The Importance of N-terminal Polycysteine and Polybasic Sequences for G14α and G16α Palmitoylation, Plasma Membrane Localization, and Signaling Function*

Plasma membrane targeting of G protein α (Gα) subunits is essential for competent receptor-to-G protein signaling. Many Gα are tethered to the plasma membrane by covalent lipid modifications at their N terminus. Additionally, it is hypothesized that Gq family members (Gqα,G11α,G14α, and G16α) in particular utilize a polybasic sequence of amino acids in their N terminus to promote membrane attachment and protein palmitoylation. However, this hypothesis has not been tested, and nothing is known about other mechanisms that control subcellular localization and signaling properties of G14α and G16α. Here we report critical biochemical factors that mediate membrane attachment and signaling function of G14α and G16α. We find that G14α and G16α are palmitoylated at distinct polycysteine sequences in their N termini and that the polycysteine sequence along with the adjacent polybasic region are both important for G16α-mediated signaling at the plasma membrane. Surprisingly, the isolated N termini of G14α and G16α expressed as peptides fused to enhanced green fluorescent protein each exhibit differential requirements for palmitoylation and membrane targeting; individual cysteine residues, but not the polybasic regions, determine lipid modification and subcellular localization. However, full-length G16α, more so than G14α, displays a functional dependence on single cysteines for membrane localization and activity, and its full signaling potential depends on the integrity of the polybasic sequence. Together, these findings indicate that G14α and G16α are palmitoylated at distinct polycysteine sequences, and that the adjacent polybasic domain is not required for Gα palmitoylation but is important for localization and functional activity of heterotrimeric G proteins.

Heterotrimeric G protein localization at the plasma membrane is critical for transmembrane signaling from activated receptors to linked effector molecules. Both the G␣ and G␥ subunits of the heterotrimer bind the lipid bilayer to position the G protein for activation by cognate G protein-coupled receptors (1,2). In response to receptor-induced activation with GTP, the G protein subunits redistribute and selectively stimulate membrane-bound effector molecules (reviewed in Refs. 3,4). The four G␣ subunits of the G q family, G q ␣, G 11 ␣, G 14 ␣, and G 16 ␣, transfer receptor-generated signals to phospholipase C-␤ isozymes 1-4 (PLC-␤1-4) 2 (reviewed in Ref. 5) and other protein binding partners (reviewed in Ref. 6). Activated G␣ subunits stimulate the enzymatic activity of PLC-␤, triggering the breakdown of the membrane lipid phosphatidylinositol (4,5)-bisphosphate into the second messenger molecules inositol (1,4,5) trisphosphate and diacylglycerol. G q ␣ and G 11 ␣ that are not associated with the membrane display impaired or abolished capacity to become activated by receptor or to activate PLC-␤ (7)(8)(9)(10)(11)(12)(13). Whether this also is the case for G 14 ␣ and G 16 ␣ is unknown because relatively few studies to date have focused on the signaling properties of these poorly understood G q family members.
Unlike the membrane-spanning receptors that directly activate G proteins, the soluble heterotrimeric G protein complex relies on extrinsic forces to mediate localization at the plasma membrane. Long chain lipid modifications of both the G␣ and G␥ subunits directly interact with lipid bilayers and stabilize the protein complex at the plasma membrane. In general, G␣ subunits are targeted for lipid modification at their N terminus, but the profiles of lipid incorporation vary among G␣ subtypes with regard to the fatty acid moiety and the sites of attachment (1). G q ␣ and G 11 ␣ both incorporate palmitate at two N-terminal cysteine residues (Cys-9 and Cys-10) (7, 14 -16). Palmitate is a 16-carbon saturated fatty acid that is bound post-translationally to regulate localization and function of target proteins (reviewed in Refs. 17,18). Unpalmitoylated G q ␣ and G 11 ␣ (C/A or C/S site mutants) are cytosolic and cannot mediate receptorto-effector signaling compared with wild-type G q/11 ␣ (7-9, 12,13,16,19).
Nothing is known about the lipidation states of G 14 ␣ and G 16 ␣ or of other biochemical mechanisms that may regulate G 14 ␣ and G 16 ␣ plasma membrane localization and signaling function. Within the region of the N-terminal ␣-helix, the sequences of G 14 ␣ and G 16 ␣ are notably different from G q ␣ and G 11 ␣ (Fig. 1A). Although G q ␣ and G 11 ␣ are 83% identical in their first 40 amino acids, G 14 ␣ and G 16 ␣ share only 65 and 35% identity, respectively, with G q ␣ across the same region. Furthermore, G 14 ␣ and G 16 ␣ are unique among G␣ subunits in that they each have three cysteine residues in their N-terminal domain that are putative sites for palmitoylation. Consequently, the N termini of G 14 ␣ and G 16 ␣ may be differentially involved in the biochemical regulation of membrane localization and signaling function of these particular G␣ proteins.
Current evidence supports a dual requirement for G␣ palmitoylation and heterotrimer formation to drive plasma membrane localization of G q ␣ and G 11 ␣ (11, 12, 19 -21). Recent models also suggest that G q family members in particular may utilize multiple positively charged basic residues in the G␣ N terminus as a third signal for membrane targeting and attachment. Three-dimensional molecular modeling of G q ␣, G 11 ␣, G 14 ␣, and G 16 ␣ predicts a cluster of basic amino acids in the N terminus of each G␣ to fold in such a way so as to form a positively charged patch on the protein surface (22). These residues are expected to align on one face of the G␣ N-terminal ␣-helix in a position favorable for ionic interactions with anionic phospholipids in the plasma membrane. It also has been postulated that this polybasic region is a necessary signal for G␣ palmitoylation (22). Palmitoylation occurs at cellular membranes, and palmitoyltransferase activity appears to be enriched in the plasma membrane specifically (17,23). Therefore, G␣ localization at the plasma membrane mediated by the electrostatic potential of the polybasic region may be a prerequisite for protein palmitoylation. These ideas, however, have never been tested experimentally.
In this study, we identify critical G␣-specific N-terminal amino acids and biochemical factors that mediate membrane localization and signaling function of G 14 ␣ and G 16 ␣. We show that G 14 ␣ and G 16 ␣ are palmitoylated at distinct polycysteine sequences in their N termini and that the G proteins utilize these sequences for plasma membrane association and effector activation. Furthermore, G 16 ␣ exhibits a functional dependence on the N-terminal polybasic domain in addition to the polycysteine sequence to maintain membrane localization and biological activity. With a comprehensive comparison of these properties in the isolated G␣ N termini and in the full-length heterotrimeric G proteins, we report the unexpected finding that the full-length G␣ proteins exhibit differential dependence on the polycysteine and polybasic sequences for localization and function than do the N termini when expressed alone as EGFP fusion proteins. Importantly, we show for the first time that the polybasic domain itself is not involved in palmitoylation of a proximal substrate like the adjacent polycysteine sequences in the G␣ N terminus, but instead plays a key role in localization and function of the whole heterotrimeric G protein.

EXPERIMENTAL PROCEDURES
Plasmids and Materials-Human G 14 ␣ and G 16 ␣ cDNA expression plasmids with internal Glu-Glu (EE) epitope tags (G 14 ␣, G 14 ␣-Q205L (G 14 ␣-Q/L), G 16 ␣, and G 16 ␣-Q212L (G 16 ␣-Q/L)) were purchased from University of Missouri-Rolla cDNA Resource Center, as were the G␤ 1 and G␥ 2 expression plasmids. pCDNA3.1 was from Invitrogen, and pEGFP-N1 was from Clontech. KpnI, AgeI, NotI, and SacII restriction enzymes were purchased from New England Biolabs. Primers were synthesized by Sigma Genosys and Operon Biotechnologies, Inc. The QuikChange site-directed mutagenesis kit was purchased from Stratagene. HEK 293 and HeLa cells were from ATCC. Lipofectamine TM 2000 transfection reagent was purchased from Invitrogen. Anti-FLAG M2 affinity gel and anti-FLAG M2 monoclonal antibody-peroxidase conjugate both were from Sigma. BD Living Colors TM Full-Length A.v. polyclonal antibody (anti-GFP) was from Clontech, and Glu-Glu monoclonal antibody (anti-EE) was from Covance. B861 (anti-G 16 ␣), a polyclonal antibody generated against a peptide fragment of G 15/16 ␣ and Z811, an anti-G q/11/14/16 ␣ polyclonal antibody, which recognizes each of the G␣ subunits of the G q ␣ family and was generated against a C-terminal peptide sequence shared by G q ␣ and G 11 ␣, were generous gifts from Dr. Paul Sternweis (University of Texas Southwestern Medical Center, Dallas). Protein A-Sepharose 4 Fast Flow was from GE Healthcare. Peroxidase-conjugated goat anti-mouse IgG antisera was purchased from Rockland, Inc., and peroxidase-conjugated goat anti-rabbit was from Bio-Rad. [ cDNA Constructs and Mutagenesis-PCR amplification was used to isolate the N-terminal domains of G 14 ␣ and G 16 ␣ from the full-length G␣ plasmids. Specifically, primers were designed to amplify cDNA sequences G 14 ␣-(1-102) and G 16 ␣-(1-123). G␣-specific oligomers included an N-terminal KpnI restriction site (sense oligomers) or an AgeI C-terminal site (antisense oligomers). For amplifying the N terminus of G 14 ␣ (NTG 14 ␣), the primers were as follows: 5Ј-AAGGTACCGCC-ACCATGGCCGGCTGCTG, CTGC-3Ј (sense) and 5Ј-CGAC-CGGTCCACGGCGCGCGTCCTTCTTG-3Ј (antisense). For amplifying the N terminus of G 16 ␣ (NTG 16 ␣), the primers were as follows: 5Ј-AAGGTACCGCCACCATGGCCCGCTCGCT-GACC-3Ј (sense) and 5Ј-CGACCGGTCCCCCGCGGTCCT-GCTTCTTC-3Ј (antisense). The amplified G␣ sequences were digested with KpnI and AgeI restriction enzymes and subcloned into pEGFP-N1 upstream and in-frame with the EGFP coding region.
