Palmitoylation of a G Protein ai Subunit Requires Membrane Localization Not Myristoylation"

Palmitoylation is a dynamic, post-translational modi- fication of the amino terminus of heterotrimeric G protein a subunits. Since myristoylation, Py interactions, and membrane attachment also involve the amino terminus of the G protein ail subunit, we studied the rela- tionships between palmitoylation and these events. Using COS cell transfection, the turnover of palmitate was slower on ail subunits co-expressed with P and y subunits than on the ail subunit expressed alone. Mutation of cysteine 3 of ail prevented ['Hlpalmitate but not ['Hlmyristate incorporation and decreased the mem- brane localization of this protein. This nonpalmitoylated mutant could form a heterotrimer with co-expressed Py subunits which restored its membrane localization. A nonmyristoylated ail mutant (glycine 2 to alanine) could incorporate ['Hlpalmitate when co-expressed with Py subunits and localized to the membrane. The ['Hlpalmi-tate turnover of this nonmyristoylated mutant was more rapid than seen with the wild-type ail subunit. While myristoylation is not required for palmitoylation, both myr- istoylation and Py association can slow the turnover of palmitate on ail. These results suggest that palmitoyla- tion maintains the membrane attachment of the free a subunit and changes in fly association could modulate palmitoylation during signaling.

Palmitoylation, the post-translational thioesterification of cysteine residues with palmitate, occurs on many a subunits: as, ai, ao, aq, az, aI2, and a13 (7, [10][11][12][13][14]. Residues at the amino terminus, cysteine 3 for a, and ao, and cysteine 9 and 10 for aq are critical for the modification (10,12,13). Mutation of these residues to prevent palmitoylation leads to some loss of membrane attachment, the extent of which depends on the a subunit and expression system used (10,12,13,15,16). Reduced membrane affinity may explain the decreased effector activation of the nonpalmitoylated a, and aq (13).
Palmitoylation is a reversible modification. Activation of as by agonist stimulation or cholera toxin leads to increased turnover of palmitate (15,17,18). Changes in the degree of palmitoylation may lead to changes in a subunit membrane aflinity which is observed during the GTPase cycle and allow modulation of signaling (19). Our previous finding that increased palmitate incorporation with agonist stimulation did not occur on the S49 cell H21a mutant of as, which dissociates slowly if at all from pr, suggests that py dissociation may affect palmitoylation (17). The amino terminus of a subunits is the site for both palmitoylation and py interactions (11,20) Myristoylation and palmitoylation share the common feature of increasing protein hydrophobicity. They both occur at the amino terminus of several G protein a subunits and Src family tyrosine kinases including ~5 6 ' '~ and p 5 w (21,221. Some studies have suggested that myristoylation is critical for palmitoylation and may be part of a general consensus sequence for the modification (7,15,22). The function of palmitoylation in membrane attachment could be similar t o myristoylation which markedly increases the affinity of the a subunit for Py (8,23). Alternatively, its increased hydrophobicity (16 carbon palmitate uersus 14 carbon myristate) could cause a strong direct interaction with the lipid bilayer.
The relationships between palmitoylation, myristoylation, and py association of an a subunit and their relevance to membrane localization were investigated in this study. We found that the modifications occurred independently; palmitoylation did not require myristoylation. Either palmitoylation or binding to the py complex was sufficient t o localize a functional ai subunit at the membrane. Association with Py did not require a palmitoylated a subunit, but it did slow the turnover of palmitate on the a subunit. EXPERIMENTAL PROCEDURES Vector Constructs and Site-directed Mutagenesis-The cDNAs for the rat wild-type and mutant ail subunits were cloned into the pCD-PS eukaryotic expression vector as described previously (8, 24). The construction of the GA-ai, mutant with the substitution of alanine for glycine 2 was described earlier (8). Site-directed mutagenesis was performed using the polymerase chain reaction and oligonucleotide TAG ATA CCG GCG CCG AGG TCG GCC ACC ATG GGC GCG ACA CTG AGC GC to change the ail cDNA to code for an alanine instead of cysteine 3 (C3A) using reaction conditions as described (24). The part of the 5'-untranslated region and the coding sequence between the unique restriction sites for KasI and BstXI was amplified and introduced by ligation into the aL1 cDNAin pCD-PS vector using the same unique pair of restriction enzymes. To create the GA-a, mutant, the mutation of glycine to alanine was introduced by polymerase chain reaction and the oligonucleotide primer TAG ATA GCC CGG CCG CGC TCG GCC ACC ATG GCA TGC CTC GGC. The amplified cDNA region with the mutation was ligated into the correspondent region of 0 1 , cDNA in pCD-PS vector using a unique pair of restriction enzymes, EagI and MZuI. The mutations were confirmed by nucleotide sequencing and restriction analysis. The construction of Pl, y2, and the C68S mutant of y2 cDNAs in pCDM8.1 expression vector was described previously (25).
