Regulation of Gαi Palmitoylation by Activation of the 5-Hydroxytryptamine-1A Receptor*

Nearly all α subunits of heterotrimeric GTP-binding regulatory proteins (G proteins) are palmitoylated at cysteine residues near the N terminus. A regulated cycle of palmitoylation could provide a mechanism for modulating G protein signaling by affecting protein interactions and localization of the subunit. In the present studies we utilized both [3H]palmitate incorporation and pulse-chase techniques to address the dynamics of αi palmitoylation in Chinese hamster ovary cells. Both techniques demonstrated a dose- and time-dependent change in [3H]palmitate labeling of αi upon activation of stably expressed 5-hydroxytryptamine-1A receptors by the agonist (+/−)-2-dipropylamino-8-hydroxy-1,2,3,4-tetrahydronaphthalene hydrobromide (DPAT), with an EC50 of ∼10 nm. For the incorporation assay, DPAT elicited an approximate doubling in labeling at the earliest time point measured. For the pulse-chase assay, DPAT promoted a significant loss of radiolabel almost equally as fast. These data demonstrate that the exchange of palmitate on αi is increased upon stimulation of 5-hydroxytryptamine-1A receptors through the combined processes of depalmitoylation and palmitoylation. These results provide the basis for extending the concept of regulated exchange of palmitate beyond Gs and provide a framework for exploring the specific functional attributes of the palmitoylated and depalmitoylated forms of subunit.

Nearly all ␣ subunits of heterotrimeric GTP-binding regulatory proteins (G proteins) are palmitoylated at cysteine residues near the N terminus. A regulated cycle of palmitoylation could provide a mechanism for modulating G protein signaling by affecting protein interactions and localization of the subunit. In the present studies we utilized both [ 3 50 of ϳ10 nM. For the incorporation assay, DPAT elicited an approximate doubling in labeling at the earliest time point measured. For the pulse-chase assay, DPAT promoted a significant loss of radiolabel almost equally as fast. These data demonstrate that the exchange of palmitate on ␣ i is increased upon stimulation of 5-hydroxytryptamine-1A receptors through the combined processes of depalmitoylation and palmitoylation. These results provide the basis for extending the concept of regulated exchange of palmitate beyond G s and provide a framework for exploring the specific functional attributes of the palmitoylated and depalmitoylated forms of subunit.

H]palmitate incorporation and pulse-chase techniques to address the dynamics of ␣ i palmitoylation in Chinese hamster ovary cells. Both techniques demonstrated a dose-and time-dependent change in [ 3 H]palmitate labeling of ␣ i upon activation of stably expressed 5-hydroxytryptamine-1A receptors by the agonist (؉/؊)-2-dipropylamino-8-hydroxy-1,2,3,4-tetrahydronaphthalene hydrobromide (DPAT), with an EC
Heterotrimeric guanine nucleotide-binding regulatory proteins (G proteins) 1 transduce signals from cell-surface receptors to intracellular effectors via multiple pathways. The phenomenon of signal transduction is relevant to both normal and abnormal physiological processes, prompting significant advances toward understanding how signal amplitude and duration are regulated at the level of receptor, G protein, and effector. Modulation of signaling has been suggested to occur through direct protein interactions (1) and by dynamic modifications that include phosphorylation (2,3) and palmitoylation (4 -9).
Palmitoylation is a reversible post-translational modification that links the fatty acid palmitate to cysteine residues by a thioester bond. Nearly all ␣ subunits of the four G protein families, G s , G i , G q , and G 12 , are palmitoylated at cysteine residues near the N terminus (10 -14). G i family ␣ subunits, which are palmitoylated at Cys 3 , are also co-translationally N-myristoylated at Gly 2 via an irreversible amide bond (9). Protein-bound palmitate has widespread functional significance, influencing protein interactions and membrane localization for G protein ␣ subunits and other proteins including G protein-coupled receptors and kinases (15)(16)(17), Ras proteins (18 -20), nonreceptor tyrosine kinases (21)(22)(23), and endothelial nitric-oxide synthase (24,25).
