Palmitoylation Regulates Regulator of G-protein Signaling (RGS) 16 Function

Palmitoylation is a reversible post-translational mod-ification used by cells to regulate protein activity. The regulator of G-protein signaling (RGS) proteins RGS4 and RGS16 share conserved cysteine (Cys) residues that undergo palmitoylation. In the accompanying article (Hiol, A., Davey, P. C., Osterhout, J. L., Waheed, A. A., Fischer, E. R., Chen, C. K., Milligan, G., Druey, K. M., and Jones, T. L. Z. (2003) J. Biol. Chem . 278, 19301–19308), we determined that mutation of NH 2 -terminal cysteine res- idues in RGS16 (Cys-2 and Cys-12) reduced GTPase accelerating (GAP) activity toward a 5-hydroxytryptamine (5-HT 1A )/G (cid:1) o1 receptor fusion protein in cell mem- branes. 2 -terminal acylation also permitted palmito- ylation of a cysteine residue in the RGS box of RGS16 (Cys-98). Here we investigated the role of internal palmitoylation in RGS16 localization and GAP activity. Mutation of RGS16 Cys-98 or RGS4 Cys-95 to alanine reduced GAP activity on the 5-HT 1A /G (cid:1) o1 fusion protein and regulation of adenylyl cyclase inhibition. The C98A mutation had no effect on RGS16 localization or GAP activity toward purified G-protein (cid:1) subunits. Enzymatic palmitoylation of RGS16 resulted in internal palmitoylation on residue Cys-98. Palmitoylated RGS16 or RGS4 WT but not C98A or C95A preincubated with membranes expressing 5-HT 1a /G (cid:1) o1 displayed increased GAP activ- ity over time. These results suggest that palmitoylation of a Cys residue in the RGS box is critical for RGS16 and RGS4 GAP activity and their ability to regulate G i -cou- pled signaling in mammalian cells.

The RGS protein family is large (more than 30 proteins in mammalian cells), and its members share high sequence homology within the conserved RGS domain that confers GAP activity. Regulation of RGS proteins may determine their cellto-cell specificity and preference for certain G-protein ␣ subunits or G-protein-coupled receptor-linked signaling pathways. Modulation of RGS activity may occur by post-translational modifications such as phosphorylation (4 -15) or palmitoylation (16 -23). Indeed, palmitoylation of many G-protein signaling pathway components, from the G-protein-coupled receptors itself to the RGS protein, affects their activity (24 -27).
Although the NH 2 and COOH termini were not observed in the crystal structure of RGS4 (28), it is clear that in the case of both RGS4 and RGS16, the amino terminus is also required for function (29 -32). A short amphipathic ␣-helical region in the NH 2 -terminal region of RGS16, conserved in both RGS4 and RGS5, imparts membrane localization (17,32). In a previous study, we found that RGS16 was palmitoylated on Cys-2 and Cys-12 at the NH 2 terminus (18). Mutation of Cys-2 and Cys-12 to alanine, and the resulting loss of palmitoylation on these residues, reduced the inhibition of both G i -and G q -coupled signaling normally observed in cells expressing RGS16. In the accompanying article (33), we observed that mutation of Cys-2 and Cys-12 virtually eliminated palmitate incorporation into RGS16, reduced GAP activity toward G␣ in a membrane-based assay, and mistargeted RGS16 away from lipid rafts. However, this abnormal localization could not explain the lack of RGS16 function, as disruption of lipid rafts with methyl-␤-cyclodextrin increased membrane GTPase activity in the absence of RGS16 while preserving the ability of transfected RGS16 to further increase the agonist-evoked GTPase rate. Instead, we found that NH 2 -terminal palmitoylation permitted palmitoylation of RGS16 on an internal cysteine residue in the RGS box (Cys-98), possibly by retaining the protein in the lipid rafts, which are enriched in protein acyltransferase activity.