Zkansfection a n d Radiolabeling-COS-7 monkey kidney cells were maintained and transfected using the DEAE-dextran method as described previously (26). 48 h after transfection the cells were prepared for metabolic labeling by incubation in serum-free DMEM for 2 h. Since [3H]palmitate can be metabolized to [3Hlmyristate, metabolic labeling with [3H]palmitate was performed in the presence of 50 pg/ml of cycloheximide (Calbiochem) in serum-free media for 30 min prior to and during the incubation with C3H1palmitate to prevent co-translational myristoylation. The cells were labeled for 30 min with 500 pCi of [9,10-3H]palmitate/ml (American Radiolabeled Chemicals; specific activity 60 Ci/mmol) or 300 pCi of [9,10-3Hlmyristate/m1 (DuPont NEN, specific activity 16 Ci/mmol) in 5 ml of serum-free media supplemented with 1% (v/v) dimethyl sulfoxide/75-cm2 flask. For the pulse-chase experiments, the cells were incubated for 20 min with L3H1palmitate, washed once with serum-free DMEM, and then incubated in complete DMEM containing 10% (v/v) fetal bovine serum. The cells were scraped in cold phosphate-buffered saline and centrifuged a t 2000 x g for 10 min. The pellets were resuspended in homogenization buffer and frozen at -70 "C.
Cell Fractionation-The cell pellets were homogenized by passaging 25 times through a 25-gauge needle in homogenization buffer composed of 5 mM HEPES, pH 7.4, 100 pg/ml soybean trypsin inhibitor, 0.5 pg/ml leupeptin, 2 pg/ml aprotinin, 1 mM EDTA, 0.7 pg/ml pepstatin, and 10 milliunits/ml a2-macroglobulin (Boehringer Mannheim). The cell lysate was centrifuged at 1000 x g for 3 min in an Eppendorf 5415 microcentrifuge to pellet the nuclei and unbroken cells. The supernatant was centrifuged a t 125,000 x g for 30 min at 4 "C in a Beckman TLA45 rotor. The supernatant (soluble fraction) was separated and the pellet (particulate fraction) resuspended in the buffer and recentrifuged. The washed pellet was resuspended in the original volume of homogenization buffer. The amount of protein in the particulate and soluble fractions was approximately equal.
Immunoprecipitation a n d Zmmunoblotting-The affinity-purified antibodies specific for the carboxyl-terminal decapeptide of ail,2 (AS) and a, (RM) were used for immunoblotting and immunoprecipitation (27,28). Antisera SW (29) and EDPL were used for detection of pl and y2 subunits, respectively. EDPL was raised against the carboxyl-terminal peptide of y2, corresponding to residues 47-64. Immunoblotting with detection of the primary antibodies with '251-protein A, and autoradiography was performed as described previously (24, 30). Immunoprecipitation was performed on equivalent amounts of protein in a solubilization buffer of 50 mM Tris-HCI, pH 7.5 (25 "C), 150 mM NaCl, 0.8% (w/v) Triton X-100, 0.2% (w/v) SDS, and 1 mM EDTA with a n incubation overnight a t 4 "C as described previously (8). The immunoprecipitates were recovered by incubating with protein A-Sepharose CL-4B (Pharmacia Biotech Inc.), washed, solubilized, separated by SDS-PAGE on 10% Tris-glycine gels (Novex), and prepared for fluorography.
