Amino-terminal Cysteine Residues Differentially Influence RGS4 Protein Plasma Membrane Targeting, Intracellular Trafficking, and Function*

Background: Intracellular trafficking of RGS proteins is an important determinant of their function as G-protein inhibitors. Results: Amino-terminal cysteine residues in RGS4 confer previously uncharacterized localization and trafficking activities. Conclusion: Cys-12 is required for plasma membrane targeting, whereas Cys-2 is required for localization to endosomal pools. Significance: Understanding the mechanisms governing cellular distribution of RGS4 may lead to novel strategies for regulation of its function. Regulator of G-protein signaling (RGS) proteins are potent inhibitors of heterotrimeric G-protein signaling. RGS4 attenuates G-protein activity in several tissues. Previous work demonstrated that cysteine palmitoylation on residues in the amino-terminal (Cys-2 and Cys-12) and core domains (Cys-95) of RGS4 is important for protein stability, plasma membrane targeting, and GTPase activating function. To date Cys-2 has been the priority target for RGS4 regulation by palmitoylation based on its putative role in stabilizing the RGS4 protein. Here, we investigate differences in the contribution of Cys-2 and Cys-12 to the intracellular localization and function of RGS4. Inhibition of RGS4 palmitoylation with 2-bromopalmitate dramatically reduced its localization to the plasma membrane. Similarly, mutation of the RGS4 amphipathic helix (L23D) prevented membrane localization and its Gq inhibitory function. Together, these data suggest that both RGS4 palmitoylation and the amphipathic helix domain are required for optimal plasma membrane targeting and function of RGS4. Mutation of Cys-12 decreased RGS4 membrane targeting to a similar extent as 2-bromopalmitate, resulting in complete loss of its Gq inhibitory function. Mutation of Cys-2 did not impair plasma membrane targeting but did partially impair its function as a Gq inhibitor. Comparison of the endosomal distribution pattern of wild type and mutant RGS4 proteins with TGN38 indicated that palmitoylation of these two cysteines contributes differentially to the intracellular trafficking of RGS4. These data show for the first time that Cys-2 and Cys-12 play markedly different roles in the regulation of RGS4 membrane localization, intracellular trafficking, and Gq inhibitory function via mechanisms that are unrelated to RGS4 protein stabilization.


Regulator of G-protein signaling (RGS) proteins are potent inhibitors of heterotrimeric G-protein signaling. RGS4 attenuates G-protein activity in several tissues. Previous work demonstrated that cysteine palmitoylation on residues in the aminoterminal (Cys-2 and Cys-12) and core domains (Cys-95) of RGS4
is important for protein stability, plasma membrane targeting, and GTPase activating function. To date Cys-2 has been the priority target for RGS4 regulation by palmitoylation based on its putative role in stabilizing the RGS4 protein. Here, we investigate differences in the contribution of Cys-2 and Cys-12 to the intracellular localization and function of RGS4. Inhibition of RGS4 palmitoylation with 2-bromopalmitate dramatically reduced its localization to the plasma membrane. Similarly, mutation of the RGS4 amphipathic helix (L23D) prevented membrane localization and its G q inhibitory function. Together, these data suggest that both RGS4 palmitoylation and the amphipathic helix domain are required for optimal plasma membrane targeting and function of RGS4. Mutation of Cys-12 decreased RGS4 membrane targeting to a similar extent as 2-bromopalmitate, resulting in complete loss of its G q inhibitory function. Mutation of Cys-2 did not impair plasma membrane targeting but did partially impair its function as a G q inhibitor. Comparison of the endosomal distribution pattern of wild type and mutant RGS4 proteins with TGN38 indicated that palmitoylation of these two cysteines contributes differentially to the intracellular trafficking of RGS4. These data show for the first time that Cys-2 and Cys-12 play markedly different roles in the regulation of RGS4 membrane localization, intracellular trafficking, and G q inhibitory function via mechanisms that are unrelated to RGS4 protein stabilization.
Heterotrimeric G-protein-coupled receptors mediate the responses of a wide array of hormones and neurotransmitters. In many circumstances increased G-protein activity is associated with pathophysiologic processes and disease progression. Thus, understanding the cellular mechanisms whereby G-protein signaling can be attenuated is an important step toward designing therapeutic strategies for the control and prevention of such diseases. Regulator of G-protein signaling (RGS) 2 proteins comprise a family of Ͼ35 G-protein inhibitors (1). RGS proteins function as GTPase activating proteins for G␣ subunits (2,3), thereby decreasing the lifetime of activated G-protein complexes. One member of the RGS protein superfamily, RGS4, has been shown to be critical for regulation of G-protein signaling in the brain (4), sinoatrial node (5), pancreas (6), and during tumor cell metastasis (7). Thus, defining intracellular pathways that modulate RGS4 function may have important implications for understanding its role in the modulation of neural function, heart rate regulation, insulin release, and cancer.
