N-Myristoylation and βγ Play Roles beyond Anchorage in the Palmitoylation of the G Protein αoSubunit*

Many of the α subunits of heterotrimeric GTP-binding regulatory proteins (G proteins) are palmitoylated, a modification proposed to play a key role in the stable anchorage of the subunits to the plasma membrane. Palmitoylation of α subunits from the Gi family is preceded byN-myristoylation, which alone or together with βγ probably supports a reversible interaction of the α subunit with membrane as a prerequisite to the eventual incorporation of palmitate. Previous studies have not addressed, however, the question of whether membrane association alone, carried out throughN-myristoylation, interaction with βγ, or other events, is sufficient for palmitoylation. We report here for αothat it is not. We found that N-myristoylation is required for palmitoylation at least in part because it supports events subsequent to membrane attachment. Mutants of αo designed to target the subunit to membrane without an N-myristoyl group are unable to be palmitoylated as evaluated by incorporation of [3H]palmitate. Mutants of αo unable to interact normally with βγ yet still attach to membrane demonstrate that βγ, in contrast, is not required for palmitoylation. βγ becomes necessary, however, when the N-myristoyl group is absent. Our results suggest that N-myristoylation and βγ, while almost certainly relevant to the reversible interaction of αo with membrane, also play at least partly overlapping, post-anchorage roles in palmitoylation.

Fatty acid acylation is one of the major covalent modifications of heterotrimeric GTP-binding regulatory protein (G protein) ␣ subunits (1)(2)(3)(4). For most ␣ subunits of the G i family, including ␣ i , ␣ o , and ␣ z , two distinct acylation events occur. N-Myristoylation takes place at the amino terminus and represents the attachment of a myristate through an amide bond to Gly 2 following cleavage of the initiator methionine (5,6). Palmitoylation occurs at the adjacent Cys 3 and represents the attachment of a palmitate (primarily) through a thioester bond (7,8). ␣ s , ␣ q , and ␣ 12 family members are palmitoylated but not N-myristoylated (8 -11), whereas ␣ t is N-myristoylated but not palmitoylated (12). The functional consequences of N-myristoylation and palmitoylation have been studied intensively. N-Myristoylation plays an important role in the attachment of G i family ␣ subunits to membrane (5,6), targeting of the subunits to caveolin-enriched membrane domains (13), and interactions with the ␤␥ heterodimer (14) and effectors such as adenylyl cyclase (15). Palmitoylation is relevant to the attachment of G protein ␣ subunits to membrane (10,16) and interaction of subunits with ␤␥ (17) and RGS proteins (18).
The palmitoylation of G protein ␣ subunits has drawn much attention recently, as this modification, unlike N-myristoylation, is reversible and subject to regulation. Several groups have demonstrated that the activation of G s is accompanied by an increased rate of palmitate exchange on ␣ s , probably reflecting depalmitoylation and repalmitoylation of the subunit released from ␤␥ (19 -21). Whether the exchange coincides with a release of ␣ s from membrane is subject to debate (22,23). Agonist-promoted palmitate exchange has also been reported for ␣ q (24) and ␣ i (25). The nature and reversibility of palmitoylation raise the expectation that it serves to regulate not only subunit location, but also activation/deactivation cycles and interactions of subunits with other proteins.
The basic requirements for palmitoylation remain obscure. For ␣ subunits of the G i family, three interrelated phenomena appear relevant: N-myristoylation, interaction with ␤␥, and proximity to membrane. A widely held hypothesis is that Nmyristoylation imparts a small amount of hydrophobicity to the ␣ subunit and/or facilitates interaction of the subunit with independently anchored ␤␥, therefore supporting a reversible interaction of the subunit with the plasma membrane (2); at the plasma membrane, the subunit is palmitoylated either enzymatically, i.e. by a membrane-localized palmitoyltransferase (26), or nonenzymatically (27). The relevance of the membrane is consistent with the observation that only membrane-associated subunits are found to be palmitoylated and that a mutant of ␣ i that is not N-myristoylated can nevertheless be palmitoylated when apparently brought to the membrane by overexpressed ␤␥ (28).