A FLAG epitope tag was added to the 3Ј end of the NTG␣-EGFP coding region by PCR amplification of a KpnI-NotI fragment that included the FLAG sequence. An antisense primer was designed to eliminate the stop codon of the EGFP sequence and introduce the FLAG sequence with a new 3Ј stop codon. The antisense primer was 5Ј-ATCGATCGATGCGGCCGCT-TTACTTGTCGTCGTCGTCCTTGTAGTCCTTGTACAG-CTCGTCCATGCC-3Ј and was used in combination with either the NTG 14 ␣ or NTG 16 ␣ sense primer listed previously to generate a KpnI-NotI fragment from the existing NTG␣-EGFP sequence for subcloning into pEGFP-N1.
Cell Culture and Transient Transfection-HEK 293 cells and HeLa cells were cultured in complete medium consisting of Dulbecco's modified Eagle's medium (DMEM; CellGro) supplemented with 10% fetal bovine serum (Atlanta Biologicals) and penicillin/streptomycin (CellGro). For transient transfections, Lipofectamine TM 2000 transfection reagent was used according to the manufacturer's suggested protocol. Transfections were carried out for 5 h at 37°C, followed by an overnight incubation in complete medium. Experiments were conducted 24 h after transfection.
Fractionation and Serial Extraction-Cells were transiently transfected at a density of 7.125 ϫ 10 6 cells/plate in 10-cm 2 plates with 24 g of DNA/plate. The following day, cells were lysed in hypotonic lysis buffer (50 mM HEPES, pH 8.0, 1 mM EDTA, 2 g/ml aprotinin, 1 g/ml leupeptin, 1 M phenylmethylsulfonyl fluoride) with a Teflon Dounce homogenizer (50 strokes), and broken cells were normalized to isotonic conditions (50 mM HEPES, pH 8.0, 5 mM MgCl 2 , 150 mM NaCl, 1 mM EDTA, 250 mM sucrose, 2 g/ml aprotinin, 1 g/ml leupeptin, 1 M phenylmethylsulfonyl fluoride). Nuclei and unbroken cells were pelleted at 600 ϫ g, 4°C for 10 min, and the remaining lysate was fractionated by ultracentrifugation at 100,000 ϫ g, 4°C for 30 min. A sample of supernatant was collected and saved for immunoblot analysis of the cytosol fraction. The membrane pellet was resuspended in Low Salt Buffer (50 mM HEPES, pH 8.0, 5 mM ␤-mercaptoethanol, 150 mM NaCl, 5 mM MgCl 2 , 1 mM EDTA, 1 g/ml leupeptin, 2 g/ml aprotinin, 1 M phenylmethylsulfonyl fluoride), Dounce-homogenized as before, and ultracentrifuged again. This process was repeated a third time to remove nonspecific proteins in the membrane fraction, and a sample of the washed membrane fraction was collected. Membrane and cytosol samples were separated by SDS-PAGE, transferred to nitrocellulose membranes (Millipore Corp.), and immunoblotted with antibodies for visualization.
Serial extraction of harvested membranes was performed as described previously with modifications (24,25). In brief, membranes were prepared from transfected cells by fractionation as described. Following the initial ultracentrifugation of cell lysates, membranes were washed three times in Low Salt Buffer, followed by three washes in buffer supplemented with 1 M NaCl (50 mM HEPES, pH 8.0, 1 M NaCl, 5 mM ␤-mercaptoethanol, 5 mM MgCl 2 , 1 mM EDTA, 1 g/ml leupeptin, 2 g/ml aprotinin, 1 M phenylmethylsulfonyl fluoride), and three final washes in buffer containing 1% Triton X-100 (50 mM HEPES, pH 8.0, 1% Triton X-100, 5 mM ␤-mercaptoethanol, 5 mM MgCl 2 , 1 mM EDTA, 1 g/ml leupeptin, 2 g/ml aprotinin, 1 M phenylmethylsulfonyl fluoride) for a total of nine washes. Each wash was carried out at 4°C with end-over-end rotation for 30 min, followed by ultracentrifugation at 100,000 ϫ g at 4°C for 30 min. After each spin, the membrane pellet was resuspended in the appropriate buffer and Dounce-homogenized (50 strokes). Equivalent sample volumes were collected from indicated wash supernatants and membrane pellets. Samples were loaded onto polyacrylamide gels for SDS-PAGE, and proteins were transferred to nitrocellulose membranes for immunoblot analysis. To assist the reader in interpretation of the data, Adobe Photoshop 7.0 (Adobe Systems) was used to duplicate bands representing the pelleted membrane fraction from each wash series ("P" bands) and display in adjacent lanes as total material for the next wash series ("T" bands), as both represent the same pool of protein.
Where indicated, protein expression levels were quantified by densitometry analysis using Scion Image freeware software. Briefly, protein bands were selected within identical rectangles, and the integrated density was calculated for each selection. Background density measured in an identical rectangle on a section of film separate from the protein band of interest was subtracted from each measurement.
Metabolic Labeling and Immunoprecipitation-Cells were transfected at a density of 7.125 ϫ 10 6 cells/plate in 10-cm 2 plates with a total of 24 g of DNA/plate. For experiments with the NTG␣-EGFP-FLAG fusion proteins, 24 g of the fusion protein of interest was transfected per plate. For experiments with full-length G␣, cells were transfected with 16.8 g of EEtagged G␣ Ϯ C/S, 4.8 g of G␤ 1 , and 2.4 g of G␥ 2 . Twenty four hours after transfection, cells were labeled and immunoprecipitated according to a method described in detail elsewhere with minor modifications (27,28). Specifically, cells for autoradiography were metabolically labeled with 1 mCi/ml [ 3 H]palmitic acid for 2 h at 37°C in palmitate labeling media (DMEM, 2.5% dialyzed fetal bovine serum, 1ϫ sodium pyruvate, and 1ϫ non-essential amino acids (CellGro)). Because G 14 ␣ has been recalcitrant to metabolic labeling with [ 3 H]palmitate in these studies for unknown reasons, three 10-cm 2 plates of G 14 ␣ Ϯ C/S-expressing cells were used for each labeling experiment and each immunoblotting experiment, whereas one 10-cm 2 plate of transfected cells was used for each of the other experiments. All cells were lysed and fractionated as described. Cytosol fractions were saved for immunoprecipitation.
For G␣ Ϯ C/S-expressing cells, membrane pellets were immediately resuspended in Tris Buffer without an additional wash step, and D␤M was added to both membrane and cytosol fractions to a final concentration of 2%. G 14 ␣ Ϯ C/S and G 16 ␣ Ϯ C/S were extracted from membranes and cytosol with a 3-h incubation in 2% D␤M at 4°C with end-over-end rotation. Cell debris was pelleted by ultracentrifugation (75,000 rpm, 4°C, 30 min). Prior to immunoprecipitation, soluble extracts were blocked for 2 h with 0.7 g/l bovine serum albumin (Sigma) and 50 l of protein A-Sepharose. Anti-EE monoclonal antibody was diluted 1:150 in each membrane and cytosol sample for overnight immunoprecipitation of EE-tagged G 14 ␣ and G 16 ␣ proteins at 4°C. The following day, Sepharose beads were pelleted at 200 ϫ g, 4°C for 1 min. Beads were washed three times with Tris Buffer ϩ 0.01% D␤M, then resuspended in 2ϫ Laemmli sample buffer.
All samples were heated to 95°C for 1 min and spun down at 14,000 rpm, and the entire supernatants were loaded onto 11% polyacrylamide gels (G␣ Ϯ C/S samples) or 15% polyacrylamide gels (NTG␣-EGFP-FLAG samples) for SDS-PAGE separation. Samples for immunoblot analysis were transferred to nitrocellulose membranes by Western transfer, and immunoblotting was carried out as described. Gels for autoradiography were prepared according to a method described previously (28). Fixed proteins were treated with En 3 Hance TM solution for autoradiography amplification according to manufacturer's instructions and were exposed to film at Ϫ80°C for indicated times. Hydroxylamine treatment and [ 35 S]methionine labeling were carried out as described previously (28).
