Distinct Subcellular Localization for Constitutive and Agonist-modulated Palmitoylation of the Human δ Opioid Receptor*

Protein palmitoylation is a reversible lipid modification that plays important roles for many proteins involved in signal transduction, but relatively little is known about the regulation of this modification and the cellular location where it occurs. We demonstrate that the humanδ opioid receptor is palmitoylated at two distinct cellular locations in human embryonic kidney 293 cells and undergoes dynamic regulation at one of these sites. Although palmitoylation could be readily observed for the mature receptor (Mr 55,000), [3H]palmitate incorporation into the receptor precursor (Mr 45,000) could be detected only following transport blockade with brefeldin A, nocodazole, and monensin, indicating that the modification occurs initially during or shortly after export from the endoplasmic reticulum. Blocking of palmitoylation with 2-bromopalmitate inhibited receptor cell surface expression, indicating that it is needed for efficient intracellular transport. However, cell surface biotinylation experiments showed that receptors can also be palmitoylated once they have reached the plasma membrane. At this location, palmitoylation is regulated in a receptor activation-dependent manner, as was indicated by the opioid agonist-promoted increase in the turnover of receptor-bound palmitate. This agonist-mediated effect did not require receptor-G protein coupling and occurred at the cell surface without the need for internalization or recycling. The activation-dependent modulation of receptor palmitoylation may thus contribute to the regulation of receptor function at the plasma membrane.

Palmitoylation, a post-translational lipid modification, plays an important role in the structure and function of a variety of proteins involved in cell signaling (1,2). Among them several G protein-coupled receptors (GPCRs), 3 mainly belonging to family A, have been shown to be palmitoylated on one or more cysteine residues located in the proximal region of their C-terminal tail (2). The attached palmitate has therefore been proposed to promote the formation of a fourth cytoplasmic loop through its anchoring in the membrane (3). Consequently, palmitoylation may have a profound effect on the local conformation of this domain and possibly controls interaction of GPCRs with specific regulatory proteins. It has been shown to display diverse roles in GPCR function, including G protein coupling, desensitization, trafficking, and targeting of the receptors (2).
The biochemical processes that regulate protein palmitoylation have remained poorly understood, and so far only a few enzymes involved in the formation or hydrolysis of palmitoyl-protein thioesters have been characterized (2,20). In addition, the subcellular site for palmitoylation of membrane proteins has remained elusive. For example, viral glycoproteins appear to be palmitoylated in the pre-/early Golgi compartment (21)(22)(23)(24), whereas attachment of fatty acids to cellular integral membrane proteins has been suggested to occur either in the endoplasmic reticulum (ER) (25) or at the plasma membrane (26,27).
The human ␦ opioid receptor (h␦OR) is one of the three opioid receptor subtypes (, ␦, ) that are targets for opiate drugs and endogenous opioid neuropeptides. It contains several cysteine residues in the proximal part of its C-terminal end (28,29) and thus may be palmitoylated. Because biosynthesis, maturation, and trafficking of the ␦ORs have been characterized in detail (30,31), we selected this receptor as a study model to examine the subcellular location and dynamic regulation of GPCR palmitoylation. We have demonstrated that h␦OR palmitoylation can occur at two distinct cellular locations. The newly synthesized receptors are first palmitoylated early in their biosynthetic pathway before they are transported to the cell surface, but palmitoylation/depalmitoylation cycles can also occur on the mature receptor at the plasma membrane. We also found that such constitutive receptorbound palmitate turnover is increased upon agonist-promoted receptor activation. This regulation occurred locally at the plasma membrane and did not require G protein coupling, internalization, or recycling of the receptor.
DNA Constructs and Cell Culture-The h␦OR (Gene accession number P41443) in pcDNA3 (Invitrogen) was tagged at the C-terminal end with a Flag epitope (DYKDDDDK) and expressed in stably transfected human embryonic kidney (HEK) 293S cells (HEK293S-h␦OR-Flag cells; 10.5 pmol/mg membrane protein, as determined by saturation binding using [ 3 H]bremazocine) as described previously (30). A DNA construct encoding the h␦OR with a cleavable influenza hemagglutinin signal peptide (KTIIALSYIFCLVFA), N-terminal Myc epitope (EQKLI-SEEDL) and C-terminal Flag epitope was created, and a stable cell line with inducible expression was established as described elsewhere (32,33). 4 Receptor expression in the HEK293 i -Myc-h␦OR-Flag cells was induced by adding tetracycline (0.5 g/ml; Invitrogen) into the culture medium.
Labeling Isolation of Cell Surface Receptors-Cells were biotinylated using sulfo-NHS-biotin, and cell surface receptors were isolated by a twostep purification using immobilized streptavidin and anti-FLAG M2 antibody as described (30). Alternatively, immunoprecipitated receptors were subjected to Western blot analysis using HRP-conjugated streptavidin.