The RGS4 protein is composed of two primary functional domains that are both regulated by palmitoylation. First, the carboxyl terminus contains the ϳ120-amino acid GTPase activating protein domain capable of inhibiting both G i -and G qmediated signaling. Palmitoylation of Cys-95 within the GTPase activating protein domain is believed to be important for its function (8). Second, the ϳ50-amino acid amino terminus contains an amphipathic ␣ helix membrane targeting domain that is found adjacent to two palmitoylatable cysteine residues (Cys-2 and Cys-12). Simultaneous mutagenesis of both Cys-2 and Cys-12 or mutagenesis of the amphipathic ␣ helix domain alone (9,10) has been shown to modulate protein function and plasma membrane targeting.
Until recently, however, the relative contributions of Cys-2 and Cys-12 were not independently examined. Some studies suggest that Cys-2 is the primary site of palmitoylation in RGS4 and that palmitoylation of this residue enhances RGS4 function by preventing oxidation/arginylation and subsequent degradation via the N-end rule ubiquitin-mediated proteasomal pathway (11,12). Recently Tu and co-workers (13) showed that palmitoylation of Cys-2 by the palmitoyl acyltransferases DHHC3 (and DHHC7) can increase the half-life of RGS4 and thereby show better G-protein inhibitory potential.
In contrast to the large amount of information known about Cys-2 palmitoylation and its contribution to RGS4 stabilization, very little information is available concerning the relative roles of Cys-2 and Cys-12 as determinants of RGS4 localization and intracellular trafficking. Here, we here investigate the extent to which Cys-12 palmitoylation, regardless of the palmitoylation status of Cys-2, may be an important regulator of RGS4 intracellular localization and function. Our data show for the first time that Cys-12 is more important than Cys-2 as a determinant of RGS4 plasma membrane targeting and G q inhibitory function, and we propose that this is due to its proximity to the amphipathic ␣ helix plasma membrane targeting domain. Although Cys-2 is largely dispensable for plasma membrane targeting, it does appear to play a role in the trafficking of RGS4 through different intracellular membrane compartments, and thus its palmitoylation status may be important for regulating its normal trafficking and recycling within the cell.

EXPERIMENTAL PROCEDURES
Materials-The pEYFP-C1 plasmid (BD Biosciences Clontech, Mississauga, ON) and pQE Trisystem1 (Qiagen, Mississauga, ON) expression vectors were used for expression of all RGS protein constructs. HA-tagged G q (R183C) was generously provided by P. Waedegartner (Thomas Jefferson University, Philadelphia, PA). Fluorescently tagged versions of the trans-Golgi network marker protein TGN38 were kind gifts from J. Lippincott-Schwartz (National Institutes of Health, Bethesda, MD). The monoclonal anti-His tag antibody was from Clontech (#631212), anti-HA antibody was from Roche Applied Science, Mississauga, ON, (#11666606001), and the anti-mouse secondary antibody was from GE Healthcare (#NXA931). All tissue culture media and transfection reagents were from Invitrogen and Roche Applied Science respectively. HEK293 cells (tsA-201 derivative) were a kind gift from Z-P. Feng (University of Toronto, Toronto, ON). Unless otherwise stated all other reagents and chemicals were from Sigma.
RGS4 Expression Constructs-For subcellular localization studies RGS4-YFP expression plasmids were generated in the pEYFP-C1 vector by cloning the human RGS4 cDNA into the NheI/AgeI sites to generate a carboxyl-terminal YFP fusion. Robust expression was ensured by inclusion of an optimized translation initiation signal in the context of the first methio-nine codon (GCCACCATGGCG). For functional studies RGS4-His expression constructs were generated in the pQE Trisystem1 vector by cloning RGS4 coding sequences into the NcoI/BamHI polylinker sites. Cysteine point mutations were introduced by site-directed mutagenesis, and primer sequences are available as supplemental data. All plasmid constructs were purified using the Endofree Maxi kit (Qiagen) and verified by sequencing of the complete protein-coding region. The expression of the pQE Trisystem clones were analyzed by Western blotting using anti-His-tag antibody and ECL detection.
