J Biol Chem, Vol. 275, Issue 2, 1327-1336, January 14, 2000

Interaction with G Is Required for Membrane Targeting and Palmitoylation of G_s and G_q^*

Daniel S. Evanko, Manimekalai M. Thiyagarajan, and Philip B. Wedegaertner

From the Department of Microbiology and Immunology and Kimmel Cancer Institute, Thomas Jefferson University, Philadelphia, Pennsylvania 19107

	ABSTRACT

TOP ABSTRACT INTRODUCTION EXPERIMENTAL PROCEDURES RESULTS DISCUSSION REFERENCES

Peripheral membrane proteins utilize a variety of mechanisms to attach tightly, and often reversibly, to cellular membranes. The covalent lipid modifications, myristoylation and palmitoylation, are critical for plasma membrane localization of heterotrimeric G protein  subunits. For _s and _q, two subunits that are palmitoylated but not myristoylated, we examined the importance of interacting with the G protein dimer for their proper plasma membrane localization and palmitoylation. Conserved  subunit N-terminal amino acids predicted to mediate binding to were mutated to create a series of binding region mutants expressed in HEK293 cells. These _s and _q mutants were found in soluble rather than particulate fractions, and they no longer localized to plasma membranes as demonstrated by immunofluorescence microscopy. The mutations also inhibited incorporation of radiolabeled palmitate into the proteins and abrogated their signaling ability. Additional _q mutants, which contain these mutations but are modified by both myristate and palmitate, retained their localization to plasma membranes and ability to undergo palmitoylation. These findings identify binding to as a critical membrane attachment signal for _s and _q and as a prerequisite for their palmitoylation, while myristoylation can restore membrane localization and palmitoylation of binding-deficient _q subunits.

	INTRODUCTION

TOP ABSTRACT INTRODUCTION EXPERIMENTAL PROCEDURES RESULTS DISCUSSION REFERENCES

The monomeric  and dimeric subunits of the heterotrimeric () G proteins¹ reversibly bind to one another at the cytoplasmic face of the plasma membrane where they interact with heptahelical receptors (1, 2). Upon receptor activation, the heterotrimer dissociates into a GTP-bound  subunit and a subunit, both of which can regulate effector proteins. The correct targeting of G proteins to the plasma membrane, and likely micro-domains of the plasma membrane, is required for the proper functioning of this signaling system (3). For example,  subunits defective in membrane binding are unable to couple receptor stimulation to effector activation (4-6). Although the importance of membrane localization for proper signaling has been established, the basic signals that direct the different  subunits to the plasma membrane and retain them there are not well defined.

A recent model proposes that G protein  subunits, and other peripheral membrane proteins, utilize two signals to bind tightly and stably to plasma membranes (7-10). These two signals are thought to act synergistically to promote a stable interaction between the membrane and G protein  subunit. According to the model, at least one of these signals must be involved in specifically targeting the  subunit to the plasma membrane rather than other intracellular membranes. The signals involved in targeting and binding to the membrane could be covalent lipid modifications, polybasic stretches of amino acids, or protein-protein interactions with other membrane-binding proteins (7-10).

Lipid modifications clearly function as membrane binding signals for G protein  subunits (11, 12). _i, _o, and _z contain two lipid modifications, myristoylation and palmitoylation, at their N terminus that are involved in membrane binding. Myristate, a 14-carbon fatty acid, is attached co-translationally via a stable amide bond to the N-terminal glycine residue after removal of the initiating methionine, while palmitate, a 16-carbon fatty acid, is attached post-translationally via a reversible thioester bond to a cysteine residue immediately C-terminal to the myristoylated glycine. Mutation of the N-terminal glycine abolishes myristoylation and inhibits palmitoylation, leading to a shift in localization from membrane to cytosol (13-17). Mutation of the palmitoylated cysteine results in a myristoylated but non-palmitoylated subunit that partially shifts to the cytosol and binds to intracellular membranes as well as the plasma membrane (4, 13, 18-21). These results support a two-signal model of membrane binding in which myristoylation and palmitoylation cooperate in targeting  subunits of the _i family to the plasma membrane. Furthermore, myristoylation is required for subsequent palmitoylation.