Inhibition of Internalization and Detection of Internalized Receptors-
The assay used to detect internalized receptors was modified from Cao et al. (34). Briefly, cells were first washed with warm Dulbecco's PBS, cooled on ice, and incubated with gentle agitation at 4°C for 30 min in the same buffer containing 0.3 mg/ml sulfo-NHS-S-S-biotin. Excess biotin was quenched by adding Tris-HCl, pH 7.4, to a final concentration of 50 mM, after which the incubation was continued for 10 min at 4°C. Thereafter, cells were detached and incubated in warm DMEM, 20 mM Hepes, pH 7.5, 5 mM EDTA in the presence or absence of sucrose (0.35 M) for 30 min at 37°C. Internalization was induced by adding LE (10 M) and incubating cells for 30 min at 37°C. After cooling on ice, the medium was removed by centrifugation (500 ϫ g, 10 min), and cells were washed with 20 mM Hepes, pH 8.0, 150 mM NaCl, 5 mM EDTA, 10% (v/v) FBS. Biotin that remained attached to the cell surface was removed by incubating cells on ice for 10 min in 50 glutathione, 20 mM Hepes, pH 8.0, 150 mM NaCl, 5 mM EDTA, 10% (v/v) FBS. Finally, to quench the residual glutathione, cells were incubated on ice for 10 min in 20 mM N-ethylmaleimide, 1% (w/v) bovine serum albumin, 5 mM EDTA in PBS, pH 7.4. The efficiency of biotin removal was confirmed by verifying that no biotin labeling could be detected in the absence of agonist-induced internalization following glutathione treatment.

SDS-PAGE and Western
Blotting-For SDS-PAGE (10% separating gels), samples were denatured by heating at 95°C for 2 min under nondenaturing conditions. For detection of radioactivity, gels were treated with Enhance (PerkinElmer Life Sciences) according to the manufacturer's instructions, dried, and exposed at Ϫ80°C for 20 -120 days using the Biomax MS film and intensifying screens (Kodak). For Western blotting, proteins resolved in SDS-PAGE were transferred electrophoretically onto Immobilon P membrane (Millipore), and bound proteins were probed using anti-FLAG M2 antibody as described (30,35) or HRP-conjugated streptavidin (1:40,000). The relative intensities of the bands on autoradiograms and Western blots were analyzed by densitometric scanning with Agfa Arcus II or Agfa Duoscan HiD laser scanners and quantitation using Scion Image 4.02 software, subtracting the local background from each lane.
Flow Cytometry-Cells were cooled on ice and detached from cell culture flasks with PBS. To label plasma membrane receptors, 1 ϫ 10 6 cells/sample were incubated for 30 min on ice with monoclonal anti-c-Myc antibody (1:1,000) in PBS, 1% (v/v) FBS. After the residual unbound primary antibody was removed, cells were incubated with 0.5 g/ml phycoerythrin-conjugated rat anti-mouse IgG 1 in PBS, 1% (v/v) FBS for 30 min. After washing, dead cells were labeled with 7-amino-actinomycin D (0.25 g/1 ϫ 10 6 cells) for 10 min. All incubations were performed on ice. Ten thousand cells of each sample were analyzed on a BD Biosciences FACSCalibur TM flow cytometer. Cells were excited with a 488-nm argon ion laser, and phycoerythrin and 7-amino-actinomycin D emission was detected with a 585 Ϯ 21-nm band-pass filter and a 670-nm long-pass filter, respectively. Data analysis to quantitate changes in mean surface receptor fluorescence values were performed using the CellQuest TM Pro 4.02 software (BD Biosciences). Mean fluorescence of live cells minus mean fluorescence of cells stained only with phycoerythrin-conjugated secondary antibody was used for calculations.
Data Analysis-Data were analyzed using GraphPad Prism 4.02 software. For statistical t tests, the limit of significance was set at p Ͻ 0.05, and data are presented as the mean Ϯ S.E.

