INTRODUCTION

Top Abstract Introduction Materials & Methods Results Discussion References

Heterotrimeric G proteins, composed of , , and  polypeptides, relay signals from cell surface receptors to intracellular effector enzymes and ion channels (Neer, 1995). Ligand-bound receptor catalyzes exchange of GTP for GDP on the  subunit, followed by dissociation of _GTP from the dimeric subunit. Each of these subunits can regulate effector molecules. Hydrolysis of bound GTP turns off signaling by the  subunit and allows _GDP to reassociate with and inactivate free . Lacking transmembrane segments, the G protein trimer must nonetheless localize at the inner face of the plasma membrane to transduce signals received from receptors for extracellular stimuli.

Here, we report experiments designed to elucidate the mechanisms that target G protein  subunits to the plasma membrane. Our experiments extend work in many laboratories that have established the role of covalently attached lipids in promoting the association of both  and subunits with membranes (reviewed in Casey, 1995; Wedegaertner et al., 1995; Resh, 1996; Bhatnagar and Gordon, 1997; Mumby, 1997). Mutational removal of sites of lipid attachment greatly reduces membrane avidity of G protein subunits (Jones et al., 1990; Mumby et al., 1990, 1994; Muntz et al., 1992; Wedegaertner et al., 1993; Degtyarev et al., 1994; Hallak et al., 1994; McCallum et al., 1995; Wilson and Bourne, 1995). Isoprenyl groups attached to  chains allow subunits to associate with membranes. Members of the _i family (_i1, _i2, _i3, _z, and _o) are myristoylated on the N-terminal glycine residue and palmitoylated on the adjacent cysteine (Wedegaertner et al., 1995). Several other  subunits are palmitoylated near their N termini, but are not myristoylated (Wedegaertner et al., 1995).

Myristate and palmitate differ in hydrophobicity and in the relative stability of their covalent links to proteins. Although myristate is linked to proteins by a stable amide bond, myristoylated proteins associate reversibly with membranes and dissociate from them rather easily (Peitzsch and McLaughlin, 1993; Bigay et al., 1994; Boman and Kahn, 1995; McLaughlin and Aderem, 1995), at least in part because this 14-carbon fatty acid provides insufficient hydrophobicity to anchor a protein tightly to membranes (Peitzsch and McLaughlin, 1993). The greater hydrophobicity of palmitate's 16 carbons promotes tighter association with membranes (Shahinian and Silvius, 1995), but its bond to cysteine can be cleaved by cellular thioesterases (Camp and Hofmann, 1993; Mumby, 1997), providing a potential handle for regulating association of a protein with membranes (Milligan et al., 1995; Wedegaertner et al., 1995; Mumby, 1997). Activated _s, for example, is more rapidly depalmitoylated (Degtyarev et al., 1993; Mumby, et al., 1994; Wedegaertner and Bourne, 1994) and can translocate to the cytosol (Wedegaertner et al., 1996).

In addition to providing hydrophobicity, lipid modifications increase the affinities of G protein subunits for each other and for effector molecules (Wedegaertner et al., 1995). Myristate on _i proteins (Jones et al., 1990; Linder et al., 1991) and palmitate on _s (Iiri et al., 1996) increase the affinity for binding and nonmyristoylated _i fails to inhibit adenylyl cyclase (Taussig et al., 1993). Similarly, prenylation of  increases the affinity of for  (Iñiguez-Lluhi et al., 1992) and is necessary for regulation of adenylyl cyclase (Iñiguez-Lluhi et al., 1992) and phospholipase C (Dietrich et al., 1994) by .

How do G proteins target themselves to one cell membrane rather than another? Ras proteins and kinases in the Src family provide insight into the requirements for specific membrane targeting. These proteins combine two membrane-binding signals, either two lipids or one lipid plus a cluster of basic residues, to effect stable target membrane binding (Hancock et al., 1990; Resh, 1994, 1996). To address the role of dual lipid modifications as signals for plasma membrane localization of G protein  subunits, we studied a member of the _i subfamily, _z, in which mutations removed sites for attachment of myristate, palmitate, or both. Our results support a model in which palmitate provides plasma membrane selectivity, whereas myristate and promote and stabilize both palmitoylation and membrane attachment.

	MATERIALS AND METHODS

Top Abstract Introduction Materials & Methods Results Discussion References

Materials

Glu-Glu (EE) monoclonal antibody was obtained from Onyx Pharmaceuticals (Richmond, CA), monoclonal antibody against hemagglutinin (HA) was obtained from Berkeley Antibody Co. (Berkeley, CA), and polyclonal antibodies against ₂ were obtained from Santa Cruz Biotechnology (Santa Cruz, CA). Rabbit anti-mannosidase II was kindly provided by Kelley Moremen (University of Georgia, Athens, GA). Pertussis toxin (PTX) was obtained from List Biologicals (Campbell, CA), quinpirole was obtained from Research Biochemicals (Natick, MA), and streptolysin O (SLO) was obtained from Murex Diagnostics (Norcross, GA). Fluorescein isothiocyanate (FITC) and Texas Red-conjugated secondary antibodies were obtained from Jackson Immunoresearch (West Grove, PA) and Texas Red-phalloidin and unlabeled phalloidin were obtained from Molecular Probes (Eugene, OR). All isotopes were from DuPont-New England Nuclear (Wilmington, DE).