Previous studies have not, however, addressed whether membrane association, carried out through N-myristoylation, interaction with ␤␥, or other events, is sufficient for palmitoylation. It is quite possible, for example, that N-myristoylation and/or ␤␥ is required not only prior to, but following interaction of the subunit with membrane to support additional events culminating in palmitoylation. We report here, using ␣ o , that this is in fact the case. We found that mutants of ␣ o that lack an N-myristoyl group but are still targeted to membrane are not palmitoylated. We also found that interactions with ␤␥ are not normally required for palmitoylation, but, in the absence of N-myristoylation, become necessary. The requirement for ␤␥ in lieu of N-myristoylation is not related to anchorage. We propose that N-myristoylation and ␤␥, while no doubt relevant to reversible interactions of subunits with membrane, also play redundant, post-anchorage roles in palmitoylation.

EXPERIMENTAL PROCEDURES
Plasmid Construction-Rat ␣ o1 and human 5-hydroxytryptamine 1A (5-HT 1A ) receptor (5-HT 1A R) 1 cDNAs (29,30) in pcDNA3 and human ␤ 1 and bovine ␥ 2 cDNAs (31,32) in pCMV5 were used directly or as templates for subsequent mutagenesis. The mutations G2A, S6N, G204A, and Q205L and the deletion of residues 8 -11 for ␣ o were accomplished by the polymerase chain reaction using mismatched primers. The proteins ␣ o K 10 , ␣ o G2A/K 10 , 5-HT 1A R/␣ o , and 5-HT 1A R/ ␣ o G2A were also constructed by polymerase chain reaction techniques in which overlapping ends were subsequently combined in a second round of polymerase chain reaction. The final constructs were subcloned into the HindIII and XbaI restriction sites of pcDNA3. Sequences were verified directly.
Transfection and Metabolic Labeling-Human embryonic kidney (HEK) 293 cells (American Type Culture Collection) were cultured at 37°C in a humidified atmosphere of 5% CO 2 in minimal essential medium containing 10% (v/v) fetal calf serum and supplemented with penicillin (100 units/ml) and streptomycin (100 g/ml). Transfections were carried out by calcium phosphate precipitation as described previously (33), usually with 10 g of DNA/10-cm plate of just subconfluent cells. Cells were harvested 40 -48 h thereafter. Biosynthetic labeling of proteins prior to harvesting was carried out as follows: (i) with Tran 35 Slabel (ICN), nominally referred to as [ 35 S]methionine, in methioninefree Dulbecco's modified Eagle's medium with 10% fetal calf serum (50 Ci/ml), usually initiated 8 h before harvesting; (ii) with [ 3 H]myristic acid (American Radiolabeled Chemicals) in minimal essential medium supplemented with 5 mg/ml fatty acid-free bovine serum albumin (250 Ci/ml; Sigma), usually initiated 2 h before harvesting; or (iii) with [ 3 H]palmitic acid (American Radiolabeled Chemicals) in minimal essential medium supplemented with 5 mg/ml fatty acid-free bovine serum albumin (250 Ci/ml), usually initiated 1 h before harvesting.
Cell Fractionation and Immunoprecipitation-Cells to be fractionated into membrane and cytosolic fractions were harvested in 20 mM HEPES, pH 8.0, 2 mM MgCl 2 , 1 mM EDTA, 2 g/ml aprotinin, 2 g/ml leupeptin, and 1 mM phenylmethylsulfonyl fluoride (0.5 ml/10-cm plate) and lysed at 0°C by repeated passage through a 27-gauge needle. Subsequent procedures were carried out at 4°C. Nuclei and unbroken cells were removed by centrifugation at 660 ϫ g for 5 min, and the cytosolic and membrane fractions were resolved subsequently by centrifugation at 100,000 ϫ g for 1 h. The cytosolic fraction was used directly for immunoprecipitation or immunoblotting. The membrane fraction was extracted in the lysis buffer described above, but containing 1% sodium cholate, followed by clarification at 100,000 ϫ g for 1 h. Immunoprecipitation of ␣ o and the various mutants from the cytosolic or extracted membrane fractions was accomplished by adding an equal volume of 100 mM sodium phosphate, pH 7.2, 2% deoxycholic acid, 2% Triton X-100, 1% SDS, 300 mM NaCl, 4 mM EDTA, 4 g/ml aprotinin, 4 g/ml leupeptin, and 2 mM phenylmethylsulfonyl fluoride and then, following pre-clearance with nonimmune rabbit serum and protein A-Sepharose, adding antiserum 9072 (1:50 to 1:100 final dilution) (34); antiserum 9072 was generated against the C-terminal 10 residues of ␣ o1 . Immunoprecipitates were collected by centrifugation; washed three times in 50 mM sodium phosphate, pH 7.2, 150 mM NaCl, 0.5% Triton X-100, and 2 mM EDTA; and resuspended in sample buffer for SDSpolyacrylamide gel electrophoresis (PAGE). When fractionation was not required, cells were harvested directly in 50 mM sodium phosphate, pH 7.2, 1% deoxycholic acid, 1% Triton X-100, 150 mM NaCl, 2 mM EDTA, 2 g/ml aprotinin, 2 g/ml leupeptin, and 1 mM phenylmethylsulfonyl fluoride (1 ml/10-cm plate). After 30 min on ice, the lysate was processed for immunoprecipitation as described above.