RESULTS
h␦OR Is Palmitoylated via Thioester Bonds-To determine whether the h␦OR is palmitoylated, we assessed whether it incorporates [ 3 H]palmitate during metabolic labeling. HEK293S cells expressing the C-terminally Flag-tagged receptor (HEK293S-h␦OR-Flag cells) were incubated for 60 min with [ 3 H]palmitate, and the receptor was immunoprecipitated using immobilized anti-FLAG M2 antibody and subjected to SDS-PAGE analysis and fluorography. As shown in Fig. 1A, [ 3 H]palmitate was incorporated into one major protein of M r 55,000 that comigrated with the mature receptor on Western blots (compare lanes 2 and 4) (30), whereas no labeled proteins were detected in nontransfected cells (lanes 1 and 3). This clearly demonstrates that, as is the case for many family A GPCRs, the h␦OR is modified by addition of palmitate. No distinct labeled band comigrating with the M r 45,000 receptor precursor (30) was detected, but occasionally an ill-defined [ 3 H]palmitate-labeled band migrating faster than the M r 55,000 mature receptor (ϳM r 48,000) was observed (Fig. 1A, lane 2). This species represents a partially glycosylated mature receptor form that contains only one N-linked glycan. 5 Covalent attachment of palmitate into proteins occurs primarily through a thioester linkage to cysteine residues; this linkage is chemically labile and can be cleaved by weak nucleophiles such as hydroxylamine at neutral pH. To assess the nature of the acyl linkage on the h␦OR, the immunoprecipitated [ 3 H]palmitate-labeled receptor was treated with either 1 M hydroxylamine (pH 7.2) or 1 M Tris-HCl (pH 7.2) and analyzed by SDS-PAGE, fluorography, and Western blotting. Treatment with hydroxylamine led to a loss of the 3 H signal on the fluorograms, whereas the protein remained intact (Fig. 1B, lanes 2 and 4, respectively), demonstrating that the acyl moiety on the receptor was attached via a thioester bond.
h␦OR Palmitoylation Is a Dynamic Process, Involving Newly Synthesized Receptors in Transit to the Plasma Membrane as well as Mature Receptors at the Cell Surface-The major h␦OR species that was found to incorporate [ 3 H]palmitate during the 60-min labeling period was the mature fully glycosylated form of M r 55,000 (Fig. 1A, lane 2), and this was observed even when the labeling was shortened to 10 min (see Fig.  4B). Given that the ER residency time of newly synthesized h␦ORs is relatively long (half-time of maturation ϳ 120 min) (30), the fast labeling kinetics of the mature receptor suggests that palmitoylation is a late post-translational event and might even take place once the protein has been inserted into the plasma membrane. To investigate further in which cellular subcompartment palmitoylation takes place, we first assessed which h␦OR species could incorporate [ 3 H]palmitate in cells treated with BFA. This drug interferes with protein transport to the cell surface and leads to the accumulation of newly synthesized proteins in the ER (36). When BFA was added to the labeling medium 30 min before [ 3 H]palmitate addition and maintained during the 30-min labeling period, label was incorporated into two major proteins of M r 45,000 and 55,000 ( Fig. 2A, lane 2), the label of which was removable by hydroxylamine treatment (compare lanes 1 and 2 in Fig. 2B). The smaller receptor species represents newly synthesized receptor precursors that accumulate in the ER in BFA-treated cells (30,37), 4 suggesting that these receptors were palmitoylated early in the biosynthetic pathway as their transport to the cell surface was blocked. However, because the M r 45,000 receptor species was not labeled in nontreated cells ( Fig. 2A, lane 1), it is unlikely that palmitoylation occurs to any considerable extent in the ER. To further investigate this early site of palmitate attachment to the h␦OR, the cells were treated with NOC, a microtubule-depolymerizing agent that inhibits microtubule-mediated vesicular transport and leads to accumulation of secretory proteins in pre-Golgi intermediates that fail to translocate to the Golgi (38). As seen in Fig. 2A, lane 3, the M r 45,000 receptor precursor incorporated [ 3 H]palmitate in the NOCtreated cells. Thus, these results suggest that palmitate is attached to the newly synthesized h␦ORs shortly after their export from the ER, most likely at the ER exit sites or the ER-Golgi intermediate compartment.
The observation that the M r 55,000 mature receptor was labeled in BFA-and NOC-treated cells ( Fig. 2A, lanes 2 and 3, respectively) indicates that palmitoylation can also occur after the receptor has been fully processed in the Golgi. This is suggested by the observation that a 30-min pretreatment with BFA is sufficient to completely block the transport of newly synthesized receptors to the cell surface (37). 4   determine whether this second event of palmitoylation occurs before exit from the Golgi/trans-Golgi network or at the plasma membrane, the HEK293S-h␦OR-Flag cells were biotinylated with a membrane-im-permeable reagent, sulfo-NHS-biotin, prior to [ 3 H]palmitate labeling. This allowed us to isolate receptors that were labeled at the plasma membrane by sequential purification using immobilized streptavidin and anti-FLAG M2 antibody (30). When the purified receptors were analyzed by Western blotting, only the mature M r 55,000 and M r 48,000 receptors were detected (Fig. 2C, lane 4), and as seen on lane 2, these receptors were labeled with [ 3 H]palmitate. Thus, the h␦OR palmitoylation can take place after the protein has been transported to the plasma membrane and does not only occur early in the biosynthetic pathway.