For the G i family ␣ subunits, palmitoylation decreases the affinity of ␣ i and ␣ z for the regulators of G protein signaling (RGS) proteins RGS4, GAIP, and G z GAP, and decreases the maximal rate of GTP hydrolysis of the ␣ subunit-RGS protein complex (26). Cys 3 mutations, which abrogate palmitoylation, have been reported variously to enhance D 2 receptor-mediated inhibition of adenylyl cyclase by ␣ z (27), to have no effect on D 2 receptor-mediated activation of mitogen-activated protein kinase by ␣ z (28), and to inhibit coupling of the ␣ 2A -adrenoreceptor to ␣ i (29). Cys 3 mutations also decrease the extent of subunit association with membrane (12,27), which for ␣ i (30) and ␣ z (28) is rescued by overexpression of ␤␥. It has been proposed that palmitoylation of ␣ subunits maintains membrane association as part of a dual signal (28) bilayer-trapping mechanism (31) similar to that proposed for acylated nonreceptor tyrosine kinases (32) and Ras proteins (20). Furthermore, palmitoylation may maintain ␣ subunits specifically at the plasma membrane and quite possibly in specific subdomains (33,34) because palmitoylation-deficient ␣ z mutants mistarget to internal membranes (28), as do nonpalmitoylated Gpa1 (yeast ␣ subunit) (35), fyn (22), and Ha-Ras (20).
Regulated palmitoylation could provide a mechanism for modulating G protein signaling by allowing for changes in protein interactions and localization of the ␣ subunit. Despite rigorous efforts, very little is known about the potential palmitoyltransferase (36 -38) and palmitoylesterase (39,40) enzymes that are specific for G protein ␣ subunits; characterization of enzyme regulation awaits successful purification and/or cloning of these proteins. Recent progress toward this end has been made by the identification and purification of a cytoplasmic acyl-protein thioesterase that depalmitoylates G protein ␣ subunits in vitro and in vivo (41). With respect to the addition of palmitate, no thiotransferase enzyme specific for G protein ␣ subunits has been purified to homogeneity, and it is currently unclear as to whether this reaction is an enzymatic or nonenzymatic process (42).
Superimposed on the potential regulation of enzymes that catalyze turnover of palmitate is the possibility that palmitoylation is modulated by the activation state or subcellular location of the ␣ subunit. It has been shown that the palmitoylation/depalmitoylation cycle is activation-dependent for ␣ s (43)(44)(45), the ␤ 2 -adrenergic receptor (46), and endothelial nitricoxide synthase (47). An increase in depalmitoylation of ␣ s , as measured by the release of [ 3 H]palmitate in pulse-chase assays, was demonstrated upon activation of G s through the ␤ 2 -adrenergic receptor (43,44) and by inhibition of GTPase activity by mutagenesis (43). An increase in [ 3 H]palmitate incorporation was also demonstrated following activation of G s through the ␤ 2 -adrenergic receptor (43)(44)(45) and cholera toxin (45), presumably due in part to an increase in depalmitoylation, facilitating an exchange of palmitate for [ 3 H]palmitate. Although regulated palmitoylation/depalmitoylation has been clearly demonstrated for the ␣ subunit of G s , regulated palmitoylation of ␣ subunits belonging to the G i , G q , and G 12 classes has only begun to be addressed. Recent reports have shown an increase in incorporation of [ 3 H]palmitate for ␣ q/11 (48) and ␣ i (49) upon gonadotropin-releasing hormone receptor activation in pituitary cells; for ␣ q , ␣ i , and ␣ o upon serotonin receptor activation in rat brain membranes in vitro (50); and for ␣ q and ␣ o upon ␣-adrenergic receptor activation in aortic membranes in vitro (51). The concept of palmitate exchange, however, has not been carefully evaluated in an intact cell setting.
In the present studies we have utilized both [ 3 H]palmitate incorporation and pulse-chase techniques to address the dynamics of ␣ i palmitoylation. We demonstrate specifically that the exchange of palmitate on ␣ i is increased upon stimulation of 5-hydroxytryptamine-1A (5-HT 1A ) receptors through the combined processes of depalmitoylation and palmitoylation. Our results provide evidence for a regulated cycling of palmitate and provide the basis for extending the concept of regulated exchange beyond the G s family. The rapid changes in palmitoylation/depalmitoylation suggest a role in transduction.
