Palmitoylation of the Human Prostacyclin Receptor

We have previously established that isoprenylation of the prostacyclin receptor (IP) is required for its efficient G protein coupling and effector signaling (Hayes, J. S., Lawler, O. A., Walsh, M. T., and Kinsella, B. T. (1999)J. Biol. Chem. 274, 23707–23718). In the present study, we sought to investigate whether the IP may actually be subject to palmitoylation in addition to isoprenylation and to establish the functional significance thereof. The human (h) IP was efficiently palmitoylated at Cys308 and Cys311, proximal to transmembrane domain 7 within its carboxyl-terminal (C)-tail domain, whereas Cys309 was not palmitoylated. The isoprenylation-defective hIPSSLC underwent palmitoylation but did not efficiently couple to Gs or Gq, confirming that isoprenylation is required for G protein coupling. Deletion of C-tail sequences distal to Val307 generated hIPΔ307 that was neither palmitoylated nor isoprenylated and did not efficiently couple to Gs or to Gq, whereas hIPΔ312 was palmitoylated and ably coupled to both effector systems. Conversion of Cys308, Cys309, Cys311, Cys308,309, or Cys309,311 to corresponding Ser residues, while leaving the isoprenylation CAAX motif intact, did not affect hIP coupling to Gs signaling, whereas mutation of Cys308,311 and Cys308,309,311 abolished signaling, indicating that palmitoylation of either Cys308or Cys311 is sufficient to maintain functional Gs coupling. Although mutation of Cys309 and Cys311 did not affect hIP-mediated Gq coupling, mutation of Cys308 abolished signaling, indicating a specific requirement for palmitoylation of Cys308 for Gq coupling. Consistent with this, neither hIPC308S,C309S, hIPC308S,C311S, nor hIPC308S,C309S,C311S coupled to Gq. Taken together, these data confirm that the hIP is isoprenylated and palmitoylated, and collectively these modifications modulate its G protein coupling and effector signaling. We propose that through lipid modification followed by membrane insertion, the C-tail domain of the IP may contain a double loop structure anchored by the dynamically regulated palmitoyl groups proximal to transmembrane domain 7 and by a distal farnesyl isoprenoid permanently attached to its carboxyl terminus.

Modification of proteins through the covalent attachment of lipid groups occurs within a wide variety of cellular proteins and may be involved in mediating protein-membrane and/or protein-protein interactions (1)(2)(3). Three of the most common lipid modifications are N-myristoylation, isoprenylation and thio(S)-acylation. In contrast to myristoylation and isoprenylation, which typically occur either as co-translational or as immediate post-translational events, S-acylation through attachment of palmitate to Cys residue(s), via a labile thioester bond, occurs post-translationally (1). Moreover, whereas the former two modifications remain attached until protein degradation, palmitoylation is a reversible, dynamic modification that turns over more rapidly than the protein itself and thus has the potential to be regulated (2,3). A diverse family of cellular proteins are established to be palmitoylated including ␣ subunits of the heterotrimeric G protein subunits, for example G␣ s and G␣ q , Ha-and N-Ras proteins, A kinase anchoring protein 15 and 18, endothelial nitric-oxide synthase, adenylyl cyclase, G protein-coupled receptor kinases 4 and 6, diverse members of the Src family of nonreceptor tyrosine kinases, in addition to several members of the G protein-coupled receptor (GPCR) 1 superfamily (1,2,4,5).
It has been suggested that palmitoylation is a general feature of GPCRs; ϳ80% of all known receptors contain at least one palmitoylable cysteine residue usually located between 10 and 14 amino acids downstream of transmembrane domain 7, within their intracellular carboxyl-terminal tail (C-tail) region (6). Rhodopsin was the first of its class to be identified as a target for palmitoylation, occuring within the C-tail at Cys 322 and Cys 323 (7). Moreover, Ovchinnikov et al. (7) proposed that the hydrophobic palmitate moiety becomes integrated into the lipid membrane bilayer, resulting in the formation of a putative fourth intracellular loop. More recently, the use of fluorescent fatty acid analogues with rhodopsin demonstrated that indeed this is the case (8). Numerous other GPCRs have been identified as targets for palmitoylation. With the finding that GPCRs undergo repeated cycles of palmitoylation/depalmitoylation, it is generally accepted that it may play a role in the regulation of diverse signaling cascades (1)(2)(3). For example, the ␤ 2 -adrenergic receptor (AR) is palmitoylated at Cys 341 , and palmitoylation at this site is required for ␤ 2 -AR coupling to adenylyl cyclase (9) and is increased upon exposure of cells to agonist (10). Palmitoylation of the endothelin (ET) A and ET B receptors is required for their coupling to the extracellular signal regulated kinase/ mitogen-activated protein kinase cascade (11). Mutation of Cys 442 within the C-tail of the ␣ 2A -AR receptor inhibits receptor down-regulation in response to prolonged agonist exposure (12). Palmitoylation has also been shown to regulate ligand-induced receptor internalization. Specifically, abolition of palmitoylation of the luteinizing hormone receptor results in a 2-fold increase in receptor internalization when compared with the wild type receptor. Moreover, intracellular receptor degradation was higher in nonpalmitoylated receptors (13). Thus, it is apparent from these studies that a diverse array of GPCR cellular signaling cascades may be regulated either directly or indirectly by palmitoylation.
Isoprenylation occurs by attachment of either C-15 farnesyl or C-20 geranylgeranyl isoprenoids, derived from the mevalonate/cholesterol biosynthetic pathway, via stable thioether linkages to specific carboxyl-terminal cysteine residues located in distinct "isoprenylation motifs" of proteins (14). Isoprenylation, an immediate post-translational modification, occurs in the cytoplasm, but subsequent proteolysis and carboxyl methylation, should they occur, are membrane-associated events (14). Palmitoylation of isoprenylated proteins generally occurs on cellular membranes whereby isoprenylation precedes palmitoylation (1). Many proteins are modified exclusively with palmitate, for example G␣ s , whereas others are subject to dual lipidation (1). For example, Ha-Ras and N-Ras are farnesylated and palmitoylated. Dual lipidation may serve to further increase protein hydrophobicity, resulting in the protein becoming a permanent resident of the membrane. However, in cases where palmitoylation is the second signal, the first signal may serve to allow transient interaction with membranes, thereby facilitating palmitoylation by a membrane-bound palmitoyltransferase (1,15).
The prostanoid prostacyclin signals through interaction with its signature GPCR, termed the prostacyclin receptor, mediating the inhibition of platelet aggregation and vascular smooth muscle relaxation. The prostacyclin receptor, also termed the IP (16), signals primarily through activation of adenylyl cyclase via G s (17,18) but may also couple to other G protein-effector systems including G q -dependent activation of phospholipase (PL) C and mobilization of intracellular calcium (18,19). Recently, it was established that in addition to G s , the mouse IP also couples to inhibition of adenylyl cyclase via G i and to G q /PLC activation through a switching mechanism involving its initial coupling to G s /adenylyl cyclase activation and subsequent cAMP-dependent protein kinase A phosphorylation of the mouse IP at Ser 357 within its C-tail region (20). Additionally, IPs are widely reported to undergo rapid agonist-induced desensitization, internalization, and down-regulation in human platelets and in other cell types (21)(22)(23)(24).
We have recently established that the IP may be unique among GPCRs in that it is isoprenylated through attachment of a C-15 farnesyl isoprenoid to a Cys within a highly conserved CAAX motif in its C-tail domain (18,25). Inhibition of isoprenylation of both the mouse and human (h) IP, either through site-directed mutagenesis of the critical acceptor Cys of the CAAX motif or through the use of the statin inhibitors of hydroxymethylglutaryl CoA reductase, established that although isoprenylation is not required for ligand binding by IP, it is absolutely required for receptor activation of adenylyl cyclase via G␣ s coupling and for efficient coupling to G␣ q /PLC (18,25,26). Moreover, it was also established that isoprenylation of the IP is required for efficient agonist-mediated receptor internalization (25). In the present study, we sought to investigate whether the human IP is subject to palmitoylation in addition to isoprenylation and thereafter to characterize the role of palmitoylation in receptor-G protein coupling.