Detergent Extraction-The particulate fractions from transfected COS cells were incubated for 1 h on ice (with occasional vortexing) in a buffer used for fractionation with 1% (w/v) Triton X-100, 120 mM NaCl, 0.1 mM GTP, and 10 mM MgCl,. The final protein concentration was 1 mg/ml. The samples were separated by centrifugation at 125,000 x g for 30 min a t 4 "C.

M G C T L S A
Site of myristoylation is underlined; probable site of palmitoylation Palmitoylation occurs after membrane localization with co-expres-is in boldface.
sion of Py.
for 15 min at 37 "C. Protein concentration was determined by the Bio-Rad protein assay dye kit with IgG as the standard (Bio-Rad). Thin layer chromatography and hydroxylamine treatment were performed as described earlier (12). Quantitation-Densitometry of the fluorographs was performed with a LKB 2202 UltroScan laser densitometer. Statistical data analysis was performed with commercial software, Sigmaplot (Jandel Scientific) and Statview (Abacus Concepts). The quantitative analysis of the gels with [32PlADP-ribose-labeled material and the nitrocellulose membranes treated with lZ5I-protein-A was done using a PhosphorImager (Molecular Dynamics).

RESULTS
Mutation of Cysteine 3 Prevents PHlPalrnitate Incorporation-COS cells were transfected with plasmids containing the cDNA for the wild-type ail (WT) or an ail mutant in which the coding sequence was changed to replace cysteine 3 with an alanine (C3A) ( Table I). Immunoprecipitation of fractions from cells metabolically labeled with [3H]myristate showed incorporation into both the WT and C3A proteins in both the particulate and soluble fractions (Fig. 1). The distribution of the proteins will be discussed. Incorporation of [3H]myristate, which is a co-translational event, was prevented with the protein synthesis inhibitor, cycloheximide. The C3A mutant did not incorporate L3H1palmitate ( Fig. 1). The WT protein incorporated the radiolabel only in the particulate fraction. Hydroxylamine treatment and thin layer chromatography of the radiolabel after release by alkali treatment confirmed the thioester linkage and [3H]palmitate incorporation, respectively (data not shown).
Intracellular Localization of the C3A Mutant-To determine the role of palmitoylation in membrane attachment, we compared the subcellular distribution of the WT, nonpalmitoylated C3A mutant and a mutant ail in which the glycine 2 was changed to alanine (GA, Table I). The WT is predominantly in the particulate fraction and the GA mutant which does not undergo myristoylation (8) is predominantly in the soluble fraction (Fig. 2). The localization of the C3Amutant is intermediate with approximately two-thirds of the expressed protein in the soluble fraction.
Since overexpression leads to more a subunits than By subunits (321, we evaluated the intracellular localization of the nonpalmitoylated C3A protein with co-transfection of py subunits. With co-expression of pl and y2 subunits, approximately two-thirds of the C3A protein was now in the particulate fraction (Fig. 3).
Association of the Nonpalrnitoylated C3A Mutant with Py Subunits-We tested whether the nonpalmitoylated mutant could form a heterotrimer with Py by performing PTX-catalyzed ADP-ribosylation. The a subunit is a substrate for ADPribosylation only when it is bound to py (33, 34). The soluble fractions of transfected COS celIs were incubated with purified Py subunits. Both the WT (not shown) and C3A proteins underwent maximal ADP-ribosylation at a much lower concentration of Py than needed for the GA mutant which has a decreased affinity for py (8) (Fig. 4).
For the particulate fraction, PTX ADP-ribosylation was significantly increased in the cells co-transfected with both the nonpalmitoylated C3A mutant and Py subunits compared to the vector transfected cells (Fig. 5). When this particulate fraction was pretreated with GTPyS to cause dissociation of the subunits, the PTX ADP-ribosylation was markedly decreased compared to the same particulate fraction pretreated with GTP (data not shown).