Many G protein  subunits (e.g. _s, _q, ₁₂, ₁₃), however, are not myristoylated, but are palmitoylated at N-terminal cysteines. The location of the cysteines and the surrounding sequences show little or no similarity among the different non-myristoylated  subunits. Although mutation of the palmitoylated cysteines inhibits membrane binding (5, 6), additional signals that contribute to membrane attachment and allow palmitoylation have not been identified for the non-myristoylated  subunits. (4, 12, 22), through its protein-protein interaction with , or unknown hydrophobic modifications of  (23) have been proposed, although not yet tested, to function as membrane binding signals for the non-myristoylated  subunits. The importance of subunits in helping to anchor  subunits to cellular (plasma) membranes is controversial. An early study showed that purified was required to attach purified _i and _o to phospholipid vesicles in vitro (24), and co-expression of enhances membrane association of G2A myristoylation site mutants of _i or _z (4, 21). On the other hand, short N-terminal sequences of 10 amino acids or less from Gpa1 (a G protein  subunit from Saccharomyces cerevisiae) (25), _i2, ² or Src-like tyrosine kinases (26, 27) are sufficient to target heterologous proteins to plasma membranes in the absence of interactions with . Another recent report suggested that dissociation of _s from combined with depalmitoylation of _s was not sufficient to release _s from isolated membranes, implying that neither palmitoylation nor interactions were important for stable membrane association (28).

The goal of the studies presented here was to test the importance of binding for two different non-myristoylated G protein  subunits. A series of mutations designed to disrupt binding were introduced into _s and _q, and the localization of these subunits was determined by cell fractionation and immunofluorescence microscopy after transient transfection. In addition, palmitoylation of mutant _s and _q subunits was assayed by metabolic labeling. To test the possibility that myristoylation could substitute for binding, an additional series of _q mutants containing an N-terminal site for myristoylation were created and assayed. The results of these experiments provide the first evidence that interactions with are crucial for the proper membrane localization and palmitoylation of _s and _q.

	EXPERIMENTAL PROCEDURES

TOP ABSTRACT INTRODUCTION EXPERIMENTAL PROCEDURES RESULTS DISCUSSION REFERENCES

Materials-- HEK293 cells were obtained from the American Type Culture Collection (CRL-1573). [2-³H]Inositol and [2-³H]adenine were from Amersham Pharmacia Biotech. [9,10-³H]Palmitic acid and [9,10-³H]myristic acid were from NEN Life Science Products. Dowex resin was from Bio-Rad. Tissue culture reagents were from Life Technologies, Inc. Other reagents were from Fisher Scientific and Sigma.

Plasmid Construction-- The HA epitope (DVPDYA)-tagged HA- _s-pcDNAI and HA-_q-pcDNAI were constructed and analyzed previously and described in detail (5, 29). The coding region was transferred from these pcDNAI plasmids to pcDNA3 plasmids by restriction and ligation and the subsequent constructs are hereafter referred to as _s or _q. The Stratagene QuikChange site-directed mutagenesis kit was used to replace residues thought to contact with alanines. The primers used for mutagenesis varied from 42 to 53 bases in length and the sense and antisense primers had 3' overhangs of 3 to 7 bases. _qAG was created in the same manner using sense (5'-ctcggatccatcgatctggagtccatcatgggatgctgcctg-3') and antisense (5'-gcagcatcccatgatggactccagatcgatggatccgagctc-3') oligonucleotides, which replaced Ala with Gly after the second N-terminal methionine and prevented initiation from the first methionine by replacing the first putative methionine codon six residues upstream of the second with an isoleucine codon. To create all the _qAG binding mutants, the 889-base pair EcoNI and EcoRI fragment from each of the binding mutants of _q was ligated into _qAG plasmid cut with EcoNI and EcoRI. The resulting cDNAs were verified by DNA sequencing (Kimmel Cancer Center Nucleic Acids Facility) to contain no mutations other than those desired. The R201C and R183C mutants were created by excising Eco47III-XhoI and Eco47III-NotI fragments from HA-_s-R201C-pcDNAI and HA-_q-R183C-pcDNAI (5), respectively, and ligating them into appropriately digested _s- or _q-pcDNA3 plasmids.

Cell Culture and Transfection-- HEK293 cells were propagated in Dulbecco's minimal essential medium containing 10% fetal bovine serum and Gentamicin. Unless otherwise noted, cells were plated in six-well plates at 7.0 × 10⁵ cells/well and grown for 24 h prior to transfection. 1 µg of each expression plasmid was transfected into the cells using LipofectAMINE from Life Technologies, Inc. according to the manufacturer's instructions.

Cell Fractionation-- HEK293 cells were plated at 2 × 10⁶ cells in 6-cm plates, grown for 24 h, and transfected with 3 µg of expression plasmid. 24 h after transfection, the cells were transferred to 10-cm plates and grown for 48 h. Cells were washed once with phosphate-buffered saline and then dislodged from the plate by washing and pelleted at low speed. The cell pellet was suspended in 0.5 ml of lysis buffer (50 mM Tris-HCl, pH 8, 2.5 mM MgCl₂, 1 mM EDTA, 1 mM dithiothreitol, 0.5 mM phenylmethylsulfonyl fluoride, 5 µg/ml leupeptin, 5 µg/ml aprotinin), and cells were lysed by 10 passages through a 27-gauge needle. Lysed cells were centrifuged at 400 × g for 5 min to remove nuclei and intact cells. The supernatant was centrifuged at 150,000 × g for 20 min at 4 °C in a TL-100 tabletop ultracentrifuge (Beckman Instruments). The supernatant (soluble fraction) was removed and the pellet (particulate fraction) was suspended in an equal volume of lysis buffer. Fractions were frozen at 20 °C and later analyzed by Western blotting. Samples were fractionated by 10% SDS-PAGE, transferred to PVDF-Plus (Micron Separations Inc.) using a Trans-Blot SD semi-dry electrophoretic transfer cell (Bio-Rad), and probed with 12CA5 monoclonal antibody. Bands were visualized by chemiluminescence and quantitated using a Kodak DC40 imaging system.