Expression Vectors

cDNAs in pcDNA1 encoding _z-WT, _z-G2A, _z-C3A, and _z-G2AC3A, all with the EE epitope tag, were as described (Wilson and Bourne, 1995). To generate stable cell lines the _z-coding sequences were subcloned into pcDNA3 as EcoRI fragments. pHA-mitogen-activated protein kinase (MAPK) DNA was a gift from J. Pouysségur (CNRS, Nice, France), ₂ and ₁ DNAs were gifts from Janet Robishaw, and D₂ receptor DNA was a gift from Olivier Civelli.

Cell Culture and Transfection

CHO-K1 cells were propagated in minimal essential medium  (MEM) with 10% fetal bovine serum. DNA was transiently transfected using the adenovirus method of Forsayeth and Garcia (1994). Plasmid DNAs were added to a transfection mix consisting of serum-free MEM, 80 µg/ml DEAE-dextran, and a 1:20 dilution of adenovirus stock. This solution was added to subconfluent cells on dishes. Following a 1.5-h incubation cells were shocked with 10% DMSO in phosphate-buffered saline (PBS) and refed with normal serum-containing medium. The following DNA amounts/10⁶ cells were used: 1 µg D₂R, 0.5 µg _z, 0.2 µg ₁, and 0.2 µg ₂.

Stable cell lines were generated by transfecting Chinese hamster ovary (CHO)-K1 cells with pcDNA3 constructs. Two days later cells were reseeded at dilutions of 1:100 and 1:500 and transferred into medium containing G418 at 600 µg/ml to select for stable transformants. Serial dilution of the stable pool was used to obtain cell lines derived from single cells. Stable cell lines were maintained in G418-containing medium.

Measurement of MAPK Activity

Cells were transfected in 6-well plates at 1.5 × 10⁶ cells/well, transferred to serum-free medium after 24 h, and assayed after 48 h. Cells were treated with PTX (100 ng/ml) for 4 h before agonist stimulation. Treated cells were washed once with cold PBS and lysed in buffer containing 1.0% Triton X-100. Insoluble material was removed by microcentrifugation, 5 µg of 12CA5 antibody were added to the supernatant, and samples were incubated at 4°C for 1 h. Protein A agarose (Life Technologies, Gaithersburg, MD) was added and samples were tumbled for 1 h. Immunoprecipitates were washed twice with lysis buffer and twice with kinase buffer as described (Faure and Bourne, 1995). MAPK activity was measured using myelin basic protein as substrate in an in vitro kinase reaction. Samples were run on 14% gels and dried, and radioactivity was quantified with a PhosphorImager (Molecular Dynamics, Sunnyvale, CA).

Cell Fractionation

Four 100-mm plates of CHO-K1 cells were trypsinized and spun down 48 h after transfection. Pellets were resuspended in 1 ml of lysis buffer (40 mM Tris, pH 8.0, 1 mM EDTA, 2 mM MgCl₂, 2 mM -mercaptoethanol) plus protease inhibitors (1 mM phenylmethyl sulfonyl fluoride, 1 µg/ml pepstatin, 2 µg/ml leupeptin, 4 µg/ml aprotinin) and passed 15 times through a 27-gauge needle. Following a low-speed spin to remove unlysed cells and nuclei, supernatant fractions were separated into soluble and particulate fractions by centrifugation for 30 min at 150,000 × g.

Palmitate Labeling and Turnover

One hundred-millimeter plates of CHO-K1 cells were incubated for 2 h with MEM containing 5 mM sodium pyruvate and 0.75 mCi/ml [³H]palmitate. Cells were washed with PBS and incubated for 1 h in 1 ml of lysis buffer (40 mM Tris, pH 8.0, 1 mM EGTA, 2 mM MgCl_2, 1% cholate) plus protease inhibitors. Following microcentrifugation at 14,000 rpm, _z was immunoprecipitated from the supernatant fraction with the EE monoclonal antibody (mAb) as described below. Samples were run on 10% gels. To determine hydroxylamine sensitivity, duplicate, fixed gels were treated for 12 h either with 1 M hydroxylamine (pH 7.0) or with 1 M Tris (pH 7.0). Gels were processed for fluorography with Amplify (Amersham, Arlington Heights, IL).

For measurement of palmitate turnover, cells were washed twice with chase medium (MEM plus 5 mM sodium pyruvate plus 100 µM palmitic acid) after labeling. Cells were then treated with chase medium with or without 10 µM quinpirole and incubated at 37°C. Depalmitoylation was terminated by washing cells with ice-cold PBS.

