Palmitoylation of Human EndothelinB

By site-directed mutagenesis, three cysteine residues (amino acids 402, 403, and 405) in the carboxyl terminus of human endothelinB (ETB) were identified as potential palmitoylation sites. Substitutions of all of the three cysteine residues with serine gave an unpalmitoylated mutant, C2S/C3S/C5S. When expressed in Chinese hamster ovary cells, C2S/C3S/C5S was localized on the cell surface, retained high affinities to ET-1 and ET-3, and was rapidly internalized when bound to the ligand. However, unlike the wild-type ETB, C2S/C3S/C5S transmitted neither an inhibitory effect on adenylate cyclase nor a stimulatory effect on phospholipase C, indicating a critical role of palmitoylation in the coupling with G proteins, regardless of the G protein subtypes. Truncation of the carboxyl terminus including Cys403/Cys405 gave a deletion mutant Δ403 that was palmitoylated on Cys402 and lacked the carboxyl terminus downstream to the palmitoylation site. Δ403 did transmit a stimulatory effect on phospholipase C via a pertussis toxin-insensitive G protein but it failed to transmit an inhibitory effect on adenylate cyclase. These results indicated a differential requirement for the carboxyl terminus downstream to the palmitoylation site in the coupling with G protein subtypes, i.e. it is required for the coupling with Gi but not for that with Gq.

By site-directed mutagenesis, three cysteine residues (amino acids 402, 403, and 405) in the carboxyl terminus of human endothelin B (ET B ) were identified as potential palmitoylation sites. Substitutions of all of the three cysteine residues with serine gave an unpalmitoylated mutant, C2S/C3S/C5S. When expressed in Chinese hamster ovary cells, C2S/C3S/C5S was localized on the cell surface, retained high affinities to ET-1 and ET-3, and was rapidly internalized when bound to the ligand. However, unlike the wild-type ET B , C2S/C3S/C5S transmitted neither an inhibitory effect on adenylate cyclase nor a stimulatory effect on phospholipase C, indicating a critical role of palmitoylation in the coupling with G proteins, regardless of the G protein subtypes. Truncation of the carboxyl terminus including Cys 403 /Cys 405 gave a deletion mutant ⌬403 that was palmitoylated on Cys 402 and lacked the carboxyl terminus downstream to the palmitoylation site. ⌬403 did transmit a stimulatory effect on phospholipase C via a pertussis toxin-insensitive G protein but it failed to transmit an inhibitory effect on adenylate cyclase. These results indicated a differential requirement for the carboxyl terminus downstream to the palmitoylation site in the coupling with G protein subtypes, i.e. it is required for the coupling with G i but not for that with G q .
Substitutions of the cysteine residues gave unpalmitoylated mutants of each GPCR and the role of the modification has been described, to a varying extent, on three aspects of the receptor functions; 1) ligand binding, 2) G protein activation, and 3) intracellular trafficking of the receptor molecule. To date, however, there appears to be no common rule applicable to all GPCRs on any of the three aspects. On ligand binding, the elimination of palmitoylation caused no changes in the binding characteristics of all the GPCRs examined (4,9,(12)(13)(14)(15)) except for ␤ 2 -adrenergic receptor (␤ 2 AR). The unpalmitoylated ␤ 2 AR lacked the GTP-sensitive high affinity state for agonists and this lack was ascribed to its uncoupling from G s (3,16). On G protein coupling, no effects of the elimination have been found on the capacities of m2 cholinergic (13), thyrotropin-releasing hormone (14), and luteinizing hormone/human choriogonadotropin (9) receptors to activate G proteins whereas opposite effects of it, both enhancement and inhibition, were described for rhodopsin to activate G t (17) and for ␤ 2 AR to activate G s (3), respectively. A differential requirement for the modification between G protein subtypes coupled to the same receptor has been highlighted in a study on human ET A (12). On intracellular trafficking of the receptor molecule, reduced cell surface expression was reported for unpalmitoylated mutants of thyrotropin-releasing hormone, luteinizing hormone/human choriogonadotropin, and vasopressin V2 receptors (9,14,15). Internalization of ␤ 2 AR (18) or ␣ 2 AR (19) was not affected while that of luteinizing hormone/human choriogonadotropin receptors was enhanced by the elimination of palmitoylation (9). Because of these pleiotropic effects described, the functional role of palmitoylation in each GPCR is an open question. The endothelins (ETs) are a family of potent vasoactive peptides that includes ET-1, -2, and -3 (20,21). They have a wide variety of biological effects in various tissues and cell types (22) that are mediated by specific GPCR subtypes, ET A and ET B (23,24). The two subtypes can be pharmacologically distinguished by different rank orders of affinity toward the three ET isopeptides; ET A is ET-1-selective, showing an affinity rank order of ET-1 Ն ET-2 Ͼ Ͼ ET-3, whereas ET B exhibits similar affinities to all of the three isopeptides (23,24). Both of them belong to a subfamily of GPCRs with a promiscuous nature that can activate multiple subtypes of G proteins and they can also be distinguished by selective coupling with G protein subtypes; when expressed in CHO cells, ET A couples with members of G q and G s families while ET B couples with those of G q and G i families (25,26).