Isolation of Caveolin-enriched Membrane Fractions-The detergentfree approach of Song et al. (13) was used to prepare caveolin-enriched membrane fractions. HEK293 cells in 10-cm dishes were harvested in 2 ml of 500 mM sodium carbonate, pH 11; passed through a 27-gauge needle 25 times; and then processed in a Polytron tissue grinder with three 10-s bursts and a sonicator (550 Sonic Dismembraner, Fisher) with three 20-s bursts. The homogenate was adjusted to 45% sucrose by addition of 2 ml of 90% sucrose in 25 mM Mes, pH 6.5, and 150 mM NaCl and placed in a 14 ϫ 95-mm Ultra Clear TM centrifuge tube (Beckman Instruments). Four ml of 35% sucrose in the same buffer were layered above the sample, followed by 4 ml of 5% sucrose also in the same buffer. Centrifugation was performed at 39,000 rpm for 16 h at 4°C in an SW 41 rotor (Beckman Instruments). One-ml fractions were collected, and the protein within each fraction was precipitated by addition of trichloroacetic acid to 10% at 4°C. The precipitates were collected, washed, and dissolved in sample buffer for analysis by Western blotting. Antiserum 9072 was used for detection of ␣ o , and caveolin was detected with an antiserum raised to the N terminus (Transduction Laboratories, Lexington, KY).
Miscellaneous Methods-Discontinuous SDS-PAGE (11% acrylamide) was carried out as described (34). Gels were stained with Coomassie Blue R-250, destained, and dried. For fluorography, the gels were soaked in Amplify TM (Amersham Pharmacia Biotech) for 30 min prior to drying. Dried gels were exposed to Hyperfilm TM MP film (Amersham Pharmacia Biotech) at Ϫ80°C for 20 -40 days (fluorography) or at room temperature for 4 -10 days (autoradiography). Western blotting using chemiluminescence was accomplished following the directions provided by Amersham Pharmacia Biotech. GTP␥S protection of the ␣ subunit from tryptic digestion was performed according to Berlot and Bourne (35). Pertussis toxin (PTX)-catalyzed ADP-ribosylation was conducted by the method of Carty (36). The [ 35 S]GTP␥S-binding assay was carried out according to Windh et al. (37).

