v) containing 10 mM sodium fluoride, 25 mM sodium pyrophosphate, 10 mM ATP, 1 g/ml leupeptin, 10 g/ml soybean trypsin inhibitor, 1 mM benzamidine hydrochloride, 0.5 mM phenylmethylsulfonyl fluoride, 1 mM sodium orthovanadate). Following a 15-min incubation on ice, cells were harvested and disrupted by sequentially passing through hypodermic needles of decreasing bore size (G20, G21, G23, and G26), and soluble cell lysates were harvested by centrifugation for 15 min at 13,000 ϫ g at room temperature. HA epitope-tagged IP receptors were immunoprecipitated using the anti-HA antibody (101R, 1:300 dilution) at room temperature for 2 h followed by the addition of 10 l of protein G-Sepharose 4B (Sigma) and further incubation at room temperature for 1 h. Immune complexes were collected by centrifugation at 13,000 ϫ g at room temperature for 5 min and were washed three times in 0.3 ml of Radioimmune Precipitation Buffer and were finally resuspended in 1ϫ Solubilization Buffer (10% ␤-mercaptoethanol (v/v), 2% SDS (w/v), 30% glycerol (v/v), 0.025% bromphenol blue (w/v), 50 mM Tris-HCl, pH 6.8; 60 l). Samples were boiled for 5 min and then loaded onto 10% polyacrylamide gels, analyzed by SDS-PAGE, and thereafter electroblotted onto PVDF membranes essentially as described previously (12). Electroblots were then exposed to Eastman Kodak X-Omat XAR film to detect 32 P-labeled proteins. Thereafter blots were subjected to PhosphorImager analysis, and the intensities of phosphorylation relative to basal phosphorylation were determined and were expressed in arbitrary units of intensity relative to basal levels. In parallel experiments, cells were incubated under identical conditions in the absence of [ 32 P]orthophosphate. HAmIP and HAmIP S357A receptors were immunoprecipitated, and immunoblots were screened using the anti-HA 3F10 horseradish peroxidase-conjugated antibody (1:500) to check for quantitative recovery of each receptor type. Immunoreactive proteins were visualized using the chemiluminescence detection system (12).
Co-immunoprecipations of the mIP and Its Associated G Proteins-HEK.HAmIP and HEK.HAmIP S357A cells were transiently co-transfected with the respective pCMV⅐G␣ plasmids encoding G␣ s, G␣ i , and G␣ q (25 g/10-cm dish) and pADVA (10 g/10-cm dish). Approximately 48 h post-transfection, cells were either preincubated for 10 min with 10 M H-89 or, as a control, with an equivalent volume of the vehicle HBS at 37°C in 5% CO 2 prior to stimulation with 1 M cicaprost for 10 min at 37°C in 5% CO 2 . Reactions were terminated, and HA-tagged IP receptors were immunoprecipitated using the anti-HA 101R antibody (1:300), blotted, and analyzed essentially as described previously (12). Samples were resuspended in 1ϫ Solubilization Buffer, boiled for 5 min, then analyzed by SDS-PAGE on 10% polyacrylamide gels, and thereafter electroblotted onto PVDF membranes. Membranes were then screened by immunoblot analysis using the appropriate G␣ s (K-20) (1:3000), G␣ q/11 (C-19) (1:3000), or G␣ i-1 (I-20) (1:500) antisera where horseradish peroxidase-conjugated goat anti-rabbit IgG (1:3000) was the secondary antibody used in each case. Immunoreactive proteins were visualized using the chemiluminescence detection system according to the manufacturer's recommendations.

Data Analyses
Statistical analysis was carried out using the unpaired Student's t test using GraphPad Prism Version 2.0 (GraphPad Software Inc., San Diego, CA). p values of less than or equal to 0.05 were considered to indicate a statistically significant difference.