Immunofluorescence Localization

CHO-K1 cells were grown on glass coverslips and fixed with 4% formaldehyde in PBS. For immunofluorescence, all incubations (except the final wash) were in PBS plus 5% nonfat milk plus 1% Triton X-100. Fixed cells were blocked for 30 min in this buffer prior to a 1-h incubation with primary antibody (anti-EE mAb, 20 µg/ml; rabbit anti-mannosidase II, 1:1000; rabbit anti-₂, 2 µg/ml). After three 10-min washes, cells were incubated with the appropriate secondary antibody (donkey anti-mouse FITC conjugate and donkey anti-rabbit Texas Red conjugate) at 1:100 dilution for 30 min. The final three washes were in PBS plus 1% Triton X-100 and coverslips were mounted on glass slides with ProLong Antifade (Molecular Probes, Eugene, OR). For actin staining, Texas Red-phalloidin was included in the secondary antibody incubation.

Immunofluorescence microscopy was performed with a confocal laser scanning microscope (MRC-1000, Bio-Rad Labs, Hercules, CA) using FITC and Texas Red filters. Images were processed with Adobe Photoshop.

For SLO permeabilization, cells on coverslips were incubated for 10 min at room temperature with 1 U/ml SLO in 10 mM MES (pH 6.1), 140 mM KCl, 3 mM MgCl_2, 2 mM EGTA, 0.3 M sucrose, and 1 µg/ml phalloidin (cytoskeletal stabilizer). Cells were fixed immediately after treatment.

Western Blotting and Immunoprecipitation

For Western blotting, proteins were transferred to polyvinylidene difluoride membrane (Bio-Rad Labs) and probed with EE mAb at 3 µg/ml. Secondary antibodies with conjugated horseradish peroxidase were used at 1:10,000 dilution and bands were visualized by ECL (DuPont-New England Nuclear).

For immunoprecipitation, Triton X-100 and SDS were added to cholate extracts of cells, to final concentrations of 1% and 0.5%, respectively. EE mAb (3 µg) was added and samples were tumbled for 1 h at 4°C. Protein G-Sepharose beads (Pharmacia, Pistcataway, NJ) were added and samples were tumbled for 1 h. Pelleted beads were washed three times with lysis buffer plus 1% Triton X-100 and 0.5% SDS. Beads were resuspended in SDS-polyacrylamide sample buffer.

	RESULTS

Top Abstract Introduction Materials & Methods Results Discussion References

Most experiments were performed with transiently transfected CHO-K1 cells expressing epitope-tagged _z subunits (Wilson and Bourne, 1995), including _z-WT (epitope tagged only), _z-G2A (myristoylation site mutated to alanine), _z-C3A (palmitoylation site mutated to alanine), and _z-G2AC3A (alanine substituted at both sites). The epitope tag did not disrupt signaling function or membrane attachment of _z-WT (Wilson and Bourne, 1995). With _z we have shown (Wilson and Bourne, 1995), as have other investigators with several _i family members (Mumby et al., 1990, 1994; Degtyarev et al., 1994; Hallak et al., 1994), that mutating the lipid modification sites abolishes incorporation of the corresponding lipid, and that in intact cells association of the G2A or the C3A mutant with membranes is completely lacking or impaired, respectively.

Membrane Attachment and Function

To assess signaling properties by the _z mutants, we assayed a positive signal, activation of the MAPK cascade. Plasmids encoding the four _z proteins were separately and transiently transfected into CHO-K1 cells along with plasmids encoding the D₂ dopamine receptor (D₂R), which activates _z, and epitope-tagged MAPK (HA-MAPK) to allow assay of MAPK activity after immunoprecipitation. PTX treatment inhibits D₂R-mediated activation of MAPK in cells not transfected with _z (Figure 1), presumably because the toxin inactivates all _i family members except _z; thus, D₂R-mediated signals in PTX-treated cells reflect activation of _z. In cells expressing _z-WT, quinpirole, a D₂R agonist, stimulated HA-MAPK activity in a PTX-insensitive manner (Figure 1). In cells expressing _z-G2A or _z-G2AC3A, quinpirole did not activate HA-MAPK, whereas HA-MAPK activation in cells expressing _z-C3A was comparable to that mediated by _z-WT (Figure 1). Abilities of the _z mutants to mediate agonist activation of MAPK paralleled our previous observations of their abilities to regulate a different effector pathway, inhibition of adenylyl cyclase (Wilson and Bourne, 1995), with a subtle difference: _z-C3A inhibited adenylyl cyclase even in the absence of agonist stimulation, to a much greater degree than _z-WT (Wilson and Bourne, 1995), whereas this mutant caused activation of MAPK activity only in response to agonist (Figure 1). We do not know the mechanism responsible for this apparent discrepancy, which does not affect inferences we draw from the data presented below.

View larger version (25K): [in this window] [in a new window]   Figure 1.   Activation of HA-MAPK by z and mutants. cDNAs encoding D2R and either z-WT, z-G2A, z-C3A, z-G2AC3A, or vector were transfected into CHO-K1 cells. Cells were serum starved for 24 h. After 4 h of pretreatment with or without PTX, cells were treated with serum-free medium (basal) or 10 µM quinpirole for 7 min. HA-MAPK activity was measured as described in MATERIALS AND METHODS. Bars, mean ± 2 SE of triplicate determinations. Similar results were obtained in two additional experiments.