Materials
Cicaprost was obtained from Schering AG (Berlin, Germany  (K-20) or the G␣ q/11 (C-19) antisera were obtained from Santa Cruz. QuikChange TM site-directed mutagenesis kit was purchased from Stratagene. Lovastatin was obtained from Merck.

Methods
Deletion-and Site-directed Mutagenesis of Human Prostacyclin Receptor-The plasmids pBluescript-hIP and pHM-hIP have been previously described (20,26). Deletion of the amino acids carboxyl to Val 307 within the hIP to generate the hIP ⌬307 was achieved by conversion of Cys 308 codon to a Stop 308 codon (Cys 308 to Stop 308 , TGC to TAG). Deletion mutagenesis of hIP was performed using pBluescript-hIP as a PCR template, employing Expand High Fidelity Taq DNA polymerase and the oligonucleotide primers: 5Ј-G AGA AGC TTG ATG GCG GAT TCG TGC AGG-3Ј (sense primer; nucleotides ϩ1 to ϩ18 of hIP sequences are underlined) and 5Ј-ATA TGA ATT CTA GAC CCA GAG CTT GAG TCG C-3Ј (antisense primer; the sequences complimentary to nucleotides ϩ 902 to ϩ 920 of hIP are underlined, and the mutator in-frame stop codon is in bold type). The resulting PCR-amplified cDNA was subcloned into the HindIII-EcoRI site of pHM6 (Roche) to generate the plasmid pHM-hIP ⌬307 . Deletion of the amino acids carboxyl to Leu 312 within the hIP to generate the hIP ⌬312 was achieved by conversion of Gly 313 codon to a Stop 313 codon (Gly 313 to Stop 313 , GGG to TAG). Deletion mutagenesis of hIP was performed using pBluescript-hIP as a template and the oligonucleotide primers: 5Ј-G AGA AGC TTG ATG GCG GAT TCG TGC AGG-3Ј (sense primer; nucleotides ϩ1 to ϩ18 of hIP sequences are underlined) and 5Ј-ATA TGA ATT CTA GAG GCA CAG GCA GCA GAC-3Ј (antisense primer; the sequences complimentary to nucleotides ϩ 918 to ϩ 935 of hIP are underlined, and the mutator in-frame stop codon is in bold type). The resulting PCR-amplified cDNA was subcloned into the HindIII-EcoRI site of pHM6 to generate the plasmid pHM-hIP ⌬312 .
All site-directed mutagenesis was performed using the QuikChange™ site-directed mutagenesis kit. Conversion Cys 308 of the hIP to Ser 308 was performed using pHM-hIP as a template and the oligonucleotide primers: 5Ј-CTC AAG CTC TGG GTC AGC TGC CTG TGC CTC G-3Ј (sense primer) and 5Ј-C GAG GCA CAG GCA GCT GAC CCA GAG CTT GAG-3Ј (antisense primer; the sequence complimentary to the single mutator codon is in bold type), resulting in the generation of the plasmid pHM-hIP C308S . Conversion of Cys 309 of the hIP to Ser 309 was performed using pHM-hIP as a template and the oligonucleotide primers: 5Ј-C AAG CTC TGG GTC TGC AGC CTG TGC CTC GGG CC-3Ј (sense primer) and 5Ј-GG CCC GAG GCA CAG GCT GCA GAC CCA GAG CTT G-3Ј (antisense primer; the sequence complimentary to the single mutator codon is highlighted in bold type), resulting in the generation of the plasmid pHM-hIP C309S . Conversion of Cys 311 of the hIP to Ser 311 was performed using pHM-hIP as a template and the oligonucleotide primers: 5Ј-GTC TGC TGC CTG AGC CTC GGG CCT G-3Ј (sense primer) and 5Ј-C AGG CCC GAG GCT CAG GCA GCA GAC-3Ј (antisense primer; the sequence complimentary to the single mutator codon is highlighted in bold type), resulting in the generation of the plasmid pHM-hIP C311S . Conversion of Cys 308 and Cys 309 of the hIP to Ser 308 and Ser 309 was performed using pHM-hIP C309S as a template and the oligonucleotide primers: 5Ј-CTC AAG CTC TGG GTC AGC AGC CTG TGC CTC GG-3Ј (sense primer) and 5Ј-CC GAG GCA CAG GCT GCT GAC CCA GAG CTT GAG-3Ј (antisense primer; the sequence complimentary to the single mutator codon is highlighted in bold type), resulting in the generation of the plasmid pHM-hIP C308S,C309S . Conversion of Cys 308 and Cys 311 of the hIP to Ser 308 and Ser 311 was performed using pHM-hIP C308S as a template and the oligonucleotide primers: 5Ј-GTC AGC TGC CTG AGC CTC GGG CCT G-3Ј (sense primer) and 5Ј-C AGG CCC GAG GCT CAG GCA GCT GAC-3Ј (antisense primer; the sequence complimentary to the single mutator codon is highlighted in bold type), resulting in the generation of the plasmid pHM-hIP C308S,C311S . Conversion of Cys 309 and Cys 311 of the hIP to Ser 309 and Ser 311 was performed using pHM-hIP C309S as a template and the oligonucleotide primers: 5Ј-GTC TGC AGC CTG AGC CTC GGG CCT G-3Ј (sense primer) and 5Ј-C AGG CCC GAG GCT CAG GCT GCA GAC-3Ј (antisense primer; the sequence complimentary to the single mutator codon is highlighted in bold type), resulting in the generation of the plasmid pHM-hIP C309S,C311S . Conversion of Cys 308 , Cys 309 , and Cys 311 of the hIP to Ser 308 , Ser 309 , and Ser 311 was performed using pHM-hIP C308S,C309S as a template and the oligonucleotide primers: 5Ј-GTC AGC AGC CTG AGC CTC GGG CCT G-3Ј (sense primer) and 5Ј-C AGG CCC GAG GCT CAG GCT GCT GAC-3Ј (antisense primer; the sequence complimentary to the single mutator codon is highlighted in bold type), resulting in the generation of the plasmid pHM-hIP C308S,C309S,C311S .
Alternatively, to confirm cell surface IP expression, whole cells were harvested by centrifugation at 500 ϫ g at 4°C for 5 min followed by washing three times with ice-cold PBS. Thereafter, the whole cells were resuspended in MES-KOH buffer (10 mM MES-KOH, pH 6.0, 10 mM MnCl 2 , 1 mM EDTA, 10 mM indomethacin). IP radioligand binding assays were carried out at 4°C for 1 h, using 100 g of whole cell protein in 100 l reactions in the presence of 4 nM [ 3 H]iloprost (15.3 Ci/mmol) essentially as previously described (18). As a control to determine any temperature-dependent changes in the ability of the IP to bind agonist at 4°C as opposed to 30°C, the level of [ 3 H]iloprost binding to membrane fractions (P 100 fractions) was also determined at 4°C and was compared with that which occurred at 30°C. In general, the level of [ 3 H]iloprost bound to IPs expressed in crude membranes when assayed at 4°C was reduced to 72% of that bound at 30°C and thereby accounted for the apparent reduced level of radioligand bound to whole cells, determined at 4°C, relative to that bound to crude membranes, determined at 30°C. The protein determinations were carried out using the Bradford assay (32).
In each case, the 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 (33). The levels of cAMP produced by ligand-treated cells over basal stimulation, determined in the presence of HBS, were expressed as fold stimulation relative to basal (fold increase Ϯ S.E.).