The most likely explanation for the finding that palmitate addition to the h␦OR can take place at two cellular locations is the reversible nature of the thioester bond, which can result in dynamic palmitoylation/depalmitoylation of the receptor. To find out whether the fast turnover of receptor-bound palmitate could contribute to the efficient incorporation of [ 3 H]palmitate into the mature receptors at the cell surface, we performed pulse-chase labeling experiments. The HEK293S-h␦OR-Flag cells were first labeled for 120 min and then incubated in the presence of an excess of unlabeled palmitate for varying lengths of time. At each time point cell surface proteins were biotinylated, and cell surface receptors were isolated. As can been seen in Fig. 2D, the labeling intensity of the mature cell surface receptors increased at the earliest chase times studied. This is possibly due to the presence of significant amounts of tracer in the palmitoylation donor pool (39) and/or to receptors that were labeled early after biosynthesis and transported to the cell surface during the chase. After 60 min, however, the labeling declined rapidly with an apparent half-time of 110 Ϯ 15 min (n ϭ 3). This is clearly shorter than the half-life of the cell surface receptor protein (15-20 h) as determined previously in [ 35 S]methionine/cysteine pulsechase experiments (30), indicating that palmitate that is bound to the mature cell surface receptor turns over more rapidly than the protein itself.
Palmitoylation Is Needed for Efficient Transport of Newly Synthesized Receptors to the Cell Surface-The results presented above suggest that the h␦OR can be palmitoylated at two cellular locations, early in the biosynthetic pathway as well as at the cell surface. To investigate palmitoylation of the newly synthesized receptors in more detail, we created a stably transfected HEK293 cell line (HEK293 i -Myc-h␦OR-Flag) that expresses the Myc and Flag epitope-tagged h␦OR under tetracycline induction. 4 Cells were treated with tetracycline for 7 h in the presence of BFA, and cell surface proteins were biotinylated with sulfo-NHS-biotin followed by immunoprecipitation of the receptor with anti-FLAG M2 antibody. Only the precursor form of the receptor was detected on Western blots with anti-FLAG M2 antibody (Fig. 3B, lane 1), but no cell surface receptors could be detected with HRP-conjugated streptavidin (Fig. 3C, lane 1), confirming the effective block of receptor cell surface trafficking with BFA. These intracellular receptors were able to incorporate [ 3 H]palmitate (Fig. 3A, lane 1), confirming the notion that the newly synthesized receptors are initially palmitoylated before they are transported to the cell surface. When BFA was removed and cells were incubated for 2 h in the presence of CHX, a heterogeneous band was detected both with anti-FLAG M2 antibody and HRP-conjugated streptavidin (lane 3 in Fig. 3, B and C, respectively), indicating that the intracellular receptors were able to reach the cell surface following removal of the transport block. The heterogeneity of the receptor band is most likely due to inefficient processing of the N-linked glycans by glycosyltransferases that need to be translocated from the ER to their correct locations in the Golgi after BFA removal. As seen on Fig. 3A 1 and 3). This suggests that the turnover of the ER-retained precursor-bound palmitate is  1 and 2) or Western blotting (lanes 3 and 4). The data shown are representative of two independent experiments. C, HEK293S-h␦OR-Flag cells were biotinylated with sulfo-NHS-biotin (0.5 mg/ml) at 4°C for 30 min and then labeled with [ 3 H]palmitate for 30 min at 37°C. Cellular membranes were solubilized, and receptors from 25% of the extract were immunoprecipitated using immobilized anti-FLAG M2 antibody to isolate the total pool of labeled receptors (lanes 1 and 3). The rest of the extract was used to isolate the cell surface receptors by sequential precipitations using first immobilized streptavidin and then anti-FLAG M2 antibody (lanes 2 and 4). Purified receptors were analyzed by SDS-PAGE and fluorography (lanes 1 and 2) or Western blotting (lanes 3 and 4). approximately the same as that of the cell surface receptors (compare with Fig. 2D). Furthermore, because the transit time of receptors through the Golgi is relatively fast, ϳ10 min, (30), these data indicate that a majority of the receptors that are inserted into the plasma membrane still contain palmitate that has been incorporated into the receptor during their transport thought the secretory pathway.
To investigate the potential role of palmitoylation for transport of the newly synthesized receptors to the cell surface, the HEK293 i -Myc-h␦OR-Flag cells were treated with 2-Br for 16 h before inducing receptor expression. As seen on Fig. 3A, lane 2, this treatment was able to completely block incorporation of [ 3 H]palmitate into the newly synthesized receptors without any major effect on the protein content (Fig. 3B,  lane 2), consistent with findings obtained previously for other proteins (8,25,40,41). The nonpalmitoylated receptors were able to reach the cell surface, albeit at a reduced level (Fig. 3C, compare lanes 4 and 3). This suggests that palmitoylation greatly facilitates cell surface expression of the newly synthesized receptors but is not absolutely required.