Because previous studies have implicated a role for palmitoylation of an internal cysteine residue in the GAP activity of other RGS proteins (19), we investigated palmitoylation of RGS16 at Cys-98 to determine its importance for membrane localization, protein-protein interactions, and enzymatic activity. We found that mutation of this Cys to alanine reduced * The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
ʈ To whom correspondence should be addressed: Bldg. 10 RGS16 GAP activity toward the 5-HT 1A /G␣ o1 fusion protein in membranes but did not substantially alter membrane or lipid raft localization. The analogous mutation in RGS4 resulted in a similar reduction in GAP activity. Most importantly, direct enzymatic palmitoylation of RGS16 or RGS4 by a protein acyltransferase followed by membrane preincubation markedly increased GAP activity over time. The C98A mutation also abrogated the ability of RGS16 or RGS4 to regulate G i -mediated adenylyl cyclase inhibition, confirming the significance of this internal palmitoylation site for RGS function in mammalian cells.
Immunoblotting-Cell lysates were resolved on 10 -20% Tris glycine gels and transferred to polyvinylidene difluroride membranes. After a 2-h blocking step in Tris-buffered saline plus 0.01% Tween 20 ϩ 10% milk, membranes were incubated with primary antibody (CT265, 1:1000) or anti-HA (12CA5, Roche Diagnostics) for an additional 2 h. Blots were then equilibrated with horseradish peroxidase-conjugated goat anti-rabbit or anti-mouse IgG for 1 h. Signal was detected by enhanced chemiluminescence (Super Signal, Pierce) according to the manufacturer's instructions.
Adenylyl Cyclase and Single Turnover GAP Assays-These assays were performed exactly as described previously (11).
Plasma Membrane and Detergent-resistant Membrane (DRM) Fractionation, High Affinity GTPase Assays-These procedures were carried out as described in the accompanying article (33).
Purification of Protein Acyltransferase-Purification of pPAT was performed as described in detail elsewhere. 2 Briefly, the plasma membrane from rat livers was extracted with 0.15% Triton X-100 and the PAT activity was purified over several chromatography steps, octyl-Sepharose, cerulenin analogue affinity, Q-Sepharose, and palmitoyl CoA-agarose.
Enzymatic Palmitoylation of RGS Proteins-Recombinant RGS16 or RGS4 was diluted in incubation buffer (IB) (20 mM Tris-HCl, pH 7.4, 150 mM KCl, 1 mM EDTA) at a protein concentration of 0.4 -1 mg/ml and incubated with the pPAT preparation (10 g of protein), 200 M CoA, 2 mM ATP, and 1 l of [9,10-3 H]palmitate (5 mCi/ml, 30 -60 Ci/mmol, ARC Inc.). The final reaction volume was adjusted to 100 l with IB. The reaction was stopped by the addition of sample buffer after 45 min at 30°C. Proteins were subjected to one-dimensional PAGE and stained by MicrowaveBlue (Protiga). Acylation was determined by fluorography using the Wax technique (ISC BioExpress) following the instructions of the manufacturer.
Stoichiometry of Palmitoylation-Recombinant RGS16 (260 pmol) incubated with pPAT was diluted into IB buffer that contained 0.1% bovine serum albumin and 1% sucrose. The solutions were loaded into an Ultrafree filtration unit (Amicon) previously equilibrated with IB. The samples were centrifuged at 2000 ϫ g for 20 min and the filters washed once with 400 l of 70% ethanol, then twice with IB. The insert cups were counted by liquid scintillation spectrometry, and palmitate incorporation was calculated based on observed counts and the specific activity of [ 3 H]palmitate. Stoichiometry of labeling was determined as the ratio of picomole of palmitate to picomole of RGS16 per reaction.
Clostripain Cleavage-This procedure was carried out as in the accompanying article (33) except that recombinant RGS16 (10 g) was used as the substrate.