Mouse Prostacyclin Receptor G Protein Coupling in Response
to Adenylyl Cyclase Activation-In the current study we sought to investigate the specificity of G protein:effector coupling by the mIP and to define the mechanism(s) whereby the mIP may couple to more than one G protein:effector system. In addition, as a comparative control, we investigated signaling by the ␤ 2 AR, which has been reported to switch coupling from G␣ s to G␣ i in a PKA-dependent mechanism (23). Thus, our studies were performed in HEK 293 cells stably overexpressing the mIP (HEK.mIP cells), in MEL cells that endogenously express mIP, and in HEK.␤ 2 AR cells stably overexpressing the human ␤ 2 AR (12). Initially radioligand binding assays established mIP expression levels in HEK.mIP cells and in MEL cells and ␤ 2 AR expression levels in HEK.␤ 2 AR cells (Table I). Thereafter agonist-induced cAMP generation was determined in HEK.mIP cells (Fig. 1A) and in MEL cells (Fig. 1B) or in HEK.␤ 2 AR cells ( Fig. 1C) in response to stimulation of cells with the selective IP agonist cicaprost or the ␤ 2 AR agonist isoproterenol in the absence or presence of the diterpene forskolin to detect G␣ i coupling should it occur. Stimulation of HEK.mIP (p Ͻ 0.0001) and MEL (p Ͻ 0.0001) cells resulted in significant increases in cAMP generation consistent with mIP coupling to G␣ s . Whereas incubation of cells with forskolin resulted in agonist-independent increases in cAMP generation, co-incubation with cicaprost and forskolin caused significant decreases in cAMP in both HEK.mIP (p Ͻ 0.0001) and MEL (p Ͻ 0.0001) cells. These results suggest that the mIP can couple to both G␣ s and G␣ i . Similar results were obtained upon examining cAMP generated by the ␤ 2 AR (Fig. 1C), consistent with previous reports that the ␤ 2 AR may couple to both G␣ s and G␣ i (23).
To determine whether the mIP:G␣ i coupling is dependent on the concentration of ligand used for stimulation, HEK.mIP cells were stimulated with cicaprost (1 pM-1 M) in the absence and presence of forskolin (Fig. 3A). mIP generated significant concentration-dependent increases in cAMP consistent with its coupling to G␣ s . However, in cells co-incubated with forskolin, 1 pM-1 nM cicaprost led to concentration-dependent augmentations of cAMP generation consistent with mIP coupling to G␣ s ; however, at the higher cicaprost concentrations (Ͼ1 nM), cicaprost led to a concentration-dependent reduction in cAMP generation indicative of mIP coupling to G␣ i . In the presence of PTx, cicaprost led to concentration-dependent increases in cAMP generation in the presence (Fig. 3B) and absence of forskolin (data not shown) over the complete range of cicaprost concentrations (1 pM-1 M). Moreover, in the presence of H-89, cicaprost resulted in concentration-dependent increases in cAMP generation in the presence (Fig. 3B) and absence of forskolin (data not shown) over the complete range of cicaprost concentrations providing further evidence that G␣ i coupling by mIP is dependent on PKA activation status.
Mouse Prostacyclin Receptor Coupling to PLC-We extended these studies to investigate the mechanism whereby the G scoupled mIP may also couple to the G␣ q :PLC effector system (12). Stimulation of cells with cicaprost resulted in a significant increase in IP 3 generation in HEK.mIP cells compared with control, nontransfected HEK 293 cells (Figs. 4A and 10A, p Ͻ 0.0009). Preincubation of HEK.mIP cells with PTx or GF 109203X had no significant effect on IP 3 generation (PTx, p Ͼ 0.911; GF 109203X, p Ͼ 0.668); however, preincubation with H-89 significantly decreased cicaprost-induced IP 3 generation (p Ͻ 0.005). Because the levels of cicaprost-stimulated IP 3 were quite low (1.56 Ϯ 0.13-fold increase in IP 3 in response to 1 M cicaprost), HEK.mIP cells were transiently co-transfected with G␣ q to increase the levels of measurable IP 3 (Fig. 4B). Overexpression of G␣ q was confirmed by Western blot analysis using anti-G␣ q -specific antisera (Fig. 4C). The presence of G␣ q augmented (p Ͻ 0.002) and led to a significant increase in cicaprost-induced IP 3 generation (Fig. 4B), and cicaprost-induced IP 3 generation in the presence of G␣ q was not affected by preincubation with either PTx (p Ͼ 0.812) or GF 109203X (p Ͼ 0.896). However, preincubation with H-89 abolished cicaprostmediated IP 3 production (Fig. 4B, p Ͻ 0.0001) such that the levels of IP 3 generation were not significantly different to those levels generated in control, nontransfected HEK 293 cells following cicaprost stimulation (p Ͼ 0.25).