The purposes of the current study were to identify potential palmitoylation sites of ET B and to reveal a role of the modification in ET B functions including ligand binding, cell surface expression, internalization, and G protein activation. An additional objective was to reveal a functional role of the carboxylterminal tail downstream to the palmitoylation site in the receptor functions. (Tokyo, Japan); fura-2 acetoxymethyl ester from Dojin Chemicals (Tokyo, Japan); BCA microprotein assay kit, ImmunoPure TM -immobilized avidin, and NHS-SS-biotin were from Pierce (Rockford, IL). All other chemicals were of reagent grade and were obtained commercially.

Materials-Transformer
Mutagenesis-The entire coding sequence of human ET B was subcloned into a BamHI restriction site of pUC19 and served as a template for mutagenesis using a Transformer TM site-directed mutagenesis kit (CLONTECH). The following primers were used to substitute the cysteine residue(s) with serine; 5Ј-CACCAGCAGCTTAAGCATGAC-3Ј to mutate Cys 402 , 5Ј-CTGGCACCAGCTGCATAAGCATG-3Ј to mutate Cys 403 , 5Ј-AATGACTGGCTCCAGCAGCAT-3Ј to mutate Cys 405 , 5Ј-CT-GGCACCAGCTGCTTAAGCATGAC-3Ј to mutate Cys 402 /Cys 403 and 5Ј-CTGGCTCCAGCTGCTTAAGCATGAC-3Ј to mutate Cys 402 Cys 403 -Cys 405 . See Fig. 1 for nomenclature of the mutants. The mutations were confirmed by sequencing and the cDNA fragments were subcloned into a XhoI/NotI restriction site of a mammalian expression vector pME18SfϪ. Procedures for construction of carboxyl-terminal deletion mutants (⌬400, ⌬402, and ⌬403) were described (27).
Cell Culture and Transfection-COS cells were routinely maintained in DMEM, 10% FCS at 37°C in a humidified atmosphere containing 5% CO 2 . CHO cells were maintained under the same conditions except for the use of Ham's F-12 instead of DMEM. For transient expressions, COS cells in 60-or 100-mm dishes were transfected with pME18SfϪ carrying the cDNA constructs using LipofectAMINE (Life Technologies, Inc.) according to the manufacturer's instructions. After 48 h the cells were subjected to the assays described below. For stable expressions, CHO cells were transfected with each expression plasmid together with pSVbsr r using Lipofectamine. Cell populations expressing the bsr r gene product were selected in Ham's F-12, 10% FCS containing blasticidin (10 g/ml) and clonal cell lines were isolated by colony lifting and maintained in the same selection medium.