Immunofluorescence Localization-- 24 h after transfection in six-well plates, HEK293 cells were replated on glass coverslips and grown for 48 h. Cells were fixed with 3.7% formaldehyde in phosphate-buffered saline (PBS) for 15 min and permeabilized by incubation in blocking buffer (2.5% nonfat milk and 1% Triton X-100 in PBS) for 30 min. Cells were then incubated with 12CA5 mouse monoclonal antibody (3 µg/ml) in blocking buffer for 1 h. The cells were washed with blocking buffer and incubated in a 1:100 dilution of donkey anti-mouse Texas Red conjugate (Jackson Immunoresearch Laboratories, West Grove, PA) for 30 min. The coverslips were washed with 1% Triton X-100 in PBS, rinsed in distilled water, and mounted on glass slides with 10 µl of Prolong Antifade reagent (Molecular Probes, Eugene, OR). Microscopy was performed with an Olympus BX60 microscope equipped with a 60×/NA1.4 objective and a Texas Red filter cube. Images were recorded with a Sony DKC-5000 digital camera and transferred to Adobe Photoshop for digital processing.

cAMP and Inositol Phosphate Assays-- 24 h after transfection in six-well plates, each well was reseeded into six wells of a 24-well plate with [2-³H]adenine or [2-³H]inositol (2 µCi/ml) and grown for 24 h. The cells were then assayed for their ability to induce production of cAMP or IP as described previously (5). The binding competition assay was performed in the same manner as the cAMP assay except that cells were stimulated with 10 µM quinpirole rather than UK-14304, as described (30).

Metabolic Labeling and Immunoprecipitation-- Radiolabeling of _s and _q with fatty acids was performed essentially as described (31). 48 h after transfection, HEK293 cells in a 6-cm dish were incubated with 1 ml of Dulbecco's modified Eagle's medium containing 10% fetal bovine serum, 5 mM sodium pyruvate, and [9,10-³H]palmitic acid (1 mCi/ml) or [9,10-³H]myristate (0.25 mCi/ml) for 2 h. Cells were washed once with PBS and lysed in 1 ml of radioimmune precipitation buffer (50 mM HEPES, pH 7.4, 150 mM NaCl, 0.5% sodium deoxycholate, 1% Nonidet P-40, 1 mM EDTA, 2.5 mM MgCl₂, 1 mM phenylmethylsulfonyl fluoride, 2 µg/ml leupeptin, 2 µg/ml aprotinin). Cell extracts were tumbled for 1 h at 4 °C, and nuclei and insoluble material were removed by microcentrifugation at 16,000 × g for 3 min. Samples containing HA-tagged _q or its mutants were adjusted to 0.1% SDS. 1 µg of 12CA5 antibody was added, and the samples were tumbled for 2 h at 4 °C. Next, 20 µl of Protein A/G Plus agarose (Santa Cruz Biotechnology, Santa Cruz, CA) was added, and the sample was tumbled overnight at 4 °C. The sample was centrifuged for 30 s at 200 × g to pellet the beads. The supernatant was discarded, and the beads were washed three times with 1 ml of radioimmune precipitation buffer. SDS-PAGE sample buffer containing 10 mM dithiothreitol was added to the washed beads, and the samples were heated at 65 °C for 1 min. An aliquot was analyzed by 10% SDS-PAGE. Gels were incubated for 20 min in an aqueous solution of 50% methanol and 10% acetic acid, followed by 10% ethanol and 10% acetic acid for 20 min, and, finally, Amplify (Amersham Pharmacia Biotech) for 20 min. Gels were dried and subjected to fluorography at 80 °C using Hyperfilm MP (Amersham Pharmacia Biotech).