To extend these studies, we also investigated cicaprost-induced PLC activation by monitoring mobilization of [Ca 2ϩ ] i . Consistent with previous studies (12) Thereafter to confirm that the source of cicaprost-induced [Ca 2ϩ ] i mobilization is derived from PLC/IP 3 -regulated intracellular Ca 2ϩ stores, we examined the effect of the PI-PLC inhibitor U73122 and, as a control, the phosphatidylcholinespecific phospholipase C inhibitor D-609 (31, 32) on cicaprostinduced [Ca 2ϩ ] i mobilization. Preincubation of HEK.mIP cells with U73122 abolished [Ca 2ϩ ] i mobilization by the mIP (Fig.  6B, p Ͻ 0.0001), whereas preincubation of the cells with D-609 had no significant effect on cicaprost-induced [Ca 2ϩ ] i mobilization (Fig. 6C, p Ͼ 0.07). As an additional control, preincubation of HEK.mIP cells with U73122 did not affect ionomycin-induced [Ca 2ϩ ] i mobilization ensuring that U73122 did not affect the general measurement of [Ca 2ϩ ] i mobilization in Fura2/AMpreloaded cells per se (p Ͻ 0.001, data not shown).
Whole Cell Phosphorylations of the mIP-To establish whether the mIP may be subject to cicaprost-induced PKA phosphorylation, whole cell phosphorylations of the mIP were investigated in HEK 293 cells stably overexpressing an aminoterminal extracellular HA epitope-tagged mIP (HEK.HAmIP cells). Consistent with previous reports (12,33) and Table I, the presence of the HA epitope tag did not affect the ligand binding characteristics or cell signaling properties of mIP (data not shown). Initially the specificity of the anti-HA 101R antiserum to immunoprecipitate HAmIP from HEK.HAmIP cells and not from HEK 293 cells (Fig. 7C) or HEK.mIP cells (data not shown) was confirmed. Short term exposure of the Western blot revealed the presence of two distinct immunoreactive bands (Fig. 7C, lane 1), one between 46 and 66 kDa and the other between 36 and 46 kDa, corresponding to the glycosylated and nonglycosylated forms of mIP, respectively (12,33), which appeared as a single broad band of 36 -66 kDa following prolonged exposure (data not shown). In 32 P metabolically labeled cells, cicaprost resulted in phosphorylation of a broad protein band of 36 -66 kDa, corresponding to the immunoprecipitated HAmIP in HEK.HAmIP cells (Fig. 7A, lane 2). In contrast, no phosphorylation of mIP occurred in vehicle-treated, nonstimulated HEK.HAmIP cells (Fig. 7A, lane 1) or in cicaprost-stimulated HEK 293 control cells (Fig. 7A, lane 5). Whereas preincubation of HEK.HAmIP cells with the PKC inhibitor GF 109203X did not significantly affect cicaprost-mediated mIP phosphorylation (Fig. 7A, lane 3; p Ͼ 0.05), preincubation of HEK.HAmIP cells with the PKA inhibitor H-89 (Fig. 7B, lane  4) abolished cicaprost-induced HAmIP phosphorylation (p Ͻ 0.05).
The Role of the C-tail of the mIP in Receptor:G Protein Coupling-It has previously been established that the isoprenylation-defective mIP SSLC , in which the critical Cys 414 of mIP was mutated to Ser 414 to generate mIP S414SLC , cannot couple to G␣ s to mediate cAMP increases and cannot efficiently couple to G␣ q :PLC activation (12) implying that the C-tail of the mIP may be important in receptor:G protein coupling. Thus, we sought to investigate whether the isoprenylation-defective mIP SSLC can couple to G␣ i . Consistent with previous studies (12), HEK.mIP SSLC cells exhibited a significantly impaired ability to generate cicaprost-mediated cAMP generation compared with HEK.mIP cells (Fig. 8, compare panel A to panel C).
In fact the levels of cicaprost-induced cAMP generation in HEK.mIP SSLC cells were not significantly different from that produced by the control, nontransfected HEK 293 cells over the range of 1 pM-100 nM cicaprost (data not shown), but at 1 M cicaprost a small, although significant, increase in cAMP generation was observed in HEK.mIP SSLC cells (Fig. 8A, p Ͻ 0.05). Moreover, in the presence of forskolin, cicaprost (1 M) did not result in a diminution of cAMP levels, indicative of G␣ i coupling, in HEK.mIP SSLC cells (p Ͼ 0.65) but rather showed a marginal augmentation in cAMP generation (Fig. 8A). H-89 had no effect on the level of G␣ s or G␣ i coupling by mIP SSLC (Fig. 8B), but consistent with previous data (Fig. 2B) it signif- icantly impaired mIP coupling to G␣ i (Fig. 8D). To exclude the possibility that the inability of the mIP SSLC to generate cAMP is not due to low levels of IP, radioligand binding assays were carried out on HEK.mIP SSLC membranes (Table I). No significant differences were observed between the levels of mIP expression in HEK.mIP and HEK.mIP SSLC cells (p Ͼ 0.0985).