Measurement of IP 3 Levels-Intracellular IP 3 levels were measured as previously described (29). Briefly, the cells were transiently cotransfected with either pCMV-G␣ q (25 g/10-cm dish) plus pADVA (10 g/10-cm dish) or, as controls, with the vector pCMV5 (25 g/10-cm dish) plus pADVA (10 g/10-cm dish). After 48 h, the cells were harvested, washed twice in ice-cold PBS, and then resuspended at ϳ5 ϫ 10 6 cells/ml in HBS containing 10 mM LiCl. The cells (200 l) were then preincubated at 37°C for 10 min. Thereafter, the cells were stimulated for 2 min at 37°C in the presence of cicaprost (1 M) or, for concentration response studies, were stimulated with cicaprost 10 Ϫ6 -10 Ϫ12 M. To determine basal IP 3 levels, the cells were incubated in the presence of an equivalent volume (50 l) of the vehicle HBS. The IP 3 levels produced were determined using the IP 3 binding protein assay (29). The 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 cell protein Ϯ standard error of the mean (pmol/mg Ϯ S.E.), and the 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 Intracellular Ca 2ϩ Mobilization-Measurement of intracellular Ca 2ϩ mobilization ([Ca 2ϩ ] i ) in FURA2/AM-preloaded cells was carried out essentially as previously described (27) Data Analyses-Statistical analysis was carried out using the unpaired Student's t test using the GraphPad Prism V2.0 program (GraphPad Software Inc., San Diego, CA). p values of less than or equal to 0.05 were considered to indicate a statistically significant difference. Amino acid sequence alignments were carried out using the Clustal W software (34), where sequences were aligned to show maximum homology.

RESULTS
The Role of C-tail Sequences in Signaling by the hIP-Previous studies have demonstrated that isoprenylation of the IP within its C-tail region is required for its efficient intracellular signaling (18,25). Specifically, mutation of the CAAX motif of the hIP from -C 383 SLC to -S 383 SLC abolished isoprenylation and generated hIP SSLC that exhibited identical ligand binding properties to that of the wild type hIP but failed to show significant coupling to G␣ s /adenylyl cyclase or efficient coupling to G␣ q /PLC activation (25). In contrast to these findings, another study demonstrated that hIP ⌬312 , a variant of hIP lacking a substantial portion of the C-tail including the isoprenylation CAAX motif, exhibited both G␣ s coupling at levels comparable with hIP and G␣ q coupling, albeit at somewhat reduced efficacy (19). Taken together, these studies suggest that additional elements within the C-tail, other than its requirement for isoprenylation, may influence IP-G protein coupling and may ultimately explain the observed functional differences between the hIP SSLC and the hIP ⌬312 (19,25). Hence, in the present study, we sought to establish whether the hIP may actually undergo dual lipid modification, namely palmitoylation in addition to its established isoprenylation and to investigate whether together these lipid modifications may be involved in regulating signaling by the hIP.
We initially investigated the above paradigm by comparing signaling by the hIP SSLC and the hIP ⌬312 with that of the hIP in HEK 293 stably transfected cell lines that were established under identical conditions, ruling out any possible experimental or artifactual differences such as differences in levels of receptor expression, cell type, growth, or assay conditions, for example. Thus, the effect of the selective IP agonist cicaprost on receptor-mediated intracellular signaling was investigated in HEK.hIP ⌬312 cells stably overexpressing the hIP ⌬312 and was compared with signaling in HEK.hIP and HEK.hIP SSLC cells stably expressing the wild type hIP and isoprenylationdefective hIP SSLC , respectively. Initial radioligand binding assays established that the levels of hIP, hIP ⌬312 , and hIP SSLC receptor expression in their respective crude membrane fractions or whole cells were not significantly different from each other, indicating that the mutations per se did not affect agonist binding or targeting of the respective IP receptors to the plasma membrane (Table I). Although stimulation of the hIP ⌬312 with cicaprost exhibited efficient increases in cAMP generation comparable with the hIP, levels of cAMP generation by hIP SSLC were significantly lower (Fig. 1A) and, at concentrations less than 10 Ϫ7 M, were not significantly different from those generated by control HEK 293 cells (25). Moreover, stimulation of hIP ⌬312 with cicaprost also led to efficient increases in [Ca 2ϩ ] i mobilization (Fig. 1B) and to concentration-dependent increases in inositol 1,4,5-trisphosphate (IP 3 ) generation (Fig. 1C) to levels that were not significantly different from those of the hIP (Fig. 1, B and C). Levels of [Ca 2ϩ ] i mobilization by the hIP SSLC (Fig. 1B) and IP 3 generation (data not shown) were significantly reduced relative to both the wild hIP and the hIP ⌬312 and, consistent with our previous reports (25), were not substantially greater than those generated by HEK 293 cells. Although co-transfection of G␣ q significantly augmented cicaprost-induced IP 3 generation by both the hIP (p Ͻ 0.05) and the hIP ⌬312 (p Ͻ 0.05), it did not enhance IP 3 generation by the hIP SSLC (Fig. 1D); Western blot analysis confirmed overexpression of G␣ q (Fig. 1E).
To exclude the possibility that the carboxyl-terminal four amino acids of the truncated hIP ⌬312 mutant, with the sequence -CCLC, may actually behave as a CAAX motif, facilitating its isoprenylation within its short C-tail domain and thereby possibly explaining its efficient signaling relative to the hIP SSLC , the effect of the hydroxymethylglutaryl CoA reductase inhibitor lovastatin (18) on cicaprost-mediated cAMP generation by hIP ⌬312 was investigated and was compared with Although the hIP ⌬312 is devoid of a substantial portion of the C-tail domain of the hIP (Fig. 2), it retains two highly conserved Cys residues, at Cys 308 and Cys 309 , and a semi-conserved residue at Cys 311 that is conserved in the hIP, mouse IP, rat IP but is replaced by a Tyr within the bovine IP (35)(36)(37)(38). Hence, to investigate the potential role of these conserved Cys residues in hIP signaling, deletion mutagenesis was used to generate hIP ⌬307 , a variant of the hIP devoid of amino acids 308 -386 and differing from the hIP ⌬312 in that residues 308 -312 were removed. Initial characterization of HEK.hIP ⌬307 cells by saturation radioligand binding confirmed high level receptor expression and established that the mutation per se did not appreciably affect agonist binding or plasma membrane targeting by the hIP ⌬307 (Table I). Although stimulation of hIP ⌬307 with cicaprost did result in marginal increases in cAMP generation, those levels of cAMP generation were significantly reduced compared with the hIP throughout the range of cicaprost concentrations used and, at concentrations less than 10 Ϫ7 M cicaprost, were not significantly greater than those generated in HEK 293 cells in response to cicaprost (Fig. 3A). Whereas co-transfection of G␣ s resulted in a 1.81-fold augmentation of cicaprost-induced cAMP generation in control HEK.hIP cells, co-transfection of G␣ s did not significantly alter cAMP generation in HEK.hIP ⌬307 cells ( Fig. 3B; p Ͻ 0.11). Overexpression of G␣ s in HEK.hIP ⌬307 cells was confirmed by Western blot analysis (Fig. 3C). nM. C, HEK.hIP (hIP) and HEK.hIP ⌬312 (hIP ⌬312 ) transiently co-transfected with pCMV-G␣ q were stimulated with cicaprost (10 Ϫ12 -10 Ϫ6 M) for 2 min, and the levels of IP 3 generation were measured. D, alternatively, HEK.hIP, HEK.hIP SSLC , and HEK.hIP ⌬312 cells transiently cotransfected with the vector pCMV5 (Ϫ) or with pCMV-G␣ q (ϩ) encoding G␣ q were stimulated with 1 M cicaprost. In each case, basal IP 3 levels were determined by exposing the cells to the vehicle HBS under identical incubation conditions. Levels of IP 3 produced in ligand-stimulated cells relative to vehicle-treated cells (basal IP 3 ) were expressed as fold stimulation of basal (fold increase in IP 3 Ϯ S.E., n ϭ 4). E, typical Western blot (75 g of total cellular protein analyzed) confirming overexpression of G␣ q in HEK.hIP cells transiently co-transfected with pCMV-G␣ q (ϩ) or, as a control, with pCMV5 (Ϫ). Similar results were obtained for HEK.hIP SSLC and HEK.hIP ⌬312 cells (data not shown). The position of the 46-kDa molecular mass marker is indicated to the right.