Confocal Fluorescence Microscopy-HeLa cells at a density of 1.75 ϫ 10 5 cells/well were transfected directly on sterile coverslips in 24-well plates with 0.8 g of DNA/well. Twenty four hours post-transfection, cells were fixed using an adaptation of methods described previously (29). In brief, cells were washed one time in phosphate-buffered saline and then incubated in fixation buffer (20 mM PIPES, pH 7.0, 1 mM MgCl 2 , 0.5 mM EGTA, 1 mM glutaraldehyde, 1 g/ml aprotinin, 0.1% Triton X-100, 3.7% paraformaldehyde) for 15 min at room temperature. Coverslips were mounted on slides using Vectashield mounting medium (Vector Laboratories). An LSM510 confocal laser scanning microscope (Zeiss) was used for imaging, and pictures were taken using a 100ϫ oil immersion objective. Images were processed using the Zeiss LSM image browser (version 2.801123) and Adobe Photoshop 7.0 (Adobe Systems).
Measurement of Inositol Phosphate Accumulation-Inositol phosphate formation was measured as described elsewhere (30). In brief, 4.75 ϫ 10 5 HEK 293 cells per well in 12-well plates were transfected as described with EE-tagged G␣-Q/L Ϯ C/S or pCDNA3.1, G␤ 1 and G␥ 2 cDNA in a ratio of 6:3:1 G␣:␤ 1 :␥ 2 . The following amounts of G␣ cDNA were used per well: 0.2 g of G 14 ␣-Q/L WT, -C4S, -C6S, and G 16 ␣-Q/L WT; 0.6 g of G 14 ␣-Q/L-C5S,C6S and -C4S,C5S,C6S. 0.3 g/well of all other G␣-Q/L cDNAs were used for transfection. cDNA amounts of G␤ 1 and G␥ 2 were adjusted to maintain the 6:3:1 ratio, and pCDNA3.1 was transfected to bring total amount of cDNA/well up to 0.5 g (G 14 ␣-Q/L-C5S,C6S and -C4S,C5S,C6S transfections had an approximate total of 1 g/well). cDNA amounts were determined according to similar protein expression levels determined by Western blot and densitometry analysis. Cells were transfected for 5 h and then metabolically labeled overnight in inositol-free complete medium containing 4 Ci/ml [ 3 H]inositol. The following day, cells were incubated for 50 min at 37°C in DMEM buffered with 25 mM HEPES, pH 8.0, and containing 10 mM LiCl 2 . After solubilization of cells with 20 mM formic acid and neutralization with 0.7 M NH 4 OH, [ 3 H]inositol phosphates were separated by anion exchange chromatography (AG 1-X8 Dowex, Bio-Rad) using increasing amounts of ammonium formate. Total [ 3 H]inositol phosphate content was assessed by liquid scintillation spectrometry. Cell lysates were prepared for immunoblotting by extracting protein directly in the plate wells with 1ϫ Laemmli sample buffer. Collected lysates were sonicated for 5 s, and proteins were denatured at 95°C for 5 min, and 20% of each lysate was loaded on an 11% polyacrylamide gel for electrophoretic separation, followed by Western transfer and immunoblotting. Protein levels were quantitated by densitometry analysis, and inositol phosphate accumulation was plotted as a ratio of total inositol phosphates to adjusted G␣ protein expression level.

RESULTS
The N Termini of G 14 ␣ and G 16 ␣ Are Membrane Targeting Domains-The primary objective of this study was to determine the potential for the G 14 ␣ and G 16 ␣ N termini to direct plasma membrane localization, with the specific goals of identifying hydrophobic and ionic signals within this region important for localization and G␣ signaling function. Therefore, we utilized two complementary approaches to compare biochem-ical properties that regulate the isolated G␣ N-terminal peptides and the corresponding full-length G 14 ␣ and G 16 ␣ proteins in cells. In this way, we were able to dissect membrane targeting signals that are native to the specific region of the G␣ N terminus and independent of G␤␥ interactions, and then evaluate the importance of these signals for normal biological activity of the full-length proteins.
To begin, we isolated the N-terminal domains of G 14 ␣ and G 16 ␣ by PCR ( Fig. 1) and expressed them as fusion proteins with enhanced green fluorescent protein (EGFP). The fusion proteins express robustly in HEK 293 and HeLa cells and behave as soluble, properly folded proteins based on biochemical analyses presented in this study. The secondary structure of G q ␣ has been predicted based on sequence alignments with the previously crystallized G t ␣/ i ␣ chimera (31), and this information was used to define homologous regions of the G 14 ␣ and G 16 ␣ sequences that span the ␣-helical domain at the extreme N terminus (G 14 ␣-(1-34) and G 16 ␣-(1-41); Fig. 1A). These sequences were subcloned upstream of the EGFP coding region in pEGFP-N1 to generate fusion constructs (NTG 14 ␣-EGFP-FLAG and NTG 16 ␣-EGFP-FLAG; Fig. 1B), and a FLAG epitope tag was engineered at the C termini of the fusion constructs for immunoprecipitation purposes. Expression of the fusion proteins in HEK 293 cells results in the predicted shift in molecular weight relative to EGFP alone (Fig. 1C).
To test if addition of the G␣ N termini promotes membrane localization of EGFP, we examined the subcellular fractionation properties of EGFP-FLAG and the NTG␣-EGFP-FLAG fusion proteins expressed in HEK 293 cells (Fig. 2). Immunoreactivity in samples of cytosolic and particulate fractions indicates that the N termini of G 14 ␣ and G 16 ␣ drive membrane association of the otherwise soluble EGFP ( Fig. 2A). We believe that the lower protein band in the cytosol fraction is N-terminally degraded fusion protein because this protein band is found exclusively in cytosol, similar to EGFP alone ( Fig. 2A). Confocal fluorescence imaging of the EGFP fusion proteins in HeLa cells supports the fractionation data, indicating that the fusion proteins, but not EGFP, are targeted to the plasma membrane (Fig. 2B). Similar results were obtained with HEK 293 cells (data not shown).
Hydrophobic Interactions Mediate Association between the Plasma Membrane and the N Termini of G 14 ␣ and G 16 ␣-To identify the biochemical nature of the membrane association of the NTG␣-EGFP-FLAG fusion proteins, we subjected membranes from transfected cells to serial extractions in low ionic, high ionic, and hydrophobic buffers. Membranes were washed first three times with physiological salt (100 mM NaCl) and then three times with high salt (1 M NaCl) to disrupt nonspecific and ionic protein interactions, respectively, with membrane lipids. These washes were followed by three extractions with detergent to release hydrophobic proteins in the membranes. Both NTG 14 ␣-EGFP and NTG 16 ␣-EGFP in membranes were resistant to washes in low and high salt (Fig. 3). The majority of membrane-bound NTG 14 ␣and NTG 16 ␣-EGFP was extracted completely with a single 30-min wash in 1% Triton X-100 (Fig.  3). These findings indicate that the G 14 ␣ and G 16 ␣ N-terminal domains confer hydrophobic properties to the GFP fusion proteins that mediate association with the plasma membrane. Addition of the C-terminal FLAG tag did not change extraction properties of either protein (see Fig. 6B).   16 ␣ associate with the plasma membrane through hydrophobic interactions. Serial extraction of EGFP fusion proteins from HEK 293 cell membranes. Twenty four hours after transfection, membranes expressing the EGFP fusion proteins were prepared by ultracentrifugation and subjected to serial washes in buffers supplemented with 100 mM NaCl, 1 M NaCl, or 1% Triton X-100 (TX-100) (three washes in each buffer). Equivalent sample volumes of total membranes (T), first supernatant (S1), third supernatant (S3), and post-extraction pellet (P) were resolved by SDS-PAGE (15% polyacrylamide). Proteins were transferred to nitrocellulose membranes and immunoblotted with anti-GFP antisera. To assist the reader in interpretation of the data, adjacent P and T bands were duplicated using Adobe Photoshop 7.0 (Adobe Systems) as described under "Experimental Procedures." Results are representative of at least three independent experiments. FIGURE 1. The N-terminal domain of G q family members is highly diverse. A, amino acid sequence alignment of the N termini of the G q family members. Conserved residues are indicated below the alignment with an asterisk. Palmitoylated cysteine residues (G q ␣/G 11 ␣) and putative sites for palmitoylation (G 14 ␣/G 16 ␣) are identified in boldface type. Basic positively charged residues are identified above the alignment with ؉. B, schematic illustrating the structure of the NTG␣-EGFP-FLAG fusion constructs. C, NTG␣-EGFP fusion protein expression in lysates of transfected HEK 293 cells. Cells were lysed in 1ϫ Laemmli Sample Buffer 24 h post-transfection. 20% of each lysate was loaded onto a 15% polyacrylamide gel for PAGE separation, followed by Western transfer onto nitrocellulose membranes. Membranes were probed with an anti-FLAG antibody, and chemiluminescence was used for visualization. G 14 ␣ and G 16 ␣ Are Palmitoylated at Their N Termini-Although the N-terminal domains of G 14 ␣ and G 16 ␣ exhibit hydrophobic properties in membrane preparations, the primary sequences appear to encode soluble proteins. The most likely source of hydrophobicity for soluble proteins is incorporation of long chain fatty acids. Many G␣ subunits are lipidmodified at the N terminus with myristate and/or palmitate (1), although it is unknown whether or not G 14 ␣ and G 16 ␣ incorporate lipid. G 14 ␣ and G 16 ␣ are unique among G␣ because they each have three cysteine residues in their respective N termini that are putative acceptor sites for palmitate (27,32) (Fig. 1A). BecauseG 14 ␣ and G 16 ␣, like G q ␣ and G 11 ␣, lack the ϩ2 glycine sequence requirement for myristoylation (27), we screened the G␣ N-terminal fusion proteins for incorporation of [ 3 H]palmitate. HEK 293 cells transiently expressing EGFP-FLAG, NTG 14 ␣-EGFP-FLAG, or NTG 16 ␣-EGFP-FLAG were metabolically labeled with [ 3 H]palmitic acid. Membrane and cytosolic fractions were prepared from labeled cells, and EGFP-FLAG or NTG␣-EGFP-FLAG fusion constructs were immunoprecipitated from each fraction. Although EGFP-FLAG was entirely cytosolic, NTG 14 ␣and NTG 16 ␣-EGFP-FLAG were found in both cytosol and membrane fractions by immunoblotting as shown previously ( Fig. 2A and Fig. 4A). Only membranebound NTG 14 ␣and NTG 16 ␣-EGFP-FLAG, however, yielded a 3 H radiolabel signal by autoradiography (Fig. 4A). Again, we believe the lower protein band in cytosol immunoprecipitates is fusion protein-degraded at the N terminus because it is purely cytosolic protein and does not incorporate palmitate.