Trypsin Protection Assay-- HEK293 cells were transiently transfected with expression constructs, and soluble and particulate fractions were isolated as already described. The particulate (membrane) fractions were centrifuged at 20,000 rpm for 35 min at 4 °C in a TL-100 tabletop ultracentrifuge (Beckman Instruments). The pellet was resuspended in 100 µl of solubilization buffer (20 mM Tris-HCl, pH 7.5, 100 mM NaCl, 2 mM MgCl₂, 0.1 mM EDTA, 1 mM dithiothreitol, 10% glycerol, 1% C₁₂E₁₀ (polyoxyethylene 10-lauryl ether), 0.1 mM phenylmethylsulfonyl fluoride) and incubated on ice for 30 min followed by centrifugation at 150,000 × g for 20 min at 4 °C. The supernatant was divided into separate tubes and incubated for 30 min at 30 °C with 10 µM GDP or AlF₄ mix (30 mM NaF, 150 µM AlCl₃, 10 µM GDP) (32). Samples were then treated with a 1:500 dilution of trypsin mix (100 µM GDP, 0.6 mg/ml trypsin, 1% C₁₂E₁₀) and further incubated for 30 min at 30 °C. Reactions were terminated by adding soybean trypsin inhibitor to a final concentration of 3 mg/ml. Trypsin-resistant fragments of _q were visualized by SDS-PAGE and Western blot analysis using 12CA5 antibody as described previously.

	RESULTS

TOP ABSTRACT INTRODUCTION EXPERIMENTAL PROCEDURES RESULTS DISCUSSION REFERENCES

Mutation of a Conserved Contact Region in _s and _q-- Examination of the extreme N terminus of several G protein  subunits reveals two conserved regions with a (I/L)(E/D)(K/R) triplet within the second region (Fig. 1). X-ray crystallography of the complex for G_t and G_i1 has shown that this second region contains residues that directly contact (33, 34). All of the conserved residues in this region, with the exception of (K/R), were shown to contact . In order to assess the importance of binding for proper localization and palmitoylation, we substituted alanine residues for these potentially important amino acids in _s and _q.³ Although (K/R) does not seem to contact , this highly conserved residue was also mutated in some constructs to examine its role in localization and palmitoylation of _s and _q. Interestingly, mutation of the lysine at this position in _o caused a reduction in binding, as measured by pertussis toxin-catalyzed ADP-ribosylation (35). Constructs with progressively greater mutations were made beginning within the conserved triplet and working outward as shown in Fig. 1.

View larger version (21K): [in this window] [in a new window]   Fig. 1.   Alignment of  subunit N termini showing mutations introduced into a putative binding region. The N termini of several  subunits have been aligned and conserved residues are shaded, showing two conserved regions. Amino acids modified by myristoylation or palmitoylation are in drop shadow font. Residues that were found to directly contact in crystal structures of t and i heterotrimers have been marked by asterisks. The descriptive suffixes for each of the s and q mutants are above the aligned sequences and the amino acids which were substituted with alanines in each mutant are shown by an "A" over each original residue shown in bold. The sequence encoded by the cDNA for the modified q subunit (qAG) is shown at the bottom with the starting methionine and new myristoylated glycine residue.

Although a second surface of the  subunit forms an even larger area of interaction with (33, 34), our studies concentrated on the N-terminal region for two important reasons. First, previous work by others showed that disruption of this region by deletion (36, 37), proteolysis (38), or antibody binding (39, 40) is sufficient to prevent  from interacting with . Second, mutations in the N-terminal region are unlikely to disrupt the overall three-dimensional structure of the  subunit. Crystal structures indicate that this region does not interact with other regions of , whereas the second contact surface is located within the switch I and II region where activation-dependent rearrangements occur (33, 34). In addition, _s lacking the N terminus is still capable of normal GTP binding and hydrolysis and normal effector activation in vitro (41), but the second contact region overlaps with an effector interacting surface (42-44). To facilitate immunodetection and immunoprecipitation of the  subunits, we utilized well characterized HA-tagged variants of _s and _q.

Mutation of the Contact Region Disrupts Membrane Attachment-- To determine whether this putative binding region of _s and _q is critical for membrane attachment, we transiently transfected the various  subunit constructs into HEK293 cells and determined their subcellular distributions. The transfected cells were lysed in hypotonic buffer, soluble (S) and particulate (P) fractions were separated by centrifugation and the distribution of  subunits was assessed by Western blot analysis. In these experiments the distribution of wild-type HA-tagged _s is reproducibly 60:40 P:S. As the putative binding region of _s is subjected to increasing numbers of mutations, the membrane-bound fraction progressively decreases until construct _sIEK+ partitions 20:80 P:S (Fig. 2). Inclusion of the arginine 201 to cysteine (RC) (45) activating mutation further increases the soluble fraction to give a 10:90 P:S ratio for _sRC IEK+. The same mutations were also assayed in the _q subunit which showed the same trend toward greater solubility. _q without mutations was almost entirely membrane associated (85:15 P:S) while _qIER+ was found almost exclusively in the soluble fraction (15:85 P:S) (Fig. 2). The subcellular distribution of these two subunits varied only slightly between experiments. In agreement with the _s data, the other _q mutants were always found in the soluble fractions to a greater degree than was wild type _q; however, they displayed greater variability between experiments (data not shown).