Statistical Analysis-Two tailed p values were determined by oneway analysis of variance followed by Tukey-Kramer multiple comparisons test using GraphPad Instat software. p values Ͻ0.05 were considered significant. Sigma Plot 8.0 software was used for curve fitting.

RESULTS
A Cysteine Residue in the RGS Box Is Critical for RGS16 GAP Activity-We tested the role of the conserved RGS box cysteine for the function of RGS16 and because it appeared to be a site of palmitoylation (19,33). We transfected a cell line stably expressing a 5-HT 1A /G␣ o1 (C351G) fusion protein with HA-RGS16 WT or RGS16 (C98A) plasmids. Cells were treated with pertussis toxin to block receptor coupling to endogenous G␣ i/o proteins, and membranes were prepared. Both basal and agonist-induced GTPase activity in membranes incubated with a range of 5-HT concentration were increased in the presence of WT RGS16 or compared with membranes transfected with a vector control plasmid (LacZ) as expected (37). However, GTPase activity in membranes expressing RGS16 (C98A) was decreased nearly 50% in comparison to membranes expressing WT RGS16 (FIG. 1A, upper panel). Addition of maximal 5-HT doses (10 Ϫ6 to 10 Ϫ4 M) resulted in a ϳ3.5-fold increase in activity in membranes containing WT RGS16 while the same dose resulted in only a 2.3-fold increase in activity in membranes containing RGS16 (C98A). Mutation of Cys-95 in RGS4 to alanine resulted in a similar reduction in GAP activity in this assay (FIG. 1A, lower panel). Decreased RGS16 or RGS4 levels in the membranes could not explain the loss of function of the cysteine mutants, because the amount of immunodetectable proteins was comparable (Fig. 1B). A previous study showed that the GAP activity of RGS4 was unaffected by mutation of Cys-95 to Val in a single turnover assay (19). To exclude the possibility that the reduction in transfected RGS16 (C98A) activity could be a result of mutation of cysteine to alanine independent of palmitoylation, we purified WT or C98A recombinant RGS16 from E. coli, which would not undergo posttranslational modification, and measured agonist-stimulated GTPase activity of the fusion protein in membranes in the presence of purified RGS16. Addition of WT RGS16 or C98A to membranes from cells expressing the 5-HT 1A /G␣ o1 protein enhanced GTPase activity equally in response to 5-HT even at subsaturating concentrations (Fig. 1C). In addition, we found no significant difference in the abilities of RGS16 WT or C98A to promote GTP hydrolysis by purified G␣ i in a single turnover assay (Fig. 1D). These results suggest that the loss of a cysteine thiol group was not by itself responsible for the decreased activity of RGS16 (C98A) expressed in mammalian cell membranes.
Cysteine 98 Is Essential for RGS16 and RGS4 Inhibition of G i -regulated Adenylyl Cyclase Activity-We tested the ability of the cysteine mutants to regulate G i signaling in HEK293 cells. Somatostatin decreases adenylyl cyclase activity through G␣ i activation. Because decreases in basal levels of adenylyl cyclase activity are difficult to detect, we treated the cells concurrently with isoproterenol, which stimulates adenylyl cyclase through endogenous ␤-adrenergic receptors coupled to G␣ s . Consistent with previous studies (18,38), expression of WT RGS16 or WT RGS4 inhibited the negative regulation of adenylyl cyclase activity induced by somatostatin compared with vector-transfected cells (Fig. 2). In contrast, expression of RGS16 (C98A), RGS16 (C2A/C12A), or the analogous RGS4 mutants did not alter adenylyl cyclase activity compared with control cells. These results demonstrate that neither mutant was able to regulate G i -coupled adenylyl cyclase inhibition effectively in HEK293 cells.