TABLE II Effect of lovastatin on radioligand binding
The cells were preincubated with 10 M lovastatin (ϩ) or with its vehicle (Ϫ) for 16 h prior to harvesting for cAMP assays, in response to stimulation with 1 M cicaprost. In each case, basal cAMP levels were determined by exposing the cells to the vehicle HBS under identical incubation conditions. The levels of cAMP produced in ligand-stimulated cells relative to basal cAMP levels were expressed as fold stimulation of basal (fold increase in cAMP Ϯ S.E., n ϭ 3  2. Alignment of prostacyclin receptor amino acid sequences. The deduced amino acid sequences of the C-tail regions of the human, bovine, mouse, and rat IP aligned to show maximum homology using the Clustal W software (34) are shown. The putative isoprenylation motif (-CSLC) is boxed, and downward arrows indicate the positions of amino acid residues 307 (Val 307 ) and 312 (Leu 312 ) within the hIP sequence. Throughout the alignment, gaps indicated by hyphens were inserted to optimize the alignment; identical amino acids are indicated by asterisks; conservative substitutions are indicated by colons; and semi-conservative substitutions are indicated by a periods. Sequences for the human, mouse, and rat IP are based on the published sequences, whereas those for the bovine IP were based on those submitted to the GenBank TM /EMBL Data Bank with accession number Z93039.
Thereafter, the ability of hIP ⌬307 to couple to G␣ q and to PLC activation was investigated. Levels of cicaprost-induced [Ca 2ϩ ] i mobilization by the hIP ⌬307 were significantly impaired relative to the hIP ( Additionally, stimulation of hIP ⌬307 did not result in significant increases in IP 3 generation throughout the range of cicaprost concentrations examined ( Fig. 4D; p Ͻ 0.0026), and co-transfection of G␣ q did not augment IP 3 generation in HEK.hIP ⌬307 cells compared with cells transfected with the vector pCMV5 ( Fig. 4E; p Ͻ 0.20). Overexpression of G␣ q in HEK.hIP ⌬307 cells was confirmed by Western blot analysis (Fig. 4F).
Thus, it appears that although the hIP ⌬312 exhibits near identical coupling to G s and G q to that of the hIP (p Ͼ 0.08), signaling by the hIP ⌬307 is significantly impaired (p Ͻ 0.05) and is not substantially different from that of the hIP SSLC (p Ͼ 0.18).
Palmitoylation of the Human Prostacyclin Receptor-To investigate whether the hIP and its variants are indeed palmitoylated, HEK.hIP, HEK.hIP SSLC , HEK.hIP ⌬312 , and HEK.hIP ⌬307 cells were metabolically labeled with [ 3 H]palmitic acid. As a positive control for the experimental conditions, metabolic labeling was also investigated in HEK 293 cells transiently transfected with pHM-Ha-Ras encoding Ha-Ras, a known substrate for palmitoylation (2,14). Following metabolic labeling, HA-tagged IP and Ras proteins were immunoprecipitated and analyzed by fluorography. Metabolic labeling of Ha-Ras was detected, consistent with its palmitoylation (Fig. 5, A  and B). The hIP also underwent palmitoylation as evidenced by the detection of a broad radiolabeled band between 46 and 66 kDa in the immunoprecipitates from HEK.hIP cells (Fig. 5C, lane 1) but not from the immunoprecipitates from control HEK 293 cells (Fig. 5C, lane 5) or indeed from HEK 293 cells transfected with Ha-Ras (Fig. 5A, lane 1).
Palmitoylation of both the isoprenylation-defective hIP SSLC and hIP ⌬312 was also observed (Fig. 5C, lanes 2 and 3, respectively). However, although the hIP ⌬312 was palmitoylated to levels comparable with that of the wild type hIP, the level of palmitoylation of the hIP SSLC appeared to be somewhat reduced (2.28-and 1.44-fold increase in palmitoylation relative to basal levels observed in HEK 293 cells, respectively). In contrast, palmitoylation of hIP ⌬307 was not detected (Fig. 5C, lane  4). To confirm the identities of the palmitoylated proteins to be those of the hIP and its variants and to ascertain whether failure to detect palmitoylation of the hIP ⌬307 was not due altered expression levels, following fluorographic exposure the PVDF membranes were screened with anti-HA 3F10 peroxidase-conjugated antibody (Fig. 5D). In each case, similar to that observed in the palmitoylation assays (Fig. 5C), a broad immunoreactive band of 46 -66 kDa was observed in the immunoprecipitates from HEK.hIP, HEK.hIP SSLC , HEK.hIP ⌬312 , and HEK.hIP ⌬307 cells but not from the control HEK 293 cells or from cells transfected with HA-epitope tagged Ha-Ras, thus confirming recovery of hIP, hIP SSLC , hIP ⌬312 , and hIP ⌬307 receptors (Fig. 5B). The chemical identity of the 3 H-labeled moiety attached to the hIP, hIP ⌬312 , and the hIP SSLC following metabolic labeling to be a thioester-linked palmitoyl group was confirmed by treatment of the immunoprecipitates with hydroxylamine (data not shown).
Palmitoylation and Signaling by hIP C308S , hIP C309S , and hIP C308S,C309S -Herein, we have demonstrated that whereas the hIP, hIP SSLC , and hIP ⌬312 are palmitoylated, hIP ⌬307 is not a substrate for palmitoylation, suggesting that amino acid residues 308 -312, inclusively, contain the target site(s) for palmitoylation of the hIP. The primary sequence within this defined region contains two identically conserved Cys residues, Cys 308 and Cys 309 , which may be potential targets for palmitoylation (Fig. 2). Thus, to investigate the possible involvement of Cys 308 and/or Cys 309 in palmitoylation and the potential implications thereof in the regulation of IP signaling, site-directed mutagenesis of hIP was used to mutate Cys 308 and Cys 309 residues individually and collectively to Ser residues to generate hIP C308S , hIP C309S , and hIP C308S,C309S .
Initial characterization of HEK.hIP C308S , HEK.hIP C309S , and HEK.hIP C308S,C309S cells, recombinant HEK 293 cells stably overexpressing hIP C308S , hIP C309S , and hIP C308S,C309S by saturation radioligand-binding confirmed receptor expression at levels comparable with those observed in HEK.hIP cells and also confirmed that the mutations per se did not affect the ligand binding properties of the receptors or their targeting to the plasma membrane (Table I). To ascertain whether Cys 308 and/or Cys 309 residues of the hIP are actually targets for palmitoylation, HEK.hIP, HEK.hIP C308S , HEK.hIP C309S , and HEK.hIP C308S,C309S cells were metabolically labeled with [ 3 H]palmitic acid in the presence of cicaprost. Palmitoylation of both the hIP and hIP C309S was observed as evidenced by the detection of a broad radiolabeled band between 46 and 66 kDa in the immunoprecipitates from HEK.hIP and HEK.hIP C309S cells (Fig. 6A, lanes 1 and 3, respectively). Additionally, it was noteworthy that the level of palmitoylation of the hIP (4.23-fold increase) and the hIP C309S (3.76-fold increase) were comparable, suggesting that Cys 309 may not be a target for palmitoylation. In contrast, although the hIP C308S and hIP C308S,C309S did undergo palmitoylation, levels of metabolic labeling of both FIG. 3. Analysis of cAMP generation by HEK.hIP ⌬307 cells. HEK.hIP ⌬307 (hIP ⌬307 ) and, as controls, HEK.hIP (hIP) and HEK 293 cells were stimulated with 10 Ϫ12 -10 Ϫ6 M cicaprost at 37°C for 10 min (A). Alternatively, HEK.hIP ⌬307 and HEK.hIP cells that had been transiently co-transfected with the vector pCMV5 (Ϫ) or with pCMV-G␣ s (ϩ), encoding G␣ s , were stimulated with 1 M cicaprost (B). In each case, the basal cAMP levels were determined by exposing the cells to the vehicle HBS under identical incubation conditions. The levels of cAMP produced in ligand-stimulated cells relative to basal cAMP levels were expressed as fold stimulation of basal (fold increase in cAMP Ϯ S.E., n ϭ 4). C, typical Western blot (75 g of total cellular protein analyzed) confirming overexpression of G␣ s in HEK.hIP ⌬307 cells transiently co-transfected with pCMV-G␣ s (ϩ) or, as a control, with pCMV5 (Ϫ). The position of the 46-kDa molecular mass marker is indicated to the right.
receptors were significantly reduced relative to hIP (Fig. 6A, compare lanes 2 and 4 with lane 1; 2.44-and 2.34-fold increases in palmitoylation relative to basal levels, respectively). Following fluorography/PhosphorImager analysis, the above PVDF membrane was subject to secondary screening by Western blot analysis using the anti-HA 3F10 peroxidase conjugate and confirmed equivalent expression and the recovery of hIP, hIP C308S , hIP C309S , and hIP C308S,C309S receptors in their re-spective immunoprecipitates (Fig. 6B, lanes 1-4, respectively).