View larger version (25K): [in this window] [in a new window]   Fig. 2.   Localization of binding region mutants of s and q by cell fractionation. Each of the putative binding mutants of s and q was transiently expressed in HEK293 cells and the subunits were separated into soluble and particulate fractions as described under "Experimental Procedures." The quantitated protein in each fraction is represented as a percentage of total detected soluble plus particulate s or q protein for each subunit. The variability in distribution of s subunits between experiments was not significantly more than the variability within a single experiment. The results shown are the means ± S.D. for n = 3-9 experiments, except for sRC and sRC IEK+ which represent 1 and 2 experiments, respectively.

Immunofluorescence Microscopy of _s and _q Contact Region Mutants-- To examine further the effect of binding region mutations on _s and _q localization, immunofluorescence microscopy was utilized to examine the subcellular distribution of the HA epitope-tagged _s and _q subunits in transiently transfected HEK293 cells. Wild-type _s displayed pronounced plasma membrane staining, as evidenced by the sharp signal at the cell periphery, with some diffuse staining of the cytoplasm (Fig. 3A). Mutations in the binding region caused a decrease in plasma membrane staining and a concomitant increase in cytoplasmic staining (Fig. 3, B-F). Complete loss of plasma membrane staining was only observed for _sIEK+. All other mutant subunits displayed an intermediate level of plasma membrane and cytoplasmic staining in which the level of cytoplasmic staining increased as the number of mutations increased. The _q subunit was much more sensitive to loss of plasma membrane staining in visualized cells. Although wild-type _q and _qR localized to the plasma membrane (Fig. 3, G and H), all other subunits displayed no detectable plasma membrane staining (Fig. 3, I-L). The _q subunits deficient in membrane staining appeared to be distributed diffusely throughout the cytoplasm and within the nucleus. A similar pattern of cytoplasmic and nuclear staining was noted previously for fatty acylation-deficient mutants of _z (4).

View larger version (111K): [in this window] [in a new window]   Fig. 3.   Subcellular localization of binding region mutants of s and q by immunofluorescence. HEK293 cells were transiently transfected with HA epitope-tagged s constructs: s (A) and binding mutants sK (B), sIE (C), sIE+ (D), sIEK (E), and sIEK+ (F), and q constructs: q (G) and binding mutants qR (H), qIE (I), qIE+ (J), qIER (K), and qIER+ (L). The cells were fixed and visualized as described under "Experimental Procedures."

Mutation of the Contact Region Disrupts _s and _q Signaling-- If mutation of the putative binding region in _s and _q subunits inhibits binding, then functional heterotrimer capable of being activated by receptor will not be formed. Additionally, if  subunits cannot associate with the plasma membrane, they will not be activated by receptor. When _s or _q subunits are cotransfected in HEK293 cells with ₂-AR, both G proteins can be activated by ₂-AR agonists (5, 46). This system provides a useful assay to test the various subunits for their ability to couple receptor to downstream effector. Exposure of cells expressing wild-type _s and ₂-AR to the ₂-AR agonist UK-14304 caused a 7-fold increase in cAMP accumulation (Fig. 4A). Co-expression of ₂-AR and pcDNA3 displayed a small increase in cAMP accumulation upon agonist stimulation, but it was not considered significant because these levels are barely above background. The K28A substitution (_sK) had almost no effect on _s signaling in this assay, but other mutations showed pronounced inhibitory effects. _sIE was capable of less than a 3-fold induction in cAMP, and _sIE+ and _sIEK stimulated cAMP less than 2-fold over basal levels. However, complete disruption of signaling required that all six residues be substituted with alanines. To ensure that substitution of this many residues does not disrupt the structure of _s to such an extent that it is incapable of stimulating adenylyl cyclase, we incorporated the R201C activating mutation to create _sRC IEK+. This mutation in _s resulted in constitutively elevated receptor-independent cAMP synthesis (45) (Fig. 4A, inset). The ability of the putative binding-deficient subunit, _sRC IEK+, to stimulate adenylyl cyclase was reduced by about 50% compared with _sRC, despite the fact that it is not localized at the plasma membrane where adenylyl cyclase resides (Fig. 4A, inset). This demonstrates that the fully mutated _sIEK+ is still capable of functionally coupling to effector. A cytoplasmic, non-palmitoylated _sRC C3S mutant shows a similar reduction in its ability to constitutively increase cAMP in transfected cells (5).