Assays of G q -dependent Phosphoinositide Hydrolysis-HEK293 cells (2.0 ϫ 10 6 in 6-well plates) were transfected in triplicate using of FuGENE 6 TM transfection reagent according to the manufacturer's specifications. DNA amounts used in transfection were determined empirically to ensure similar expression levels of all RGS4 constructs tested. Specifically we used 0.5 g of G q (R183C), 3 g of RGS4WT-His, 0.5 g of RGS4C2A-His, 3 g of RGS4C12A-His, 0.5 g of RGS4C2AC12A-His, 1 g of RGS4L23D-His, and an appropriate amount of empty HisStrep vector to ensure 4.0 g of total plasmid DNA/well. After 24 h cells were trypsinized, pooled, and replated in 6-well plates for analysis of inositol phosphate production (n ϭ 3 wells/construct) and corresponding protein expression determination by Western blotting. Inositol phosphate production was measured 48 h after transfection as described previously (14). Briefly, 5 h after plating, cells were washed with phosphate-buffered saline and labeled in complete Dulbecco's modified Eagle's medium (without inositol) containing 10 mM LiCl and 1.4 Ci/ml myo-[ 3 H]inositol (PerkinElmer Life Sciences) for exactly 15 h. Then cells were washed with phosphate-buffered saline, and the inositol phosphate production was stopped with 750 l of ice-cold 20 mM formic acid. The entire contents of the wells were collected and spun at 13,000 ϫ g for 15 min in a microcentrifuge at 4°C. The supernatant fraction (700 l) was neutralized with 214 l of 0.7 M NH 4 OH before proceeding to the ion exchange chromatography steps. For each well to be measured, a 3-ml Dowex resin (AG 1-X8, 200 -400-mesh, formate form) column was prepared. The entire sample was added to the column, and the unbound 3 H-labeled inositol-containing fraction was collected for determination of total 3 H-inositol loading, while the inositol phosphate-containing fraction was eluted into collection tubes using 5 ml of 1.2 M ammonium formate. 0.5 ml of each sample from total inositol-containing and inositol phosphate-containing fractions was added to 10 ml of scintillation fluid and counted. Inositol phosphate levels were expressed as the fraction of the total soluble 3 H-labeled inositol material (inositol phosphate total/inositol-containing fraction) for each sample.
Confocal Microscopy-For most experiments HEK293 cells were plated at 50% in tissue culture-treated microscopy dishes (Ibidi, #81156) and transfected with 1 g of each construct to be tested using FuGENE 6 transfection reagent as describe above. For localization of RGS4 during phosphoinositol hydrolysis experiments, constructs were transfected in the identical ratios that were used for functional analysis (outlined above). After a 20-h incubation, dishes were sent for confocal microscopy to determine their plasma membrane/cytosol localization ratio containing transfected cells. Confocal microscopy was per-formed on ϳ70% confluence live cells at 37°C using an Olympus FluoView 1000 laser-scanning confocal microscope. Images represent a single equatorial plane on the basal side of the cell obtained with a 60ϫ oil objective, 1.4 numerical aperture. Confocal images were processed with Microsoft Office 2010. Quantification of plasma and endo-membrane localization was performed in a blinded manner, with membrane/cytosol ratios measured using the Image J software package and Pearson correlation coefficient (PCC) for each endosome was calculated by the Fluo-View software. For movie data, the cells were visualized on a WaveFX Spinning-Disk confocal microscope (Quorum Technologies, Guelph, Canada), which comprises an Olympus IX81 microscope stand, a Yokogawa CSU10 spinning-disk unit, and a Hamamatsu C9100 -13 EM-CCD camera, all controlled with Volocity software. Imaging was performed using a 60ϫ/1.42NA oil immersion objective lens, 488-nm solid-state laser illumination, and an EGFP bandpass filter.
Palmitate Labeling and Detection by Click Chemistry-10 ϫ 10-cm plates of HEK293 cells stably transfected with either wild type or the C2A/C12A mutant of RGS4 were plated to ϳ70% confluence. To increase final RGS4 protein yield, each plate was transiently transfected with the same RGS4 clone that was used to make the stable lines (RGS4WT (6 g) or C2AC12A (2 mg) in 6 l of X-tremegene HP, Roche Applied Science). The RGS4 expression constructs contained a carboxyl-terminal streptavidin-binding peptide tag. 12 h post-transfection medium was changed to 1:1 DMEM:F-12 with 5% charcoal-treated fatty acid-free serum. After 24 h cells were serum-starved for 1 h in DMEM:F-12 and then incubated for 8 h in palmitic acid labeling medium (DMEM containing 10% charcoal-treated serum, 0.4% of 25 mM alkyl-17-ODYA palmitate analog). Cells were collected and lysed at 4°C using lysis buffer containing 20 mM HEPES, 150 mM NaCl, 3 mM MgCl 2 , 1% Triton-X, pH 7.4, and protease Complete Mini TM protease inhibitors (Roche Applied Science). Samples were briefly sonicated, and lysates were incubated with streptavidin-Sepharose High Performance beads (GE Healthcare) overnight at 4°C. Beads were washed 5 times with lysis buffer and once with acidic lysis buffer, pH 4.0. For click chemistry reaction, beads were mixed with lysis buffer containing 1 mM CuSO 4 , 1 mM Tris(2-carboxyethyl)phosphine, 100 M tris(benzyltriazolylmethyl)amine, and 20 M Alexa-488 azide (Invitrogen) at room temperature for 1.5 h. After the click-mediated addition of Alexa-488 to palmitoylated proteins, the beads were washed 3 times with lysis buffer, and RGS4 eluted in SDS-PAGE loading buffer before running on 12% gels. Gels were fixed in 40% methanol, 10% acetic acid for in-gel fluorescent detection of in Alexa-488 on the Typhoon imaging station. The relative amount of RGS4 protein in the eluates was examined by Western blotting for the HA epitope tag.