Computational analysis of the C-tail region of the IP revealed a putative PKA phosphorylation site in the mIP sequence where Ser 357 represents the predicted target residue for PKA phosphorylation (34). Thus, to determine the importance of this potential PKA phosphorylation motif, site-directed mutagenesis was used to generate mIP S357A , a variant of mIP in which Ser 357 was converted to Ala 357 , thereby destroying the putative PKA phosphorylation site within the C-tail of mIP. Stable cell lines overexpressing the epitope-tagged (HA) mIP S357A (HEK.HAmIP S357A ) were established and characterized by radioligand binding (Table I). Whereas through transient expression studies it was established that mutation of Ser 357 did not affect the ligand binding characteristics of the mIP S357A compared with the wild type mIP, the level of mIP S357A expression in HEK.HAmIP S357A cells was appreciably lower than that of mIP expressed in HEK.mIP or in HEK-.HAmIP cells (Table I). Thus, to ensure that any potential difference in the signaling behavior of the mIP S357A when compared with the mIP may not be due to simple differences in the relative levels of the respective receptor expression, for further studies we chose to use a second pure clonal cell line or isolate of the HEK.HAmIP cells (HEK.HAmIP isolate 2) that exhibited similar levels of mIP expression to that of the HEK. HAmIP S357A cells (Table I). Both the HEK.HAmIP and HEK.HAmIP S357A cells showed significant cicaprost-mediated, concentration-dependent increases in cAMP generation (Fig.  9A). Moreover, co-transfection of cells with G␣ s (Fig. 9B) significantly augmented cAMP generation in HEK.HAmIP cells (p Ͻ 0.003) and in HEK.HAmIP S357A cells (p Ͻ 0.0001) where G␣ s overexpression was confirmed by Western blot analysis (Fig. 9B, inset). However, whereas mIP S357A exhibited efficient G␣ s coupling (p Ͻ 0.001) unlike that of the mIP isolates, it did not exhibit G␣ i coupling as evidenced by its failure to reduce, but rather augmented, cAMP levels in cells co-stimulated with cicaprost plus forskolin when compared with cells incubated with forskolin alone (Fig. 9C). cAMP generation by the mIP S357A was not affected by the presence of H-89 (Fig. 9D) or GF 109203X (data not shown).
Next the ability of mIP S357A to couple to G␣ q and PLC activation was examined. HEK.HAmIP cells produced a concentration-dependent increase in IP 3 generation upon stimulation with 1 nM-10 M cicaprost (Fig. 10A, p Ͻ 0.001 at 10 M cicaprost). In contrast, neither the HEK.HAmIP S357A or the nontransfected HEK 293 cells showed any significant increases in IP 3 even at the highest concentration used (p Ͼ 0.249 at 10 M and p Ͼ 0.314 at 10 M, respectively). Moreover, co-transfection of cells with G␣ q did not augment cicaprost-mediated IP 3 generation in HEK.HAmIP S357A cells (data not shown). Stimulation of both the mIP and the mIP S357A did lead to significant increases in [Ca 2ϩ ] i mobilization (Fig. 10B); however, the level of [Ca 2ϩ ] i mobilization by the mIP S357A was significantly less than that of the mIP (p Ͻ 0.003) and was only marginally greater than that mobilized in nontransfected HEK 293 cells in response to cicaprost (Ref. 12 and data not shown).
Whole Cell Phosphorylations of the mIP S357A -Whole cell phosphorylations in HEK.HAmIP S357A cells (Fig. 7B)  firmed that HAmIP S357A did undergo increased phosphorylation, represented as a broad phosphoprotein band of 36 -66 kDa, in response to stimulation with cicaprost (Fig. 7B, compare lanes 1 and 2). However, the level of HAmIP S357A phosphorylation was not significantly different from that observed in HEK 293 cells (Fig. 7A, lane 5; p Ͼ 0.05) but was substantially less when compared with that of the HAmIP (Fig. 7,  compare panel A, lane 2 to panel B, lane 2; p Ͻ 0.05) despite the fact that HAmIP S357A was expressed at levels similar to that of the HAmIP (Fig. 7C, lanes 1 and 2; Table I). Moreover, preincubation of HEK.HAmIP S357A cells with GF 109203X or H-89 ( Fig. 7B, lanes 3 and 4, respectively) had no effect on the level of mIP S357A phosphorylation in cicaprost-stimulated cells.