The particulate fraction of the cells transfected with the nonpalmitoylated C3A mutant alone showed a small increase in PTX ADP-ribosylation compared to the vector transfected cells (Fig. 5, open bar), but there was no further increase in ADPribosylation in the presence of purified Py subunits as was seen for the WTai, protein (Fig. 5, filled bars). To test whether this lack of effect with purified Py subunits was due to improper folding of the transfected a subunit, detergent solubilization   particulate fraction (Fig. 3). We then tested whether this nonmyristoylated protein would undergo palmitoylation. Immunoprecipitation of the particulate fractions of COS cells transfected with GA, P, and y plasmids and radiolabeled with ["Hlpalmitate showed incorporation of palmitate into the GA protein (Fig. 6A ). ['HlPalmitate incorporation and membrane localization of the GA protein was dependent on the amount of P and y plasmids transfected (Fig. 6, B and C) and Py protein expressed (data not shown). Incorporation of palmitate was not seen when GA was expressed alone, with the p and y vectors separately, or with a p and a mutant y subunit the latter of which does not undergo isoprenylation or membrane localization (25).
In some experiments co-transfection of COS cells with the p and y plasmids led to a slight increase in the amount of the endogenous ai protein in the particulate fraction which could be detected because the GA protein migrates slower on SDS-PAGE (Figs. 3 and 6 0 . However, transfection of COS cells with the vector and 30 pg of P and y plasmids did not show an increase in ["Hlpalmitate incorporation in the endogenous ai subunits (Fig. 6D). The particulate fraction of cells transfected with the GA mutant and Py had increased PTX ADP-ribosylation compared to cells transfected only with the vector or the GAmutant (data not shown).
To further investigate the substrate requirements for palmitoylation, the amino-terminal glycine of a, was mutated to alanine. Since a, does not undergo myristoylation, this mutation was made to assess the significance of adjacent amino acids on palmitoylation. This GAa, protein incorporated ['Hlpalmitate and was localized to the particulate fraction (data not shown).

Palmitate Turnover in the Presence of Py Subunits-To
determine the turnover of palmitate, we performed a pulsechase experiment with COS cells transfected with GA+Py, WT+py, and WT alone and incubated with ['Hlpalmitate (Fig.  7). The nonmyristoylated GA protein had a rapid turnover of palmitate with a half-life (t1,J of 13 min compared to a t,,2 of 41 min for WT and t,,, of 52 min for WT+Py. The t,,2 is the mean of two independent experiments. Turnover was also assessed by incorporation of ["Hlpalmitate after a 20-min incubation as a measure of the exchange of unlabeled palmitate with ['Hlpalmitate. The amount of incorporation in the WT ail protein co-expressed with Py was 73% f 8% (mean f S.E.) of the incorporation into the WT alone based on four independent experiments (Student's t test, p < 0.005).
The amount of expressed WT protein was nearly the same with and without Py co-expression based on quantitation of immunoblots. Since palmitate is incorporated into both the endogenous and transfected a subunits, these values do not represent the absolute differences in palmitate turnover for the free versus heterotrimeric a subunit but do indicate that co-expression of Py subunits could slow palmitate turnover on the a, subunit.

DISCUSSION
The interplay of acylation with membrane attachment and protein interactions has biologic significance for many signal transduction proteins. For the heterotrimeric G protein ail, myristoylation and palmitoylation occur in a region important for both protein interaction ( f l y ) and membrane attachment. We found that mutation of cysteine 3 prevented palmitoylation and led to loss of membrane attachment. This nonpalmitoylated mutant could form a heterotrimer with the Py subunits which restored its membrane localization. Since palmitoylation is reversible, we investigated the regulation of this modification. Membrane localization and cysteine 3 were crucial for palmitoylation. Myristoylation was not required for palmitoylation, but myristoylation and Py association could slow the ail palmitate turnover.
Palmitoylation and Membrane Attachment-Palmitoylation, myristoylation, and Py association all contributed to the membrane attachment of the ail subunit; membrane attachment being the greatest with the occurrence of all three but possible with any two. After translation, the 14 carbon myristoyl group can provide a marginal affinity for membranes as tested with acylated peptides (37) and evidenced by the distribution of myristoylated proteins to both the membrane and cytosol (38). The myristoylated ail and C3Aproteins are possibly localized at membranes by the presence of endogenous Py subunits or weakly attach through the myristate group. Palmitoylation may stabilize the membrane localization of the ail subunit, whereas the nonpalmitoylated C3A mutant, when it is free of By, may either dissociate from the membrane or form aggregates. The amount of the nonpalmitoylated C3A mutant we found in the particulate fraction without Py co-transfection did not reflect true membrane affinity because most of the protein was nonfunctional and possibly aggregated.