[ 125 I]ET-1 Binding Assay-Assays using intact cells or membrane preparations were done exactly as described in Refs. 28  Biotinylation of ET-1-ET-1 was biotinylated using ImmunoPure TM NHS-SS-biotin (Pierce) according to the manufacturer's instructions. In brief, 48 nmol of ET-1 (in 480 l of 50 mM NaHCO 3 , pH 8.5, containing 0.02% Triton X-100) was mixed with 480 nmol of NHS-SS-biotin (in 30 l of the same buffer) and the reaction was let go for 3 h at room temperature. After the addition of another 480 nmol of NHS-SS-biotin, the reaction was further let go for 16 h. Biotinylated ET-1 was separated from naive ET-1 by high performance liquid chromatography using a C18 column (4.6 mm ϫ 10 cm, Waters) as described (29). The biological activity of the biotinylated ET-1 was verified by its ability to induce a transient increase of [Ca 2ϩ ] i in CHO cells expressing wild-type (wt) ET B (not shown).
Affinity Purification of Receptor Proteins-After the labeling with [ 35 S]Cys/Met or [ 3 H]palmitic acid, COS cells were harvested by incubation in PBS, 1 mM EGTA. The cells from each well were centrifuged and resuspended in 0.5 ml of PBS containing biotinylated ET-1 (100 nM) and incubated for 60 min at 25°C. The cells were centrifuged and then lysed by incubation for 2 h at 4°C in 0.5 ml of the lysis buffer (20 mM sodium phosphate buffer, pH 7.4, 130 mM NaCl, 1 mM EDTA, 0.2 mM phenymethylsulfonyl fluoride, 10 g/ml pepstatin, 10 g/ml leupeptin, 0.25% CHAPS, and 0.4% digitonin). After removing insoluble materials by centrifugation at 100,000 ϫ g for 1 h at 4°C, 30 l of avidin-agarose (50% (v/v) slurry in the lysis buffer) was added to the supernatant and the reaction was let go at 4°C for 16 h. The agarose-avidin-biotin-ET-1-receptor complex was recovered by centrifugation and then extensively washed with ice-cold lysis buffer containing high (0.5 M) or low (0.03 M) concentrations of NaCl. The receptor protein was eluted from the resulting pellet by incubation in 25 l of 0.2 M 2-mercaptoethanol at room temperature for 20 min. The recovered proteins were subjected to 12% SDS-PAGE under reducing conditions. The dried gels were exposed to the imaging plates for 2 days for [ 35 S]Cys/Met-labeled proteins or for 14 days for [ 3 H]palmitic acid-labeled proteins and the autoradiographs were developed with a BAS2000 image analyzer (Fujitsu, Tokyo, Japan).
Cyclic AMP Formation-Cells at ϳ50% confluence in 48-well plates were incubated for 16 h with or without PTX (50 ng/ml). The cells were washed with PBS and then incubated at 37°C for 10 min with 0.3 ml of PBS containing 3-isobuthyl-1-methylxanthine (1 mM). They were then stimulated for 10 min with forskolin (100 M) alone or simultaneously with forskolin and ET-1. The reaction was halted by addition of 10% (w/v) trichloroacetic acid, and the cAMP content in the trichloroacetic acid-soluble cell extract was measured using a radioimmunoassay kit (Amersham).
Phosphoinositide Breakdown-CHO cells in 24-well plates were incubated for 24 h in Ham's F-12, 10% FCS containing myo-[ 3 H]inositol (5 Ci/ml). Where indicated, PTX (50 ng/ml) was added to the labeling medium for the last 16 h. The cells were washed with Krebs-Hensleit buffer with LiCl (110 mM NaCl, 4.5 mM KCl, 1.3 mM CaCl 2 , 1.2 mM KH 2 PO 4 , 1.2 mM MgSO 4 , 25 mM NaHCO 3 , 11.7 mM glucose, 5 mM HEPES, pH 7.4, and 10 mM LiCl) equilibrated with 5% CO 2 and then incubated in 250 l of the same buffer. ET-1 (50-l solutions in Krebs-Hensleit) was added at various concentrations and the plates were kept in a CO 2 incubator for 30 min. The reactions were terminated by adding ice-cold 10% perchloric acid (100 l/well). The following procedures including neutralization of the extracts and separation of [ 3 H]inositol phosphates ([ 3 H]IPs) by anion-exchange chromatography were done exactly as described (30).