To study the significance of palmitate incorporation more quantitatively, we assessed the appearance of newly synthesized receptors at the cell surface by flow cytometry. As seen in Fig. 3D, blocking of palmitoylation with 2-Br decreased cell surface expression of the receptors by 61.4 Ϯ 0.4% (compare the first and fourth bar), supporting the notion that palmitoylation has a substantial facilitating effect. This was also demonstrated by the fact that there was a significant decrease in the number of cell surface receptors (35.6 Ϯ 2.0%) even when 2-Br was added to the culture medium only during the 2-h chase incubation (compare the first and second bar in Fig. 3D). In contrast, adding an excess of palmitate into the culture medium during the chase had no effect (compare the first and third bar in Fig. 3D).
h␦OR Palmitoylation Is Modulated by Opioid Agonists-The fact that palmitoylation of the h␦OR appears to be a reversible and a dynamic modification that can also occur at the cell surface prompted us to investigate whether it could be modulated by opioid agonists.  1 and 2, right panel). This LE-induced increase in [ 3 H]palmitate incorporation into the receptor was rapid, being clearly apparent as early as 10 min following receptor stimulation, the earliest labeling time point studied (Fig. 4B). [ 3 H]Palmitate incorporation into the M r 55,000 mature receptor was also increased in the presence of another peptidic opioid agonist, DPDPE, but not in the presence of a ␤ 2 -adrenergic FIGURE 3. Addition of palmitate to newly synthesized h␦ORs is needed for their efficient transport to the cell surface. A-C, HEK293 i -Myc-h␦OR-Flag cells expressing the Myc and Flag epitope-tagged h␦OR in a tetracycline-inducible manner were pretreated or not with 2-Br (2-Br Pre, 100 M) for 16 h as indicated. Receptor expression was induced by adding tetracycline (0.5 g/ml) to the culture medium, and transport of newly synthesized receptors to the cell surface was inhibited with BFA (5 g/ml). Cells were incubated for 7 h in the continued absence or presence of 2-Br and then were labeled with [ 3 H]palmitate for 60 min at 37°C (A). Alternatively, cells were treated as described in A without adding [ 3 H]palmitate, and cell surface proteins were labeled with sulfo-NHS-biotin (0.5 mg/ml) at 4°C for 30 min (B and C ). Cells were then harvested immediately (lanes 1 and 2) or after a 2-h chase in the absence of BFA, tetracycline, and 2-Br but in the presence of CHX (20 g/ml) and palmitate (100 M) (lanes 3 and 4). Total cellular membranes were isolated and extracted in DDM, and receptors were isolated by immunoprecipitation. Aliquots of the eluates were analyzed by SDS-PAGE and fluorography receptor agonist, isoproterenol (Fig. 4A, left panel, lanes 3 and 4, respectively), establishing the specificity of the agonist-promoted incorporation of the radiolabel. Taken together, these findings suggest that opioid agonists are able to modulate receptor palmitoylation, possibly by increasing the turnover rate of h␦OR-bound palmitate (exchange of nonlabeled receptor-bound palmitate for labeled palmitate).
To test this idea more directly, pulse-chase studies were performed. The HEK293S-h␦OR-Flag cells were labeled with [ 3 H]palmitate for 120 min followed by incubation with 100 M unlabeled palmitate in the presence or absence of 10 M LE for varying lengths of time. The agonist was added following a 60-min preincubation to avoid the initial increase in receptor labeling (see Fig. 2D). As shown in Fig. 5, LE promoted a greater decrease in receptor-bound [ 3 H]palmitate at all time points tested without any change in the amount of receptor protein (data not shown). The maximal effect of the agonist treatment was observed following 30 min of chase, the remaining receptor-bound [ 3 H]palmitate in LE-treated cells being only 56 Ϯ 11% (n ϭ 5) of that in the control cells. Because the potential confounding effects of recycling of the palmitate donor pool (39) were not taken into account in these studies, one cannot determine the absolute turnover rates. Nevertheless, the data clearly demonstrate that agonist stimulation increases palmitate turnover on the receptor. Combined with the observation that an opioid but not an adrenergic agonist increased incorporation of [ 3 H]palmitate during pulse labeling of the h␦OR (Fig. 4A), these results suggest that the state of palmitoylation is dynamically regulated in a receptor-activation dependent manner. We therefore investigated whether the agonistmodulated receptor palmitoylation requires coupling of the receptor to G proteins and modulation of its downstream effectors. For that purpose, we tested the effect of PTX on the LE-induced increase in incorporation of [ 3 H]palmitate into the receptor. Treatment with PTX, which efficiently uncoupled the h␦OR from G i proteins as confirmed by its ability to block the LE-induced inhibition of cyclic AMP accumulation (data not shown), was without effect on the LE-promoted increase in [ 3 H]palmitate incorporation into the mature M r 55,000 receptor (Fig.  6, compare lanes 1 and 3 with lanes 2 and 4). This indicates that agonistmodulated palmitoylation of the h␦OR requires receptor activation but is independent of G protein coupling.