Thioester bonds that attach palmitate to target proteins are sensitive to NH 2 OH cleavage (28). Therefore, to confirm the source of the autoradiograph band as a [ 3 H]palmitate signal, we subjected gels from a similar labeling experiment to washes with either 0.5 M Tris, pH 7, or 0.5 M NH 2 OH, pH 7. After washing immunoprecipitates with NH 2 OH, the 3 H radiolabel signals for NTG 14 ␣and NTG 16 ␣-EGFP-FLAG were lost, whereas Tris Buffer did not affect either signal (Fig. 4B), thereby confirming the attachment of [ 3 H]palmitate as the source of the signals. Cells were labeled in parallel with [ 35 S]methionine to control for loss of protein during gel washes.
Palmitoylation Requires Different Sequence-specific Sites in G 14 ␣ and G 16 ␣ N Termini-With three cysteine residues each in the N termini of G 14 ␣ and G 16 ␣, there are seven combinations of cysteines that potentially could be important for palmitoylation. Two cysteines of the three in G 14 ␣ and G 16 ␣ (Cys-5,Cys-6 and Cys-9,Cys-10, respectively) are conserved across G q family members (Fig. 1A). Each of these cysteines in G q ␣ and G 11 ␣ are palmitoylated because mutation of both residues together is necessary to eliminate palmitoylation (7,16). For this reason, we began identification of sites important for palmitoylation by using site-directed mutagenesis to change the pair of conserved cysteines in NTG 14 ␣and NTG 16 ␣-EGFP-FLAG to serines. Similar to G q ␣ and G 11 ␣, the double cysteine-to-serine (C/S) point mutants of both fusion constructs failed to incorporate palmitate (Fig. 5). Accordingly, mutation of all three cysteine residues in both NTG 14 ␣and NTG 16 ␣-EGFP-FLAG also eliminated palmitoylation. These results established the requirement of one or both of the conserved cysteines for palmitoylation of the N termini of G 14 ␣ and G 16 ␣. The necessity of the third, nonconserved cysteine residue, however, was unresolved. Unexpectedly, we found that this third cysteine is differentially required for palmitoylation of the G 14 ␣ and G 16 ␣ N termini. Mutating each cysteine individually in NTG 14 ␣-EGFP-FLAG revealed that all three amino Twenty four hours after transfection, cells expressing indicated proteins were metabolically labeled for 2 h with 1 mCi/ml [ 3 H]palmitic acid. Cell lysates were separated into membrane (M) and cytosol (C) fractions by ultracentrifugation at 100,000 ϫ g. EGFP fusion proteins were extracted from membranes with 2% D␤M (3 h, 4°C) and immunoprecipitated from each fraction overnight at 4°C with anti-FLAG affinity gel. Immunoprecipitates were resolved by SDS-PAGE (15% polyacrylamide). Top box, Western blot analysis of cytosol-and membrane-immunoprecipitated proteins with an anti-FLAG antibody. Bottom box, autoradiography of labeled proteins. Proteins were fixed as described under "Experimental Procedures" and were exposed to film for 5 days at Ϫ80°C. Results are representative of at least three independent experiments. B, NH 2 OH sensitivity of autoradiography band. To verify [ 3 H]palmitic acid as the source of the autoradiography signal, membrane immunoprecipitates from transfected cells were washed with NH 2 OH to cleave thioester linkages. Cells were labeled in parallel with [ 35 S]methionine to control for protein loss during washes. Proteins were fixed and exposed as described in A. acids are required for palmitoylation of the G 14 ␣ N-terminal domain. Similar analysis of the cysteines in the N terminus of G 16 ␣ indicated that whereas both Cys-9 and Cys-10 are necessary for palmitoylation, the third cysteine (Cys-13) is dispensable under these conditions (Fig. 5). These findings are summarized in Table 1. Of note, NTG 14 ␣and NTG 16 ␣-EGFP-FLAG fusion proteins containing multisite C/S mutations remain in the membrane fraction to a degree (Fig. 5). We believe this is nonspecific membrane association because these mutants do not incorporate palmitate. Furthermore, much of this protein partitions to the cytosol with additional washes in physiological salt (see Fig. 6B).
The Palmitoylation State of the G␣ N Termini Correlates with Plasma Membrane Localization Patterns-Consistent with a role for palmitoylation in promoting membrane attachment, immunoprecipitation studies revealed a corresponding decrease in fusion protein in cell membranes for all cysteine point mutants displaying a loss of palmitoylation (Fig. 5). To verify these findings, we used confocal fluorescence imaging to examine subcellular localization of the C/S mutants in intact HeLa cells. Consistent with our earlier results, each of the C/S single or multisite mutants of NTG 14 ␣and NTG 16 ␣-EGFP-FLAG that failed to incorporate palmitate also did not localize to the plasma membrane but instead displayed diffuse cytosolic and nuclear staining (supplemental Fig. 1 and Table 1). Of note, the single cysteine point mutation of NTG 16 ␣-EGFP-FLAG that did not interfere with protein palmitoylation, C13S, retained the plasma membrane targeting pattern seen with the wild-type G 16 ␣ fusion construct.

Critical Cysteine Residues, but Not the Polybasic Region, Are Necessary for Plasma Membrane Binding of G␣ N Termini-
The addition of palmitate to the EGFP fusion proteins provides a mechanism for membrane binding and likely is responsible for the hydrophobic nature of the membrane interaction shown in Fig. 3. However, it also has been postulated that basic residues in the N termini of G 14 ␣ and G 16 ␣ form a positively charged structural patch on one surface of the N-terminal ␣-helix and help mediate membrane binding through ionic interactions (22), although this has never been tested experimentally. To test this hypothesis, we introduced combinations of multiple point mutations in the polybasic regions of NTG 14 ␣-EGFP-FLAG and NTG 16 ␣-EGFP-FLAG that eliminated increasing amounts of this charge. Ultimately, we selectively changed nine and five basic amino acids in NTG 14 ␣-EGFP-FLAG and NTG 16 ␣-EGFP-FLAG, respectively, to glutamine (NTG 14 ␣-9Q and NTG 16 ␣-5Q; Fig. 6A). Cell membranes expressing wild-type fusion protein, nonpalmitoylated C/S point mutants, or polybasic region point mutants were subjected to serial protein extraction with washes of increasing concentrations of salt or detergent. The nonpalmitoylated C/S fusion proteins were washed from the membranes with a single rinse in physiological salt buffer (150 mM NaCl; Fig. 6B). Residual NTG 14 ␣-C4S,C5S,C6S-EGFP-FLAG was removed in a single high salt wash, and whereas a relatively small amount of NTG 16 ␣-C9S,C10S,C13S-EGFP-FLAG was retained in membranes until detergent treatment, we suggest that this residual protein is aggregated and diluted to undetectable levels following multiple washes. The large majority of both proteins, however, were removed in the initial wash, suggesting that associa-

Subcellular localization patterns and palmitoylation state of NTG␣-EGFP-FLAG mutants
HeLa cells on coverslips were transfected with indicated DNAs and fixed 24 h post-transfection for confocal microscopic fluorescence imaging (see "Experimental Procedures"). Results are based on data presented in Fig. 4 and supplemental Fig.  1. ϩϩϩ indicates most intense fluorescence, ϩ indicates relatively minimal fluorescence, Ϫ indicates no fluorescence. The specific R/Q and K/Q mutations are defined under "Experimental Procedures." Observations were based on many cells observed in at least three separate transfections of each DNA construct. The palmitoylation state of each construct is based on data presented in Fig. 5 and was determined as described under "Experimental Procedures." ND indicates no data; constructs were not tested for incorporation of palmitate.