mIP and mIP S357A Receptor:G Protein Interactions-To investigate the G protein interactions of the mIP and the mIP S357A more closely, HEK.HAmIP (Fig. 11A) and HEK.HA-mIP S357A (Fig. 11B) cells were transiently transfected with the cDNA encoding the ␣ subunit of G␣ s (Fig. 11, panels A  and B, lanes 1 and 2), G␣ i (Fig. 11, panels A and B, lanes 3  and 4), or G␣ q (Fig. 11, panels A and B, lanes 5 and 6). Receptor:G protein interactions were examined by immunoprecipitating the HAmIP or HAmIP S357A receptor using the anti-HA 101R antisera followed by screening the immunoprecipitates for co-precipitation of the respective G␣ subunit using anti-G␣-specific antisera. Whereas nonstimulated, vehicle (HBS)-treated cells showed no receptor:G protein interaction (data not shown), stimulation of both HEK.HAmIP and HEK.HAmIP S357A cells with cicaprost (1 M) for 10 min permitted co-precipitation of G␣ s . However, whereas stimulation of HEK.HAmIP cells with cicaprost resulted in co-precipitation of HAmIP with G␣ i and G␣ q , precipitation of neither G␣ i nor G␣ q along with HAmIP S357A was observed. The presence of HAmIP and HAmIP S357A in the immunoprecipitates was confirmed by screening the blots with the anti-HA 3F10 horseradish peroxidase-conjugated antibody (data not shown). Moreover, pretreatment of HEK.HAmIP cells with H-89 blocked co-precipitation of G␣ i (Fig. 11C, lanes 3 and 4) and G␣ q (Fig. 11C, lanes 5 and 6) but had no effect on co-precipitation of G␣ s along with HAmIP (Fig. 11C, lanes  1 and 2). DISCUSSION Prostacyclin is central to the regulation of vascular hemostasis and may also mediate other somewhat diverse cellular actions under both physiologic and pathophysiologic settings (3,8,35). Mice deficient in the IP exhibit altered pain perception and inflammatory reactions as well as exhibiting an impaired response to thrombotic stimuli (4). Whereas IP is primarily coupled to G s -dependent adenylyl cyclase activation, ligand activation of multiple signaling pathways by IP agonists has been identified, such as in mouse BNu2cl3 mast cells (36), in mouse preadipocyte Ob1771 cells (37), in human erythroleu-  A and B, lanes 2 and 5). Additionally, HEK.HAmIP isolate 2 cells and HEK.HAmIP S357A cells were preincubated with 50 nM GF 109203X (panels A and B, lane 3 A and B, lane 4) for 10 min prior to stimulation with 1 M cicaprost for 10 min. Thereafter HA epitope-tagged IP receptors were immunoprecipitated using the anti-HA antibody 101R. Immunoprecipitates were resolved by SDS-PAGE, electroblotted onto PVDF membranes, and then exposed to Xomat XAR-5 film (Kodak) for 8 -10 days. Thereafter blots were subjected to PhosphorImage analysis, and the intensities of cicaprost-mediated mIP and mIP S357A phosphorylation relative to basal phosphorylation in the presence of HBS were determined and expressed in arbitrary units as follows: mIP: 1 M cicaprost, 11. kemia cells (13,15,38), and in human megakaryoblastic leukemia (MEG01) cells (39). Coupling of IP to changes in intracellular Ca 2ϩ , rather than to activation of adenylyl cyclase, has been reported in rabbit cortical collecting ducts (40), whereas the IP expressed in the rat medullary thick ascending limb coupled exclusively to G i -mediated inhibition of adenylyl cyclases rather than to G s -mediated adenylyl cyclase activation (16). Whereas molecular cloning has confirmed the existence of single IPs in human, mouse, rat, and bovine tissues (11,(41)(42)(43), the cloned human and mouse IP are reported to couple to activation of both adenylyl cyclase and PLC␤ effectors (10 -13, 41). To date, IP isoforms are not known to exist unlike other prostanoid receptors, and thus it appears that differential mRNA splicing mechanisms per se cannot account for the mul-tiple patterns of IP signaling within a given species (3,44). Thus, the key objective of the present study was to investigate the molecular mechanisms whereby a single IP may couple to more than one G protein:effector system by examining in detail the signaling properties of the mIP either stably overexpressed in HEK 293 (HEK.mIP) cells or in native MEL cells.