Agonist-induced Increase in the Turnover Rate of h␦OR-bound Palmitate Occurs at the Cell Surface and Does Not Require Internalization-Because
LE is a peptidic agonist that cannot cross the plasma membrane (37), it is expected that the increase in the turnover rate of receptor-bound palmitate occurs at the cell surface. To verify this hypothesis, we first assessed the effect of LE treatment on [ 3 H]palmitate incorporation into the h␦OR in HEK293S-h␦OR-Flag cells treated with BFA. As seen in Fig. 7A, LE was able to enhance labeling of the mature M r 55,000 receptor, whereas it did not increase that of the intracellularly retained M r 45,000 receptor precursor (compare lanes 3 and 4). Similar results were also obtained using MON (Fig. 7A, lanes 5 and 6), a drug that interferes with protein transport, leading to accumulation of newly synthesized proteins in the medial Golgi (42). In another set of experiments, the cells were labeled with [ 3 H]palmitate in the presence or absence of the agonist after biotinylating the cell surface receptors with sulfo-NHS-biotin. As seen in Fig. 7B, the agonist-induced increase in palmitate incorporation was apparent in the cell surface pool of receptors (compare lanes 3 and 4).
The data presented above indicate that dynamic regulation of receptor palmitoylation can occur at the cell surface. However, because binding of opioid agonists to their cognate receptors leads to rapid internalization (31), enhancement in the palmitate turnover might require internalization of the receptors followed by their recycling to the cell  surface. To test this possibility, we assessed whether inhibition of receptor internalization by sucrose treatment could blunt the agonist-enhanced incorporation of [ 3 H]palmitate. To confirm that the sucrose treatment inhibited internalization of the h␦OR in the HEK293S-h␦OR-Flag cells, we used a biochemical assay to measure internalization (34). For this purpose, cells were biotinylated using a cell-impermeable reagent, sulfo-NHS-S-S-biotin, and following a 30-min pretreatment with or without sucrose, LE was added and the cells were incubated for another 30 min. The remaining biotin at the cell surface was then removed by a cell-impermeable reducing agent, glutathione, allowing detection of the internalized receptors by their resistance to cleavage. Analysis of the immunoprecipitated receptors on Western blot using HRP-conjugated streptavidin revealed that agonist treatment induced internalization of the receptor (Fig. 8A, compare lanes 3 and 4), which was efficiently inhibited with sucrose (Fig. 8A, compare lanes 6 and 4) (70.7 Ϯ 4.7%, n ϭ 3). For unknown reasons, sucrose treatment led to a decrease in [ 3 H]palmitate incorporation into the receptor both at basal and agonist-stimulated conditions. However, the agonist treatment still promoted a 1.6-fold increase in receptor [ 3 H]palmitate incorporation (Fig. 8B, compare lanes 3 and 4) that was comparable with the 1.4-fold increase observed in the sucrose-nontreated cells (Fig. 8B, compare  lanes 1 and 2). This indicates that the agonist-induced enhancement in the turnover of receptor-bound palmitate is not dependent on internalization of the receptors. Similarly, recycling of the internalized receptors to the cell surface appears not to be necessary for the agonist-mediated effect on the h␦OR palmitoylation. This was shown by the inability of MON, a drug that inhibits recycling of cell surface receptors (42), to block the agonist-enhanced incorporation of [ 3 H]palmitate into the M r 55,000 receptor (see Fig. 7A, lanes 5 and 6). Taken together, these data support the notion that agonist-induced palmitoylation/depalmitoylation cycle of the h␦OR occurs at the plasma membrane and does not require internalization or cycling of the receptors between the plasma membrane and endocytic vesicles.

DISCUSSION
The results of the present study indicate that the h␦OR is modified by post-translational addition of palmitate, thus joining the growing number of membrane proteins that have been shown to be palmitoylated. We demonstrate that the newly synthesized h␦ORs are palmitoylated constitutively while in transit to the cell surface and that the modification is needed for their efficient transport. However, because of the dynamic nature of the modification, the receptor-bound palmitate turns over rapidly, and binding of an agonist to the receptor increases the turnover rate even further in a receptor activation-dependent manner at the cell surface. This agonist-mediated effect on receptor palmitoylation did not require receptor-G protein coupling, internalization, or recycling of the receptor, thus suggesting that it is a consequence of conformational changes promoted by agonist binding. Such plasma membrane-restricted regulation of receptor palmitoylation provides new insights into the potential role of this modification and of the enzymatic machinery involved.  1 and 2) or with (lanes 3 and 4) PTX (100 ng/ml) for 16 h and labeled with [ 3 H]palmitate for 30 min at 37°C in the absence (lanes 1 and 3) or presence (lanes 2 and 4) of LE (10 M). Cellular membranes were solubilized, and immunoprecipitated receptors were analyzed by SDS-PAGE and fluorography. The fluorogaph shown is representative of two independent experiments.  Determining the cellular location in which palmitoylation of membrane proteins takes place has proven to be a difficult task, and several cellular locations for this modification have been proposed. For example, palmitoylation of the immunoglobulin E receptor (27), transferrin receptor (26), and caveolin-1 (43) is suggested to occur at the plasma membrane, whereas several other membrane proteins appear to be modified prior to the final processing of their N-and O-linked oligosaccharides in the Golgi (21,22,24). In addition, ␣-bungarotoxin receptors were found to be palmitoylated in the ER during their assembly (25). These apparently contradictory results may result from the fact that proteins may move rapidly from one cellular site to another, which makes it difficult to pinpoint the cellular location in which addition of palmitate into the protein initially occurs. Another possibility is that membrane proteins may be palmitoylated in more than one location within the cell because of the reversibility of the modification. Our results support the latter possibility and demonstrate directly that the h␦OR is palmitoylated constitutively at two cellular locations.