Plasma membrane
Cytosol a Palmitate incorporation ϩϩϩ ϩ Yes a Cytosol localization includes apparent nuclear localization. Twenty four hours after transfection, membranes expressing the EGFP fusion proteins were prepared by ultracentrifugation at 100,000 ϫ g and subjected to serial washes in buffers supplemented with 150 mM NaCl, 1 M NaCl, or 1% Triton X-100 (TX-100) (three washes in each buffer). Equivalent sample volumes of total membranes (T), first supernatant (S1), third supernatant (S3), and post-extraction pellet (P) were resolved by SDS-PAGE (15% polyacrylamide). Western blot analysis was performed using an anti-FLAG antibody. To assist the reader in interpretation of the data, adjacent P and T bands were duplicated using Adobe Photoshop 7.0 (Adobe Systems) as described under "Experimental Procedures." tion with membranes in the absence of palmitoylation is because of nonspecific interactions. By contrast, mutation of all or most of the basic charges in the G 14 ␣ or the G 16 ␣ N terminus, respectively, did not affect the biochemical properties of the fusion proteins in this assay. The polybasic region mutants each behaved similarly to wild type and associated with the membrane fraction in a detergent-sensitive manner (Fig. 6B).

The N-terminal Polybasic Regions in G 14 ␣ and G 16 ␣ Are Dispensable for Plasma Membrane Localization and Palmitoylation of the Fusion Constructs-
The polybasic region mutants retain association with cellular membranes in simple cell fractionation. To test whether disruption of the polybasic region interferes with plasma membrane targeting specifically, we examined the subcellular localization of the constructs in intact HeLa cells by confocal fluorescence imaging. Both NTG 14 ␣-9Q and NTG 16 ␣-5Q are targeted primarily to the plasma membrane and not other intracellular membranes (Fig. 7). We also noted that the polybasic mutants appear to display decreased nuclear staining compared with their wild-type counterparts, particularly for NTG 16 ␣-EGFP-FLAG. Although nuclear markers were not used to verify nuclear localization of the wildtype fusion proteins, the pattern of intracellular fluorescence is consistent with recent work from our laboratory in which we identified the nuclear compartment in the same cells by Hoechst 33258 staining (29). It is possible that elimination of the basic charges in the N termini of G 14 ␣ and G 16 ␣ interrupts an artificial nuclear localization signal that may be utilized by these GFP fusion proteins, as polybasic sequences have been reported to target proteins to the nucleus (reviewed in Ref. 33). Altered nuclear staining, however, was the only apparent difference in localization patterns of the polybasic region mutants.
As mentioned previously, various constructs with increasing numbers of polybasic point mutations were generated for both NTG 14 ␣and NTG 16 ␣-EGFP-FLAG. These constructs and their subcellular distributions and palmitoylation state are listed in Table 1. The specific mutations are denoted under "Experimental Procedures." Of note, plasma membrane localization patterns did not change for any of the polybasic mutants of NTG 14 ␣or NTG 16 ␣-EGFP-FLAG; each of the polybasic region mutants showed nearly exclusive localization at the plasma membrane (Table 1).
Based on the hydrophobic properties the mutant proteins displayed in membrane extraction experiments (Fig. 6B), we then tested the polybasic region mutants for incorporation of palmitate. Cells transiently expressing NTG 14 ␣-9Q or NTG 16 ␣-5Q were metabolically labeled with [ 3 H]palmitic acid, and the fusion constructs were isolated from prepared membranes by immunoprecipitation. Both NTG 14 ␣-9Q and NTG 16 ␣-5Q incorporated palmitate like the wild-type G 14 ␣ or G 16 ␣ N-terminal domains (Fig. 7). We did note that consistently less NTG 14 ␣-9Q was immunoprecipitated from membranes compared with the amount of wild-type NTG 14 ␣-EGFP-FLAG immunoprecipitated under the same conditions, and this corresponded to a less robust autoradiography band when cells were labeled with [ 3 H]palmitic acid. We believe the decreased recovery of NTG 14 ␣-EGFP-FLAG was because of altered interaction with the anti-FLAG affinity gel used for immunoprecipitation rather than to decreased expression or altered lipid modification of NTG 14 ␣-9Q. This is supported by normal expression levels and membrane extraction properties of NTG 14 ␣-9Q demonstrated in imaging and extraction experiments ( Fig. 6B and Fig. 7).
Full-length G 14 ␣ and G 16 ␣ Are Palmitoylated at N-terminal Cysteine Residues-To compare our findings with the NTG␣-EGFP-FLAG fusion proteins, we examined [ 3 H]palmitate labeling of full-length G 14 ␣ and G 16 ␣ in intact cells (Fig. 8). G q family members are unusual among G␣ proteins in that they are intrinsically unstable and exhibit a strong propensity to misfold and aggregate when expressed alone as recombinant protein.
Full-length G q ␣ and G 11 ␣ aggregate and are inactive when Cell lysates were separated into membrane and cytosol fractions by ultracentrifugation at 100,000 ϫ g, and FLAGtagged proteins were extracted from membranes with 2% D␤M (3 h, 4°C) and immunoprecipitated from membrane extracts overnight at 4°C with anti-FLAG affinity gel. Immunoprecipitates were resolved by SDS-PAGE (15% polyacrylamide). Top boxes, Western blot analysis of membrane-immunoprecipitated proteins with an anti-GFP antibody. Bottom boxes, autoradiography of labeled proteins. Proteins were fixed as described under "Experimental Procedures" and were exposed to film for 5 days at Ϫ80°C. Results represent at least two independent experiments with each DNA construct. FIGURE 8. Full-length G 14 ␣ and G 16 ␣ are palmitoylated at N-terminal cysteine residues. In vivo palmitoylation of full-length G␣ proteins in HEK 293 cells and the effects of multisite C/S mutations. Cells were co-transfected with the G␤ 1 and G␥ 2 constructs and the indicated wild-type (WT) or C/S mutants of G 14 ␣ and G 16 ␣. Twenty four hours after transfection, cells were labeled with [ 3 H]palmitic acid for 2 h. Cell lysates were separated into membrane and cytosol fractions by ultracentrifugation at 100,000 ϫ g, and EE-tagged G␣ proteins were extracted from membranes with 2% D␤M (3 h, 4°C) and immunoprecipitated from each fraction with an anti-EE antibody overnight at 4°C. Immunoprecipitates were resolved by SDS-PAGE (11% polyacrylamide). Top boxes, Western blot analysis of cytosol (C)-and membrane (M)-immunoprecipitated proteins with Z811, an anti-G q/11/14/16 ␣ antibody. Bottom boxes, autoradiography of labeled proteins. Proteins were fixed as described under "Experimental Procedures" and were exposed to film for an average of 35 days at Ϫ80°C. Results are representative of at least two independent experiments with each DNA construct. expressed in either bacteria or eukaryotic cells and only fold properly as active protein when co-expressed with G␤␥ or when expressed as a chimera with G i ␣ (34,35). For these reasons, we co-expressed G␤ 1 and G␥ 2 with EE-tagged versions of G 14 ␣ and G 16 ␣ in HEK 293 cells for biochemical analyses of the full-length proteins. We transfected the G␣␤␥ subunit cDNAs in a ratio of 7:2:1 as has been done similarly for other functional analyses of G q ␣ in cell expression systems (19,21,36). The EE-tagged G␣ constructs were expressed in HEK 293 cells and immunoprecipitated with an anti-EE antibody from cytosol and membrane fractions 24 h later. Both G 14 ␣ and G 16 ␣ from membranes incorporated palmitate, whereas multisite C/S mutations eliminated palmitoylation in both G␣ (Fig. 8).