Western Blotting-Proteins were transferred to (Trans-Blot, Bio-Rad) nitrocellulose membrane. Membranes were blocked for 1 h in 0.1% Tween 20 and 5% bovine serum albumin. Primary antibodies were added to 5% BSA at concentrations provided by the vendor's instructions and incubated with membranes overnight at 4°C before removing by washing. Horseradish peroxidase-linked secondary mouse or rabbit antibody in 5%BSA was added for 2 h before washing and signal detection using Super Signal West Pico Chemiluminescent Substrate (Thermo Scientific).
Statistical Analysis-One-way ANOVA with a Tukey's post hoc test was used to determine statistical significance between the groups. Where correlation coefficients were compared Fisher's r to zЈ transformation was employed. *, p Ͻ 0.05.

RESULTS
The amino-terminal domain of RGS4 was previously shown to be important for its plasma membrane targeting in yeast cells (9,10). This domain contains an amphipathic ␣ helical domain and two palmitoylatable cysteine residues (Cys-2 and Cys-12) as potential plasma membrane targeting motifs. In a heterologous yeast system, the amphipathic helix domain of RGS4 was shown to be necessary and sufficient for plasma membrane targeting and G-protein inhibition. These data suggested a diminutive role for amino-terminal palmitoylation in regulating RGS4 function in lower eukaryotes. In a more autologous human cell line, however, the determinants that control human RGS4 plasma membrane targeting, function, and intracellular trafficking have not been carefully examined. In particular, very little is known about the intracellular trafficking of RGS4 between plasma membrane and endosomal or cytosolic pools.
Wild type RGS4-YFP strongly localized to the plasma membrane as well at bright endosome-like structures (ϳ40% of HEK293 cells examined) in laser-scanning confocal microscopy experiments (Fig. 1A). Spinning-disc confocal analysis of the wild type clone also revealed a high level of RGS4-containing endosome activity including the slower-moving structures described above and numerous less fluorescent rapidly moving endosomes that were not readily discernible by the laser-scanning technique (supplemental Movie 1). Mutation of Leu-23 within the amphipathic helix domain (L23D) resulted in complete redistribution of RGS4 from the plasma membrane to the cytosol and an apparent loss of RGS4 localization to the endosomal compartment. Spinning-disc confocal microscopy of L23D confirmed loss of RGS4 within both the bright slow moving and rapidly moving endosome pools. (supplemental Movie 2).
Compared with controls, expression of G q (R183C) markedly increased the production of inositol phosphates in HEK293 cells. Co-expression of wild type RGS4 with G q (R183C) resulted in an ϳ80% reduction in inositol phosphate production. Consistent with the notion that proper plasma membrane targeting is required for its function, the mislocalized L23D mutant was unable to inhibit G q (R183C)-mediated inositol phosphate production compared with wild type RGS4 (Fig. 1B). Thus, the L23D mutant could be used as a nonfunctional RGS transfection control for other mutants to be analyzed.
We focused first on whether palmitoylation of the amino terminus was an important determinant of its plasma membrane localization and function in mammalian cells. Indeed, inhibition of endogenous palmitoyl-CoA transferases with 2-bromopalmitate (2-BP) resulted in marked redistribution of RGS4 from the plasma membrane to the cytosol ( Fig. 2A). To determine the potential contribution of Cys-2 and Cys-12 palmitoylation to the membrane localization of RGS4, we studied the localization of individual (C2A or C12A) and combined (C2A/C12A) alanine mutations. Supporting the notion that palmitoylation of the amino terminus is a key determinant of human RGS4 localization in human cells, the combined mutation of Cys-2 and Cys-12 dramatically decreased plasma membrane:cytosol localization ratio (Fig. 2, B and C). Indeed, the extent of mislocalization of the combined mutant approached that of the L23D clone. Evidence that a loss of palmitoylation may explain the mislocalization of the mutant RGS4 protein came from palmitate-labeling studies, where introduction of the C2A/C12A mutant all but abolished incorporation of the palmitic acid analog 17-octadecynoic acid (Fig. 2D). Also of interest was the observation that, like L23D, the C2A/C12A mutant show dramatically reduced endosomal localization when it was examined by spinning disk microscopy (supplemental Movie 3). Surprisingly, mutation of Cys-12 alone (C12A), but not Cys-2 (C2A), was sufficient to disrupt the plasma membrane targeting of RGS4. Notably, the effect of the cysteine mutations on the plasma membrane:cytosol ratio was similar in both the absence (Fig. 2) and the presence (Fig. 3) of coexpressed G q R183C. It did appear, however, that active G q modestly increased the plasma membrane:cytosol ratio of all the wild type and cysteine mutants, presumably via its ability to recruit some RGS4 from the cytosolic pool.