View larger version (21K): [in this window] [in a new window]   Fig. 4.   Ability of G protein binding region mutants to stimulate effectors. HEK293 cells were transiently transfected with 500 ng of 2-AR-pCMV4 plus 500 ng of pcDNA3 or pcDNA3 containing cDNA encoding each of the  subunits. A, cells expressing s subunits were treated with 10 µM UK-14304 and 1 mM isobutylmethylxanthine (shaded bars) or isobutylmethylxanthine alone (open bars). The inset shows the results of constitutively active subunits of s containing the R201C mutation. The y axis scale of the inset corresponds to the scale of the larger graph. The level of cAMP was determined as described under "Experimental Procedures." B, cells expressing q subunits were treated with 10 µM UK-14304 (shaded bars) or carrier (open bars). The level of inositol phosphate was determined as described under "Experimental Procedures." Background counts were subtracted from the raw counts, and the level of stimulated wild-type subunit was arbitrarily set at 100. The results shown are the means ± S.D. for one experiment assayed in triplicate. Duplicate experiments exhibited similar results.

binding region mutants of _q also lost the ability to couple an activated receptor to effector stimulation. However, the defect was more severe with _q mutants compared with _s mutants. The only _q subunits capable of coupling receptor to effector were wild-type _q and _qR (Fig. 4B). All other putative binding mutants were completely unable to stimulate IP production following agonist activation. In contrast to _s, the addition of a constitutively activating R183C mutation (46) to all _q mutants, with the exception of _qR, failed to increase IP levels (data not shown). Previously, a lack of constitutive activity was also observed with a cytoplasmic, non-palmitoylated _qC9S,C10S mutant (5).

The Ability of  Subunits to Localize at the Plasma Membrane Determines Their Palmitoylation State-- To determine whether loss of plasma membrane localization and signaling correlates with decreased palmitoylation of the  subunits, we compared the ability of the different binding mutants of _s and _q to incorporate radiolabeled palmitate. Cells transiently transfected with the HA epitope-tagged  subunits were metabolically labeled with [³H]palmitic acid and immunoprecipitated with the 12CA5 monoclonal antibody. Because extremely low levels of palmitate were incorporated into wild type _s after expression in HEK293 cells in these studies, COS cells were used to assay _s palmitoylation (Fig. 5A). Palmitoylation of _q was performed using transfection of either COS (data not shown) or HEK293 cells (Fig. 5B), and the two cell lines gave similar results. Palmitoylation levels were visualized by fluorography of immunoprecipitated proteins after SDS-PAGE fractionation. The amount of radiolabeled palmitate incorporated in _s increased in the _sK and _sIE mutants, fell to wild-type levels in _sIE+ and _sIEK, and was greatly reduced in _sIEK+ (Fig. 5A). Palmitoylation of the binding-deficient mutants of the _q subunits correlated strongly with membrane localization (Figs. 2 and 3, G-L) and functional activity (Fig. 4B). _q and _qR incorporated equal levels of radiolabeled palmitate (Fig. 5B). Palmitate incorporation into _qIE was dramatically reduced, while palmitoylation of _qIE+, _qIER, and _qIER+ was reduced even further when corrected for expression levels (Fig. 5B). Thus, despite the fact that cysteine sites of palmitoylation are retained in all mutants, the loss of critical side chains in the N-terminal binding regions of _s and _q can adversely affect palmitoylation.

View larger version (21K): [in this window] [in a new window]   Fig. 5.   Palmitoylation of binding region mutants of s and q. The palmitoylation state of each of the binding mutant  subunits was measured by transiently transfecting COS7 cells with s subunit expression constructs (A) and HEK293 cells with q subunit expression constructs (B) and metabolically labeling the subunits with [3H]palmitic acid as described under "Experimental Procedures." The upper panel shows the radiolabeled palmitate incorporated by each subunit and visualized by fluorography (A, 21-day exposure; B, 50-day exposure). Aliquots of each radiolabeled subunit were analyzed by Western blotting (lower panel) as described. Similar results were obtained in two separate experiments with the q subunits in HEK293 and COS7 cells.

Myristoylation Restores Plasma Membrane Localization and Palmitoylation of Mutant _q-- To investigate whether membrane binding could be recovered through the use of another membrane anchor, a variant _q (_qAG), was engineered with a site for myristoylation (Fig. 1). The various binding region mutations (Fig. 1) were introduced into the _qAG background. The amino acid sequence of the extreme N terminus of _qAG is consistent with a consensus site for myristoylation (47, 48), and, indeed, directs myristoylation of _qAG and _qAG containing binding region mutations (Fig. 7, lower panel), as described below.