Similar to its N-terminal fusion protein counterpart, nonpalmitoylated G 16 ␣-C9S,C10S showed enhanced cytosolic accumulation and reduced membrane binding compared with wildtype G 16 ␣. The distribution of G 14 ␣-C4S,C5S,C6S between membranes and cytosol, however, appeared identical to that of wild-type G 14 ␣. G 14 ␣ in particular has been recalcitrant to metabolic labeling with [ 3 H]palmitate in these studies for unknown reasons, and so the amount of G 14 ␣-expressing cells was increased by 3-fold for these experiments, with a proportional increase in [ 3 H]palmitate label. Therefore, the large amount of cytosolic and membrane-associated G 14 ␣ likely reflects the 3-fold concentration of protein in each fraction. We hypothesize that the membrane association of G 14 ␣-C4S,C5S,C6S is nonspecific and includes nonfunctional, aggregated protein because this mutant does not incorporate palmitate and is relatively inactive in functional assays of second messenger signaling (see Fig. 10, A and B). Furthermore, confocal microscopic imaging of both G 14 ␣-C4S,C5S,C6S and G 16 ␣-C9S,C10S,C13S indicates that these proteins are not present at the plasma membrane but rather are concentrated in discrete intracellular locations, likely as aggregated protein or as targets for proteolytic degradation (data not shown). In support of this idea, one study reports that G q ␣ and G 16 ␣ are targeted for degradation through the proteasome, and a nonpalmitoylated C/S mutant of G q ␣, G q ␣-CCSS, is more rapidly degraded by proteasomal enzymes than is wild-type G q ␣ (37). Whether or not the subcellular locations where we observe G 14 ␣-C4S,C5S,C6S and G 16 ␣-C9S,C10S are proteasomes, however, remains to be tested.

Individual Cysteine Residues and the G 16 ␣ N-terminal Polybasic Sequence Differentially Affect G␣ Membrane Binding and
Palmitoylation-To determine the importance of individual cysteine residues for G␣ palmitoylation and membrane association, each of the single C/S mutations was introduced into full-length G 14 ␣ and G 16 ␣, and these mutants were co-expressed with G␤ 1 ␥ 2 in cells. Membrane and cytosol distribution patterns of the various C/S mutants were different depending on the particular G␣ and were surprisingly different from the results acquired with the NTG␣-EGFP proteins (Fig. 9). Individual C/S mutations in full-length G 16 ␣ had a greater apparent effect on the amount of G␣ in the membrane fraction than did single-site C/S mutations in G 14 ␣. Although a large amount of G 16 ␣ Ϯ C/S consistently was cytosolic, each C/S mutation, including C13S, resulted in a decrease in membrane-bound G 16 ␣ (Fig. 9B). By contrast, the membrane pool of G 14 ␣ was only slightly decreased by single-site mutations in the N terminus, with the C5S mutation having the greatest apparent effect (Fig. 9A). These observations are supported by densitometry analysis of relative G␣ protein expression levels in membrane and cytosol fractions (data not shown). Contrary to our findings with the N-terminal fusion constructs harboring the same mutations (see Fig. 5), palmitoylation of G 14 ␣ and G 16 ␣ was preserved in each of the individual C/S G␣ mutants (Fig. 9). Although the signal from [ 3 H]palmitate was visibly reduced for each protein, the signal strength for most of the C/S mutants appeared consistent with the reduction in protein associated with the membrane fraction as represented in the corresponding immunoblots. These findings are summarized in Table 2.
We also noted that G 14 ␣ Ϯ C/S and G 16 ␣ Ϯ C/S immunoprecipitated from cytosol fractions incorporated [ 3 H]palmitate. Ultracentrifugation should have separated all cellular membranes from the cytosol, so it is possible that this represents a small amount of G␣ Ϯ C/S that is palmitoylated but is not associated with intracellular vesicles or other lipid bilayers. We hypothesize that this cytosolic pool of G 14 ␣ and G 16 ␣ represents protein being trafficked through the cytosol by an unidentified mechanism, although it should be noted that the [ 3 H]palmitate signal is relatively weak compared with the much larger amount of protein recovered from the cytosol fraction as represented in the corresponding immunoblots, and this signal may in fact be an experimental artifact. Of note, the smeared signals in the cytosol lanes for G 14 ␣ and G 14 ␣-C4S represent extensive nonspecific immunoprecipitation of cytosolic proteins in this particular experiment, as this smear was evenly distributed throughout the length of each lane (Fig. 9A, 1st box, and other data not shown). Twenty four hours after transfection, cells were labeled with [ 3 H]palmitic acid for 2 h. Cell lysates were separated into membrane (M) and cytosol (C) fractions by ultracentrifugation at 100,000 ϫ g, and EE-tagged G␣ proteins were extracted from membranes and cytosol (not including cytosol for G 14 Ϯ C4S) with 2% D␤M (3 h, 4°C) and immunoprecipitated from each fraction with an anti-EE antibody overnight at 4°C. Immunoprecipitates were resolved by SDS-PAGE (11% polyacrylamide). Top boxes, Western blot analysis of cytosoland membrane-immunoprecipitated proteins with indicated antibodies. Bottom boxes, autoradiography of labeled proteins. Proteins were fixed as described under "Experimental Procedures" and were exposed to film at Ϫ80°C for the following times: G 14 ␣ Ϯ C4S, 36 days; G 14 ␣ Ϯ C5S, 65 days; G4␣ Ϯ C6S, 54 days; G 16 ␣ Ϯ C/S, 39 days. Because of the 3-fold increase in starting material of G 14 ␣ Ϯ C/S-expressing cells (see "Experimental Procedures"), the G 14 ␣-C/S mutants were analyzed in separate experiments, each with an internal positive control for palmitoylation (wild type G 14 ␣).
Additionally, we tested the biochemical properties of a polybasic region mutant of full-length G 16 ␣ by mutating the same five basic amino acids that are mutated in the NTG 16 ␣-5Q construct (G16␣-5Q, see Fig. 6A). Like each of the C/S mutants of G 16 ␣, the G 16 ␣-5Q construct exhibited diminished membrane binding (Fig. 9B), although by densitometry analysis, the relative amount of G 16 ␣-5Q protein in the membrane fraction was only slightly less than relative levels of membrane-bound wild type G 16 ␣ (data not shown). Importantly, G 16 ␣-5Q was still a target for palmitoylation (Fig. 9B).
Palmitoylated Polycysteine Sequences and N-terminal Positively Charged Amino Acids Are Important for G␣-mediated Activation of Inositol Lipid Signaling in Intact Cells-Both G 14 ␣ and G 16 ␣ directly activate PLC-␤ to initiate phosphatidylinositol (4,5)-bisphosphate hydrolysis and inositol lipid signaling (5,6). To see if the same sequences important for palmitoylation and plasma membrane association of G 14 ␣ and G 16 ␣ also are important for the signaling function of the full-length proteins, we introduced second-site C/S mutations into backgrounds of mutationally activated G 14 ␣ and G 16 ␣. These constructs contain a point mutation (Q/L) in the guanine nucleotide binding domain that abrogates the GTPase activity of the G␣ protein (38,39) and allows for constitutive effector stimulation (G 14 ␣-Q/L and G 16 ␣-Q/L). As a read-out of recombinant G 14 ␣-Q/Land G 16 ␣-Q/L-induced PLC-␤ activation, we measured accumulation of radiolabeled inositol phosphates in transfected HEK 293 cells following an overnight labeling period with myo-[ 3 H]inositol (Fig. 10). Although expression of G 14 ␣-Q/L stimulated inositol phosphate production more than 7-fold over unstimulated levels, mutation of two or three cysteines inhibited inositol phosphate production equal to or slightly above basal levels, respectively (Fig. 10, A and B; Table 2). Of note, we observed that G 14 ␣-Q/L-C4S,C5S,C6S was consistently expressed at much lower levels than the other G 14 ␣-Q/L proteins, and transfection of increasing amounts of this cDNA indicated an upper limit to expressed protein levels that was markedly reduced relative to the other G␣ proteins. We speculate that this likely is because of increased degradation of the triple C/S mutant of G 14 ␣-Q/L, because similar findings have been reported for the double C/S mutant of G q ␣ (37). We believe this accounts for the apparent increase in activity of G 14 ␣-Q/L-C4S,C5S,C6S above basal when adjusted for protein expression levels and does not reflect a real difference in activity between the double and triple C/S mutants of G 14 ␣-Q/L (Fig. 10, A and B).
Like the double mutant of G 14 ␣-Q/L, G 16 ␣-Q/L-C9S,C10S also exhibited minimal stimulation of PLC-␤ activity, whereas G 16 ␣-Q/L increased accumulation of inositol phosphates by 5-fold (Fig. 10, C and D; Table 2). These data are supported by confocal microscopic immunofluorescence images that show localization of wild-type G 14 ␣-Q/L and G 16 ␣-Q/L at the plasma membrane and in discrete intracellular locations but G 14 ␣-Q/ L-C4S,C5S,C6S and G 16 ␣-Q/L-C9S,C10S,C13S in intracellular spaces only and not at the plasma membrane (data not shown).