Thereafter, cicaprost-induced signaling by the HEK.hIP C308S , HEK.hIP C309S , and HEK.hIP C308S,C309S cells was examined and was compared with that of the HEK.hIP and nontransfected HEK 293 cells. Stimulation of hIP C308S , hIP C309S , and hIP C308S,C309S each resulted in significant, concentration-dependent increase in cAMP generation to levels that were not significantly different from the hIP, throughout the range of cicaprost concentrations examined ( Fig. 7A; p Ͼ  , n ϭ 4). D, HEK.hIP ⌬307 (hIP ⌬307 ) and HEK.hIP (hIP) cells transiently co-transfected with pCMV-G␣ q were stimulated with cicaprost (10 Ϫ12 -10 Ϫ6 M). E, alternatively, HEK.hIP ⌬307 and HEK.hIP cells transiently co-transfected with pCMV5 (Ϫ) or with pCMV-G␣ q (ϩ) were stimulated with 1 M cicaprost. The basal IP 3 levels were determined by exposing the cells to the vehicle HBS under identical incubation conditions. The levels of IP 3 produced in ligand-stimulated cells relative to vehicle-treated cells (basal IP 3 ) were expressed as fold stimulation of basal (fold increase in IP 3 Ϯ S.E., n ϭ 4). F, typical Western blot (75 g of total cellular protein analyzed) confirming overexpression of G␣ q in HEK.hIP ⌬307 cells transiently co-transfected with pCMV-G␣ q (ϩ) or, as a control, with pCMV5 (Ϫ). The position of the 46-kDa molecular mass marker is indicated to the right. . Thereafter, the HA-tagged hIP, hIP SSLC , hIP ⌬312 , and hIP ⌬307 receptors or HA-tagged Ha-Ras were immunoprecipitated using anti-HA 101r antibody, with HEK 293 cells serving as negative control. Immunoprecipitates were resolved by SDS-PAGE followed by electroblotting onto PVDF membrane. The blots were then soaked in Amplify for 30 min followed by autoradiography for 60 days at Ϫ70°C (A and C). Thereafter, blots were subjected to PhosphorImager analysis, and the intensities of cicaprost-mediated palmitoylation of hIP, hIP SSLC , hIP ⌬312 , and hIP ⌬307 were determined and expressed, in arbitrary units, as fold increases in palmitoylation relative to basal levels detected in HEK 293 cells, as follows: hIP, 4.1-fold; hIP SSLC , 1.44-fold; hIP ⌬312 , 2.28-fold; and hIP ⌬307 , 1.0-fold. Subsequently, the blots were screened by immunoblot analysis using the anti-HA 3F10 peroxidase conjugate followed by chemiluminescent detection (B and D). . Thereafter, the HA-tagged hIP, hIP C308S , hIP C309S , and hIP C308S,C309S receptors were immunoprecipitated using anti-HA 101r antibody with HEK 293 cells serving as a negative control. Immunoprecipitates were resolved by SDS-PAGE followed by electroblotting onto PVDF membrane. The blots were then soaked in Amplify for 30 min followed by autoradiography for 60 days at Ϫ70°C. The positions of the molecular mass markers (kDa) are indicated to the left and right of A and B, respectively. The data presented are representative of three independent experiments. Thereafter, the blots were subjected to densitometric analysis, and the intensities of cicaprost-mediated palmitoylation of hIP, hIP C308S , hIP C309S , and hIP C308S,309S were determined and expressed, in arbitrary units, as fold increases in palmitoylation relative to basal levels detected in HEK 293 cells, as follows: hIP, 4.23-fold; hIP C308S , 2.44-fold; hIP C309S , 3.76-fold; and hIP C308S,309S , 2.34-fold. Thereafter, the blots were screened by immunoblot analysis using the anti-HA 3F10 peroxidase-conjugated antibody followed by chemiluminescent detection (B). 0.05). Moreover, co-transfection of cells with G␣ s resulted in significant augmentations in cAMP generation in HEK.hIP C308S , HEK.hIP C309S , and HEK.hIP C308S,C309S cells to levels comparable with the control HEK.hIP cells (Fig. 7B). Overexpression of G␣ s was confirmed by Western blot analysis (Fig. 7C). Similar to that of the hIP, preincubation of HEK.hIP C308S , HEK.hIP C309S , and HEK.hIP C308S,C309S cells with lovastatin significantly impaired cicaprost-induced cAMP generation, confirming that the hIP C308S , hIP C309S , and hIP C308S,C309S are indeed subject to protein isoprenylation (Table II).
Whereas stimulation of HEK.hIP C309S cells exhibited increases in IP 3 generation that were not significantly different from that generated by HEK.hIP cells throughout the range of cicaprost concentrations examined ( Fig. 8A; p ϭ 0.96), stimulation of HEK.hIP C308S and HEK.hIP C308S,C309S cells failed to induce measurable increases in IP 3 generation, even at the 1 M cicaprost (p Ͻ 0.05). Moreover, co-transfection with G␣ q significantly augmented IP 3 generation in HEK.hIP C309S (p Ͻ 0.05) and HEK.hIP (p Ͻ 0.05) cells but did not affect IP 3 generation by HEK.hIP C308S (p ϭ 0.65) or HEK.hIP C308S,C309S (p Ͻ 0.78) cells (Fig. 8B). Similarly, stimulation of hIP C309S with cicaprost yielded concentration-dependent increases in [Ca 2ϩ ] i mobilization to levels that were not significantly different from those generated by the hIP (Fig. 8C; p ϭ 0.68). However, levels of [Ca 2ϩ ] i mobilization by the hIP C308S (p ϭ 0.0048) and the hIP C308S,C309S (p ϭ 0.0039) were significantly reduced relative to the hIP throughout the range of cicaprost concentration (Fig. 8C). Moreover, co-transfection of cells with G␣ q resulted in a significant augmentation of [Ca 2ϩ ] i mobilization in HEK.hIP (p ϭ 0.0001) and HEK.hIP C309S (p ϭ 0.0001) cells, but not in HEK.hIP C308S (p ϭ 0.39) and HEK.hIP C308S,C309S (p ϭ 0.41) cells (Fig. 8D). Overexpression of G␣ q was confirmed by Western blot analysis (Fig. 8E). Thus, taken together, as the metabolic labeling of hIP C308S and hIP C308S,C309S by [ 3 H]palmitic acid is reduced relative to hIP and hIP C309S , it is proposed that Cys 308 , but not Cys 309 , is a target site for palmitoylation and that palmitoylation at Cys 308 is required for efficient hIP coupling to G q /PLC activation but may not be required for G s /adenylyl cyclase activation.