Functional studies for the RGS4 cysteine mutant series mostly paralleled the plasma membrane localization data. Specifically, compared with the wild type protein, the mislocalized C12A and C2A/C12A proteins showed virtually no G q inhibitory function, whereas the C2A protein retained some G q inhibitory function (Fig. 4). In an attempt to explain the differential importance of Cys-2 and Cys-12 with respect to RGS4 localization and function, we used a helical net projection to map the positions of Cys-2 and Cys-12 relative to the stretch of amino acids in RGS4 containing its amphipathic helix domain (amino acid residues 12-30) (Fig. 5). Based on the relative proximity of Cys-12 to this helix domain and its position immedi- HEK293 cells were co-transfected with vector control or constitutively active G q (R183C) and the indicated HisStrep-tagged RGS4 construct. Relative expression levels of RGS4-HisStrep proteins and G q (R183C) were determined by immunoblotting. After overnight labeling, inositol phosphate production was measured as described under "Experimental Procedures." Values indicate absolute inositol phosphate/total soluble inositol ratios and are the means of five independent experiments each performed in triplicate. Raw cpm data are presented in supplemental Table IA. S.E. are indicated by error bars. One-way ANOVA with a Tukey's post hoc test was used to determine differences between groups (*, p Ͻ 0.001).

FIGURE 2. Defining the effect of palmitoylation and amino-terminal cysteine residues mutation on the plasma membrane targeting of RGS4.
A, localization of the wild type RGS4-YFP construct in the presence and absence of the palmitoyl-CoA transferase inhibitor 2-BP was examined by confocal microscopy as described above. B, localization of different YFPtagged cysteine mutants of RGS4 fusion constructs in HEK293 cells is shown. Scale bars represent 1 m. C, the ratio of the RGS4-YFP signal between the cytosol and plasma membrane was analyzed by densitometry using ImageJ software. Shown are the means ratio of n Ͼ 80 cells. S.E. are indicated by error bars. D, ODYA-17 (palmitic acid analog) labeling of wild type and cysteine mutants of RGS4 is shown. The upper panel shows the extent of palmitoylation labeling in the indicated RGS4 constructs as measured by epitope-tag pulldown, click chemistry adduction of Alexa-488 and in-gel fluorescence. The control sample is from HEK cells not transfected with RGS4. The lower panel shows Western blot analysis of the pulldown eluates analyzed in the panel above. Where necessary, one-way ANOVA with a Tukey's post hoc test was used to determine differences between groups (*, p Ͻ 0.01). AUGUST 17, 2012 • VOLUME 287 • NUMBER 34

JOURNAL OF BIOLOGICAL CHEMISTRY 28969
ately adjacent to the putative membrane-binding surface, it seems likely that palmitoylation of Cys-12 would have a greater impact on the length of the hydrophobic interface of the helix than would palmitoylation of Cys-2. Previous work on the RGS2 protein showed that the length of the aliphatic interface had a profound effect on its plasma membrane localization and G q inhibitory function (15).
Because palmitoylation regulates intracellular trafficking of multiple components of the G-protein-coupled receptor signaling pathways including G-protein ␣ subunits (16), RGS proteins (17), and some RGS protein-binding proteins (18), we next examined the extent to which palmitoylation on Cys-2 or Cys-12 affects the intracellular trafficking of RGS4. The spinning-disk confocal data discussed above provided preliminary evidence that mutation of Cys-2 and Cys-12 may alter endosomal localization of RGS4. We next used laser scanning confocal to more carefully quantify this effect. RGS4-YFP was observed in discreet intracellular compartments in ϳ 40% of the cells examined (Fig. 6). The addition of 2-BP dramatically reduced targeting of RGS4 to the endosome compartment (data not shown). Notably, Cys-2 and Cys-12 seem to contribute dif-ferentially to the distribution of RGS4 within the intracellular pool. Mutation of Cys-12 reduced the percentage of cells showing endosomes to ϳ15%, whereas mutation of Cys-2 nearly completely disrupted RGS4 localization to endosomal structures (Fig. 6).