Consistent with the changes individual C/S mutations cause in G␣ membrane association properties (see Fig. 9), single C/S mutations in G 16 ␣-Q/L had a greater impact on PLC-␤ activation than did single C/S mutations in G 14 ␣-Q/L. Interestingly, whereas the C13S mutation was without effect on membrane distribution and palmitoylation of NT G 16 ␣-EGFP-FLAG, the same mutation reduced function of G 16 ␣-Q/L by ϳ50% (Fig.  10, C and D; Table 2). Mutation of the polybasic sequence in G 16 ␣-Q/L resulted in a retained capacity of G 16 ␣-Q/L-5Q to induce a slight measurable increase in inositol phosphate accumulation over basal levels, although despite similar G␣ expression levels, this protein was less active than all the G 16 ␣-Q/L single-site C/S mutants (Fig. 10, C and D; Table 2).

DISCUSSION
Proper positioning of G␣ at the plasma membrane is the critical first level of regulation for G protein-mediated signal transduction. Despite this importance, molecular mechanisms underlying plasma membrane targeting of most G␣ are only partially understood, and for G 14 ␣ and G 16 ␣, nothing at all is known about how these G␣ are functionally regulated. We had two primary goals with this study. The first was to determine whether the isolated N termini of G 14 ␣ and G 16 ␣ alone possess the capacity to drive plasma membrane localization of soluble protein, and to identify possible lipid signals within this region that mediate this process. The second goal was to evaluate roles for the adjacent N-terminal polybasic regions in determining G␣ membrane localization, lipid modification, and functional properties. Here we show that the polycysteine and polybasic motifs in the N termini of G 14 ␣ and G 16 ␣ serve essential yet differential roles in G␣ palmitoylation, membrane localization, and signaling function.
The Isolated N termini of G 14 ␣ and G 16 ␣ Utilize Distinct Polycysteine Sequences but Not the Polybasic Region for Plasma TABLE 2 Membrane association, palmitoylation, and signaling properties of G 14 ␣ and G 16 ␣ site mutants HEK 293 cells were co-transfected with G␤ 1 and G␥ 2 and the indicated G␣ DNAs, followed by biochemical analysis at 24 h post-transfection (see "Experimental Procedures"). Evaluations of membrane association and palmitoylation are based on subcellular fractionation results and autoradiography results, respectively, as presented in Figs. 8 and 9. For membrane association, protein levels are indicated by ϩ and are based on intensity of immunoreactivity in membrane fractions. Additional or fewer ϩ indicate more or less immunoreactivity, respectively, relative to each wild-type (WT) G␣ protein.

Membrane association
ϩ Yes ϩ a G␣ Ϯ C/S constructs contain a constitutively activating point mutation (Q/L) in the GTPase domain. b Intracellular protein verified by confocal immunofluorescence imaging; data not shown.

Membrane Localization and Palmitoylation in Cells-With
complementary biochemical and confocal microscopic imaging experiments, we have determined that the isolated N termini of G 14 ␣ and G 16 ␣ are palmitoylated at polycysteine sequences, and palmitoylation of these sequences alone provides membrane targeting and binding properties sufficient to direct a cytosolic protein, in this case EGFP, to the plasma membrane. Surprisingly, the sequences required for palmitate incorporation are not conserved across the two G␣ peptides; NTG 14 ␣-EGFP-FLAG requires three contiguous cysteines (Cys-4, Cys-5,and Cys-6), whereas NTG 16 ␣-EGFP-FLAG requires only Cys-9 and Cys-10,but not the downstream Cys-13. Palmitoylation of the G 14 ␣ and G 16 ␣ N termini is the necessary and sufficient signal for plasma membrane localization of the EGFP fusion proteins because interruption of the sequences required for palmitoylation completely eliminates the targeting function of each G␣ N terminus. By contrast, disruption of the polybasic region, in part or entirely, had no effect on plasma membrane targeting or binding of either NTG 14 ␣or NTG 16 ␣-EGFP-FLAG. Furthermore, the polybasic region is not a signal for palmitoylation of the EGFP fusion proteins, indicating that this region per se is not required for productive interaction with an acyltransferase. Unexpectedly, we found that the isolated N termini of G 14 ␣ and G 16 ␣ and their full-length G␣ counterparts exhibit differential dependence on the polycysteine sequences and the polybasic region for localization and palmitoylation, suggesting that there exist distinct molecular requirements for palmitoylation and membrane localization of a targeted peptide sequence versus a large, multimeric G protein heterotrimer complex. Specifically, individual C/S mutations had a dominant negative effect on palmitoylation and plasma membrane localization of the EGFP fusion proteins, whereas loss of the polybasic region was without effect. By contrast, the same C/S mutations only partially altered the properties of full-length G␣ in biochemical and functional assays, whereas the polybasic region of G 16 ␣ was required for normal levels of membrane binding and G 16 ␣-mediated PLC-␤ activation. The primary molecular property underlying the biochemical behavior of full-length G␣, which likely is absent for the EGFP fusion proteins, is the interaction in cells with the G␤␥ dimer. N-terminal point mutations in G q ␣ and Gs␣ that prevent binding to G␤␥ also prevent G␣ palmitoylation and plasma membrane targeting of the G protein (10,19), suggesting that heterotrimer formation is a prerequisite to G␣ palmitoylation and membrane trafficking. The results presented here suggest that co-expressed G␤␥, which also is lipidmodified, can promote palmitoylation and can drive plasma membrane localization of G␣-C/S to an extent. These findings are consistent with reports showing that the other G q family members, G q ␣ and G 11 ␣, are targeted for palmitoylation and exhibit normal signaling capacity providing at least one of the two conserved N-terminal cysteine residues (Cys-9 or Cys-10) is intact (7,8,16).
Palmitoylation of Distinct N-terminal Polycysteine Sequences Is Critical for G 14 ␣ and G 16 ␣ Membrane Localization and Functional Activity-Unequivocally, the different polycysteine sequences in the N termini of G 14 ␣ and G 16 ␣ are required for G␣ palmitoylation and plasma membrane localization necessary for efficient induction of PLC-␤ signaling pathways in cells. Alteration of these sequences at any one of the three cysteines in each results in reduced interaction with cellular mem-  A and B) and G 16 ␣ (C and D). Cells were co-transfected with the indicated G␣ cDNA and G␤ 1 and G␥ 2 constructs in a 6:3:1 ratio as described under "Experimental Procedures." After a 5-h transfection period, cells were metabolically labeled overnight with 4 Ci/ml myo-[ 3 H]inositol. Following a 50-min incubation at 37°C in 10 mM LiCl 2 , cells were solubilized with 20 mM formic acid, and lysates were neutralized with 0.7 M NH 4 OH. [ 3 H]Inositol phosphate fractions were separated by anion exchange chromatography, and total [ 3 H]inositol phosphate content was assessed by liquid scintillation spectrometry. The data are expressed as a ratio of total inositol phosphates (InsP) to adjusted G␣ protein expression level. Data in plots A and C are presented as total inositol phosphates from a single representative experiment (mean cpm Ϯ S.E.; each point performed in triplicate). Paired representative immunoblots of transfected cell lysates indicate relative expression levels of each protein (A and C). Data in plots B and D represent pooled data from three independent experiments and are presented as fold increase in total inositol phosphates over basal levels (mean fold increase Ϯ S.E.; n ϭ 3 for B and D). All data have been normalized to adjusted protein expression level measured by densitometry analysis of corresponding Western blots. Protein bands by Western blot were quantitated, and wild type protein was adjusted to a value of 1.0. All other protein density levels were determined relative to wild type. branes and impaired functional responses, and loss of at least two of the three cysteine residues in either G 14 ␣ or G 16 ␣ eliminates signaling function entirely. In general, the functional activity of G 16 ␣-Q/L is more sensitive to a single C/S point mutation than is the activity of G 14 ␣-Q/L. A small subset of each C/S single mutant of G 16 ␣ binds membranes and is palmitoylated, although in functional assays, the two preserved cysteine residues are not sufficient to maintain normal levels of signaling by G 16 ␣-Q/L; for G 16 ␣ in particular, all three N-terminal cysteine residues are necessary for full signaling function. A single C/S mutation in G 14 ␣, on the other hand, has a lesser impact on membrane binding and effector activation, although at least two of the three cysteines are required for any detectable functional activity. Consistently, the palmitoylation state of G 14 ␣ Ϯ C/S or G 16 ␣ Ϯ C/S and the capacity for constitutive PLC-␤ activation closely correlate with the apparent distribution of each G protein in the membrane fraction. Surprisingly, although the third cysteine in the G 16 ␣ polycysteine sequence, Cys-13, is unimportant for palmitoylation and plasma membrane targeting of the isolated G 16 ␣ N terminus, mutating this residue in full-length G 16 ␣ reduces membrane binding and signaling capacity to approximately the same extent as the C10S mutation in G 16 ␣, suggesting that whereas Cys-13 is not palmitoylated in the context of the G 16 ␣ N-terminal peptide, this residue is of functional importance for the G protein as a whole. Whether or not there are differences in the molecular requirements for acyltransferase activity involving the variant polycysteine sequences in G 14 ␣ and G 16 ␣, which could account for the observed effects of different C/S mutations, should be examined in future studies.