Analysis of Palmitoylation and Signaling Properties of hIP C308S,C309S,C311S , hIP C311S , hIP C308S,C311S , and hIP C309S,C311S -Analysis of the palmitoylation status of hIP C308S,C309S (Fig. 6A, lane 4) demonstrated that whereas a significant reduction in metabolic labeling of hIP C308S,C309S by [ 3 H]palmitic acid was evident relative to hIP (p Ͻ 0.05), the levels of metabolic labeling detected were consistently ϳ2-fold greater than that detected in control cells, suggesting that hIP C308S,C309S may be palmitoylated at another Cys residue distinct from Cys 308 and Cys 309 . It was postulated that the remaining semi-conserved Cys residue, Cys 311 , located between amino acids 308 and 312 may be palmitoylated, accounting for the higher than basal levels of metabolic labeling of the hIP C308S,C309S . Thus, to investigate whether Cys 311 may indeed be palmitoylated, site-directed mutagenesis was used to generate hIP C308S,C309S,C311S .
Initial characterization of HEK.hIP C308S,C309S,C311S cells by saturation radioligand binding confirmed high level receptor expression at levels comparable with those observed in HEK.hIP cells (Table I). Metabolic labeling of HEK.hIP C308S,C309S,C311S cells with [ 3 H]palmitic acid established that, unlike the hIP (Fig.  9A, lane 1), the hIP C308S,C309S,C311S was not palmitoylated (Fig.  9A, lane 2). Subsequent screening of the PVDF membrane by immunoblot analysis using the anti-HA 3F10 peroxidase conjugate followed by chemiluminescence detection confirmed that both hIP C308S,C309S,C311S and hIP were expressed and immunoprecipitated at comparable levels (Fig. 9B, lanes 1 and  2, respectively).
To further confirm that Cys 308 and Cys 311 , but not Cys 309 , were subject to palmitoylation, site-directed mutagenesis was used to generate hIP C311S , hIP C308S,C311S , and hIP C309S,C311S . Saturation radioligand binding assays again confirmed high level receptor expression in HEK.hIP C311S , HEK.hIP C308S,C311S , and HEK.hIP C309S,C311S cells at levels comparable with those observed in HEK.hIP cells and confirmed that the mutations per se did not affect the ligand binding properties of the receptors or their targeting to the plasma membrane (Table I). Metabolic labeling of HEK.hIP C308S,C311S cells with [ 3 H]palmitic acid established that, like the hIP C308S,C309S,C311S (Fig. 9A, lane 2), the hIP C308S,C311S was not palmitoylated (Fig. 9A, lane 4). Although the hIP C311S and hIP C309S,C311S did undergo palmitoylation, the levels of metabolic labeling of both receptors were significantly reduced relative to hIP (Fig. 9A, compare lanes 3 and 5 with lane 1; 2.11-and 2.43-fold increases in palmitoylation relative to basal levels, respectively). Subsequent secondary screening of the PVDF membrane by immunoblot analysis using the anti-HA 3F10 peroxidase conjugate followed by chemiluminescence detection confirmed that hIP, hIP C311S , hIP C308S,C311S , and hIP C309S,C311S were expressed and immunoprecipitated at comparable levels (Fig. 9B, lanes 1 and 3-5, respectively). Taken together, these data confirm that, similar to Cys 308 , Cys 311 is also subject to palmitoylation. Moreover, these data also demonstrate that Cys 309 is not a site of palmitoylation, because no palmitate was incorporated into hIP C308S,C311S . FIG. 7. Analysis of cAMP generation by HEK.hIP C308S , HEK.hIP C309S , and HEK.hIP C308S,C309S cells. HEK.hIP C308S (hIP C308S ), HEK.hIP C309S (hIP C309S ), HEK.hIP C308S,C309S (hIP C308S,C309S ), and, as positive controls, HEK.hIP (hIP) cells were stimulated with 10 Ϫ12 -10 Ϫ6 M cicaprost at 37°C for 10 min (A). Alternatively, HEK.hIP C308S HEK.hIP C309S , HEK.hIP C308S,C309S , and HEK.hIP cells transiently co-transfected with pCMV5 (Ϫ) or with pCMV-G␣ s (ϩ), encoding G␣ s , were stimulated with 1 M cicaprost (B). In each case, basal cAMP levels were determined by exposing the cells to the vehicle HBS under identical incubation conditions. The levels of cAMP produced in ligand-stimulated cells relative to basal cAMP levels were expressed as fold stimulation of basal (fold increase in cAMP Ϯ S.E., n ϭ 4). C, typical Western blot (75 g of total cellular protein analyzed) confirming overexpression of G␣ s in HEK.hIP C308S cells transiently co-transfected with pCMV-G␣ s (ϩ) or, as a control, with pCMV5 (Ϫ). Similarly, overexpression of G␣ s in HEK.hIP C309S and HEK.hIP C308S,C309S cells was also confirmed (data not shown). The position of the 46-kDa molecular mass marker is indicated to the right.
Thereafter, cicaprost-induced signaling by hIP C311S , hIP C308S,C311S , hIP C309S,C311S , and hIP C308S,C309S,C311S was examined and compared with signaling by the hIP. Stimulation of hIP C311S and hIP C309S,C311S each resulted in significant, concentration-dependent increase in cAMP generation to levels that were not significantly different from the hIP throughout the range of cicaprost concentrations examined ( Fig. 10A; p Ͼ 0.05). On the other hand, levels of cAMP generation by hIP C308S,C311S and hIP C308S,C309S,C311S were significantly lower than those levels generated by the hIP throughout the range of cicaprost concentrations examined ( Fig. 10A; p Ͻ 0.05) and, in fact, were not significantly different from those levels generated by the hIP SSLC (p Ͼ 0.88; 25) or the hIP ⌬307 cells (p Ͼ 0.18; Fig. 3A). Moreover, co-transfection of cells with G␣ s significantly augmented cAMP generation by the hIP C311S and the hIP C309S,C311S to levels comparable with the hIP (Fig. 10B) but failed to significantly augment cAMP generation by hIP C308S,C311S or by the hIP C308S,C309S,C311S (Fig. 10B). Positive overexpression of G␣ s was confirmed by Western blot analysis (Fig. 10E). Similar to that of the hIP, preincubation of HEK.hIP C311S , HEK.hIP C309S,C311S cells with lovastatin significantly impaired cicaprost-induced cAMP generation, confirming that the hIP C308S , hIP C311S , and hIP C308S,C311S are indeed subject to protein isoprenylation (Table II).
Stimulation of hIP C311S and hIP C309S,C311S with cicaprost yielded concentration-dependent increases in [Ca 2ϩ ] i mobilization to levels that were not significantly different from those generated by the hIP (Fig. 10C; p ϭ 0.83). However, levels of [Ca 2ϩ ] i mobilization by the hIP C308S,C311S (p ϭ 0.0034) and the hIP C308S,C309S,C311S (p ϭ 0.003) were significantly reduced relative to the hIP throughout the range of cicaprost concentrations (Fig. 10C) and were not significantly different from those levels generated by the hIP SSLC (p Ͼ 0.05) (25) or the hIP ⌬307 (p Ͼ 0.05; Fig. 4B). Moreover, co-transfection of cells with G␣ q significantly augmented of cicaprost-induced [Ca 2ϩ ] i mobilization by the hIP C311S (p ϭ 0.0003) and the hIP C309S,C311S (p ϭ 0.0008) but failed to significantly augment signaling by the hIP C308S,C311S (p ϭ 0.58) or by the hIP C308S,C309S,C311S (p ϭ 0.69). Western blot analysis confirmed overexpression of G␣ q (Fig. 10F).  , n ϭ 4). E, typical Western blot (75 g of total cellular protein analyzed) confirming overexpression of G␣ q in HEK.hIP C308S cells transiently co-transfected with pCMV-G␣ q (ϩ) or, as a control, with pCMV5 (Ϫ). Similarly, overexpression of G␣ q in HEK.hIP C309S and HEK.hIP C308S,C309S cells was also confirmed (data not shown). The position of the 46-kDa molecular mass marker is indicated to the right. . Thereafter, the HA-tagged hIP, hIP C308,308,311S , hIP C311S , hIP C308S,C311S , and hIP C309S,C311S receptors were immunoprecipitated using anti-HA 101r antibody with HEK 293 cells serving as a negative control. The immunoprecipitates were resolved by SDS-PAGE followed by electroblotting onto PVDF membrane; the blots were then soaked in Amplify for 30 min followed by fluorography for 60 days at Ϫ70°C. Thereafter, the blots were subjected to densitometric analysis, and the intensities of cicaprost-mediated palmitoylation of hIP, hIP C308,308,311S , hIP C311S , hIP C308S,C311S , and hIP C309S,C311S were determined and expressed, in arbitrary units as fold increases in palmitoylation relative to basal levels detected in HEK 293 cells, as follows: hIP, 4.23-fold; hIP C308S,C309S,C311S , 1.03-fold; hIP C311S , 2.11-fold, hIP C308S,311S , 1.08-fold; and hIP C309S,311S , 2.43-fold. Subsequently, the blots were screened by immunoblot analysis using the anti-HA 3F10 peroxidase conjugate followed by chemiluminescent detection (B). The positions of the molecular mass markers (kDa) are indicated to the left and right of A and B, respectively. The data presented are representative of three independent experiments.