The newly synthesized h␦ORs were initially palmitoylated within the secretory pathway before they were inserted into the plasma membrane. This was demonstrated by the ability of the M r 45,000 receptor precursor to incorporate [ 3 H]palmitate in BFA-, NOC-, and MON-treated cells in which protein transport to the cell surface was blocked. However, in nontreated cells that expressed the receptor constitutively, only the mature M r 55,000, and 48,000 receptor forms were labeled, indicating that normally palmitoylated precursor forms do not accumulate. This suggests that the initial h␦OR palmitoylation is a late post-translational modification, occurring either concomitantly or just before the N-and O-linked oligosaccharides are being processed in the Golgi. The most likely sites are the ER exit sites or the ER-Golgi intermediate compartment. Our interpretations are in agreement with those made previously for viral glycoproteins, which are palmitoylated either in the ER-Golgi intermediate compartment or in the cis-Golgi (22)(23)(24). Similarly, rhodopsin has been proposed to be palmitoylated early in the secretory pathway (44).
Efficient incorporation of [ 3 H]palmitate into the ER-retained receptor precursors in BFA-treated but not in nontreated cells could be due to retrotranslocation of Golgi-resident protein palmitoyltransferases to the ER. This possibility is, however, unlikely, because the M r 45,000 receptor precursors were also palmitoylated in NOC-treated cells, a treatment that does not promote rapid retrotranslocation of Golgi-resident enzymes (38). An alternative possibility is that the receptor precursors may be palmitoylated in BFA-treated cells because of their recycling between the ER and the ER exit sites, a pathway that apparently remains functional following BFA treatment (45). In any case, h␦ORs that are unable to leave the ER and accumulate in this compartment before being degraded (35) do not appear to undergo palmitoylation. Indeed, receptor precursors failed to incorporate [ 3 H]palmitate to any detectable level in nontreated cells despite the fact that a substantial proportion of the receptors exist in the precursor form at steady state. This notion is consistent with previous findings showing that ER-retained mutants of the luteinizing hormone receptor and rhodopsin are not palmitoylated (46,47). Interestingly, permanently misfolded vesicular stomatitis virus G proteins have been shown to be unable to enter the ER exit sites (48), which may be the case for the misfolded/incompletely folded GPCRs as well. This might explain the difference in palmitoylation observed for the ER-retained misfolded/incompletely folded receptors and wild type folding-competent receptors in BFA-treated cells.
The addition of palmitate to the newly synthesized h␦ORs prior to their plasma membrane insertion suggests that the modification may have a role in the transport process. This was supported by the fact that blocking the palmitoylation with 2-Br reduced the number of receptors that reached the cell surface. This observation is in line with previously published studies (25, 41, 49 -51). For example, blocking palmitoylation with 2-Br or mutation of the three C-terminal palmitoylated cysteines of the CCR5 receptor was found to lead to intracellular accumulation of the protein (41,51).
The occurrence of palmitoylation at later steps in the h␦OR life cycle was first suggested by the fast labeling kinetics of the mature, fully processed receptors (M r 48,000 and M r 55,000), which were the only [ 3 H]palmitate-labeled receptor species observed even during very short labeling periods in cells expressing the receptor constitutively. Indeed, these data indicated that [ 3 H]palmitate could still be incorporated into the receptor following the final processing of the carbohydrates, pointing to the trans-Golgi network and/or the plasma membrane as potential sites for active palmitoylation. The finding that the mature receptors were able to incorporate [ 3 H]palmitate even following treatment with the protein transport blockers BFA, NOC, and MON is also consistent with late palmitate incorporation. However, it made the trans-Golgi network an unlikely location for the observed palmitoylation reaction because the transit time through the Golgi has been shown to be very short (ϳ10 min) (30). Thus all of the receptors should have exited the Golgi before the labeling was initiated because the blocks were initiated 15 min (NOC) or 30 min (BFA and MON) before adding [ 3 H]palmitate. Because the transport of newly synthesized receptors was blocked before the labeling, it follows that the M r 55,000 and, to a lesser extent, the M r 48,000 receptor species were palmitoylated after they reached the plasma membrane. This interpretation was confirmed in experiments in which plasma membrane receptors were biotinylated prior to metabolic labeling. Detection of streptavidin-purified receptors that had incorporated [ 3 H]palmitate proved that palmitoylation of the h␦OR is not entirely dependent on newly synthesized receptors and can occur once the receptors have reached the cell surface. This is in agreement with previous studies showing that red blood cells and platelets support protein palmitoylation in the absence of the Golgi or ER (52,53), as do isolated nerve growth cones (54) and retinal outer segments (55). This is also consistent with the finding that the h␦OR-bound palmitate turns over faster than the protein itself, indicating that the mature receptor is subject to a dynamic constitutive palmitoylation/depalmitoylation cycle at the plasma membrane.