Labeling of ␣ o with [ 3 H]Palmitate Normally Requires N-
Myristoylation-The observation that mutation of Gly 2 (a residue that composes part of a consensus sequence for N-myristoylation) in ␣ i subunits abolishes palmitoylation as well as N-myristoylation (20,38,39) has led to the hypothesis that N-myristoylation, as a cotranslational modification, is a prerequisite to palmitoylation, a post-translational modification. To more firmly establish the relationship between N-myristoylation and palmitoylation, we initiated studies with another mutation, S6N, using ␣ o expressed in HEK293 cells. HEK293 cells lack ␣ o and therefore provide a suitable setting for the analysis of heterologously expressed ␣ o and mutants of ␣ o . Ser 6 , like Gly 2 , is a consensus element for N-myristoylation (40,41 Mutational approaches are subject to the criticism that a mutation, while inhibiting N-myristoylation, might also have a direct bearing on nonenzymatic palmitoylation or substrate recognition by a palmitoyltransferase apart from N-myristoylation. We therefore studied the effects of a chemical inhibitor of N-myristoyltransferase, 2-hydroxymyristic acid (2-HMA) (42,43 A small proportion of ␣ o G2A and ␣ o S6N usually cofractionates with membrane (see Fig. 1, for example). This can be the result of nonspecific association due to subunit denaturation; however, we found that GTP␥S protects this population of subunit from degradation by trypsin, allowing only the 2-kDa (N-terminal) clip characteristic of a conformationally sound subunit (data not shown). Because ␣ o G2A and ␣ o S6N that cofractionate with the membrane do not incorporate [ 3 H]palmitate, as noted previously, we hypothesized that the absence of an N-myristoyl group and possibly a consequent decrease in affinity of the subunits for ␤␥ might be the underlying defect. We therefore determined whether the G204A mutation, which would increase the affinity of the ␣ subunit for ␤␥, would confer to the G2A and S6N forms of ␣ o the capacity to be palmitoylated. We found that introduction of the G204A mutation did not change the distribution of ␣ o G2A and ␣ o S6N between membrane and cytosolic fractions (Fig. 3, upper panel). However, although ␣ o G2A and ␣ o S6N did not incorporate [ 3 H]palmitate, ␣ o G2A/G204A and ␣ o S6N/G204A did (lower panel). The results lend support to the concept that events in addition to membrane attachment are required for [ 3 H]palmitate labeling, and they specifically demonstrate that a non-myristoylated form of subunit can be palmitoylated if made to assume a conformation more refractory to activation and/or more conducive to interaction with ␤␥. The extent of labeling in the case of ␣ o G2A/G204A and ␣ o S6N/G204A was severalfold less than that achieved with wild-type ␣ o (data not shown), due probably to the relatively greater amount of the latter associated with membrane.
We evaluated the capacity of ␣ o and the various mutants to be [ 32 P]ADP-ribosylated by PTX, a reaction that requires at least a transient interaction of the ␣ subunit with ␤␥ (46, 47). ␣ o G204A, as expected based on its presumed high affinity for ␤␥, was a better substrate for ADP-ribosylation than ␣ o (Fig.   FIG. 1. Distribution, N-  4A). ␣ o Q205L was also a good substrate, which was surprising, as an active form of the subunit would not be expected to interact well with ␤␥. The ADP-ribosylation of Gln-to-Leu mutants expressed in mammalian cells has not been evaluated previously. ␣ o G2A and ␣ o S6N were not substrates for ADPribosylation (Fig. 4B), a result consistent with the observation that the N-myristoyl group is relevant to interaction with ␤␥ (14). Introduction of the G204A mutation restored the capacity of ␣ o G2A and ␣ o S6N to be ADP-ribosylated, as it did the capacity of these subunits to be palmitoylated. The extent of ADP-ribosylation was less for ␣ o S6N/G204A than for ␣ o G2A/ G204A, which might be explained by a greater impact of the S6N mutation on interactions with ␤␥. The results with the G2A and S6N mutants imply that the loss of [ 3 H]palmitate labeling observed when the N-myristoyl group is absent can be reversed by G204A through an enhanced interaction with ␤␥.
Labeling of ␣ o with [ 3 H]Palmitate Does Not Require Normal Interactions with ␤␥, Unless Myristate Is Absent-The above observations suggest a role for ␤␥ in palmitoylation. To begin addressing this role in more detail, we constructed a mutant of ␣ o previously noted to associate only poorly with ␤␥ (48). This mutant, ␣ o ⌬8 -11 (lacking Glu 8 -Ala 11 ), was, as expected, a poor substrate for PTX-catalyzed ADP-ribosylation (Fig. 5A). The mutant behaved nearly identically to ␣ o in terms of membrane/ cytosol distribution and [ 3 H]myristate and [ 3 H]palmitate labeling (Fig. 5B). Therefore, whereas a mutation enhancing interaction of ␣ o with ␤␥ enhanced [ 3 H]palmitate labeling (G204A; see above), a mutation attenuating interaction but maintaining N-myristoylation and membrane anchorage (␣ o ⌬8 -11) had no effect. This result indicates that [ 3 H]palmitate labeling of ␣ o does not require fully normal interactions of ␣ o with ␤␥.