) or 10 M H-89 (panels
Consistent with mIP coupling to G␣ s , stimulation of HEK. mIP and MEL cells resulted in significant cicaprost-induced increases in cAMP generation. However, contrary to previous reports, mIP expressed in both cell types also coupled to inhibition of adenylyl cyclase in a PTx-sensitive, G␣ i -dependent manner (data relating to PTx studies in MEL cells are not shown). The PKA inhibitor H-89 abolished mIP:G␣ i coupling in HEK.mIP cells but had no effect on mIP:G␣ s coupling. On the other hand, the PKC inhibitor GF 109203X had no effect on mIP coupling to either G␣ s or G␣ i . To establish whether the specificity of mIP coupling to G␣ s or G␣ i was concentration-dependent, HEK.mIP cells were stimulated with a range of cicaprost concentrations in the absence and presence of forskolin. In the absence of forskolin, mIP generated significant, concentrationdependent increases in measurable cAMP consistent with its coupling to G␣ s . However, in the presence of forskolin, stimulation of mIP with low concentrations of cicaprost (Ͻ1 nM) generated significant concentration-dependent increases in measurable cAMP consistent with its coupling to G␣ s , whereas at higher concentrations (Ͼ1 nM) cicaprost yielded a concentra-tiondependent reduction in cAMP generation indicative of mIP coupling to G␣ i . Preincubation with PTx or H-89, but not GF 109203X (data not shown), blocked mIP-mediated G␣ i coupling but had no effect on mIP coupling to G␣ s over the range of cicaprost concentrations used. Taken together these data indicate that mIP can couple to both G␣ s and to PTx-sensitive G␣ i to mediate activation and inhibition of adenylyl cyclase, respectively. Coupling of mIP to G␣ i , but not G␣ s , appears to require a critical concentration of cAMP and is dependent on PKA activation.
We extended our studies to investigate the mechanism whereby the G s -coupled mIP may also couple to PLC activation. Cicaprost stimulation of HEK.mIP cells led to significant increases in IP 3 generation and mobilization of [Ca 2ϩ ] i . H-89, but not PTx or GF 109203X, almost completely abolished cicaprostmediated signaling indicating that mIP-mediated IP 3 generation and [Ca 2ϩ ] i mobilization occurred in a PTx-insensitive, PKC-independent, PKA-dependent mechanism(s). Moreover, overexpression of the carboxyl-terminal residues of ␤ARK1, to sequester G␤␥ subunits, confirmed that cicaprost-mediated IP 3 generation (data not shown) or [Ca 2ϩ ] i mobilization were not through G␤␥-stimulated mechanisms such as through G␤␥ regulation of PLC␤ isozymes (45). Consistent with these data, transient overexpression of G␣ q led to significant augmentations of cicaprost-mediated IP 3 generation and mobilization of [Ca 2ϩ ] i (Ref. 12 and data not shown) in a PTx-insensitive, PKC-independent, PKA-dependent mechanism(s). Furthermore, the PI-PLC inhibitor U73122 (32) abolished cicaprostinduced [Ca 2ϩ ] i mobilization confirming that the primary source of mIP-mediated [Ca 2ϩ ] i mobilization is from IP 3 -regulated intracellular stores. Thus, taken together these data indicate that mIP, in addition to coupling to G s and G i , can couple to PLC and that PLC coupling occurs in a G␣ q -mediated, PTxinsensitive, PKC-independent, but PKA-dependent, manner. Additionally, coupling of mIP to both G i and G q is dependent on its prior coupling to G s to stimulate cAMP-dependent PKA activation. Moreover, subsequent whole cell phosphorylation assays established that the mIP is subject to cicaprost-mediated phosphorylation in a GF 109203X-independent, H-89-dependent manner indicating that mIP coupling to G␣ i and G␣ q may involve direct PKA-mediated phosphorylation of the mIP itself.
We have previously established that mIP may be unique among GPCRs in that it is isoprenylated within its C-tail domain (12,13). Disruption of isoprenylation through sitedirected mutagenesis to generate mIP SSLC established that isoprenylation of mIP is required for its efficient G␣ s -and G␣ q -mediated activation of adenylyl cyclase and PLC, respectively, suggesting that the C-tail of the mIP may be involved in receptor:G protein interaction (12). Consistent with previous studies (12), the isoprenylation-defective mIP SSLC exhibited significantly diminished coupling to both adenylyl cyclase and PLC activation compared with the wild type mIP. Moreover, mIP SSLC failed to couple to G␣ i to mediate inhibition of forskolin-induced cAMP generation indicating that (a) the C-tail domain of mIP may be involved in receptor:G␣ i interaction and/or that (b) failure of mIP SSLC to couple to G␣ i may be due to its initial inability to couple to G␣ s to stimulate PKA-mediated mIP phosphorylation and subsequent coupling to G␣ i and indeed to G␣ q .