The same _z mutants that fail to signal (Figure 1) also fail to associate with membranes (Figure 2A). Separation of CHO-K1 cell homogenates into particulate (P100) and soluble (S100) fractions shows that _z-WT is associated exclusively with the particulate fraction and that _z-C3A partitions into both fractions, whereas _z-G2A and _z-G2AC3A are almost entirely soluble (Figure 2A).

View larger version (26K): [in this window] [in a new window]   Figure 2.   Particulate versus soluble distribution of z mutants with or without . CHO-K1 cells were transiently transfected with z mutants alone (A) or transfected with z mutants plus 1 and 2 (B). Crude particulate (P) and soluble (S) fractions were prepared as described in MATERIALS AND METHODS. Equivalent proportions of each fraction were analyzed by SDS-PAGE and Western blotting with the EE mAb.

Despite an intact palmitoylation site, _z-G2A does not incorporate palmitate. Metabolic labeling with [³H]palmitate shows that _z-WT is palmitoylated, while none of the mutants, including _z-G2A, incorporate palmitate (Figure 3). These results confirm previous reports (Degtyarev et al., 1994; Hallak et al., 1994; Mumby et al., 1994; Wilson and Bourne, 1995) that G2A mutants of _i family members incorporate neither myristate nor palmitate.

View larger version (20K): [in this window] [in a new window]   Figure 3.   Incorporation of palmitate by z mutants. CHO-K1 cells, transiently transfected with z mutants, were incubated with 0.75 mCi/ml [3H]palmitic acid for 2 h. z was immunoprecipitated from total cell extracts with the EE mAb followed by SDS-PAGE. Identical gels were treated with either 1 M Tris (pH 7.0) or 1 M hydroxylamine (pH 7.0) for 12 h. Gels were then processed for fluorography (19-d exposure).

Immunofluorescence Microscopy Reveals Distinct Patterns

We examined the subcellular localization of both stably and transiently expressed _z mutants by indirect immunofluorescence. The two kinds of expression produced identical staining patterns, thus ruling out artifacts that might result from short-term overexpression. The epitope tag allowed specific detection of the mutants with the EE antibody; very low background staining is seen in untransfected cells (Figure 4A).

View larger version (124K): [in this window] [in a new window]   Figure 4.   Subcellular localization of WT and mutant z in stably transfected cells. CHO-K1 cells were stably transfected with pcDNA3 vector alone (A), z-WT (B), z-G2A (C), z-C3A (D), or z-G2AC3A (E). Cells were plated onto glass coverslips and fixed with 4% formaldehyde in PBS. Permeabilized cells were incubated with EE mAb followed by incubation with FITC-conjugated antimouse antibody. Images were obtained by confocal laser scanning microscopy. Bar, 10 µm

Immunofluorescence revealed that _z without palmitate is partially mistargeted to intracellular membranes, whereas _z mutants without myristate or without sites for either myristate or palmitate do not associate with membranes or with any other structure in the cytoplasm. _z-WT shows characteristic plasma membrane localization, with uniform staining across the cell surface that is enhanced between adjacent cells (Figure 4B). A plasma membrane protein, Na⁺-K⁺-ATPase, shows the same pattern (our unpublished results). _z-C3A, which incorporates myristate but not palmitate, shows a more complex distribution pattern. Unlike _z-WT, _z-C3A associates with internal structures clustered around the nucleus (Figure 4D). This partial mistargeting of _z-C3A significantly diminishes but does not abolish its association with the plasma membrane (Figure 4D). The perinuclear and plasma membrane staining pattern is seen with both transiently and stably transfected _z-C3A (our unpublished results). _z-G2A (Figure 4C) and _z-G2AC3A (Figure 4E) show no plasma membrane staining; instead, these mutants are diffusely distributed throughout the cytosol and nucleus.

Immunofluorescence of cells treated with SLO confirmed that _z-G2A is cytosolic, whereas _z-WT and _z-C3A are membrane associated. SLO, a bacterial cytolysin, specifically perforates the plasma membrane, allowing cytoplasmic proteins to escape but leaving intracellular membranes intact (Miller and Moore, 1991). SLO treatment abolishes the apparent cytosolic distribution of _z-G2A, leaving only the nuclear staining (Figure 5, C and E). _z-G2AC3A shows an identical pattern after SLO treatment (our unpublished results). In contrast, SLO treatment does not affect the distribution of _z-WT (Figure 5A) or of _z-C3A (Figure 5G). The latter result (Figure 5G) confirms that the staining pattern seen with _z-C3A represents _z associated not only with internal cellular membranes but also with the plasma membrane (in addition to the internal staining, note enhanced staining between adjacent cells). Staining of actin filaments (Texas Red-palloidin) (Figure 5, B, D, and H) and cis-Golgi (mannosidase II antibody; Figure 5F) shows that cellular structures other than the nucleus are intact after SLO treatment.