We wished to corroborate this result and therefore devised a mutant in which 10 lysine residues were inserted between Ser 6 and Ala 7 (␣ o K 10 ). We anticipated (and examined subsequently below) that the 10-lysine insertion, like ␣ o ⌬8 -11, would disrupt interaction of the subunit with ␤␥. The 10-lysine insert was also chosen to permit a set of experiments with ␣ o G2A to confirm the role of N-myristoylation in palmitoylation for even the membrane-attached subunit, as suggested by the G2A and S6N mutants. Polylysine domains support the attachment of various proteins to membrane, probably by virtue of electro-static interactions with phospholipids (49 -51); we felt that the 10-lysine insert might similarly support attachment of the largely cytosolic, non-myristoylated G2A form of subunit to membrane. The 10-lysine insert did not affect the folding of ␣ o or ␣ o G2A as evaluated by GTP␥S protection from tryptic digestion (data not shown). Consistent with previous experiments, ϳ80% of ␣ o cofractionated with membranes, whereas most but not all of ␣ o G2A cofractionated with cytosol (Fig. 6A) (Fig. 6B). Therefore, the 10lysine insert, which is predicted to disrupt normal interactions of ␣ o with ␤␥, was not deleterious to N-myristoylation or palmitoylation. This result suggests, again, that normal interactions with ␤␥ are not required for ␣ o to be palmitoylated. In the case of the G2A mutant, however, the 10-lysine insert did not restore the capacity of the subunit (␣ o G2A/K 10 ) to be palmitoylated, despite the large fraction of subunit targeted to membrane. This result confirms a requirement for N-myristoylation in palmitoylation despite membrane attachment.
The failure of the 10-lysine insert to rescue palmitoylation of ␣ o G2A might have resulted from incorrect targeting of the subunit. Caveolae are specialized plasma membrane domains that represent a major storage or reaction site for molecules involved in G protein-coupled signal transduction pathways, including G protein ␣ and ␤␥ subunits (13, 52-55). We therefore used cofractionation with caveolin-enriched domains as an assay of appropriate targeting. The results of a typical sucrose step gradient centrifugation that separates caveolin-enriched domains from the bulk of cellular membranes and cytosolic proteins are shown in Fig. 7. Caveolin, a marker for caveolae, was localized predominantly to fractions 4 and 5. In agreement with a previous report (13), the bulk of ␣ o cofractionated with caveolin. So, too, did ␣ o K 10 . ␣ o G2A was found with soluble protein (i.e. fractions 9 -12), and ␣ o G2A/K 10 partitioned between soluble protein and caveolin-enriched fractions. These data imply that ␣ o , ␣ o K 10 , and most of ␣ o G2A/K 10 that is not cytosolic target correctly.