Analysis of the primary sequence of the mIP revealed the presence of a putative PKA phosphorylation site within the C-tail region of the mIP where Ser 357 represents the predicted target residue for phosphorylation (34). Thus, through the use of site-directed mutagenesis, the role of Ser 357 was investigated. Whereas mIP S357A exhibited G␣ s activation of adenylyl  A and B, lanes 3 and 4), and G␣ q (panels A and B, lanes 5 and 6). Alternatively HEK.HAmIP isolate 2 cells (panel C) transiently co-transfected for 48 h with G␣ s (panel C, lanes 1 and 2), G␣ i (panel C, lanes 3 and 4), and G␣ q (panel C, lanes 5 and 6) were preincubated for 10 min with vehicle (Ϫ) or with 10 M H-89 (ϩ). Thereafter cells were stimulated with 1 M cicaprost for 10 min, and HA epitope-tagged receptors were immunoprecipitated using the anti-HA antibody 101R. Immunoprecipitations were resolved by SDS-PAGE and electroblotted onto PVDF membranes. Membranes were screened for the presence of G␣ s (panels A-C, lanes 1 and 2), G␣ i (Panels A-C, lanes 3 and 4), or G␣ q (panels A-C, lanes 5 and 6) using the respective anti-G␣ subunit antisera essentially as described under "Experimental Procedures," and immunoreactive bands were visualized by chemiluminescence detection. Data presented are representative of three independent experiments. The positions of the molecular mass markers (kDa) are indicated to the right of panels A-C.
cyclase similar to that of the wild type mIP, it failed to couple to G␣ i or to G␣ q to mediate inhibition of adenylyl cyclase and activation of PLC, respectively. The fact that mIP, but not mIP S357A , exhibited cicaprost-mediated G␣ i and G␣ q coupling confirmed that cicaprost is acting through the mIP itself and not through another, perhaps related prostanoid, receptor to mediate these effects. Unlike that which occurred for the wild type mIP, whole cell phosphorylation assays established that mIP S357A did not undergo significant agonist-induced PKA phosphorylation. Finally, cicaprost stimulation of cells overexpressing HA epitope-tagged forms of the mIP and mIP S357A permitted co-immunoprecipitation of G␣ s confirming direct, agonist-induced interaction of both mIP and mIP S357A with G␣ s . On the other hand, whereas stimulation of HEK.HAmIP cells with cicaprost resulted in co-precipitation of HAmIP with both G␣ i and G␣ q , co-precipitation of neither G␣ i nor G␣ q along with HAmIP S357A was observed. Furthermore, in support of this, inhibition of PKA blocked co-precipitation of G␣ i and G␣ q with HAmIP but had no effect on co-precipitation of G␣ s with HAmIP suggesting that for mIP interaction with G␣ i and/or G␣ q to occur mIP must be phosphorylated on Ser 357 by PKA. Thus, taken together the data presented in this study provide a mechanism whereby the classical G␣ s -coupled mIP may couple both to G␣ i -mediated inhibition of adenylyl cyclase and to G␣ q -mediated activation of PLC. This mechanism, as outlined in the model presented in Fig. 12, requires initial mIP coupling to G␣ s leading to elevation of cAMP and PKA activation. Thereafter PKA-mediated phosphorylation of mIP at Ser 357 within the C-tail domain of mIP in turn favors its coupling to G␣ i , leading to inhibition of adenylyl cyclase, and to G␣ q , leading to activation of PLC.
The proposed mechanism of mIP coupling to both G␣ s and to G␣ i is not unlike that which has been previously reported for the ␤ 2 AR-mediated switching from G␣ s to G␣ i (21,23). In their model, the ␤ 2 AR can switch coupling from G␣ s to G␣ i in a mechanism involving initial G␣ s activation of adenylyl cyclase and subsequent PKA phosphorylation within the third intracellular loop of the ␤ 2 AR; the phosphorylated ␤ 2 AR, in turn, exhibited diminished coupling to G␣ s and increased coupling to G␣ i thereby switching its coupling from G␣ s to G␣ i. (21,23,46). In our studies, we identified Ser 357 within the C-tail of the mIP itself as being the critical residue for PKA phosphorylation and, hence, regulating the G␣ s to G␣ i coupling mechanism. Our studies do not rule out the possibility that other components of the signaling pathway may also be targets for the PKA-mediated phosphorylation and contribute to the differential signaling event; however, the fact that mIP S357A can couple efficiently to G s to generate cAMP and activate PKA but not to G i makes it unlikely that other signaling elements are involved. One of the manifestations of the ␤ 2 AR switching its coupling from G s is that the G i -coupled receptor can stimulate activation of the mitogen-activated protein kinase cascades in a PTxsensitive, G␤␥-and Src-mediated cascade (23). Whether a similar mechanism applies to the G i -coupled mIP will require further investigation.