DISCUSSION
In the current study, through a combination of deletion and site-directed mutagenesis, in vivo metabolic labeling studies, and the use of pharmacologic inhibitors of protein isoprenylation, we investigated the requirement for palmitoylation and isoprenylation for IP function and intracellular signaling. To this end, we initially compared signaling by the wild type hIP, the isoprenylation-defective hIP SSLC , and the deletion mutant hIP ⌬312 . Similar to the hIP, the hIP ⌬312 efficiently coupled to G␣ s and G␣ q (Fig. 1). In contrast, consistent with our previous findings (25), signaling by the hIP SSLC was not substantially different from that of control HEK 293 cells (Fig. 1). Although the hydroxymethylglutaryl CoA reductase inhibitor lovastatin, an effective inhibitor of isoprenylation (18), impaired signaling by the hIP, it did not affect signaling by the hIP ⌬312 , confirming that the truncated C-tail of the hIP ⌬312 (with the carboxyl sequence -CCLC) is not actually modified by isoprenylation. Similar to that of the hIP SSLC , the hIP ⌬307 exhibited significantly impaired coupling to G␣ s and G␣ q (Fig. 3). Whole cell (in vivo) metabolic labeling studies established that the hIP is indeed palmitoylated, and through a combination of site-directed and deletion mutagenesis, we have confirmed that this occurs at both Cys 308 and Cys 311 (Figs. 5, 6, and 9). Although palmitoylation of the hIP was not required for its ligand binding (Table I), palmitoylation at either Cys 308 or Cys 311 is sufficient to maintain hIP coupling to G s /adenylyl cyclase, whereas mutation of both palmitoylation sites completely abolishes that signaling (Figs. 7 and 10). On the other hand, palmitoylation of Cys 308 is specifically required for G q coupling and PLC activation as indicated by the failure of the hIP C308S to mediate increases in IP 3 generation or [Ca 2ϩ ] i mobilization (Fig. 8). Moreover, similar to that of the hIP, signaling by the hIP C308S , hIP C309S , hIP C311S , hIP C308S,C309S , and hIP C309S,C311S was significantly inhibited by lovastatin (Table  II), confirming that these receptors are indeed isoprenylated at Cys 383 within the CAAX motif; on the other hand, the fact that lovastatin did not affect signaling by the hIP C308S,C311S and hIP C308S,C309S,C311S was reflective of the fact that these two palmitoylation-defective receptors did not mediate substantial increases in cAMP generation, even in the absence of lovastatin. Whereas conceptually lovastatin may inhibit hIP signaling indirectly through inhibition of G␤␥ geranylgeranylation, the fact that lovastatin had no affect on signaling by the hIP ⌬312 (Table II) confirms that lovastatin is not indirectly functioning through inhibition of G␤␥ isoprenylation and function but rather is specifically impairing IP isoprenylation and thereby directly inhibiting IP (hIP, hIP C308S , hIP C309S , hIP C311S , hIP C308S,C309S , and hIP C309S,C311S ) signaling.
A number of independent studies have demonstrated that impairment of receptor palmitoylation, such as in the case of the CCR5 receptor, may result in accumulation of GPCRs in intracellular stores and in the lowering of cell surface receptor expression (39). In the present study, for the hIP and its various mutant forms, we have established that all stable cell lines under study express equivalent cell surface receptor numbers, thereby confirming that neither palmitoylation nor, indeed, isoprenylation per se alter the ability of the hIP to reach the plasma membrane. Thus, differences in IP signaling by the palmitoylation-defective mutants may not be attributed to altered/reduced cell surface receptor expression but rather are most likely due to direct alteration in their palmitoylation status. Although our metabolic labeling studies clearly demonstrate that the hIP is palmitoylated at Cys 308 and Cys 311 and our intracellular signaling studies have highlighted the importance of those palmitoyl moieties in facilitating G protein coupling/effector signaling, in the absence of an appropriate phar- , and, as positive controls HEK.hIP (hIP) cells were preloaded with FURA2/AM prior to stimulation with 10 Ϫ8 -10 Ϫ5 M cicaprost (C). Alternatively, the cells that had been transiently co-transfected with either the control vector pCMV5 (Ϫ) or with pCMV-G␣ q (ϩ) were stimulated with 1 M cicaprost (D). The data were calculated as the mean changes in intracellular calcium mobilized (⌬[Ca 2ϩ ] i , nM Ϯ S.E., n ϭ 4). E and F, typical Western blot (75 g of total cellular protein analyzed) confirming overexpression of G␣ s and G␣ q , respectively, in HEK.hIP C308S,C309S,C311S cells transiently co-transfected with pCMV-G␣ s (ϩG␣ s ), pCMV-G␣ q (ϩG␣ q ) or, as a control, with pCMV5 (ϪG␣ s /ϪG␣ q ). Similar data was obtained for overexpression of G␣ s and G␣ q in HEK.hIP C311S , HEK.hIP C308S,C311S , and HEK.hIP C309S,C311S cells (data not shown). The position of the 46-kDa molecular mass marker is indicated to the right of E and F. macologic inhibitor of palmitoylation, our studies cannot exclude the possibility that it is the presence of Cys residues at those sites, as opposed to the palmitoyl-Cys residues, that are critical in mediating that signaling.
Thus, through our previous studies (18,25,26) and those outlined in the present study involving the hIP, it is evident that the IP is certainly unique among GPCRs thus far characterized in that it is isoprenylated and palmitoylated, and together these dual lipidations modulate IP coupling to G s -and G q -regulated effector systems but do not affect agonist binding. In attempting to put forward a model (Fig. 11) to describe the relative involvements of isoprenylation and palmitoylation in regulating IP signaling, we have drawn from the prototypic Ha-Ras in assuming that isoprenylation precedes palmitoylation (14) and that both types of lipid modifications mediate protein-membrane rather than protein-protein interactions. However, we preface our model by stating that this may not strictly be accurate because in the case of the hIP SSLC , for example, palmitoylation occurs in the absence of isoprenylation, and further experiments are required to precisely define or identify the nature of the individual lipid interactants. However, despite these limitations, we now propose a model, as outlined in Fig. 11, to explain the role of isoprenylation and palmitoylation in mediating hIP coupling to G s /adenylyl cyclase and G q /PLC effector signaling. Additionally, through the model outlined below, we propose to explain the apparent paradox as to why the hIP ⌬312 but not the hIP SSLC mediates efficient signaling to both G s /adenylyl cyclase and G q /PLC effector systems.