View larger version (99K): [in this window] [in a new window]   Figure 5.   Immunofluorescence pattern of z following SLO treatment. CHO-K1 cell lines stably expressing z-WT (A and B), z-G2A (C-F), or z-C3A (G and H) were treated with SLO as described in MATERIALS AND METHODS and fixed. z was detected by incubation with EE mAb followed by FITC-conjugated antimouse antibody (A, C, E, and G). The same cells were stained with Texas Red-conjugated phalloidin to visualize actin (B, D, and H) or with rabbit anti-mannosidase II followed by Texas Red-conjugated antirabbit antibodies to visualize Golgi (F). Images were obtained by confocal laser scanning microscopy. Bar, 10 µm

Effects of Overexpressing

What role does play in targeting of _z and anchoring it to membranes? Overexpression of with _z-G2A and _z-G2AC3A revealed that facilitates palmitoylation, membrane association, and receptor-mediated signaling of _z. As shown (Figure 1), _z-G2A and _z-G2AC3A fail to mediate MAPK stimulation following activation of the D₂R. In contrast, coexpression of with these mutants rescues the ability of _z-G2A to signal (Figure 6). Quinpirole stimulates HA-MAPK activity only when _z-G2A is coexpressed with . _z-G2AC3A fails to signal regardless of whether is coexpressed (Figure 6).

View larger version (20K): [in this window] [in a new window]   Figure 6.   Effect of coexpressing on activation of HA-MAPK by z-G2A and z-G2AC3A. D2R was transiently transfected into CHO-K1 cells with vector, z-G2A, z-G2A plus 1 and 2, z-G2AC3A, or z-G2AC3A plus 1 and 2. Cells were serum starved for 24 h and pretreated with PTX for 4 h before stimulating with serum-free medium (basal) or 10 µM quinpirole for 7 min. HA-MAPK activity was assayed as described in MATERIALS AND METHODS. Bars, mean ± 2 SE of triplicate determinations. Similar results were obtained in two separate experiments.

To determine whether coexpression of affects the ability of _z to associate with membranes, cells expressing each of the _z proteins with or without were separated into S100 and P100 fractions. coexpression increases the amount of _z associated with the particulate fraction (Figure 2). This effect is especially marked in the case of _z-G2A, which is entirely cytosolic when expressed alone but associates significantly with the particulate fraction when is coexpressed; this shift to the particulate fraction parallels the effect of in restoring to _z-G2A the ability to transmit hormonal signals (Figure 6). _z-G2AC3A, which fails to signal even in the presence of coexpressed (Figure 6), also shifts to the P100 fraction in cells coexpressing , but consistently to a lower extent than does _z-G2A (Figure 2).

Does membrane association promoted by affect the palmitoylation state of _z-G2A? Coexpression of greatly increases palmitoylation of _z-G2A, but palmitate labeling of _z-WT is unchanged by excess (Figure 7A). All of the [³H]palmitate on _z-G2A is found in the particulate fraction (Figure 7B), in keeping with the idea that palmitate plays a role in membrane association.

View larger version (19K): [in this window] [in a new window]   Figure 7.   Effect of on palmitate incorporation by z-G2A and the cellular distribution of palmitoylated z-G2A. (A) z-WT and z-G2A were separately transfected into CHO-K1 cells without or with 1 and 2. Palmitate incorporation was determined as described in the legend of Figure 3. A third identical gel was analyzed by Western blotting with the EE mAb. (B) CHO-K1 cells transfected with z-G2A, 1 and 2 were labeled with [3H]palmitate for 2 h. Following radiolabeling, particulate (P) and soluble (S) extracts were prepared. After immunoprecipitation with the EE mAb, proteins were resolved by SDS-PAGE and visualized by fluorography (21-d exposure); an identical gel was analyzed by Western blotting with the EE mAb.

Targets _z to the Plasma Membrane

Immunofluorescence microscopy of cells transiently coexpressing _z mutants and revealed that targets _z to the plasma membrane. _z-G2A shows two different staining patterns, depending on whether is coexpressed. As observed in stable cell lines (Figure 4C), _z-G2A alone localizes diffusely throughout the cytosol and nucleus with the intensity of fluorescent signal diminishing toward the edges where the cell is thinnest (Figure 8A). In cells coexpressing , however, _z-G2A is localized over the entire cell surface, extending uniformly to the cell edges, as well as in the nucleus (Figure 8B). The nuclear staining probably reflects solubility of a fraction of _z-G2A, as shown by subcellular fractionation (Figure 2A). To confirm association of _z-G2A with the plasma membrane, cells were treated with SLO. Although SLO treatment abolishes cytosolic staining of _z-G2A expressed without exogenous (Figure 8C), _z-G2A coexpressed with remains associated with the plasma membrane, and no staining of intracellular membranes is observed (Figure 8E). _z-G2AC3A coexpressed with shows a pattern similar to that seen with _z-G2A, but with apparently less robust staining at the plasma membrane (our unpublished results).