Many G s -coupled GPCRs can activate both cAMP and IP 3 / Ca 2ϩ signaling pathways, and three mechanisms have been proposed to explain this dual G protein coupling. One mechanism is independent GPCR coupling to G s and to a G q member, such as G 15 or G 16 , to initiate independent activation of adenylyl cyclase and PLC␤ (24). The other two mechanisms involve cAMP-regulated PKA activation of PLC␤ or PKA-induced switching of receptor coupling from G s to G i (24). In the latter mechanism, such as that reported to occur for the vasoactive intestinal peptide receptor, both PLC-mediated IP 3 generation and mobilization of [Ca 2ϩ ] i are regulated through PTx-sensitive, G i -derived G␤␥ subunits rather than through G q -coupled mechanisms (24). In our studies investigating mIP, as stated, we established that PKA phosphorylation of Ser 357 within the C-tail of the mIP itself as the critical residue to regulate coupling to both G␣ i and G␣ q ; thereafter the phosphorylated mIP can independently regulate G i -and G q -mediated inhibition of adenylyl cyclase and activation of PLC, respectively. Thus, our mechanism differs from that proposed for the vasoactive intestinal peptide receptor in that mIP-mediated PLC activation is PTx-insensitive, G i -and G␤␥-independent. Moreover, mIP-mediated PLC activation occurs in a PKA-dependent, G q -coupled mechanism suggesting that mIP must undergo PKA-mediated switching from G s to G q to activate PLC activation as opposed to independently coupling to G s and G q members as proposed in the first mechanism (24). Similar to that previously stated for mIP:G␣ i coupling, our studies do not fully exclude the possibility that other components of the signaling pathway, such as PLC␤, may also be targets for the PKA-mediated phosphorylation as proposed in the second mechanism (24); however, the fact that mIP S357A can couple efficiently to G s but does not couple to G q :PLC activation makes it unlikely.
Several GPCRs are known to activate dual or multiple signaling systems, such as the thrombin receptor, the histamine H 2 receptor, or the parathyroid receptor to name but a few (47). In many instances, this can occur in a species-or cell typespecific manner. However, the molecular mechanisms whereby a given GPCR may couple to multiple G protein-coupled effectors are, on the whole, poorly understood. Our studies extend those previously reported for the ␤ 2 AR (23) and for the vasoactive peptide receptor (24) and establish that second messenger kinase-, such as PKA-mediated switching is not restricted to G s /G i mechanisms but may also be extended to dual switching from G s to both G i and G q mechanisms independently. Moreover, our studies add further evidence to the growing awareness that mechanisms such as GPCR phosphorylation and internalization, previously thought to be exclusively asso- FIG. 12. Model of mIP coupling to G␣ s -mediated activation of adenylyl cyclase (AC) (a), G␣ i -coupled inhibition of adenylyl cyclase (b), and G␣ q -coupled activation of PI-PLC (c). a, ligand activation of the mIP stimulates G␣ s -mediated activation of adenylyl cyclase leading to increases in cAMP generation and, in turn, activation of PKA. b, activated PKA phosphorylates the mIP at Ser 357 favoring mIP:G␣ i coupling leading to decreases in adenylyl cyclase activity and, in turn, decreases in cAMP generation. PKA-mediated mIP phosphorylation and coupling to PTx-sensitive G␣ i may be inhibited by H-89. c, alternatively PKA-mediated phosphorylation of mIP at Ser 357 may promote mIP:G␣ q coupling leading to activation of PI-PLC and, in turn, leading to increases in IP 3 generation and mobilization of [Ca 2ϩ ] i . PKA-mediated mIP phosphorylation and coupling to G␣ q may be inhibited by H-89, and the PI-PLC inhibitor U73122 blocks mIP-mediated IP 3 generation and mobilization of [Ca 2ϩ ] i . ciated with the desensitization of GPCR signaling, may trigger activation of other signaling events, such as switching receptor coupling from one G protein-coupled cascade to other G proteincoupled cascade(s), or to initiation of other signaling events, such as activation of mitogen-activated protein kinase cascades and/or transactivation of growth factor signaling (21,22