Measurement of [Ca 2ϩ ] i -CHO cells in 100-mm dishes were incubated for 16 h with or without PTX (50 ng/ml) and dispersed by incubation in PBS, 1 mM EGTA. The following procedures including fura-2-loading and measurement of [Ca 2ϩ ] i with a CAF-110 spectrofluorometer (Japan Spectroscopy Inc., Tokyo, Japan) were exactly as described (30).
Cy5 Labeling of ET-1-ET-1 was labeled with a fluorescent dye Cy5 using a Fluorolink TM Cy5 reactive dye pack (Amersham) according to the manufacturer's instructions. In brief, 48 nmol of ET-1 in 1 ml of 0.1 M sodium carbonate buffer, pH 9.3, was applied to a vial containing the dye. The reaction was let go for 3 h at room temperature and the Cy5-labeled ET-1 was separated from naive ET-1 by high performance liquid chromatography using a C18 column as described above for the purification of biotinylated ET-1. The biological activity of Cy5-labeled FIG. 1. Nomenclature of the human ET B mutants. Aligned are the amino acids sequences of the carboxyl-terminal tail of the wild-type human ET B , five substitution mutants and three deletion mutants. The amino acid numbers of the four cysteine residues are indicated. Boxed are the serine residues substituted with the corresponding cysteine residues. TMVII, the seventh transmembrane domain.

ET-1 was verified by its ability to induce a transient increase of [Ca 2ϩ ] i in CHO cells expressing wtET B (not shown).
In Situ Binding and Internalization of Cy5-labeled ET-1-CHO cells grown on poly-L-lysine-coated glass coverslips were washed with icecold binding buffer (140 mM NaCl, 4 mM KCl, 1 mM CaCl 2 , 1 mM Na 2 HPO 4 , 1 mM MgCl 2 , 25 mM HEPES, pH 7.4, 11.7 mM glucose, 0.1% bovine serum albumin) and then incubated at 4°C for 2 h in the same buffer containing 10 nM Cy5-labeled ET-1. After washing with ice-cold binding buffer, the fluorescent images of the cells were obtained with a MRC1024 laser-scanning confocal microscope (Bio-Rad, Osaka, Japan). To facilitate internalization of the bound ligand, the cells were then incubated in the binding buffer at 37°C and the images were obtained at the time indicated.
Statistical Analysis-Student's t test was used for the statistical analysis of the results. p values of Ͻ0.05 were considered to be significant.

Identification of the Potential Palmitoylation Sites of ET B -
When COS cells were transfected with pME/wtET B and then metabolically labeled with [ 35 S]Cys/Met, two radioactive proteins with the average molecular sizes of 52 and 34 kDa were affinity-purified from the cell lysate with biotinylated ET-1 (Fig. 2a). The specificity of either band was verified by their absence in the lysate from vector-transfected cells and by their disappearance in the presence of excess unlabeled ET-1 in the binding step. Kozuka et al. (29) purified endogenous ET B from bovine lung membrane preparations in essentially the same way and proved that the 52-and 34-kDa species correspond to a full-length intact receptor and a proteolytic derivative with amino-terminal truncation, respectively. When the cells were labeled with [ 3 H]palmitic acid, the radioactivity was incorporated into both bands (Fig. 2b), indicating the palmitoylation of ET B .
Potential palmitoylation sites of ET B were then determined by expression and affinity purification of a series of mutant receptors with substitutions of cysteine residues. All the substitution mutant receptors were successfully expressed in and purified from the transfected cells as judged by the recovery of [ 35 S]Cys/Met-labeled proteins (Fig. 2a). Of the four cysteine residues at the carboxyl-terminal juxtamembrane portion of ET B , Cys 402 is highly conserved among GPCRs and corresponds to the site that was proven to be palmitoylated in some of them (1). However, a single substitution of Cys 402 with serine did not inhibit the [ 3 H]palmitic acid incorporation. Neither that of Cys 403 nor Cys 405 affected the incorporation. In contrast, simultaneous substitutions of Cys 402 and Cys 403 resulted in an apparent decrease and further substitution of Cys 405 resulted in a complete disappearance. Thus, we concluded that Cys 402 , Cys 403 , and Cys 405 , but not Cys 400 are the potential palmitoylation sites of ET B .