View larger version (124K): [in this window] [in a new window]   Figure 8.   Effect of on the subcellular localization of z-G2A. CHO-K1 cells transiently transfected with z-G2A (A, C, and D) or z-G2A plus 1 and 2 (B, E, and F) were fixed after no treatment (A and B) or after treatment with SLO (C-F). z-G2A was detected by incubation with the EE mAb followed by incubation with FITC-conjugated antimouse antibody (A-C and E). Texas Red-conjugated phalloidin-stained actin filaments (D and F). Bar, 10 µm

Costaining for  and  revealed that _z and localize in distinct but partially overlapping locations. Cells expressing either _z-WT plus or _z-G2A plus were stained for _z, using the anti-EE mAb, and for , using a ₂ specific polyclonal antibody. Although the _z subunits associate predominantly with the plasma membrane (Figure 9, A and C),  subunits localize to extensive perinuclear structures in the cytoplasm in addition to the plasma membrane (Figure 9, B and D). Other observations support the inference that subunits stably localize in at least two cellular locations: First, polyclonal antisera show identical staining patterns for endogenous  and  (our unpublished results). In addition, cells coexpressing recombinant ₂ and a tagged ₁ subunit show overlapping staining, seen on the plasma membrane and intracellular membranes (our unpublished results).

View larger version (100K): [in this window] [in a new window]   Figure 9.   Subcellular localization of z and . CHO-K1 cells transiently transfected with 1 and 2 plus z-WT (A and B) or z-G2A (C and D) were fixed in 4% formaldehyde. z was detected by incubation with EE mAb followed by FITC-conjugated antimouse antibody (A and C). 2 was detected by incubation with polyclonal antiserum against 2 followed by Texas Red antirabbit antibody (B and D). Bar, 10 µm.

Myristate Stabilizes Palmitate on _z

Is the palmitoylation state of _z dynamically regulated, as reported (Wedegaertner and Bourne, 1994) in the case of _s, which is rapidly depalmitoylated upon activation? Does the adjacent myristate on _z affect palmitate turnover? We measured the rate of depalmitoylation of _z-WT, with or without D₂R stimulation, by metabolic labeling with [³H]palmitate followed by a chase with excess unlabeled palmitate. Unlike previous observations (Wedegaertner and Bourne, 1994) with _s, depalmitoylation of _z-WT is quite slow and is not increased by treatment with agonist (Figure 10A). To address whether stability of the palmitate linkage depends upon myristate, we coexpressed with _z-G2A to induce palmitoylation and membrane association. Compared with _z-WT, _z-G2A loses its palmitate label much more rapidly; agonist treatment greatly accelerates this rate (Figure 10B). Taken together, these results suggest that myristate stabilizes the palmitate on _z-WT; removing myristate produces an _z that behaves like _s.

View larger version (27K): [in this window] [in a new window]   Figure 10.   Turnover of palmitate on z-WT and z-G2A plus . Cells were transfected with either D2R and z-WT (A) or D2R, z-G2A, 1, and 2 (B). Cells were incubated with 0.75 mCi/ml [3H]palmitate for 2 h. After radiolabeling cells were incubated in chase medium in the presence or absence of 10 µM quinpirole. At indicated times, cells (100-mm plate) were harvested and z was immunoprecipitated from the total cell extract. Depalmitoylation was determined by densitometry of the fluorographs (A, 12-d exposure; B, 9-d exposure).

	DISCUSSION

Top Abstract Introduction Materials & Methods Results Discussion References

In lieu of amino acid sequences that traverse the membrane bilayer, many proteins use covalently attached lipids as anchors for mooring to cellular membranes. Accumulating evidence indicates that the attached lipid may provide not only a hydrophobic anchor but also a signal that specifies localization of the protein to a distinct membrane compartment in the cell. In many cases, two attached lipids cooperate to trap the protein in a membrane compartment. As outlined by several investigators (Cadwallader et al., 1994; Shahinian and Silvius, 1995; Resh, 1996), the two-lipid bilayer-trapping mechanism involves three key elements: 1) Together lipids A and B moor a protein much more securely to the membrane bilayer than does either lipid alone; thus, attachment of lipid B to a protein that is already modified by lipid A can trap the protein on the bilayer. 2) A membrane compartment becomes the target for localization of a protein when the enzyme for attaching lipid B is itself restricted to that compartment. 3) Lipid A both promotes attachment of lipid B and reinforces tight association with the membrane.

Our experiments suggest that for _z palmitate functions as lipid B, providing plasma membrane specificity, whereas myristate, lipid A, promotes palmitoylation and stable membrane association. As an extension of the two-signal hypothesis, we show that the signal provided by lipid A can be provided by an interacting protein, in that can substitute for myristate to allow palmitoylation-dependent trapping of _z on the plasma membrane. We propose that similarly provides signal A for  subunits that are palmitoylated but not myristoylated, such as _s, _q, 12, and 13.

Palmitate Can Serve as Lipid B

Two observations support the inference that palmitate is the signal that specifies plasma membrane localization of _z: 1) _z-C3A, which incorporates myristate but not palmitate, partially mislocalizes to intracellular membranes (Figure 4D). 2) When coexpressed with , _z-G2A is palmitoylated and associates specifically with the plasma membrane, as indicated by immunofluorescence staining in SLO-treated cells (Figure 8E); palmitoylation and the consequent association of this mutant with the plasma membrane restore its ability to transduce hormonal signals.