Cysteine 98 Is Not Required for RGS16 Plasma Membrane or Lipid Raft Localization-We delineated the membrane localization of RGS16 (C98A) by cellular fractionation and immunoblotting as previously described (33) to determine whether redistribution of the protein in membranes could account for its lack of function. The amounts of RGS16 WT or C98A found in the total membrane fraction (T) as well as the plasma membrane fraction (P) were comparable (Fig. 3A). The integrity of the fractions was verified by enrichment of the marker Na ϩ / K ϩ -ATPase in the plasma membrane fraction. We then assessed the distribution of RGS16 (C98A) within the membrane fraction by OptiPrep gradient centrifugation. We found RGS16 (C98A) in fraction 1, which contains DRMs, indicated by caveolin and G␣ i immunoreactivity (Fig. 3B). Because the distribution of the C98A mutant was similar to WT RGS16 (33), this result suggests that the C98A mutation did not substantially alter the distribution of RGS16 in the plasma membrane or its localization to lipid rafts.
Enzymatic Palmitoylation of RGS16 -The aforementioned results suggest that the poor function of RGS16 (C98A) was not because of the cysteine to alanine mutation itself or a decrease in RGS16 found at the membrane. Therefore, we investigated whether palmitoylation on the RGS box cysteine could affect function of RGS16 in vitro. In the accompanying article (33), we metabolically labeled cells expressing RGS16 WT or C98A with [ 3 H]palmitate and immunoprecipitated RGS16. We then treated immunoprecipitates with the protease clostripain, which was expected to yield a 5-kDa fragment containing Cys-98. Consistent with this prediction, we observed palmitate incorporation in a 5-kDa band only in immunoprecipitates of RGS16 WT but not C98A. To assess whether this residue underwent enzymatic palmitoylation directly, we chemically acylated recombinant RGS16 with partially purified PAT from rat liver membranes. Incubation of RGS16 WT or C98A with pPAT and [ 3 H]palmitoyl-CoA led to incorporation of tritium in a band with a molecular mass consistent with RGS16 (ϳ30 kDa, Fig. 4A, left panel), confirming that the NH 2 -terminal cysteines were the major sites of palmitoylation. RGS4 demonstrated a similar pattern of palmitoylation after incubation with the pPAT preparation (FIG. 4A, right panel). After clostripain cleavage of WT RGS16, we observed [ 3 H]palmitate incorporation in a 5-kDa band only in samples containing RGS16 WT but not C98A. Coomassie Blue staining revealed an identical pattern of cleavage of RGS16 WT, C2A/C12A, and C98A, indicating that the 5-kDa band was present in the digest of C98A but was not palmitoylated (Fig. 4B). The stoichiometry of palmitate incorporation was ϳ2.6 Ϯ 0.1 mol of palmitate to 1 mol of RGS16 (WT), which correlates well with the 3 predicted palmitoylation sites (mean Ϯ S.E. of three independent experiments). Moreover, each mutant demonstrated incorporation of [ 3 H]palmitate congruent with the number of target cysteine residues present (0.7 Ϯ 0.1 mol of palmitate/mol of RGS16 C2A/C12A; 1.5 Ϯ 0.03 mol of palmitate/mol of RGS16 (C98A)). These results suggest that Cys-98 is a direct site of palmitoylation.
Effect of Enzymatic Palmitoylation on RGS16 GAP Activity-We first assessed GAP activity of RGS16 preincubated with

FIG. 3. Cys-98 is not required for RGS16 localization at the plasma membrane or in lipid rafts.
A, plasma membranes were isolated from cells expressing either RGS16 WT or C98A. Proteins (40 g) were separated by SDS-PAGE and immunoblotted for RGS16 (arrows, left panel) or Na ϩ /K ϩ -ATPase (a plasma membrane marker, arrowheads, right panel). T, total membrane; and P, plasma membrane fractions. B, cells expressing RGS16 were lysed in detergent buffer and subjected to OptiPrep gradient centrifugation. Fractions were collected (fraction 1 is from the top of the gradient, the "floating" fraction) and subjected to SDS-PAGE before immunoblotting with antibodies against RGS16 (top), caveolin (second from top) or G␣ i (third from top) (markers for DRMs), and Na ϩ /K ϩ -ATPase (bottom), which is a marker for a plasma membrane protein not found predominantly in rafts.