G-protein ␣ subunits cycle between the plasma membrane and the Golgi compartment. Palmitoylation at the level of the Golgi is a key step in this process (16). We wondered whether the Cys-2 and Cys-12 palmitoylation sites on RGS4 may likewise be a component of its cycling between the plasma membrane and Golgi. Indeed RGS4 was reported to be associated with both the plasma membrane and Golgi compartments (19). We thus examined the potential for RGS4 to colocalize with TGN38, a trans-Golgi compartment marker known to traffic constitutively between the Golgi and plasma membrane (20). In cells expressing TGN38-CFP, it is localized primarily to the trans-Golgi region as expected, but its expression was also observed in endosomal structures proximal to the Golgi. Although strong trans-Golgi colocalization was not observed for wild type RGS4-YFP, the fluorescent endosomes containing RGS4 frequently coincided with vesicles containing TGN38-CFP (Fig. 7A). Indeed, when the entire RGS4-YFP endosome population was examined for its colocalization with TGN38-CFP, a PCC of 0.32 was obtained. Consistent with a differential role for Cys-2 and Cys-12 in the intracellular trafficking of RGS4, endosomes containing the C12A mutant showed a high extent of colocalization with TGN38-CFP (PCC ϭ 0.75), whereas endosomes containing C2A or the double mutant C2A/C12A were poorly colocalized with the TGN38 pool (PCC  G q (R183C). A, localization of different YFP-tagged cysteine mutants of RGS4 fusion constructs in HEK293 cells in the presence of co-expressed G q (R183C) was examined by confocal microscopy as described above. Scale bars represent 1 m. B, the ratio of the RGS4-YFP signal between the cytosol and plasma membrane was analyzed as above (n Ͼ 30). One-way ANOVA with a Tukey's post hoc test was used to determine differences between groups. S.E. are indicated by arrow bars (*, p Ͻ 0.05). FIGURE 4. Mutation of Cys-2 and Cys-12 exert differential effects on RGS4 G q inhibitory activity. Inositol phosphate production was measured using [ 3 H]myoinositol labeling from triplicate wells for each transfection condition. Briefly, HEK293 cells were co-transfected with constitutively active G q (R183C) and the indicated HisStrep-tagged RGS4 construct. Relative expression levels of RGS4-HisStrep proteins and G q (R183C) were determined by immunoblotting. After overnight labeling, inositol phosphate production was measured as described under "Experimental Procedures." Values indicate the mean inositol phosphate/total soluble inositol ratio relative to that for the internal RGS4-inactive control (L23D) and are the mean of five independent experiments performed on separate days. Raw cpm data are presented in supplemental Table IB. S.E. are indicated by error bars. NS, not significant. One-way ANOVA with a Tukey's post hoc test was used to determine differences between groups (*, p Ͻ 0.05).
values of Ϫ0.1 and Ϫ0.3, respectively) (Fig. 7B). Together, these data suggest that putative palmitoylation of Cys-2 or Cys-12 differentially affects the trafficking of RGS4 along endosome recycling circuits that normally allow the traffic of RGS4 and other similarly recycled proteins such as TGN38 between the plasma membrane and Golgi.
The potential significance of the difference in TGN38 colocalization between wild type RGS4-and C12A-containing endosomes was examined in Fig. 8. When scatter plots of endosome colocalization coefficient versus diameter were examined, two interesting patterns emerged. First, the wild type protein has a vastly different endosome size distribution profile compared with C12A. In particular, the wild type protein is observed much more commonly in small sized (Ͻ1 m) endosomes than is the C12A protein (73 versus 35% of total endosomes respectively). Secondly, when the size distribution of endosomes with high and low TGN38 colocalization coefficients is compared, we observe that for the wild type clone there is a large (39% of total endosomes) pool of small sized endosomes that does not colocalize with TGN38. The mutation of Cys-12 seems to prevent the localization of RGS4 to this smallsized TGN38-deficient pool as most Cys-12 containing endosomes contain TGN38 irrespective of their size (Fig. 8). Taken together, these data are consistent with a model where palmitoylation at either Cys-2 or Cys-12 can differentially regulate the intracellular trafficking of RGS4. Mutation at Cys-2 almost completely eliminates its endosomal localization, whereas mutation at Cys-12 impairs its localization to TGN-38-refractory endosomes without affecting its targeting to a TGN38 containing endosomal compartment.

DISCUSSION
Over the last decade palmitoylation has been shown to be a key regulator of signaling protein localization and function. Specifically, this reversible post-translational modification of cysteine residues regulates the intracellular trafficking of proteins such as Ras, G␣ i , G␣ q , and Src-family kinases and helps promote their proper intracellular trafficking (19,21). Potential sites of palmitoylation are often clustered together within protein domains, and in many cases they are believed to work in concert to promote proper intracellular targeting. Such is believed to be the case for the tyrosine kinases Fyn, Yes (21), and Shown is a two-schematic representation of the ␣ helical RGS4 membrane association domain. Arrows denote putative palmitoylation sites (Cys-2 and Cys-12) in the RGS4 amino terminus. Shading indicates the length of the hydrophobic surface on the amphipathic ␣-helix of RGS4. Aliphatic and nonpolar aromatic residues are shown as black, polar residues are in white, and palmitoylated cysteine residues are indicated by a filled black box surrounding the residue. Note how Cys-12 palmitoylation may be predicted to increase the length of the hydrophobic surface by at least one turn of the helix. Lck (22) that undergo palmitoylation at cysteine residues 3 and 6 in their amino-terminal targeting domain (23). It has been similarly suggested that a pair of closely spaced cysteines (Cys-2 and Cys-12) found in the amino-terminal domains of RGS4, RGS5, and RGS16 may work in concert to promote their proper localization and G-protein inhibitory function (24). We show here the surprising result, however, that Cys-2 and Cys-12 appear to act independently to modulate plasma membrane targeting, intracellular trafficking, and function of RGS4 in mammalian cells.