Palmitate also provides the specific signal for association of Ras proteins and Src kinases with the plasma membrane (Hancock et al., 1990; Cadwallader et al., 1994; Schroeder et al., 1996; van't Hof and Resh, 1997; Wolven et al., 1997; Zlatkine et al., 1997). In the case of H-ras, a prenyl group at the carboxyl terminus normally serves as lipid A, and attachment of palmitate as lipid B is required for targeting to the plasma membrane (Hancock et al., 1990; Cadwallader et al., 1994). Removal of the palmitoylation site on H-ras, combined with the addition of an N-terminal myristate, causes the protein to be mistargeted to numerous membranes (Cadwallader et al., 1994); palmitoylation in combination with either myristate or a prenyl group, however, causes H-ras to localize at the plasma membrane (Cadwallader et al., 1994). Similarly, synthetic cell-permeant myristoylated peptides are rapidly palmitoylated and localize at the plasma membrane when added to cultured cells, but only if the peptide contains a palmitoylation site resembling the sites in proteins (G protein  subunits and Src kinases) that are modified by palmitate as well as myristate (Schroeder et al., 1996). Fyn kinase, which is both myristoylated and palmitoylated, targets rapidly and efficiently to the plasma membrane following synthesis (van't Hof and Resh, 1997). Fyn mutants that are only myristoylated, however, move slowly to membranes (van't Hof and Resh, 1997) and mislocalize to intracellular membranes (Wolven et al., 1997).

It is likely that palmitate is attached to _z, Fyn kinase, and H-ras by enzymes that are themselves restricted to the plasma membrane. Indeed, a recently described palmitoyltransferase (Dunphy et al., 1996), partially purified on the basis of its ability to palmitoylate G protein  subunits, is enriched in plasma membrane fractions. Our observations with _z mutants are consistent with the idea that palmitoylation occurs at the plasma membrane. Thus, the G2A mutant of _z, as previously reported for _i (Degtyarev et al., 1994), is soluble and fails to incorporate palmitate, but coexpression with partially restores association with membranes and palmitoylation. The inference that palmitoylation of _z-G2A occurs specifically at the plasma membrane accords with immunolocalization of the mutant and its ability to transduce hormonal signals when coexpressed with . Moreover, is found in internal membranes as well as the plasma membrane (Figure 9, B and D), but delivers _z-G2A to the plasma membrane only, as would be the case if provides the nonspecific membrane association signal (replacing lipid A, myristate), whereas lipid B, palmitate, provides the specific signal. Similarly, two different lipids can serve as lipid A for H-ras (Cadwallader et al., 1994): Mutations that prevent prenylation of H-ras inhibit its palmitoylation (Hancock et al., 1989), but a myristoylation site at the N terminus can substitute for the prenyl group in permitting modification by palmitate (Cadwallader et al., 1994).

Signal A Promotes Palmitoylation

An essential feature of the two-lipid bilayer-trapping mechanism is that the nonspecific signal, lipid A, promotes and stabilizes attachment of lipid B. How does myristateor its substitute, coexpressed facilitate palmitoylation and stable trapping of _z and other G protein  subunits at the plasma membrane? Available evidence suggests that myristate and not only promote the forward palmitoylation reaction but also stabilize the link between palmitate and the  subunit. Myristate and could promote palmitoylation indirectly by delivering  subunits to membranes and increasing the concentration of protein substrate accessible to a membrane-bound palmitoyltransferase. More direct roles for and myristate are suggested by the observations that: 1) A palmitoyltransferase enriched in plasma membrane fractions prefers as a substrate an  subunit complexed to in comparison to the free  subunit (Dunphy et al., 1996). 2) A second palmitoyltransferase activity, recently characterized (Berthiaume and Resh, 1995), attaches palmitate only to myristoylated proteins, including G protein  subunits.

Other observations indicate that myristate and also inhibit depalmitoylation of _z. Thus, _z-WT is depalmitoylated much more slowly than is _z-G2A, which is modified by palmitate but not myristate, in cells coexpressing (Figure 10). Moreover, receptor activation increases the rate of turnover of the palmitate attached to _s (Wedegaertner and Bourne, 1994) and of the palmitate attached to _z-G2A in cells that coexpress (Figure 10). Because activation of the  subunit causes it to dissociate from , these effects of receptor activation on palmitate turnover suggest that release of increases the susceptibility of  to attack by cellular thioesterases. Indeed, pure inhibits depalmitoylation of pure _s by an esterase in vitro (Iiri et al., 1996). Three-dimensional structures of G protein trimers (Wall et al., 1995; Lambright et al., 1996) suggest a possible explanation for these effects of and myristate. The extreme N terminus of the  subunit, to which both myristate and palmitate are attached, is located in close proximity to the dimer and to its prenylation site at the C terminus of the  polypeptide. The proximity of two additional modifying lipids may limit the accessibility of palmitate to esterases, thus stabilizing its attachment to the protein.