We used an additional method of targeting ␣ o to membrane without N-myristoylation to test further the apparent requirement of palmitoylation for N-myristoylation following attachment of subunit to membrane. The ␣ o and ␣ o G2A subunits were appended to the C terminus of human 5-HT 1A R, which possesses a seven-transmembrane domain motif and is targeted exclusively to membrane. Similar constructs of the ␣ 2A -adrenoreceptor and ␣ i1 and the ␤ 2 -adrenoreceptor and ␣ s , when expressed in intact cells, are sensitive to extracellular ligands (56,57), and G protein-coupled receptors similarly fused to green fluorescent protein target predominantly to the plasma membrane (58,59). The fusion precludes N-myristoylation of the ␣ subunits. Cys 417 and Cys 420 of 5-HT 1A R, which represent potential sites of palmitoylation (60), were changed to alanines. Fig. 8A shows the distribution of 5-HT 1A R/␣ o and 5-HT 1A R/ ␣ o G2A. A prominent band at ϳ100 kDa, present in membrane only, was evident for both chimeric proteins. The mobility of the chimeras was less than predicted from cDNA (85 kDa), a difference that probably reflects glycosylation of the receptor. The functional integrity of the appended ␣ o and ␣ o G2A subunits was demonstrated in two ways. First, GTP␥S protected the subunit from proteolysis by trypsin (i.e. a released ϳ37-kDa ␣ o fragment was stabilized by GTP␥S; data not shown). Second,

Overexpressed ␤␥ Rescues the Labeling of ␣ o with [ 3 H]Palmitate when the N-Myristoyl Group Is Absent-
The results with ␣ o G2A/K 10 and 5-HT 1A R/␣ o support the assertion that an ␣ o subunit that cannot be N-myristoylated is not palmitoylated, even though it is targeted to membrane. However, it might be argued that the proteins are not positioned correctly on the membrane, whether due to the absence of the N-myristoyl group or not. Because N-myristoylation is relevant to the interaction of ␣ o with ␤␥, we evaluated PTX-catalyzed [ 32 P]ADPribosylation of the two mutants, as we did the other mutants (see above). We also tested whether overexpression of ␤␥ might promote [ 3 H]palmitate labeling. The data in Fig. 9A imply that the variant subunits indeed interact poorly with ␤␥, i.e. insertion of 10 lysines or linkage of ␣ o to 5-HT 1A R resulted in a much diminished incorporation of [ 32 P]ADP-ribose. ␣ o G2A/K 10 and 5-HT 1A R/␣ o alone did not incorporate [ 3 H]palmitate (Fig. 9B), as noted previously. However, coexpression of exogenous ␤␥ promoted labeling. The lack of an N-myristoyl group therefore normally precludes [ 3 H]palmitate labeling, even though the subunit is associated with membrane, but the defect can be overcome by overexpression of ␤␥. The mutant/chimeric subunits are therefore not resistant to assuming a conformation or positioning on the membrane conducive to palmitoylation.

DISCUSSION
That N-myristoylation, interactions with ␤␥, and proximity to membrane are interrelated events has made it difficult to sort out the contributions of any single one of these phenomena to palmitoylation of ␣ i family members. For the most part, proximity to the plasma membrane has been considered the key event, with palmitoylation occurring subsequently to establish a stable anchorage. Proximity can be achieved by Nmyristoylation alone (5, 6), by an N-myristoylation-facilitated interaction of ␣ with independently anchored ␤␥ (14), or, in lieu of N-myristoylation, by overexpression of ␤␥ (28). Studies re- garding the requirements for N-myristoylation, however, have relied almost solely on mutations that, while inhibiting Nmyristoylation, might have other effects. Furthermore, whether proximity represents the only requirement fulfilled by N-myristoylation and/or ␤␥ is unclear. The results presented here demonstrate that (i) palmitoylation indeed requires Nmyristoylation under normal circumstances, which was supported previously with a G2A mutation and is confirmed here by an S6N mutation and 2-HMA; (ii) palmitoylation is sensitive to the conformational status of ␣ o , demonstrated by the G204A and Q205L mutants; (iii) a conformation of ␣ o that increases affinity for ␤␥ overcomes the absence of an N-myristoyl group, shown with ␣ o G2A/G204A and ␣ o S6N/G204A; although (iv) palmitoylation does not usually require a normal interaction of ␣ o with ␤␥, shown with ␣ o ⌬8 -11 and ␣ o K 10 ; and (v) interactions with ␤␥ are required when the N-myristoyl group is absent, demonstrated with ␣ o G2A/K 10 and 5-HT 1A R/␣ o Ϯ ␤␥. We propose that N-myristoylation and ␤␥, while almost certainly relevant to achieving proximity of the subunit to membrane, also play redundant roles in palmitoylation following anchorage.
The S6N mutation and 2-HMA were used to help address the concern that the effects of G2A on palmitoylation in previous studies were not so much due to inhibiting N-myristoylation (and consequently palmitoylation) as they were to a disruption of palmitoylation directly at the adjacent Cys 3 . S6N conforms to Asn 6 in ␣ s and was selected in an attempt to conserve N-terminal function; ␣ s contains Cys 3 , but is not N-myristoylated (5). 2-HMA is an inhibitor of N-myristoylation that obviates the use of mutations altogether. Our results, in which manipulations inhibit both N-myristoylation and palmitoylation, agree partly with those of Galbiati et al. (61). These investigators found that an S6D mutation in ␣ i1 , like the S6N mutation in ␣ o here, inhibited N-myristoylation and palmitoylation. However, they did not observe an effect of 2-HMA on palmitoylation and consequently concluded that G2A and S6D mutations caused a deformation of the N terminus. Although the effects of 2-HMA in our study were less dramatic than those achieved with G2A or S6N, they reinforced the sequential relationship between N-myristoylation and palmitoylation implied by the two mutations.