pPAT in a solution assay measuring single turnover GTP hydrolysis by purified G␣ o . We treated RGS16 or RGS4 with pPAT or boiled pPAT and found that boiling the enzyme dramatically reduced the amount of RGS16 palmitoylation, suggesting that the process was predominantly enzymatic (data not shown). We observed no significant difference in GAP activity toward G␣ o (RGS16) or G␣ i (RGS4) between RGS proteins treated with pPAT and untreated proteins (Fig. 5A) or RGS proteins preincubated with boiled pPAT in the single turnover assay (data not shown). We next measured high affinity GTPase activity of the 5-HT 1A /G␣ o1 fusion protein in membranes incubated with recombinant RGS16 treated with pPAT. When assessed with a single time point (representing a 20-min incubation with 10 Ϫ6 M 5-HT), there was little difference in GAP activity between RGS16 and pPAT-treated RGS16 (data not shown). However, because recent studies suggest that RGS proteins may bind to membranes in a time-dependent fashion and that this process may be affected by palmitoylation (22), we preincubated membranes with recombinant RGS16 or RGS4 for various times before agonist stimulation to determine whether palmitoylation affected RGS membrane binding and GAP activity. The GAP activity of RGS16 or RGS4 that had been treated with pPAT and palmitate increased significantly with longer membrane preincubation times (Fig. 5, B-C). In contrast, membranes incubated with either the pPAT preparation alone or untreated RGS proteins failed to exhibit any significant time-dependent increase in GTPase activity. Although the GAP activity of untreated RGS16 did not increase over time (Fig. 5B), RGS16-containing membranes displayed significantly higher absolute GTPase activity compared with membranes incubated with pPAT alone (2.05 Ϯ 0.21-fold). In contrast, activity of membranes containing pPAT was equiva-lent to buffer-treated membranes (0.98 Ϯ 0.19-fold), indicating that the pPAT preparation lacked intrinsic GTPase or GAP activity. These results suggest that palmitoylation of RGS16 and RGS4 augments GAP activity toward a membrane-bound G␣ subunit and that this enhancement is facilitated by RGSmembrane interaction over time.
To confirm the sites of RGS16 palmitoylation and their relative roles in GAP activity, we first measured incorporation of [ 3 H]palmitate into pPAT-treated, recombinant RGS16 WT or C2A/C12A and C98A mutants with or without cleavage with clostripain (Fig. 5C, right panel). Levels of tritium incorporation into full-length RGS16 (30 kDa) were compatible with the number of palmitoylated residues. No palmitoylation of the ϳ5-kDa band containing Cys-98 was observed with the C98A mutant after clostripain treatment, whereas the palmitoylated fragment containing Cys-98 was clearly visualized with the C2A/C12A mutant. Palmitoylation of Cys-98 in the C2A/C12A mutant was expected in this assay because recombinant RGS16 was incubated with pPAT in vitro. To determine which site of palmitoylation was responsible for increased RGS16 GAP activity induced by membrane interaction, we treated recombinant RGS16 WT, C2A/C12A, and C98A with pPAT and then preincubated proteins with 5-HT 1A /G␣ o1 membranes before measuring GTPase activity in response to 5-HT. Treatment of the C2A/C12A mutant with pPAT led to a preincubation time-dependent increase in GAP activity that was mildly reduced compared with WT RGS16. After a 2-h preincubation with WT RGS16, agonist-induced GTPase activity (normalized to GTPase activity without agonist addition) of membranes was ϳ17 pmol/mg/min, whereas GTPase activity of membranes preincubated with the C2A/C12A mutant was reduced ϳ36% (11 pmol/mg/min). However, GTPase activity of membranes preincubated with pPAT-treated C98A for 2 h was reduced nearly 70% compared with membranes preincubated with pPATtreated, WT RGS16 (5 pmol/mg/min) (Fig. 5C). Similarly, the GTPase activity of membranes incubated with pPAT-treated WT RGS4 for 2 h was ϳ7 pmol/mg/min, whereas activity in membranes containing pPAT-treated RGS4 (C95A) was ϳ1.5 pmol/mg/min, an 80% decrease. Therefore, although overall palmitate incorporation was much greater for the RGS16 C98A mutant compared with the C2A/C12A mutant (30-kDa band in panel C), palmitoylation increased the GAP activity of C2A/ C12A to a greater extent than C98A. Absolute levels of GAP activity were much lower for palmitoylated RGS16/RGS4 (C98A/C95A) compared with WT RGS proteins, and the activity of these mutants treated with pPAT did not increase substantially over time. From these data, we conclude that although NH 2 -terminal cysteine residues are major sites of RGS16 and RGS4 palmitoylation, internal RGS box cysteine palmitoylation is critical for full GAP activity. DISCUSSION Studies of palmitoylated residues of RGS proteins have implicated a direct effect of palmitoylation on RGS GAP activity (18,19,22). In the accompanying article (33), we demonstrated that NH 2 -terminal palmitoylation was required for RGS16 GAP activity in cellular membranes as well as its targeting to lipid rafts. Although NH 2 -terminal palmitoylation of RGS16 is not required for plasma membrane localization, it may play a role in the orientation of RGS16 within the membrane or, most intriguingly, in the facilitation of RGS16 palmitoylation on Cys-98, because we found enhanced PAT activity in DRM fractions. The question that remains is how does this internal palmitoylation affect the function of RGS16?
We investigated the mechanism and specificity of palmitoylation of cysteine 98 for RGS16 GAP activity and localization. We found that RGS16 Cys-98 is palmitoylated, similar to the

FIG. 4. RGS16 is palmitoylated on Cys-98 in vitro.
A, recombinant, untagged RGS16 (left panel) or His 6 RGS4 (right panel) was purified from E. coli as described and treated with partially purified PAT in the presence of [ 3 H]palmitate. RGS16 was subjected to cleavage with clostripain, which is predicted to generate a 5-kDa polypeptide containing Cys-98 (arrow). Proteins were then separated by SDS-PAGE, and gels were dried and subjected to fluorography. B, RGS16 WT or mutants (10 g) were digested with clostripain or not and separated by SDS-PAGE. Gel was visualized with Coomassie Blue stain and compared with Mark 12 molecular weight standards (Invitrogen).
FIG . 5. RGS palmitoylation increases GAP activity. A, GTPase activity of recombinant G␣ o (250 nM) was assayed during a single catalytic turnover in the presence of buffer alone or RGS16 (50 -100 nM), either untreated or treated with pPAT as indicated. Similar experiments were performed for RGS4 (50 nM) except that G␣ i1 (250 nM) was used. Graphs represent the mean Ϯ S.E. of four independent experiments. B, recombinant RGS16 WT or RGS4 WT was treated with the pPAT preparation or buffer alone as described in the legend to Fig. 4 and then internal palmitoylation on Cys-95 in RGS4 that has been described previously (19). Most importantly, direct enzymatic palmitoylation enhanced RGS16 and RGS4 GAP activity to a similar degree in a membrane-based assay. Mutation of Cys-98 to alanine did not affect RGS16 localization in membrane fractions or lipid rafts but resulted in a loss of incorporation of palmitate on a 5-kDa peptide containing the Cys-98 residue, significantly reducing the increase in RGS16 GAP activity induced by prolonged membrane interactions. Furthermore, mutation of Cys-98 in RGS16 or Cys-95 in RGS4 eliminated inhibition of somatostatin signaling by these RGS proteins to a similar extent as mutation of NH 2 -terminal palmitoylated cysteine residues (Cys-2 and Cys-12). In agreement with these results, mutation of the analogous cysteine in RGS10, which is the sole site of palmitoylation, abolishes its ability to inhibit signaling evoked by stimulation of the gonadotropin hormonereleasing hormone receptor (GnHR) (23). In addition, autopalmitoylation of RGS10 markedly potentiates its GAP activity when it is reconstituted in phospholipid vesicles containing both receptor and G-protein, which most closely approximates our membrane-based GTPase assay (19).