_qAG was found exclusively in the membrane fraction for all binding region mutants including _qAG IE and _qAG IER+ (Fig. 6A). To determine if these myristoylated and potentially binding-defective subunits are correctly targeted to the plasma membrane, they were visualized by immunofluorescence microscopy. We detected no discernible difference in localization between _q and _qAG within intact cells (Fig. 3G versus Fig. 6B). Therefore, neither removal of residues 1-6, mutation of alanine to glycine nor myristoylation itself alters membrane targeting of _q. Visualization of _qAG IE and _qAG IER+ (Fig. 6, C and D) showed that they too are correctly localized at the plasma membrane and not at internal membranes. Thus, myristoylation does indeed seem capable of substituting for the loss of putative binding ability in targeting the subunits to the plasma membrane.

View larger version (52K): [in this window] [in a new window]   Fig. 6.   Localization of binding region mutants of myristoylated qAG by cell fractionation and immunofluorescence. Each of the putative binding mutants of HA epitope-tagged qAG was transiently expressed in HEK293 cells. A, S and P extracts were prepared as described, resolved by SDS-PAGE, and visualized by Western blotting with 12CA5 antibody. The results shown with qAG, qAG IE, and qAG IER+ are representative of all the other myristoylated subunits and similar results were obtained in two different experiments, although the level of expression did vary slightly between experiments. The asterisk marks the position of the 42-kDa  subunit bands. The subcellular localization of three different subunits, qAG (B), qAG IE (C), and qAG IER+ (D) was visualized by immunofluorescence as described under "Experimental Procedures."

Our previous experiments demonstrated a correlation between membrane binding and the palmitoylation state of the subunit, and mutant _q or _s subunits containing a myristoylation sequence from _t but lacking palmitoylatable cysteines (5) fail to properly localize to plasma membranes.⁴ This suggests that the myristoylated _qAG subunits would be fully palmitoylated since they are all associated with the plasma membrane. The myristoylated _qAG variants were assayed to determine their palmitoylation state and all of them were found to be palmitoylated to the same degree regardless of mutations in the binding region (Fig. 7, upper panel). This suggests that membrane binding is sufficient for palmitoylation of  subunits.

View larger version (49K): [in this window] [in a new window]   Fig. 7.   Palmitoylation and myristoylation of binding region mutants of q and qAG. The myristoylation and palmitoylation state of each of the binding mutant qAG subunits was measured by transiently transfecting HEK293 cells with qAG subunit expression constructs and metabolically labeling the subunits with [3H]palmitic acid or [3H]myristic acid as described under "Experimental Procedures." The amount of radiolabeled palmitate (upper panel) and myristate (lower panel) incorporated by each subunit was visualized by fluorography after a 30-day exposure. Identical gels were treated with hydroxylamine (data not shown). The myristate signal was unaffected by hydroxylamine treatment, while the palmitate signal was greatly decreased in all samples. [3H]palmitate-labeled q, qAG, and qAG containing N-terminal mutations displayed identical sensitivities to hydroxylamine.

Although myristoylation restored plasma membrane localization and palmitoylation to binding region mutants of _q (Fig. 7), _qAG subunits containing binding region mutations were unable to couple activated ₂-AR to increased inositol phosphate production (data not shown). This defect was not due to the alanine to glycine mutation since _qAG worked as well as _q in this signaling assay (data not shown). Instead, this likely reflects the inability of these mutants to bind and form a functional -receptor complex. To verify that the _qAG mutants are properly folded, capable of binding GTP, and attaining their activated conformation, the mutant with the most mutations, _qAG IER+, was tested in a trypsin protection assay. This widely used assay takes advantage of the fact that activated (GTP- or GDP·AlF₄-bound) G protein  subunits acquire resistance to trypsin digestion due to an activation-dependent conformational change (38, 49). In this assay, trypsin treatment removes only a small portion of the N terminus from activated  subunits, whereas unactivated (GDP-bound)  subunits are degraded to smaller fragments. In the assay described here, binding of GDP·AlF₄ was much more efficient than GTPS (data not shown) at promoting protection from trypsin cleavage, probably due to the slow rate of _q guanine nucleotide exchange (50). The results show that both _qAG and _qAG IER+ were protected from proteolysis by trypsin after incubation with GDP·AlF₄ (Fig. 8), indicating that multiple mutations in the N terminus of _q do not affect the ability of the protein to bind guanine nucleotides and undergo an activation-dependent conformational change.

View larger version (30K): [in this window] [in a new window]   Fig. 8.   Trypsin protection of myristoylated qAG constructs. qAG and qAG IER+ were transiently expressed in HEK293 cells, and the subunits were separated into soluble and particulate fractions as already described. The particulate fraction was utilized in a trypsin protection assay as described under "Experimental Procedures" to verify that these subunits could still attain an activated conformation upon binding a GTP analog. The left and right panels show the results of incubating qAG and qAG IER+, respectively, with GDP and AlF4 and digesting them with trypsin. The size markers correspond to molecular masses of 44 and 33 kDa.