We set out to characterize the relative contribution of determinants within RGS4 that are necessary for its plasma membrane targeting and function in mammalian cells. Consistent with previous studies that showed the relative importance of the amphipathic helical domain (9); the L23D mutant remained completely cytosolic with no detectable membrane localization. It was interesting to observe that inhibition of RGS4 palmitoylation by 2-bromoplamitate also results in a complete disruption of plasma membrane targeting. Thus, the amphipathic helix domain and cysteine palmitoylation both appear necessary but not sufficient for optimal plasma membrane targeting of human RGS4 in human cells. Together these data indicate that the G-protein binding domain (RGS box) with its previously reported palmitoylation site at Cys-95 may not be a primary determinant of RGS4 plasma membrane localization. However, the observation that C2A/C12A shows low, but appreciable levels of tonic membrane localization compared with L23D and WT after 2-BP addition suggests that Cys-95 may contribute a minor component of RGS4 membrane targeting capacity.
The data herein show that Cys-12, the palmitoylatable residue adjacent to the amphipathic helix domain, is the most important cysteine residue for plasma membrane targeting of RGS4. By contrast, Cys-2 residue appears to be dispensable for membrane localization of RGS4. The simplest explanation for these data is that palmitoylation of Cys-12 increases the length of the hydrophobic surface of the amphipathic ␣ helix to promote greater steady-state association with the inner leaflet of the plasma membrane. In an analogous situation, an extended hydrophobic surface of the RGS2 amphipathic helix was shown to be critical for its strong localization to the cell membrane (15). Circular dichroism studies have shown that peptides containing the amphipathic helical domain of RGS4 maintain a disordered (random coil) structure until presented with lipid vesicles whose phospholipid composition resembles the anionic plasma membrane inner leaflet (9). Whereas Cys-12 palmitoylation may be able to cooperate with the adjacent hydrophobic amino acids to promote formation of an extended helix domain (one additional turn), it may be that Cys-2 palmitoylation is located at too great a distance from the core helix to promote extended helix formation. It is prudent, however, to consider the alternative possibility that Cys-12, in its palmitoylated or non-palmitoylated state, works independently from the amphipathic helix to promote association with an unknown plasma membrane protein or protein complex. Consistent with the notion that membrane targeting is necessary for RGS4 function, mutation of Cys-12 in the context of both the C12A and C2A/C12A reduced RGS4 G q inhibitory function to a level similar to that of the L23D mutation. Functional data for the C2A mutant, however, suggest that membrane localization alone may not be sufficient for optimal RGS4 activity. Specifically, despite the strong membrane localization of the C2A, this clone also showed a measurable decrease in its G q inhibitory function, suggesting that Cys-2 may contribute to an additional previously uncharacterized functional activity to RGS4. It is important to note that this novel activity appears to be independent from the known protein-stabilizing effects of Cys-2 palmitoylation (11, 12) as protein expression levels were nor-FIGURE 7. Colocalization of RGS4-containing endosomes with TGN38 is differentially affected by Cys-2 and Cys-12. Localization of RGS4-containing endosomal structures with the trans-Golgi-endosome marker TGN38 (TGN38-CFP) was examined by co-transfection in HEK293 cells followed by fluorescent microscopy of live cells. A, WT, upper and lower panels highlight the existence of both strongly (cell 1) and poorly (cell 2) colocalized endosomes in different cells). B, C2A typically showed poorly colocalized endosomes. C, C12A typically showed well colocalized endosomes. Using the excitation and emission discrimination capabilities of the Olympus FV1000 confocal microscope, RGS4 (red pseudocolor) and TGN38 (green pseudocolor) images were collected from the same confocal plane. Merged images indicate areas of potential colocalization (shown in yellow). Scale bars represent 1 m. D, PCCs for RGS4-containing endosomes with TGN38 expression were determined using Olympus Fluo-View software. Shown are the mean PCC values (WT, n ϭ 84; C12A, n ϭ 70; C2A, n ϭ 12; C2AC12A, n ϭ 12) pooled from four independent experiments. Fisher's r to zЈ transformation was employed to determine differences between correlation coefficients of RGS4WT and C12A. *, p Ͻ 0.0001. S.E. are indicated by error bars. malized at the transfection stage for these experiments. Instead, it may be that Cys-2 in its palmitoylated or non-palmitoylated states promotes interaction between RGS4 and Homer2 (25) or a similar docking protein that might increase its G-protein inhibitory activity. It will be interesting to determine the residues on RGS4 that are required for its interaction with Homer2 and whether palmitoylation enhances this interaction.