The correlative decrease in membrane binding of G 14 ␣ or G 16 ␣ with a single C/S mutation and functional activity of G 14 ␣-Q/L and G 16 ␣-Q/L with a single C/S mutation could reflect diminished palmitate incorporation. C/S mutations inherently reduce the available putative palmitate acceptor sites on G␣ and therefore may alter the stoichiometry of palmitate-bound G␣ in the membrane. Fewer palmitate molecules could reduce membrane affinity or increase the rate of depalmitoylation of the G␣ protein. Whether or not palmitate binds each cysteine residue in question is unknown. The metabolic labeling techniques used in this study indicate the palmitoylation state of isolated proteins but cannot provide quantitative information regarding stoichiometry or identify specific sites of lipid incorporation in the protein target. Although the single C/S mutants of both G 14 ␣ and G 16 ␣ display much weaker signals of [ 3 H]palmitate incorporation relative to wild-type G␣, we can conclude that these proteins are substrates for palmitoylation, but not that fewer lipid molecules are bound. The functional effects of each single C/S mutation in G 14 ␣ and G 16 ␣ suggest that all three residues in each G␣ may be separate binding sites for palmitate. Our findings with the isolated N terminus of G 14 ␣ support this idea, although experiments with NTG 16 ␣-EGFP-FLAG suggest that Cys-13 in particular is not important for lipid modification of the upstream dual cysteine sequence in G 16 ␣ and may not be a site of palmitate incorporation, at least in the context of the isolated N terminus. If in fact all three cysteine residues in G 14 ␣ are palmitoylated as our findings suggest, then G 14 ␣ has a unique profile of lipid incorpora-tion among all heterotrimeric G proteins, as no other G␣ subunits have been shown to have three palmitate-binding sites.
The Polybasic Region Is Important for G 16 ␣ Signaling Activity at the Plasma Membrane-Our studies with full-length G␣ provide the first evidence that the polybasic region in the N terminus of G 16 ␣ is an important determinant of G␣ signaling function but not palmitoylation. These findings support a proposed multisignal model for plasma membrane targeting of palmitoylated G␣ proteins like those of the G q family in which surface electrostatic charges and lipid modifications together directly bind the protein to membrane lipids (22). Such a mechanism is utilized by proteins like RGS4 and RGS16, which require both palmitoylation and positively charged protein surfaces in their N-terminal ␣-helix for functional interaction with the plasma membrane (40 -42). Other signaling molecules, including the nonreceptor tyrosine kinase c-Src, the small G protein K-Ras4B, and neuromodulin/GAP-43, also rely on hydrophobic lipid molecules and polybasic motifs for binding to the plasma membrane and cannot actively participate in signaling reactions without the cooperativity of these dual membrane targeting mechanisms (43)(44)(45)(46)(47)(48). Similarly, full-length G 16 ␣ lacking most of this positive charge in the N terminus is markedly reduced in the membrane fraction and exhibits minimal functional activity. The polybasic region in G 14 ␣ includes several additional positively charged amino acids (nine total) compared with the polybasic region of G 16 ␣ (six total). Although we were able to compare the effects of substitution mutations in the polybasic region on localization and palmitoylation of the isolated N termini of G 14 ␣ and G 16 ␣, functional experiments with full-length G 14 ␣ were not included due to the limiting technical barriers of introducing all nine point mutations into the full-length G 14 ␣ sequence. As G q ␣ and G 11 ␣ each contain 10 positively charged amino acids in their N termini (Fig. 1A), it will be important to identify the functional importance of these larger regions of basic charge for the other members of the G q family.
Implications for G Protein Membrane Localization and Signaling Function-Previous work indicates that G␣ membrane localization and signaling function is dictated by lipid-modifying groups on the G␣ N terminus and by interaction with the G␤␥ dimer. Here we provide evidence that the polybasic region on G 16 ␣ also may contribute to membrane binding and signaling activity of palmitoylated G␣. Precisely how these residues are utilized by the protein, however, is unclear. Palmitoylation of the isolated N termini of G 14 ␣ and G 16 ␣ and full-length G 16 ␣ do not require these residues, but full-length G 16 ␣ has altered subcellular localization and diminished function without them. Therefore, it is likely that this sequence of amino acids is not an intrinsic requirement for palmitoylation of targeted G␣ cysteines, but instead it may indirectly affect palmitoylation by regulating localization of the G protein at the plasma membrane.
The polybasic region potentially may be important for maintaining G 16 ␣ localization at the membrane during palmitate turnover, because palmitoylation of G proteins has been shown to increase during active signaling cycles (49 -51). A growing body of evidence demonstrates that G q ␣, G 11 ␣, and other palmitoylated G␣ like G s ␣ dynamically associate with the plasma membrane during signaling cycles. Agonist activation induces redistribution of G 11 ␣ laterally within the membrane, followed by internalization and, depending on treatment time, either membrane recycling or degradation of G␣ (52,53). Real time imaging of GFP-tagged G q ␣ shows only a minor translocation to cytosol following receptor stimulation or direct G protein activation with AlF 4 Ϫ (13), and other studies show that direct activation or enzymatic depalmitoylation of G s ␣ and G q ␣ does not release these G␣ from membranes (54 -57). Together these reports suggest that conditions sufficient for palmitate turnover on targeted G␣, in some cases, can induce spatially limited changes in membrane localization of G␣ without causing an extensive loss of G protein from the plasma membrane. It is possible that electrostatic interactions between membrane lipids and the positively charged surface of the G␣ N terminus may be one of the molecular properties preserving G␣ localization at the plasma membrane in these studies.
It is also likely that persistent interaction with G␤␥ during G␣-mediated effector activation contributes to membrane localization of the G␣ that is repeatedly undergoing depalmitoylation and palmitate reloading. A recent study shows that site mutants of G q ␣ that cannot bind G␤␥ also cannot activate PLC-␤ and inositol lipid signaling, even when targeted to the plasma membrane by a myristoylated peptide sequence (36), and adds to a growing list of reports demonstrating that some heterotrimeric G protein subunits may not fully dissociate in vivo during the GTPase cycle as previously thought (58 -60). Although neither the N-terminal cysteines nor the polybasic region amino acids are among the contact sites in the G␣ N terminus for G␤␥ binding based on available heterotrimer crystal structures (31,61), it is important to consider that the G 14 ␣ and G 16 ␣ point mutations utilized in this study could introduce structural changes in the G␣ N terminus that alter the affinity of the G␣-G␤␥ interaction, and this deficiency could contribute to the observed reduction in G␣ function and membrane localization. Reduced interaction with G␤␥ in fact may explain the localization and functional differences of the single-site C/S mutants of G 14 ␣ Ϯ Q/L and G 16 ␣ Ϯ Q/L, because G␤␥ binding is required for G␣ palmitoylation (10,19); a less stable heterotrimer could result in reduced G␣ palmitoylation, and both mechanisms could account for reduced G␣ functional activity at the plasma membrane. However, our data suggest that the functional differences measured for the polybasic region mutant of G 16 ␣ are independent of G␤␥ based on the capacity for G 16 ␣-5Q to incorporate palmitate like wild-type G 16 ␣. Of note, mutation of Arg-21 in G o ␣ was shown to inhibit heterotrimer formation by an unknown mechanism (62), but the corresponding amino acid in G q ␣, Arg-27, was dispensable for G q ␣ palmitoylation and therefore does not seem to be involved in G␤␥ binding (10). Interestingly, G q ␣-R27A exhibited a measurable decrease in receptor-stimulated PLC-␤ activation (10). This study, together with the results presented here, supports the idea that N-terminal basic residues in G q family members have a discrete role in mediating G␣ functional activity at the plasma membrane.
Conclusions-In this study we have established that distinct polycysteine sequences in the N-terminal domains are critical for palmitoylation of G 14 ␣ and G 16 ␣ and are sufficient for plasma membrane binding when expressed alone. The fulllength G proteins utilize a combination of biochemical mechanisms, including, but not restricted to, palmitoylation of the G␣ N terminus for membrane localization and signaling function. N-terminal basic residues, which include an hypothesized positively charged basic patch on the G␣ surface, appear to be important for fully functional G 16 ␣-mediated inositol lipid signaling at the cell membrane but not for palmitoylation of the adjacent polycysteine sequences. Mutation of these basic amino acids in G 16 ␣ or of N-terminal cysteine residues in G 14 ␣ and G 16 ␣ reduces the cellular pool of active and palmitoylated membrane-bound G␣ to varying degrees. Although the precise roles of G␣ palmitoylation, heterotrimer formation, and the charged G␣ polybasic domain during the G protein signaling cycle remain to be resolved, this study presents the first evidence that G 14 ␣ and G 16 ␣ function may be regulated by all three of these mechanisms.