Ligand-binding Properties of the Wild-type and Mutant Receptors Expressed in CHO Cells-That all the mutant receptors with cysteine substitutions were successfully affinity-purified by binding of biotinylated ET-1 to intact cells suggested that palmitoylation is required neither for the ligand binding nor the cell surface expression of ET B . [ 125 I]ET-1 binding assays on intact COS cells expressing the wild-type or mutant receptors failed to reveal any differences in the binding characteristics (data not shown), giving a supportive evidence to the notions. To further confirm these and to explore a functional significance of palmitoylation in the receptor trafficking and signal transduction, the mutant receptors were stably expressed in CHO cells. This cell line was adopted because we had already shown that wtET B , when stably expressed in CHO cells, directly couples with members of both G i and G q families to inhibit adenylate cyclase and activate phospholipase C (PLC), respectively (26).
By co-transfecting CHO cells with each expression plasmid and pSVbsr r and then selecting for resistance against blasticidin, we obtained more than three individual clonal cell lines that stably expressed each receptor construct. The specificity of the binding was verified by its disappearance in the presence of excess unlabeled ET-1 and also by its absence in native CHO cells (Fig. 3). Subsequent incubation at 37°C  elicited internalization of the surface-bound ligand within minutes and, after 30 min, the ligand showed a patchy distribution both below the plasma membrane and around the nucleus, presumably being localized in lysozomes. As shown in Fig. 4, there were no apparent differences between wtET B and C2S/ C3S/C5S in the localization of Cy5-ET-1.
Failure of an Unpalmitoylated Mutant to Activate G i -To reveal a functional significance of palmitoylation in the coupling with G i , we tested the abilities of the mutant receptors to transmit an inhibitory effect on adenylate cyclase (Fig. 5). In CHO/wtET B cells, ET-1 caused a dose-dependent inhibition of forskolin-stimulated cAMP formation with EC 50 values of 58 Ϯ 1 pM (mean Ϯ S.E., n ϭ 3) and the maximum inhibition to ϳ40% of control. This effect was abolished by pretreatment of the cells with PTX (50 ng/ml for 16 h) as reported previously (26). An unpalmitoylated mutant C2S/C3S/C5S totally failed to transmit this effect while C2S/C3S did transmit the effect with EC 50 values of 63 Ϯ 6 pM (n ϭ 3) and the maximum inhibition to ϳ35%, both of which were comparable with those obtained for wtET B . Also comparable with the effect transmitted by wtET B were those transmitted by C2S, C3S, or C5S (data not shown). These results suggested a critical role of palmitoylation of ET B in the coupling with G i .
Failure of an Unpalmitoylated Mutant to Activate G q -To reveal a functional significance of palmitoylation in the coupling with G q , we tested the abilities of the mutant receptors to transmit a stimulatory effect on PLC (Fig. 6). In CHO/wtET B cells, ET-1 caused a dose-dependent stimulation of [ 3 H]IPs accumulation with EC 50 values of 2.5 Ϯ 0.9 nM (n ϭ 3) and the maximum effect of ϳ2.5-fold increase. PTX treatment of the cells partially inhibited the ET-1-induced accumulation, suggesting an involvement of both PTX-sensitive and -insensitive G-proteins, most likely the members of G i and G q families, respectively. Neither PTX-sensitive nor -insensitive increase in the accumulations was observed in cells expressing C2S/C3S/ C5S while both were observed in cells expressing C2S/C3S. [Ca 2ϩ ] i measurement with fura-2-loaded cells reproduced the findings (Fig. 7). Both PTX-sensitive and -insensitive increases were detected in cells expressing wtET B or C2S/C3S while neither was detected in cells expressing C2S/C3S/C5S. Both of the effects of ET-1 on [ 3 H]IPs accumulation and [Ca 2ϩ ] i in cells expressing C2S, C3S, or C5S were indistinguishable from those in cells expressing wtET B (data not shown). These results suggested a critical role of palmitoylation in the coupling with G q as in the case of G i .