Many palmitoylated proteins shuttle continuously between the plasma membrane and other intracellular compartments (16, 21, 26 -28). Because 2-BP treatment of RGS4-transfected cells completely prevented its localization to the endosomal compartment, we asked whether Cys-2 and Cys-12 may contribute to endosomal trafficking and localization of RGS4. The C2A mutant showed dramatically impaired localization to the endosome population consistent with the idea that endosome localization is dependent on Cys-2 palmitoylation. If endosome localization is required for a posttranslational modification or protein interaction that serves to increase RGS4 function, then its impaired localization to the endosome pool could explain its impaired inhibition of G q signaling. It will be of future interest to determine whether RGS4 activity promoting modifications such as PKG/PKA-dependent phosphorylation (29), Ca 2ϩ /calmodulin binding (30), spinophilin (31), and/or Homer2 interaction (25) require recycling of RGS4 through an endosomal compartment. By contrast, the C12A mutation had a quite distinct effect on endosomal localization. Although this mutant targeted TGN38-containing endosomes with a similar efficiency as the wild type protein, it was completely absent from the TGN38-deficient pool of small-sized endosomes that is normally populated by the wild type protein. To the extent that trafficking of RGS4 through endosome compartments that do not contain TGN38 may be important for its G-protein function, this observation might further serve to explain the lack of inhibitory activity that we observed for the C12A mutants.
The fact that the L23D amphipathic ␣-helix mutant did not populate either endosomal compartment is consistent with the possibility that these previously unreported structures are bounded by lipid bilayers. The nature of the compartment that is populated by both wild type and C12A mutant proteins is of much interest. Its high TGN38 expression and localization deep within the cytosol suggest that it may be an intracellular sorting compartment related to the trans-Golgi network. To our knowledge no such compartments have been previously described for fluorescently tagged TGN38 constructs. Clearly, Cys-12 is not required for localization to this endosomal structure. Taken together these data are consistent with a model whereby sequential palmitoylation of Cys-2 and Cys-12 may be required to promote proper trafficking of RGS4 between the plasma membrane, Golgi, and endosomal compartments. Specifically, Cys-2 palmitoylation would allow access to the endosomal TGN38-containing pool, and Cys-12 palmitoylation presumably occurring within that compartment would be required for exiting that TGN38 containing pool. The precedent for this type of sequential trafficking model are proteins such as H-or N-Ras where palmitoylation was shown to be required for them to migrate out of the Golgi (32,33). This model would also suggest that the different Golgi-associated and endosomal compartments through which RGS4 and other palmitoylated proteins traffic may contain palmitoyl transferases (DHHC family members) with different substrate specificities or compartmentalizations to facilitate efficient movement of proteins from one compartment to another.
In summary we show here for the first time that two putative palmitoylation sites in the RGS4 amino terminus, Cys-2 and Cys-12, may have unique and complementary activities with respect to mediating intracellular localization and G q inhibitory function. Cys-12 appears to be more important for plasma membrane targeting and endosomal trafficking, whereas Cys-2 appears more important for trafficking within intracellular pathways. It will, therefore, be of future interest to characterize the effects of the two palmitoyl-CoA transferases known to palmitoylate RGS4 (DHHC3 and DHHC7) on its localization trafficking and function to determine whether there may exist a specificity of these enzymes for Cys-2 or Cys-12. Moreover, extending the characterization of the endosomal pools of RGS4 might be critical for understanding their importance in the functionality of RGS4. Last, given the similarity of the aminoterminal domains of RGS5 and RGS16, it will be of interest to determine whether Cys-2 and Cys-12 play similar complimen-FIGURE 8. Distribution of RGS4-containing endosomes by endosome size and TGN38 colocalization coefficient. Scatter plot for RGS4-containing endosomes comparing diameter and extent of colocalization with TGN38 colocalization (PCC). Plotted data were only available for WT and C12A, as C2A and C2A/C12A constructs localized very poorly to the endosome pool. The plot represents pooled data from four independent experiments. Each data point represents a single endosome. All RGS4-containing endosomes identified by microscopy were examined for their colocalization with TGN38. RGS4-containing endosomes were arbitrarily sorted into strongly (PCC Ͼ 0.4) and weakly (PCC Ͻ 0.2) TGN38-colocalized pools. Pool distribution profiles of RGS4-containing endosomes varied greatly between the WT and C12A constructs. AUGUST 17, 2012 • VOLUME 287 • NUMBER 34 tary roles in the regulation of these closely related RGS protein family members.