Collectively, these studies firmly establish a prominent role for NH 2 -terminal palmitoylation as well as palmitoylation of a cysteine residue in the RGS box for the function of RGS4, RGS10, and RGS16. Tu and colleagues (19) reported that autopalmitoylation of RGS4 on Cys-95 inhibited GAP activity in single turnover or proteoliposome-based assays. In contrast, GAP activity of enzymatically palmitoylated RGS16 or RGS4 incubated with mammalian cell membranes expressing a receptor/G␣ GTPase markedly increases over time. The procedures utilized by Tu et al. (19) differ from the current study in several important ways that might partially account for the discrepant findings.
First, Tu et al. (19) mutated the RGS box cysteine to valine in RGS4 whereas we used cysteine to alanine mutants of both RGS4 and RGS16. Valine is a more hydrophobic residue than alanine; as a result, this substitution could affect membrane interactions independent of palmitoylation. Second, we performed enzymatic palmitoylation at physiological pH (7.4) for a short period of time (45 min), which did not appear to alter the intrinsic GAP activity in single turnover assays (Fig. 5A). In contrast, autopalmitoylation required higher pH (8) and longer incubation times to achieve significant palmitate incorporation (2-6 h). The greater efficiency of palmitoylation achieved using the pPAT preparation is reflected in the stoichiometry of labeling. WT RGS16 incorporated close to 3 mol of palmitate/mol, suggesting 3 palmitoylation sites. In contrast, the lower efficiency of autopalmitoylation (2 mol/mol) may not represent stoichiometry in mammalian cells.
Third, the composition of proteoliposomes used in the former study could affect also RGS function. Phosphatidic acid, and, to a lesser extent, phosphatidylserine, inhibits RGS4 GAP activity, which is dependent on the NH 2 terminus of RGS4 (39). Tu et al. (19) utilized proteoliposomes comprised of phosphatidylserine/phosphatidylethanolamine/cholesterol hemisuccinate in a ratio of 8:5:1. It is entirely plausible that phosphatidylserine in these proportions has a more pronounced inhibitory effect on RGS4 palmitoylated at the NH 2 terminus. Last, additional proteins in mammalian membrane preparations absent in lipid vesicles could synergize with palmitate to affect RGS conformation or activity. For example, RGS16 interacts with a membrane glycerophosphodiester phosphodiesterase, MIR16, through the NH 2 -terminal region of the RGS domain (40). Although the effect of MIR16 binding on RGS16 GAP activity has not been studied, this interaction could also contribute to RGS16 anchorage or orientation within the membrane. In addition, RGS16 undergoes phosphorylation on both serine and tyrosine residues (10,11). RGS16 interacts with the receptor for epidermal growth factor, which mediates RGS16 tyrosine phosphorylation and enhances RGS16 GAP activity. Phosphorylation and palmitoylation may act together to enhance the GAP activity of RGS16 in membranes.
Palmitoylation at the internal cysteine may be involved in the optimal placement of RGS16 or RGS4 within the membrane after an initial docking step. Tu et al. (22) found that RGS4 binds vesicles through its NH 2 terminus in a dynamic but ultimately irreversible fashion, which enhanced the GAP activity of RGS4 WT but not RGS4 (C95V) over time. Although the effect of palmitoylation was not reported in these studies, the results are consistent with our experiments demonstrating that the internal palmitoylation site in RGS16 and RGS4 is critical for the incremental increase in GAP activity induced by membrane interactions. Consistent with this notion, RGS10, which lacks the NH 2 -terminal ␣-helix and NH 2 -terminal palmitoylation sites that promote phospholipid vesicle binding, also displays relatively fixed GAP activity over time similar to RGS4 (C95V) (22), or RGS4/RGS16 (C95/98A) (this study).