Palmitoylation may also be important in stabilizing the a subunit at its proper intracellular location. Cadwallader and colleagues (39) have shown that myristoylation can lead to nonspecific membrane attachment of Ras mutants whereas palmitoylation was needed for plasma membrane attachment.
The role of palmitoylation in membrane attachment may be different for the myristoylated and nonmyristoylated a subunits. Mutation of the third cysteine in ail or a, (10,15,16) led to a significantly greater decrease in membrane attachment than the same mutation in a, (12,15). The proximity of the myristate and palmitate groups on the ai and a, subunits suggests that they may be part of the same membrane attachment domain. Alternatively, palmitoylation on as and aq may act independently of another membrane attachment domain. Since this domain has not been identified, it is possible that the epitope tagging of a, and aq by Wedegaertner et al. (13) could have altered this domain and explain the marked decrease in membrane attachment of the nonpalmitoylated a, and aq in their study.
Palmitoylation, Py Association, and Myristoylation-Unlike myristoylation which markedly increases the affinity of ai and a, for Py (8,23), palmitoylation was not crucial for heterotrimer formation. Consequently, the decreased Py affinity of a GAa, mutant described by van der Neut and colleagues (40) is unlikely to be associated with altered palmitoylation. In any case, we found that this mutant could undergo palmitoylation.
The importance of myristoylation may be to primarily affect the structure of the a subunit enhancing Py affinity, whereas the importance of palmitoylation may be to cause binding of the a subunit to the lipid bilayer. While this study did not address whether palmitoylated a subunits can form a heterotrimer, indirect evidence suggests that they do. The H21a mutant of a, which is inactive and does not dissociate from Py has a basal incorporation of [3H]palmitate similar to the wild-type a, (17).
The C3A mutation in ail did not prevent myristoylation of glycine 2. Myristoylation also occurs on the nonpalmitoylated C3A or C3S mutants of a, (10,15). These results are consistent with myristoylation being a co-translational event in the cytosol and palmitoylation occurring later at membranes (38, 41). The third residue is not a critical component of the amino-terminal consensus sequence for N-myristoyl transferase activity (38).
Regulation of Palmitoylation-In this study, the crucial determinants of ai palmitoylation appeared to be the third cysteine residue and membrane localization. This third residue is probably the actual site of palmitoylation since it is the sole cysteine within the first 21 amino acids shown to have the site of palmitoylation (11). Previous reports have indicated that myristoylation may be required for palmitoylation because myristoylation defective G2A mutants of a, and a, did not undergo palmitoylation (7,15). However, in this study a nonmyristoylated mutant could incorporate [3Hlpalmitate when it was localized at the membrane after co-expression with Py. The frequent occurrence of palmitoylation with either myristoylation at the amino terminus (22) or with isoprenylation at the carboxyl terminus (42) may be due to the myristate and isoprenoid groups first allowing membrane attachment and then positioning the cysteine close to the membrane.
Regulation of the availability of the cysteine thiol group to a palmitoyl CoA could occur from both the local and global protein conformation. The adjacent amino acids may facilitate the reaction by creating a favorable secondary structure and increasing membrane interactions. Dynamic regulation of palmitoylation can occur from global conformational changes which change the distance of the cysteine from the membrane or change the closeness of the Py subunits. The rapid exchange of the palmitate on the nonmyristoylated GA protein is an example of both levels of regulation. The loss of the myristate on the adjacent glycine could provide more exposure of the cysteine. The decreased Py affinity, probably resulting from a change in amino-terminal conformation secondary to a lack of myristoylation, could also contribute to rapid turnover of palmitate.
For a subunits, the Py subunits may be the primary regulator of palmitoylation. The increased palmitate turnover of a, upon activation could be the result of decreased Py affinity due to the conformational changes of the a subunit upon binding of GTP. However, more studies are needed for a full understanding of the regulation and function of a subunit palmitoylation. Its reversibility and effects on membrane affinity suggest it has a prominent role in the regulation of G protein signal transduction.