In this model (Fig. 11), we propose that the newly translated hIP undergoes isoprenylation through attachment of a farnesyl isoprenoid to Cys 383 within its CAAX motif. Insertion of the farnesyl group into the lipid bilayer results in the formation of a fourth intracellular loop (IC4 C-tail ) within the C-tail domain of the hIP and also positions Cys 308 and Cys 311 in the proximity of the membrane, making them available for palmitoylation by the membrane-bound palmitoyl transferase (15). Palmitoylation of Cys 308 and Cys 311 residues, which may be enhanced on agonist binding, and subsequent membrane insertion of those palmitoyl groups results in the formation of a double loop structure (loop A and loop B, amino-and carboxyl-terminal to the palmitoylated Cys 308 and Cys 311 , respectively) within the IC4 C-tail and is required for hIP coupling to both G s -and G qmediated effector signaling. The hIP ⌬312 , although not undergoing isoprenylation, does undergo palmitoylation at Cys 308 and Cys 311 . We suggest that following membrane insertion of the palmitoyl groups, loop A, but not loop B, is formed within the truncated C-tail domain of the hIP ⌬312 . Thus, from studies with hIP ⌬312 , it appears that loop A is critical for both G s and G q coupling. Paradoxically, the hIP SSLC , although not undergoing isoprenylation, undergoes palmitoylation at Cys 308 and Cys 311 and, following membrane insertion, results in the formation of loop A, and not loop B; however, unlike the hIP ⌬312 , the hIP SSLC does contain those constituent amino acid residues of loop B (312-386), which we propose may either (i) physically impair hIP SSLC :G s and G q interaction or (ii) destabilize or shield loop A within the hIP SSLC , thereby inhibiting signaling by the hIP SSLC . Through co-immunoprecipitation studies, we have previously established that the hIP SSLC does not physically associate with either G␣ s or G␣ q (25), thereby suggesting that indeed amino acid residues 312-386 within the hIP SSLC may serve to impair receptor-G protein interaction. Neither the palmitoylation-defective deletion mutant hIP ⌬307 nor the sitedirected mutants hIP C308S,C311S and hIP C308S,C309S,C311S contain either loop A or loop B within their C-tail domains and hence cannot mediate efficient coupling to G s or G q effector signaling. The hIP C308S , which is predicted to contain a modified loop A and a loop B, can couple to G␣ s -mediated adenylyl cyclase activation but cannot couple to G␣ q /PLC activation, suggesting that there is a strict requirement for palmitoylation at Cys 308 for G q , but not G s , coupling.
While recognizing that the IP is somewhat unparalleled among well characterized GPCRs, the proposed double loop A and loop B structure within the C-tail domain of the hIP is not unlike that which has recently been proposed to occur for the 5-hydroxytryptamine (4A) (5-HT 4(a) ) receptor (40). In their study, Ponimaskin et al. (40) established that, like many other GPCRs, the 5-HT 4(a) receptor is subject to palmitoylation at two adjacent Cys residues (Cys 328 and Cys 329 ) located proximal to transmembrane domain 7 but is also palmitoylated at Cys 386 , which is located one amino acid residue from the C terminus. Membrane insertion of the palmitates at the proximal Cys 328 and Cys 329 and distal Cys 386 would generate a dynamic (agonist-regulated) double loop structure within the C-tail of the FIG. 11. Model of lipid modification of the hIP. The newly translated, seven transmembrane domain (I-VII) integral plasma membrane (PM)-bound hIP contains a CAAX motif within its C-tail domain with the sequence -C 383 SLC (CSLC). Isoprenylation of the hIP through attachment of a farnesyl isoprenoid (jagged tail symbol) to Cys 383 followed by membrane insertion results in the formation of fourth intracellular loop within its C-tail domain (IC4 C-tail ). Subsequent palmitoylation (1) at Cys 308 and Cys 311 within its IC4 C-tail followed by membrane insertion results in the formation of a double loop structure, referred to as loop A and loop B, within the C-tail domain, which is required for efficient hIP coupling to G s and G q and for effector signaling. The hIP ⌬312 retains loop A, but not loop B, within its C-tail domain and exhibits efficient coupling to G s and G q . Neither the deletion mutant hIP ⌬307 nor hIP C308S,C311S contain loop A or loop B sequences and hence cannot efficiently couple to G s -or G q -mediated effector signaling. The hIP C308S , although not undergoing palmitoylation at Ser 308 (S), does undergo palmitoylation at Cys 311 and hence contains a modified loop A and loop B within its C-tail domain. The hIP C311S , although not undergoing palmitoylation at Ser 311 , does undergo palmitoylation at Cys 308 and hence contains loop A and a modified loop B within its C-tail domain (not shown). Although palmitoylation at either Cys 308 or Cys 311 alone is sufficient to maintain G␣ s coupling, there is a strict requirement for palmitoylation of Cys 308 , but not of Cys 311 , for efficient G␣ q coupling. The hIP SSLC , although not undergoing isoprenylation, does undergo palmitoylation at Cys 308 and Cys 311 which following membrane insertion, results in the formation of loop A, but not loop B, within its C-tail domain; the presence of loop B sequences not involved in loop B formation either (i) physically impairs G s and G q interaction or (ii) destabilizes loop A, thereby impairing G s and G q coupling and effector signaling. 5-HT 4(a) receptor. Although palmitoylation-deficient mutants of 5-HT 4(a) receptor were indistinguishable from the wild type receptor in mediating G protein/effector signaling, mutation of the proximal residues converted the inactive receptor (R) into the active (R*) form and thereby promoted agonist-independent or constitutive activation of the 5-HT 4(a) receptor.
The finding that modulation of the palmitoylation status of the hIP significantly affects its G protein coupling characteristics is indeed consistent with data generated for several GPCRs including rhodopsin, the ␤ 2 -AR, as well as for the ET A and ET B receptors (7,9,41). Functional characterization of the nonpalmitoylated ␤ 2 -AR and the ET B receptor revealed that palmitoylation is essential for agonist-stimulated coupling to G s and to both G q and G i proteins, respectively (9,41). However, analysis of the nonpalmitoylated ET A receptor mutant revealed that agonist-mediated G s coupling was unaffected, whereas signaling through G q was abolished (42). Similar to that of the ET A receptor, we have established that mutation of Cys 308 significantly impaired hIP C308S coupling to G␣ q -regulated but not to G␣ s -regulated effector signaling, providing further evidence that palmitoylation may differentially regulate distinct signal transduction pathways.
As stated, various studies have indeed indicated that the functional implications of palmitoylation are both individual/ specific to the given GPCR and may also be diverse (1)(2)(3). Although agonist stimulation has been reported to enhance palmitoylation of the ␤ 2 -AR (10) and of the 5-HT 4(a) receptor (40), it is also widely reported to enhance the rate of the palmitate turnover relative to protein turnover, such as in the case of the ␤ 2 -AR (43) and the ␣ 2A AR (44). Further experiments are required to investigate whether there is enhanced agonist-induced turnover of palmitate, relative to protein turnover, in the case of the hIP. In addition to its widely documented role in modulating G protein coupling (1-3), palmitoylation has been shown to influence the phosphorylation, desensitization, and/or internalization status of various GPCRs (12,13,39,45,46). For example, it has been suggested that the location of the palmitoylated Cys residue(s) in GPCRs may be strategic, being juxtaposed between the G protein-binding site and amino acids that may be phosphorylated and play a role in desensitization (3). Additionally, mutation of Cys 341 of the ␤ 2 -AR significantly impairs coupling to G s /adenylyl cyclase but generated a receptor that is highly phosphorylated in its basal state, suggesting that palmitoylation of Cys 341 plays a central role in maintaining the ␤ 2 -AR in a state capable of interacting with the G protein and, hence, with its effector system (9,40,45). Whether an analogous regulatory system applies to the hIP remains to be investigated; however, it is noteworthy that the hIP is subject to agonist-mediated phosphorylation and desensitization through a protein kinase C-dependent mechanism whereby Ser 328 , within the putative loop B (Fig. 11), has been identified as the target residue for phosphorylation (19,30). Additionally, we have established that isoprenylation of the hIP is required for its efficient agonist-induced internalization (25), and hence it is indeed likely that, either in combination with isoprenylation or alone, palmitoylation may also regulate the processes that contribute to agonist-induced internalization and desensitization.