Alternatively, palmitoylation could induce a conformational change in RGS16 or RGS4 that promotes G-protein binding and catalysis. Because enzymatic palmitoylation did not significantly affect RGS GAP activity in solution, it may fail to induce a gross structural change in the protein. However, because RGS palmitoylation could occur prior to G␣ interactions, the FIG. 6. Relationship between internal palmitoylation and the RGS-G protein interface. A ribbon representation of the crystal structure of RGS4 (green) bound to G␣ i (gray) demonstrating the close proximity of Cys-95 in RGS4 (orange) to the residues between ␣3 and ␣4 of RGS4 (purple) that interact with switch region I (yellow) of G␣ i . The membrane-facing side of the complex is predicted to be in the foreground. Created using Protein Data Bank number 1AGR and visualized with ViewerLite TM 5.0 (Accelrys Inc.). preincubated for various time periods with HEK293 membranes expressing the 5-HT 1A /G␣ o1 fusion protein. GTPase activity of the membranes was determined after a 20-min incubation with 5-HT (10 Ϫ6 M). Values shown are the GTPase activity for each preincubation time subtracted from the zero preincubation value (activity increase in picomole/mg/min, representing the mean Ϯ S.E. of three to four independent experiments). C, effect of palmitoylation on RGS16 and RGS4 Cys mutants. The experiment was performed as in B, except that identical concentrations of pPAT-treated RGS16 WT, RGS16 C2A/C12A, or RGS16 (C98A) and RGS4 WT or RGS4 (C95A) were compared directly. Values represent normalized GTPase activity (raw activity divided by the activity of 5-HT-treated membranes with no additions incubated for the equivalent amount of time (mean Ϯ S.E. of three to five independent experiments). Activity correlates with palmitoylation of the Cys-98 residue in RGS16 as revealed by incorporation of [ 3 H]palmitate into a 5-kDa band (upper right panel) generated by clostripain cleavage of full-length (30-kDa band, left panel) RGS16 WT or C2A/C12A but not C98A. structure of RGS16 or RGS4 in the membrane may be different before and after G-protein binding. Cys-95 in RGS4 lies on helix 4 in close proximity to the loop between helices 3 and 4, which forms part of the binding surface with G␣ switch region I. A palmitate group on Cys-95 in RGS4 or Cys-98 in RGS16 could re-orient the adjacent contact residues (particularly Tyr-87, Ser-88, Glu-90, and Asn-91 in human RGS16) to facilitate more efficient binding, which would be predicted to enhance GAP activity (28,41) (Fig. 6). A recent study demonstrated that mutation of Glu-89 (mouse RGS16) to lysine did not substantially affect GAP activity toward G␣ i in solution, but nearly abolished GAP activity toward a related G␣ subunit, transducin, in rod outer segment membranes. In addition, RGS16 (E89K) was unable to regulate ␣ 2 -adrenoreceptor-mediated inhibition of BK Ca channel activity in rat uterine myocytes (42). These results suggest a distinct interaction between RGS16 helix 4 and G␣ switch I that may only be evident in membranes. This binding surface could be affected by palmitoylation of the resident cysteine residue in this region.
Finally, the poor activity or RGS16 (C98A) in functional assays despite its normal localization to lipid rafts lend further support to the view that abnormal DRM localization of either RGS16 (C98A) or (C2A/C12A) cannot account for their lack of activity. Our results show that palmitoylation plays an important positive regulatory role in the GAP activity of both RGS16 and RGS4. In the future, it will be of great interest to study the regulation and activity of RGS proteins in their native cellular context, which should further illuminate specific physiological functions.