Expression and Palmitoylation of the Carboxyl-terminal Deletion Mutants of ET B in COS Cells-We have constructed a series of carboxyl-terminal deletion mutants of ET B and
showed that ET-1 induced a [Ca 2ϩ ] i response in Ltk Ϫ cells expressing a mutant ⌬403 but not in cells expressing ⌬402 or ⌬400 (27). It was, however, left unknown whether the lack of response was due to a lack of palmitoylation or due to that of the carboxyl terminus per se. To resolve the issue and further explore a functional role of the carboxyl terminus, the deletion mutants were expressed and subjected to the same assays as described.
All the three deletion mutants were successfully expressed in and purified from the transfected COS cells as judged by   (Fig. 8a).
[ 3 H]Palmitic acid was metabolically incorporated into ⌬403 but not into ⌬402 or ⌬400 (Fig. 8b), as expected from the notion that Cys 402 , Cys 403 , and Cys 405 , but not Cys 400 are the potential palmitoylation sites of ET B .
Ligand Binding and Intracellular Trafficking of the Carboxyl-terminal Deletion Mutants Expressed in CHO Cells-Binding parameters obtained from saturation isotherms with [ 125 I]ET-1 on representative CHO cell clones stably expressing the deletion mutants are listed in Table I. Consistent with our previous results on Ltk Ϫ cells expressing these mutant receptors (27), all of them retained a high affinity to ET-1 and also that to ET-3 as judged from the competition binding experiments of [ 125 I]ET-1 with ET-3 (data not shown). In situ binding and internalization assays with Cy5-ET-1 on these mutant receptors also failed to reveal any apparent differences from the behavior of wtET B (data not shown).
A Palmitoylated Deletion Mutant ⌬403 Coupled with G q but Not with G i -ET-1-induced signaling was examined in cells expressing an unpalmitoylated mutant ⌬402 or a palmitoylated mutant ⌬403. ET-1 failed to inhibit forskolin-induced cAMP formation in CHO cells expressing either receptor (Fig.  9), suggesting a lack of coupling with G i . ET-1 also failed to stimulate ϳ2.5-fold increase, both of which were comparable with those obtained in cells expressing wtET B . A distinct difference in the responses elicited by ⌬403 from those elicited by wtET B was the lack of PTX-sensitive components both in [ 3 H]IPs accumulation (Fig. 10) and [Ca 2ϩ ] i response (Fig. 11), suggesting that the responses of the cells expressing ⌬403 were mediated solely by a member(s) of the G q family. DISCUSSION We have demonstrated that human ET B is covalently modified by thioesterification of palmitic acid and that the potential palmitoylation sites are the cysteine residues at amino acids 402, 403, and 405. The results obtained in the present study, however, did not indicate which of the three potential sites were actually palmitoylated in wtET B but suggested that palmitoylation of the individual cysteines was not an independent event but both alternative and hierarchical modifications of the three residues were taking place. Mutation of individual cysteines did not affect the level of [ 3 H]palmitic acid incorporation but that of two caused a significant reduction (Fig. 2b), suggesting that, in the wild-type receptor, not all of the three but two of them are palmitoylated and that, in the singlesubstitution mutants (C2S, C3S, and C5S), the remaining two cysteines were alternatively palmitoylated. The reduction of [ 3 H]palmitic acid incorporation in the double mutant C2S/C3S was more than 80% on densitometry suggesting the presence of a hierarchical order between Cys 402 /Cys 403 and Cys 405 , i.e. palmitoylation of either Cys 402 or Cys 403 may be a prerequisite for the efficient palmitoylation of Cys 405 . Hierarchical modification of two potential palmitoylation sites has been demonstrated so far only for rhodopsin (5). Identification of the actual palmitoylation sites in wtET B as well as verification of the alternative/hierarchical palmitoylation must await further studies that employ chemical or enzymatic methods to detect the modification of individual cysteines.