The notion that dynamic palmitoylation events occur at the cell surface was further supported by the ability of the membrane-impermeable opioid agonists LE and DPDPE to increase the turnover of the receptorbound palmitate in pulse and pulse-chase labeling experiments. Interestingly, the LE-promoted increase in [ 3 H]palmitate incorporation into the receptor was observed even when the receptor-G protein coupling was impaired by PTX, ruling out the contribution of classical G proteinmediated signaling pathways in the control of the palmitoylation/depalmitoylation cycle. Although one cannot exclude the potential role of accessory proteins that could be recruited to the receptor in a G proteinindependent manner, the simplest hypothesis is to propose that the conformational changes promoted by the agonist increase the accessibility of the palmitoylated residues to palmitoyl-protein thioesterases and/or transferases.
Activation-dependent increase in the turnover of receptor-bound palmitate has been demonstrated previously for the ␤ 2 -adrenergic (5), ␣ 2A -adrenergic (15), and serotonin 5-HT4a receptors (17,18). Similarly, the ␣-subunits of the G s and G i proteins have been found to undergo activation-dependent cycles of palmitoylation/depalmitoylation (4, 9 -11). This activation-dependent modulation of palmitoylation of the two key components in the GPCR signal transduction pathway argues that regulation of the palmitoylation state of these proteins has an important functional role. Among the roles that have been suggested (2) and could be controlled by palmitoylation/depalmitoylation cycles at the plasma membrane, modulation of receptor-G protein coupling efficacy as well as receptor desensitization and internalization come to mind. However, the activation-dependent palmitoylation/depalmitoylation cycle of the h␦OR did not require receptor internalization, because inhibiting endocytosis with sucrose did not prevent the LE-mediated increase in [ 3 H]palmitate incorporation. Similarly, receptor recycling from the endosomes to the plasma membrane was not found to be required for efficient agonist-promoted turnover of the receptor-bound palmitate, as the LE-mediated increase in [ 3 H]palmitate incorporation was not affected by MON. These results are in contrast to the regulation of another post-translational modification of GPCRs, namely phosphorylation. Indeed, at least for the ␤ 2 -adrenergic receptor, dephosphorylation has been shown to occur in the endosomes before the receptor is recycled back to the plasma membrane (56). This apparent difference in the cellular location of regulated palmitoylation and phosphorylation cycles of GPCRs may reflect differences in the subcellular distribution of the participating enzymes.
Membrane-bound protein palmitoyltransferase activities have been identified in the mammalian ER, Golgi, and plasma membrane fractions (21,57,58), but the instability of the enzymatic activity has hampered purification and molecular identification of the enzymes. Recently, however, a family of proteins containing a conserved DHHC cysteinerich domain has emerged as potential enzymes capable of catalyzing the transfer of palmitate onto a variety of substrates (59 -65). Interestingly, one of these proteins found in yeast, Akr1p, was found to have a role in pheromone signaling and to interact with the C-terminal tail of the a-factor receptor (Ste3p) and G protein ␤-subunit (Ste4p) (66 -68). However, it is not known whether Akr1p is able to function as a palmitoyl transferase for the a-factor receptor. Similarly, none of the DHHCcontaining proteins has been found to catalyze palmitoylation of mammalian GPCRs to date.
In contrast to potential protein palmitoyltransferases, the molecular characterization of palmitoyl-protein thioesterases has been less successful. Only one putative mammalian cytosolic protein thioesterase, the acyl-protein thioesterase 1, has been purified and cloned (69). This enzyme was found to catalyze depalmitoylation of the G protein ␣-subunits, p21 ras , and endothelial nitric-oxide synthase (69 -71). However, no additional substrates have been identified since these original reports. The identity of the enzyme responsible for agonist-mediated h␦OR depalmitoylation thus remains an open question.
Taken together, the results of the present study indicate that palmitoylation of the h␦OR is a highly dynamic and reversible modification, occurring constitutively while receptors are transported to the cell surface, and that addition of palmitate into the newly synthesized receptors appears important for the efficient intracellular transport of these proteins. Furthermore, because h␦OR palmitoylation also appears to be regulated at the plasma membrane in an activation-dependent manner, it is likely to have an important role in agonist-induced signal transduction.