The data here indicate that an association of ␣ o with membrane, even perhaps a stable anchorage, is not sufficient for [ 3 H]palmitate labeling. Small but easily distinguished populations of ␣ o G2A and ␣ o S6N, for example, often cofractionated with membrane, but did not incorporate [ 3 H]palmitate. Moreover, ␣ o Q205L cofractionated entirely with membrane, but incorporated only a small amount of radiolabel. In the case of membrane-associated ␣ o G2A and ␣ o S6N, the underlying defect would appear to be the lack of an N-myristoyl group or the consequent diminished affinity for ␤␥. The defect is less clear for ␣ o Q205L, as N-myristoylation was unaffected by the mutation, and the interaction with ␤␥, at least as intimated by PTX-catalyzed [ 32 P]ADP-ribosylation, appeared to be normal if not enhanced. We suspect that the active conformation itself assumed by ␣ o Q205L explains the diminished [ 3 H]palmitate labeling and that this conformation overrides the positive inputs provided by N-myristoylation and/or ␤␥ interaction. A GTPase-deficient form of ␣ s (␣ s R201C) has similarly been shown to label poorly with [ 3 H]palmitate (19). Unlike ␣ s R201C, however, no net translocation of ␣ o Q205L to cytosol was observed to occur, consistent with observations for ␣ i1 Q204L (22). The effects of G2A, S6N, and Q205L on N-myristoylation, ␤␥ interaction, or conformation may either abrogate palmitoylation directly or regulate depalmitoylation in such a manner as to inhibit radiolabel exchange or to cause a net decrease in palmitate associated with the subunits.
In stark contrast to ␣ o Q205L, ␣ o G204A was characterized by an enhanced labeling with [ 3 H]palmitate. ␣ o G204A, by reference to other Gly-to-Ala mutant ␣ subunits (44,45), resists entering into an active conformation and binds ␤␥ particularly tightly. Thus, the inactive conformation per se or the heightened interaction with ␤␥ enhances palmitoylation and/or protects the palmitoylated subunit from depalmitoylation; we assume that isotopic equilibrium has been reached and that the effects are not due to an enhanced turnover of label alone. The enhancement in [ 3 H]palmitate labeling achieved with ␣ o G204A was not limited to wild-type ␣ o . ␣ o G2A and ␣ o S6N, for which incorporation of [ 3 H]palmitate was not normally evident, incorporated [ 3 H]palmitate when mutated to additionally include G204A. Thus, we found that non-myristoylated subunits are capable of being palmitoylated. This has been shown previously for G2A mutants upon overexpression of ␤␥, wherein the amount of subunit associated with membrane increased (28), but our data demonstrate that it also occurs where changes in membrane association are not observed. Our results demonstrate that N-myristoylation is required for [ 3 H]palmitate labeling of ␣ o under normal circumstances and that mutations favoring inactive conformations and/or interaction with ␤␥ can enhance palmitate incorporation into ␣ o and rescue that of mutants devoid of the N-myristoyl group. Experiments with additional mutants were performed to test further the notion that N-myristoylation and ␤␥ exert roles in palmitoylation not only prior to, as usually acknowledged, but following anchorage of subunit to membrane. The mutants that were utilized toward this end included ␣ o ⌬8 -11, ␣ o K 10 , and 5-HT 1A R/␣ o . These experiments together demonstrate that Nmyristoylation or normal interactions with ␤␥, but not both, is required for [ 3 H]palmitate labeling and that this requirement exists regardless of alterations in membrane attachment: ␣ o ⌬8 -11 and ␣ o K 10 , which are N-myristoylated but interact poorly with ␤␥, incorporate [ 3 H]palmitate; ␣ o G2A/G204A and ␣ o S6N/G204A, which likely interact with ␤␥ but are not Nmyristoylated, also incorporate [ 3 H]palmitate; and ␣ o G2A, ␣ o S6N, ␣ o G2A/K 10 , and 5-HT 1A R/␣ o , which neither are N-myristoylated nor interact well with ␤␥, do not incorporate [ 3 H]palmitate unless (in the case of at least ␣ o G2A/K 10 and 5-HT 1A R/␣ o ) ␤␥ is overexpressed. There were hardly any differences in the extent to which ␣ o , ␣ o ⌬8 -11, ␣ o K 10 , and 5-HT 1A R/␣ o cofractionated with membrane, i.e. most if not all of these subunits were anchored. A substantial fraction of ␣ o G2A/ K 10 also cofractionated with membrane. Issues of membrane identity notwithstanding, the requirements for N-myristoylation or ␤␥ therefore appear to extend beyond those relating to proximity and anchorage.