Also left unaddressed in the present study was the possible regulation of the palmitoylation level by receptor activation as has been described for ␤ 2 AR (31) or D 1 dopaminergic receptor (7). Because the expressed receptors were recovered after [ 3 H]palmitic acid labeling and washing the cells, we could at least conclude that wtET B as well as the various mutants were constitutively palmitoylated, without agonist-stimulation. However, because of the use of biotin-labeled agonist in the purification step, we did not assess the effect of receptor activation on the palmitoylation level. Alternative purification procedures including epitope-tagged receptors or specific antibodies are required to pursuit the issue.
[ 125 I]ET-1 binding assays on the wild-type and various mutant receptors gave K d and B max values within similar ranges regardless of the presence or absence of palmitoylation, suggesting that the overall integrity of the ligand-binding surface did not depend on the modification. These results are in line with the data obtained in many GPCRs (4,9,(12)(13)(14)(15) except for ␤ 2 AR. In the case of ␤ 2 AR, the lack of palmitoylation eliminated the GTP-sensitive high affinity state of the receptor, secondary to its uncoupling from G s (3,16). There was indeed a ϳ4-fold difference in the various mutant receptors used in the present study (Table I). It is, however, unlikely that the apparent difference in the affinities was due to the presence or absence of a GTP-sensitive high affinity state of the receptor, because GTP␥S failed to affect the [ 125 I]ET-1 binding to wtET B as well as the various mutant receptors (data not shown). The absence of a GTP-sensitive high affinity state of wtET B may imply that the receptor does not couple to G proteins prior to agonist binding as has been suggested for ET A (32). In the present study, a precise reason for the apparent difference in the binding affinities between the various mutants was left unknown. It is, however, at least clear that the differences in the G protein coupling capacities of the various mutants cannot be ascribed to the differences in the ligand binding affinities, because of the all or none feature of the coupling (discussed below).
In situ binding assays with Cy5-ET-1 (Fig. 4) indicated that palmitoylation was not required for the overall sequestration (cell surface expression and internalization) of wtET B . The binding assay used, however, is not quantitative and the intracellular traffic of the receptor molecule from the site of synthesis (endoplasmic reticulum) to the plasma membrane could not be assessed with this assay. Therefore, it is still an open question whether the intracellular traffic of wtET B is actively regulated by palmitoylation of the receptor molecule.
The most distinct finding of the present study was the failure of unpalmitoylated mutant receptors to activate the G proteindependent signaling pathways. The unpalmitoylated mutant C2S/C3S/C5S totally failed to transmit an inhibitory effect on adenylate cyclase (Fig. 5) and a stimulatory effect on PLC (Figs. 6 and 7) while C2S/C3S retained the signaling activities comparable with those of wtET B . These results indicated that either the presence of Cys 405 per se or the palmitoylation of the same residue in C2S/C3S was required for the G protein coupling. Because the specific requirement for the presence of Cys 405 can be excluded by the unaltered signaling activities of the mutant C5S, the signaling activities of C2S/C3S must be ascribed to the palmitoylation of Cys 405 . The unaltered signaling activities of C5S can in turn be ascribed to the palmitoylation of Cys 402 and Cys 403 . Therefore, we conclude that palmitoylation of at least one of the three potential sites is required for the G protein coupling, regardless of the G protein subtypes.
The activation of signaling pathways to adenylate cyclase and PLC by various receptors was an all or none phenomenon and the quantitative relationship between the palmitoylation level and the signaling effects was not detected in the present study. Both the inhibition of cAMP formation and the stimulation of [ 3 H]IPs accumulations caused by C2S/C3S were comparable with those by wtET B (and other single-substitution mutants) (Figs. 5 and 6) despite the apparently reduced palmitoylation level of C2S/C3S (Fig. 2b). A precise mechanism for C2S/C3S to cause the maximum effects is left unknown, however, possible explanations include the intracellular amplification of the signal by sequential interactions of receptor-G protein-effector molecules.