Our observations with _z are consistent with the bilayer-trapping mechanism for specific membrane localization as proposed by other investigators (Shahinian and Silvius, 1995; Resh, 1996). After myristate is attached (cotranslationally, in the cytoplasm), _z can associate, weakly and nonspecifically, with any cellular membrane. Transient binding of myristoylated _z to the plasma membrane and association with make _z a potential substrate for posttranslational palmitoylation; attachment of palmitate traps the protein at the plasma membrane (Shahinian and Silvius, 1995; Schroeder et al., 1996) rather than at other membranes, presumably because the G-specific transferase that catalyzes palmitoylation is itself located at the plasma membrane (Dunphy et al., 1996). Ultimately, plasma membrane association of _z is maintained through strong but nonspecific hydrophobicity provided by both myristate and palmitate. A similar pathway has been described for Fyn kinase (van't Hof and Resh, 1997); synthesis and myristoylation of Fyn in the cytoplasm is followed by rapid palmitoylation and association with the plasma membrane (van't Hof and Resh, 1997).

Several groups have reported localization of G protein  subunits and several other signaling proteins to specialized regions of the plasma membrane, including caveolae and/or detergent-insoluble complexes (Sargiacomo et al., 1993; Mineo et al., 1996). In many cases, localization to these microdomains appears to depend on palmitoylation (Shenoy-Scaria et al., 1994; García-Cardeña et al., 1996 and references in Resh, 1996; Mumby, 1997). Palmitoylation-dependent targeting to a microdomain of the plasma membrane may explain the ability of coexpressed to restore signaling function to _z-G2A but not to _z-G2AC3A. Although delivers some fraction of both proteins to membranes, only _z-G2A is palmitoylated. Although our experiments did not address this issue, perhaps palmitate targets _z-G2A to plasma membrane microdomains where signaling can take place.

Other G protein  Subunits

Although _z and other members of the _i family are both myristoylated and palmitoylated, the N termini of other  subunits, _s, _q, ₁₁, ₁₂, and ₁₃, are palmitoylated but lack sites for myristoylation. In the absence of myristate, a source of hydrophobicity and a stabilizer of palmitoylation, how does palmitate become attached to these other  subunits? The mere presence of a palmitoylation site does not suffice for membrane attachment. For example, an _s/Fyn kinase chimera, in which residues from the N terminus of _s (including its palmitoylation site) replace the N-terminal myristoylation and palmitoylation sites of Fyn, is not palmitoylated and does not associate with membranes (van't Hof and Resh, 1997). If a palmitoylation site alone is not enough, how does an  subunit lacking myristate avail itself of palmitate and localize at the plasma membrane?

We propose that the subunit performs the missing functions of lipid A for such an  subunit. In this proposed mechanism, exactly as in the case of _z-G2A, association of with nonmyristoylated  subunits promotes their palmitoylation and stable association with the plasma membrane; the necessary hydrophobicity is supplied by the prenyl group of , whereas association of with the N terminus of the  subunit helps to stabilize attached palmitate.

The presence or absence of myristate on an  subunit may determine its ability to dissociate from the plasma membrane upon hormonal activation. Activation of _s, a nonmyristoylated protein, accelerates its depalmitoylation (Mumby et al., 1994; Wedegaertner and Bourne, 1994) and causes its translocation from the plasma membrane into cytosol (Wedegaertner et al., 1996). GTP-induced dissociation of _s from probably accounts for both of these effects, as suggested previously (Wedegaertner and Bourne, 1994; Iiri et al., 1996; Wedegaertner, et al., 1996). Rapid depalmitoylation and membrane dissociation are not general phenomena for G protein  subunits, as shown by the very slow rate of palmitate turnover on _z-WT, with or without agonist stimulation (Figure 10A). The presence of myristate may ensure that activated _z remains palmitoylated and bound to the plasma membrane even when bound GTP reduces its affinity for . Indeed, neither receptor activation nor mutational activation caused detectable translocation of _z-WT into the cytosol (our unpublished results).

How do G protein subunits localize at the plasma membrane? In terms of the two-signal bilayer-trapping mechanism, the prenyl group on the  polypeptide probably serves as lipid A for , as does the C terminal farnesyl modification for H-ras (Cadwallader et al., 1994). Lipid A provides a nonspecific hydrophobic signal for attachment to membranes, but what about lipid B? For , an obvious candidate for this role is the G protein  subunit; in this case, palmitate attached to the  subunit would play the role of lipid B, targeting to the plasma membrane via its association with the  subunit. Thus, the affinities of  and for associating with one another would allow each protein to participate in targeting the other to the plasma membrane by supplying lipid A (a prenyl group) and  by supplying lipid B, palmitate. In keeping with the idea that  helps to retain at the plasma membrane, genetic deletion of the  subunit in S. cerevisiae shifts to cytosol and internal membranes of the yeast cell, markedly reducing its concentration at the plasma membrane (Hirschman et al., 1997). In this scenario,  and subunits recipro