The requirement for an N-myristoyl group in [ 3 H]palmitate labeling of ␣ o already attached to membrane is probably not related to its role in facilitating interaction of the subunit with ␤␥. ␣ o ⌬8 -11 and ␣ o K 10 , whose interactions with ␤␥ are disrupted, incorporate [ 3 H]palmitate to the same extent as ␣ o itself. The role of N-myristoylation may instead involve targeting of the membrane-attached subunit to a relevant microdomain or the performance of some other post-attachment event. Fishburn et al. (62) have proposed a scheme in which the N-myristoyl group supports attachment of ␣ z to internal membranes for expedited transfer to the plasma membrane and subsequent palmitoylation. Something of this nature may occur for ␣ o . However, our data with ␣ o G2A/K 10 suggest appro-priate plasma membrane targeting for at least a fraction of this non-myristoylated subunit, and there is good reason to believe that this is also the case with at least the 5-HT 1A R/␣ o chimeras based on the signaling and targeting properties of other receptor/protein fusion constructs in the intact cell (56 -59). It is more likely that the N-myristoyl group is recognized by a palmitoyltransferase as part of the substrate or that it positions the N terminus of the ␣ subunit in such a manner as to facilitate palmitoylation. This interpretation is consistent with the findings of Dunphy et al. (26) using bovine brain membrane extracts as a source of palmitoyltransferase activity and purified ␣ i1 , wherein N-myristoylated ␣ i was found to be the preferred substrate. It is also conceivable that N-myristoylation protects ␣ o against depalmitoylation, as suggested by Morales et al. with ␣ z (63). However, in mutants of ␣ o that lack the N-myristoyl group and that do not interact with ␤␥, we found no incorporation of [ 3 H]palmitate whatsoever, implying that, if protection against an esterase is the case, the esterase is extremely active. That the non-myristoylated subunits might be fully palmitoylated but unable to exchange the lipid for [ 3 H]palmitate is a formal possibility, in which case the Nmyristoyl group might enhance labeling by actually increasing esterase activity. Regardless, it is quite clear that N-myristoylation serves a role in regulating palmitoylation that exceeds that of providing a loose association with membrane.
Although normal interactions with ␤␥ appear not to be required for [ 3 H]palmitate labeling under normal circumstances, these interactions become crucial when the N-myristoyl group is absent. This was first suggested by the observation that G204A rescues palmitoylation of ␣ o G2A and ␣ o S6N and later demonstrated by the fact that ␤␥ rescues palmitoylation of ␣ o G2A/K 10 and 5-HT 1A R/␣ o G2A. ␤␥ has been proposed to help target ␣ subunits to the plasma membrane (62). However, as discussed above, the mutant subunits are probably already targeted correctly. We therefore suspect that ␤␥ performs essentially the same function as N-myristoylation: it enhances recognition of the ␣ subunit by a palmitoyltransferase, prevents depalmitoylation by an unusually active esterase, or enhances turnover of palmitate. These actions may be direct or related to changes in ␣ subunit conformation. Protection from an esterase is a commonly assumed role from studies with ␣ s (19 -21). However, Dunphy et al. (26) have shown an effect of ␤␥ on palmitoylation of purified ␣ i1 in vitro. The requirement for ␤␥ in our studies becomes evident only when the N-myristoyl group is absent.