Cell Culture-Chinese hamster ovary (CHO) cells expressing the human 5-HT 1A receptor (XbaI-BamHI restriction fragment of G21 (54) in pcDNA1/neo) were provided as a gift from Dr. Perry B. Molinoff. Cells were maintained in monolayer culture in HAM F-12 with L-glutamine supplemented with 10% charcoal-treated fetal bovine serum, 100 units/ml penicillin, 100 g/ml streptomycin, and 400 g/ml Geneticin ® at 37°C in a humidified atmosphere of 5% CO 2 . Single 100-mm plates were used for each treatment. Cells were split 2-4 days prior to each experiment and grown to 70 -100% confluency.
Cell Fractionation-Plates were placed on ice, the medium was aspirated, and the cells were washed twice with ice-cold Dulbecco's phosphate-buffered saline followed by harvesting with a cell scraper into 500 l of ice-cold 20 mM HEPES, pH 7.2, 2 mM MgCl 2 , 1 mM EDTA, pH 7.2, 2 g/ml aprotinin, and 2 g/ml leupeptin (HME/PI). After 5 min on ice, cells were quick frozen in dry ice/EtOH and stored at Ϫ80°C until use. Following quick thawing, cells were homogenized by 15 passes through a 26-gauge needle/1-cc syringe on ice. Homogenates were centrifuged at 1000 ϫ g for 5 min at 4°C to pellet nuclei and unbroken cells. The Incorporation of radiolabel was assessed by immunoprecipitation of ␣ i from subsequently prepared membranes followed by SDS-PAGE/fluorography (inset) and densitometry. A representative of two independent experiments is shown.
supernatant was centrifuged at 100,000 ϫ g for 60 min at 4°C to pellet membranes. The pellet was washed twice and resuspended in 500 l of HME/PI by 10 passes through a 26-gauge needle/1-cc syringe on ice.
SDS-Polyacrylamide Gel Electrophoresis (SDS-PAGE), Fluorography, and Densitometry-Samples and prestained molecular weight markers were electrophoresed on 9% SDS-polyacrylamide gels. Gels were stained with 50% methanol, 10% acetic acid, and 0.25% Coomassie ® Blue R-250 and destained with 30% methanol and 10% acetic acid followed by 10% acetic acid alone. For 3 H experiments, gels were subsequently incubated in Amplify TM for 30 min at room temperature. For all experiments, gels were then dried at 80°C for 60 min, exposed to preflashed Hyperfilm ( 3 H experiments at Ϫ80°C, 35 S experiments at room temperature), and quantified by densitometry using multiple exposures and comparisons to standard curves to verify that all bands were within film linear range.

RESULTS
To address the hypothesis that activation of a G i -coupled receptor induces changes in palmitoylation of ␣ i , we carried out experiments using CHO cells stably expressing the 5-HT 1A receptor. The pharmacological properties of the 5-HT 1A receptor have been extensively characterized, and it is quite clear from direct measurements of G protein activation (55) and PTX sensitivity of effector regulation that the 5-HT 1A receptor couples primarily to members of the G i family. The CHO cells used here contain endogenous ␣ i but not ␣ o or ␣ z as determined by Western blotting (56). Receptor levels were ϳ3 pmol/mg as determined by Scatchard analysis of [ 125 I]p-MPPI saturation binding (57).
To evaluate the possibility of an agonist-promoted incorporation of palmitate into ␣ i , CHO cells were incubated for 60 min with [ 3 H]palmitate in the absence or presence of increasing concentrations of DPAT, a 5-HT 1A receptor-selective agonist (58). Incorporation of radiolabel was assessed by immunoprecipitation of ␣ i from subsequently prepared membranes fol- lowed by SDS-PAGE and fluorography. As shown in Fig. 1, DPAT elicited a dose-dependent increase in radiolabel incorporation into ␣ i ; the EC 50 for DPAT was ϳ10 Ϫ8 M. DPAT was used at 1 M in all subsequent studies.