Mutations Lead to Loss of Binding in Myristoylated _qAG-- Our experimental results so far have suggested that mutation of the residues in the N-terminal binding region of _s and _q interfere with the binding of to  subunits. However, the changes seen in localization and signaling could be caused by something other than a loss of binding. To demonstrate that mutant _q subunits are indeed defective in interacting with , we employed an indirect assay by testing their ability to sequester and thus inhibit -mediated signaling. This assay takes advantage of the ability of subunits, released from G_i by agonist activation of a G_i-coupled receptor, to enhance stimulation of _s-stimulated adenylyl cyclase II (ACII) in transfected HEK293 cells. As demonstrated previously with _t (30), _q can inhibit stimulation of ACII-mediated cAMP production (Fig. 9, compare GFP control versus _q bar), most likely by binding to and sequestering free . In contrast, _qIER+ had no effect on the -dependent stimulation of cAMP in this model assay. Because _qIER+ is not found at the plasma membrane, its lack of inhibition may simply be due to its inability to co-localize with free . To address this, we turned to _qAG since binding region mutations in the _qAG background did not compromise localization to plasma membranes (Fig. 6). _qAG is able to compete for binding of , and this results in a 70% reduction in cAMP production (Fig. 9). _qAG subunits with mutations in the binding region are unable to fully compete with ACII for binding, and the introduction of multiple mutations results in  subunit with no more affinity for than GFP protein, transfected as a control. Thus, these results provide evidence that the mutations inhibit interaction of  with as predicted from the crystal structures.

View larger version (36K): [in this window] [in a new window]   Fig. 9.   Inhibition of signaling by qAG mutants. HEK293 cells were transiently co-transfected with 200 ng of ACII-pcDNAI, 200 ng of dopamine type 2 receptor-pcDNAI, 50 ng of sQL-pcDNAI, and 500 ng of pcDNA3 containing cDNA encoding the competitor protein. The competitor protein consisted of various q and qAG subunits or GFP, which served as a negative control. The results are represented as the percentage of increase in cAMP upon agonist treatment. The results shown are the means ± S.D. for two or three experiments assayed in triplicate.

	DISCUSSION

TOP ABSTRACT INTRODUCTION EXPERIMENTAL PROCEDURES RESULTS DISCUSSION REFERENCES

The experiments presented here demonstrate that mutation of several residues in the N-terminal binding region of _s and _q disrupts the membrane localization and palmitoylation of these non-myristoylated G protein  subunits. The potency of this effect was shown to increase as more residues were mutated in _s and _q. The extent of disruption in plasma membrane localization was determined both by cell fractionation and by immunofluorescence localization of each subunit transiently expressed in HEK293 cells. In addition, our results show that introduction of a site for myristoylation in _q can restore plasma membrane localization and palmitoylation to binding region mutants. The residues mutated in this study are highly conserved throughout the family of heterotrimeric G protein  subunits and are located at identical positions to residues in _t and _i1 previously shown to contact in the crystal structures of heterotrimers (Fig. 1) (33, 34). In addition, _q binding region mutants failed to inhibit -mediated signaling in an in vivo assay designed to measure the ability of these subunits to bind and sequester , consistent with the proposal that these mutations are affecting the interaction of  with . Finally, the ability of _s binding region mutants containing an activating mutation to retain stimulation of cAMP production and the ability of _q mutants to be protected from trypsin digestion in an aluminum fluoride-dependent manner argue that the binding region mutations did not disrupt the overall structure of the  subunits.

It is informative to consider these results in terms of the two-signal model for membrane binding. This recently proposed model for membrane targeting of lipid-modified proteins proposes that two membrane attachment signals are required for stable membrane binding of the modified protein (7-12). Further, one of these signals should direct the protein to the correct cellular membrane, in this case, the plasma membrane. For G protein  subunits of the _i family (_i, _o, and _z), it appears that these two signals are myristoylation and palmitoylation. The model states that co-translationally added myristate functions as the first signal and provides a general membrane targeting signal for the _i subunits. When the myristoylated proteins contact the plasma membrane, a palmitoyl transferase proposed to reside there (51, 52) adds a palmitate molecule, which acts synergistically with myristate to give a thermodynamically strong and kinetically long-lived association with the membrane (8, 51). However, members of the _s, _q, and ₁₂ families are not myristoylated; thus, according to the two-signal model, something else must function as the first signal to get these subunits to their correct place at the plasma membrane. Our results provide compelling evidence that interaction of _s and _q with , which is prenylated and capable of stably binding to the plasma membrane, provides this signal. Mutants of _s and _q deficient in binding ability failed to associate with the plasma membrane or incorporate radiolabeled palmitate. It appears that the  subunits must associate with that is already membrane-bound or being transported to the plasma membrane before they can undergo palmitoylation and stably associate with the plasma membrane. Although the relative levels of