In addition to the critical role of palmitoylation in the G protein coupling, the present study also revealed a differential requirement for the carboxyl-terminal tail downstream to the palmitoylation site by G protein subtypes. The palmitoylated deletion mutant ⌬403 failed to transmit an inhibitory effect on adenylate cyclase (Fig. 9) but did transmit a stimulatory effect on PLC (Figs. 10 and 11) indicating that the carboxyl-terminal tail downstream to the palmitoylation site was required for the coupling with G i but not for that with G q . To reveal a role of the carboxyl terminus in ET B signaling, Aquilla et al. (33) constructed a deletion mutant which terminates within the seventh transmembrane domain and showed a lack of a capacity of this mutant to activate various cellular kinases. The critical role of palmitoylation and that of the carboxyl-terminal tail in the G protein coupling described here are consistent with their findings.
The requirement for palmitoylation in the G protein coupling has so far been documented for ␤ 2 AR (16) and ET A (12). In the case of ␤ 2 AR, the decreased coupling of unpalmitoylated mutant receptors was linked to an increased phosphorylation of the carboxyl tail of the receptor and not to the formation of a fourth intracellular loop (ICLIV) (16). Indeed there are as many as 10 putative phosphorylation sites in the carboxyl-terminal tail of ET B (34). However, both deletion mutants ⌬402 and ⌬403 lacked the potential phosphorylation sites in the carboxyl terminus and the difference of their abilities to activate G q depended on the presence or absence of the potential palmitoylation site Cys 402 . Therefore, it is unlikely that the lack of a capacity of unpalmitoylated mutants was secondary to an altered phosphorylation state of the carboxyl terminus. Although the data presented does not exclude the possible regulation of G protein coupling by phosphorylation, it favors a structural requirement for the formation of ICLIV in the G protein coupling of ET B .
Comparison of the data obtained here on ET B and that reported on ET A (12) revealed some features shared by these receptor subtypes. In both cases, the receptors appeared to be constitutively palmitoylated and palmitoylation was not essential for the ligand binding capacities. Another feature shared by ET A and ET B is the absolute requirement for palmitoylation in the coupling with G proteins of the G q family. This is, at present, a feature unique to these receptor subtypes; whether it is shared by any other GPCRs coupled with G q is a subject for the future study. A distinct difference between ET A and ET B lies in the structural basis for the coupling of ET A with G s and that for the coupling of ET B with G i . Palmitoylation was required for ET B -G i interaction but not for ET A -G s interaction. Using chimeric receptors between ET A and ET B , we have shown that ICLII of ET A and ICLIII of ET B are the major determinants for the selective coupling of each receptor subtype with G s and G i , respectively (26). The requirement for palmitoylation in ET B -G i interaction suggested an involvement of ICLVI either in selection or activation of G i . Also suggested from the data obtained from deletion mutants was an involvement of the cytoplasmic free tail. Thus, in the case of ET B -G i interaction, all of the three intracellular domains of the receptor, ICLIII, VI, and the cytoplasmic free tail appear to be involved.
Recently, a splice variant of human ET B was identified by molecular cloning (35). It is formed by a substitution of a large part of the carboxyl-terminal tail and the newly identified carboxyl-terminal sequence lacks any potential palmitoylation sites. When expressed in cultured cells, the splice variant retained ligand binding capacities but apparently lacked a capacity to activate G proteins, giving rise to a hypothesis that it may represent the "spare" ET B , the presence of which has been predicted by some functional studies (36,37). An obvious explanation for the failure of this splice variant to activate G proteins is a lack of palmitoylation. If this is the case, it raises a possibility of a novel mechanism to adjust cells' responses by alternative expression of palmitoylation-positive and -negative GPCR variants.
In conclusion, we have identified the potential palmitoylation sites of human ET B and revealed a critical role of the modification in the coupling with G proteins. The relevance of these findings to the functional defects of ET B variant will be clarified in the future study.