If regulated palmitoylation plays a role in signal transduction, we predicted that it would be evident at time points that correlate with effector regulation. CHO cells were incubated for 5, 30, 60, and 120 min with [ 3 H]palmitate in the absence and presence of DPAT to evaluate the time course of [ 3 H]palmitate incorporation into ␣ i . As shown in Fig. 2A, incorporation of radiolabel increased over time in the absence of agonist (vehicle), representing "basal" turnover of palmitate. DPAT resulted in an increase in radiolabel incorporation at all time points. The data are expressed as percent change in Fig. 2B. Here, DPAT is seen to elicit the maximal difference beginning at the earliest measurable time point (5 min); this difference was sustained up to 60 min. The narrowing between stimulated and unstimulated values by 120 min probably represents an approach to equilibrium. Control experiments in which cells were pretreated with DPAT for 120 min prior to labeling excluded the possibilities that DPAT instability or 5-HT 1A receptor desensitization accounted for the converging values.
To establish the specificity of the agonist-promoted incorporation of [ 3 H]palmitate into ␣ i , CHO cells were incubated for varying lengths of time with [ 3 H]palmitate in the presence of DPAT and MPPI, a 5-HT 1A receptor-selective antagonist (59), or DPAT and PTX, a bacterial toxin that ADP ribosylates ␣ i in the heterotrimeric form and prevents receptor-G protein coupling. MPPI blocked the increase in incorporation of radiolabel elicited by DPAT at all time points (Fig. 2A). MPPI alone resulted in a small decrease in incorporation (data not shown), which may have resulted from its weak inverse-agonist activity or from inhibition of residual serotonin activity in the charcoaltreated serum. PTX similarly inhibited the increase in incorporation of radiolabel elicited by DPAT, suggesting that receptor-G protein coupling is important for regulated palmitate turnover. Of note, when cells were treated with PTX alone, radiolabel incorporation was reduced below basal for ␣ i , though not ␣ s (data not shown), suggesting that receptor-G protein coupling also contributes to basal palmitate turnover itself. Labeling was nevertheless observed to some extent despite the complete ADP ribosylation of ␣ i (as indicated by a shift in electrophoretic mobility), therefore factors beyond receptor-G protein coupling likely contribute to basal turnover. Both MPPI and PTX significantly reduced the percent change elicited by DPAT (Fig. 2B). Taken together, these data demonstrate that the agonist-dependent increase in palmitate turnover on ␣ i is because of 5-HT 1A receptor activation and is dependent on receptor-G protein coupling.
As shown in Fig. 3, several controls attested to the specificity of [ 3 H]palmitate incorporation. No signal was detected when pre-immune antibody was used instead of the ␣ i -directed antibody (Fig. 3A, lanes 1 and 2). There was no significant change in incorporation of radiolabel into membrane proteins other than ␣ i (data not shown) or into immunoprecipitated ␣ s ( lanes  10 and 11), indicating that DPAT does not alter equilibration of Remaining radiolabel was assessed by immunoprecipitation of ␣ i from subsequently prepared membranes followed by SDS-PAGE/fluorography (inset) and densitometry. A representative of two independent experiments is shown. radiolabel within available intracellular pools. DPAT did not increase ␣ i synthesis as measured by [ 35 S]methionine labeling (lanes 7 and 9), nor did it alter the amount of ␣ i in the immunoprecipitate as determined by immunoblotting (data not shown). To further exclude the possibility of an increase in ␣ i synthesis, we tested whether the increase in incorporation of [ 3 H]palmitate was evident in the absence of protein synthesis. Cycloheximide was sufficient to block protein synthesis as measured by [ 35 S]methionine (lanes 7 and 8) but did not affect the increase in ␣ i radiolabeling promoted by DPAT (lanes 5 and  6). These data indicate that de novo palmitoylation of newly synthesized ␣ i was not a contributing factor, and further that 5-HT 1A receptor up-regulation (57) did not account for our observations.
The cycloheximide data also suggest that the increase in ␣ i labeling promoted by DPAT is not because of an increase in co-translational myristoylation of newly synthesized proteins with [ 3 H]myristate metabolized from [ 3 H]palmitate. To confirm that our measurements were reflective of palmitoylation, gels containing immunoprecipitated labeled ␣ i were treated with hydroxylamine. The incorporation of label was almost completely sensitive to hydroxylamine (Fig. 3B, lanes 4 and 5), which indicates fatty acid linkage via a thioester bond and is consistent with palmitate incorporation. As a negative control, [ 3 H]myristate, which is linked via an amide bond to ␣ i , was resistant to hydroxylamine as expected (lanes 3 and 6).
We went on to evaluate [ 3 H]palmitate release from ␣ i by pulse-chase techniques to ascertain whether the increase in [ 3 H]palmitate incorporation in fact represented turnover of palmitate. CHO cells were incubated with [ 3 H]palmitate for 4 h (pulse) followed by a 2-h incubation with 100 M unlabeled palmitate in the absence or presence of increasing concentrations of DPAT (chase). The remaining [ 3 H]palmitate was assessed by immunoprecipitation of ␣ i from membranes followed by SDS-PAGE and fluorography. As presented in Fig. 4, DPAT caused a dose-dependent enhancement of release of radiolabel from ␣ i . The EC 50 for DPAT was ϳ10 Ϫ8 M, similar to that obtained from the incorporation assay.
To determine the time course and specificity of the agonistpromoted release of [ 3 H]palmitate, CHO cells were incubated with [ 3 H]palmitate for 3-4 h followed by incubation with 100 M unlabeled palmitate in the presence of either DPAT or DPAT plus MPPI for varying lengths of time. As shown in Fig.  5A, in the absence of agonist (vehicle) radiolabel was released from ␣ i over time, representative of basal depalmitoylation. DPAT promoted a greater loss of [ 3 H]palmitate at all time points beyond 5 min. This effect was blocked by MPPI. The data are expressed as percent change versus vehicle in Fig. 5B. Here, MPPI significantly inhibited the percent change elicited by DPAT. The somewhat slower attainment of the maximal response elicited by DPAT, as compared with that measured by the [ 3 H]palmitate incorporation experiments, was possibly due to the continued incorporation of residual [ 3 H]palmitate at the onset of the chase. The pulse-chase experiments demonstrate an agonist-promoted depalmitoylation of ␣ i . These data, taken together with the incorporation experiments that demonstrated an agonist-promoted incorporation of radiolabel, strongly support an increase in palmitate turnover upon ␣ i activation.

DISCUSSION
Palmitoylation of G protein ␣ subunits is emerging as a key player in modulating subunit location and subunit-protein interactions. The roles of palmitoylation have been investigated mainly by a loss of function approach involving subunits that lack the cysteine acceptor site for palmitate or subunits that are depalmitoylated by chemical or enzymatic means. How-ever, the regulation of ␣ i palmitoylation in an activation-dependent manner has not been critically assessed. We demonstrate here that palmitoylation of ␣ i is subject to regulation through the 5-HT 1A receptor and that this regulation specifically entails an enhanced exchange of palmitate through the combined processes of depalmitoylation and palmitoylation. These data provide the basis for extending the concept of regulated exchange of palmitate beyond G s and provide a framework for exploring the specific functional attributes of the palmitoylated and depalmitoylated forms of subunit.
Incorporation and pulse-chase experiments demonstrated a dose-and time-dependent change in [ 3 H]palmitate labeling of ␣ i upon activation of the 5-HT 1A receptor by DPAT. For both types of experiments, the EC 50 for DPAT was ϳ10 nM and maximal changes were achieved by 1 M. These values are consistent with those previously reported for effector regulation through the 5-HT 1A receptor (56,60,61). For the incorporation assay, DPAT elicited an increase in labeling with the maximal percent change evident by 5 min. For the pulse-chase assay, DPAT promoted a loss of radiolabel with the maximal percent change evident by 120 min. The changes in labeling of ␣ i were specific to receptor stimulation, as they were inhibited by the antagonist MPPI. They were additionally specific to receptor-G i coupling, as they were inhibited by PTX. Nonspecific labeling, changes in ␣ i synthesis or immunoprecipitation, and alterations in radioligand uptake were excluded. The increase in labeling did not represent co-translational myristoylation or compensatory up-regulation of the 5-HT 1A receptor. Sensitivity to hydroxylamine confirmed the nature of palmitate attachment as a thioester bond.
The combined use of incorporation and pulse-chase techniques was important. The incorporation assay does not distinguish an enhanced exchange of palmitate from a net increase in palmitate. Nor does the pulse-chase assay distinguish an enhanced exchange from a net decrease. Detection of agonistpromoted changes by both assays, however, supports an enhanced exchange. These observations nevertheless do not preclude a simultaneous change in stoichiometry. For ␣ s , however, the stoichiometry was reported to be unaffected by agonist (62).
Our results parallel previous observations of an increased turnover of palmitate on ␣ s upon activation through the ␤ 2adrenergic receptor (43,44) and by inhibition of GTPase activity by mutagenesis (43). Also consistent with our results are recent reports of increases in [ 3 H]palmitate incorporation for ␣ q/11 (48), ␣ s , and ␣ i (49) upon gonadotropin-releasing hormone receptor activation in pituitary cells. The changes in these cells, however, may have been attributable to an increase in subunit synthesis, as demonstrated subsequently for ␣ q/11 in gonadotropes (63). Increases in [ 3 H]palmitate incorporation have also been observed for ␣ q , ␣ s , ␣ i , and ␣ o upon incubation of rat brain membranes with serotonin in vitro (50), and for ␣ q and ␣ o upon incubation of aortic membranes with phenylephrine in vitro (51); however, neither study addressed the dynamics of palmitoylation in intact cells. In studies pertaining to ␣ i , specificity, de novo palmitoylation, and depalmitoylation were not directly evaluated. Agonist stimulation of the D 2 dopamine receptor in CHO-K1 cells did not promote depalmitoylation of wild-type ␣ z , another member of the G i family (28). Differences in receptor expression and coupling efficiencies may contribute to the differences in results, or regulated palmitate turnover may not occur for all G i family ␣ subunits.
The status of ␣ subunit palmitoylation may contribute positively or negatively to signal transduction. Recent data for G i family ␣ subunits suggest an interplay between palmitoylation and the GTPase activity promoted by RGS proteins. Palmitoylation of ␣ i1 inhibits the affinity of RGS4 for ␣ i1 , as well as the GAP activity for the subunit; similar results were obtained with GAIP and ␣ i and with G z GAP and ␣ z (26). Data for ␣ s subunits suggest that palmitoylation contributes to subunit distribution between plasma membrane and cytosol. A temporal correlation, for example, between activation-dependent depalmitoylation (43) and increases in the soluble/cytosolic subunit (64,65) has been demonstrated. Consistent with these data, nonpalmitoylated ␣ s has a reduced hydrophobicity and a reduced affinity for ␤␥ in vitro (66), and palmitoylation-defective (C3S) (65,67) and constitutively active (R201C) (64, 65) ␣ s mutants are localized to cytosol; in contrast however, other mutants maintain membrane association (10,68), and activation-dependent changes in solubility are not always detected (62). For ␣ i , membrane association is maintained for activated subunits (68), similar to that seen for ␣ o (69) and ␣ z (27), and depalmitoylation of ␣ i by acyl-protein thioesterase does not alter cytosolic levels (68). These data suggest that ␣ i may remain at the plasma membrane upon depalmitoylation but do not preclude alterations in localization of the subunit to mi-crodomains, which may function to attenuate or promote signaling. Several investigators have reported that G proteins are concentrated in caveolin-enriched domains (33,34), as well as other subdomains with low buoyant density and distinct immunofluorescence patterns (68,70).
Possible mechanisms for activation-dependent depalmitoylation may involve changes in ␣ subunit conformation or accessibility upon binding or releasing ␤␥ or guanine nucleotide. The current model favors an increase in susceptibility to a constitutively active palmitoylesterase upon dissociation of ␣ from ␤␥ (41,43). In vitro biochemical data support this model in that ␤␥ protects ␣ s -GDP but not ␣ s -GTP from depalmitoylation by a recombinant esterase (66). Moreover, heterotrimeric G␣ that is purified (41) or present in cell extracts (43) exhibits an increase in depalmitoylation upon AlF 4 Ϫ activation. Monomeric ␣ s -GTP and monomeric ␣ s -GDP are depalmitoylated at the same rate in vitro, suggesting that the nucleotide-dependent changes in ␣ subunit conformation are not contributing factors (66). Despite compelling evidence for this model, these data do not preclude the possibility of additional regulation of palmitoylation downstream of ␣ and ␤␥ dissociation (71).