Palmitoylation of a Conserved Cysteine in the Regulator of G Protein Signaling (RGS) Domain Modulates the GTPase-activating Activity of RGS4 and RGS10*

RGS4 and RGS10 expressed in Sf9 cells are palmitoylated at a conserved Cys residue (Cys95 in RGS4, Cys66 in RGS10) in the regulator of G protein signaling (RGS) domain that is also autopalmitoylated when the purified proteins are incubated with palmitoyl-CoA. RGS4 also autopalmitoylates at a previously identified cellular palmitoylation site, either Cys2 or Cys12. The C2A/C12A mutation essentially eliminates both autopalmitoylation and cellular [3H]palmitate labeling of Cys95. Membrane-bound RGS4 is palmitoylated both at Cys95 and Cys2/12, but cytosolic RGS4 is not palmitoylated. RGS4 and RGS10 are GTPase-activating proteins (GAPs) for the Gi and Gq families of G proteins. Palmitoylation of Cys95 on RGS4 or Cys66 on RGS10 inhibits GAP activity 80–100% toward either Gαi or Gαzin a single-turnover, solution-based assay. In contrast, when GAP activity was assayed as acceleration of steady-state GTPase in receptor-G protein proteoliposomes, palmitoylation of RGS10 potentiated GAP activity ≥20-fold. Palmitoylation near the N terminus of C95V RGS4 did not alter GAP activity toward soluble Gαz and increased Gz GAP activity about 2-fold in the vesicle-based assay. Dual palmitoylation of wild-type RGS4 remained inhibitory. RGS protein palmitoylation is thus multi-site, complex in its control, and either inhibitory or stimulatory depending on the RGS protein and its sites of palmitoylation.

rate-limiting event in palmitate turnover. Duncan and Gilman (2) recently identified a palmitoyl-protein thioesterase that is probably the enzyme that depalmitoylates G␣ subunits and which may also depalmitoylate other proteins. The enzyme that transfers palmitate to proteins has not been clearly identified. Dunphy et al. (3) partially purified an activity from hepatic membranes that accelerates palmitoylation of G␣ using Pal-CoA as donor, potentially a protein-palmitoyl transferase. However, G␣ subunits can autopalmitoylate in vitro at the physiologically correct Cys residues (4), and G protein-coupled receptors may also autopalmitoylate (5,6). Autopalmitoylation is somewhat slower than palmitoylation in vivo, but the correlation between Cys residues that autopalmitoylate in vitro and those that are naturally palmitoylated suggests that autopalmitoylation is at least involved in the physiological process. The Cys residues that selectively autopalmitoylate may do so because their thiol groups have an unusually low pI, because they are at relatively hydrophobic surfaces and are thus exposed to Pal-CoA, or because these proteins contain ancillary residues that catalyze the reaction. The ability of a specific residue to autopalmitoylate would provide selectivity to the palmitoylation process, and the protein studied by Dunphy et al. (3) might act non-selectively either as a Pal-CoA carrier protein or as a non-selective transferase.
The functions of protein palmitoylation is an active area of study. In the case of p21ras and G␣ subunits, palmitoylation helps anchor the proteins to the plasma membrane (1,7,8). Palmitoylation of G␣ subunits also enhances their binding to G␤␥ and desensitizes them to the GAP activity of RGS proteins (9,10). Palmitoylation has been reported to enhance the activity of a G protein-coupled receptor kinase (11), and mutations of the palmitoylated Cys residue in the C-terminal region of G protein-coupled receptors themselves have resulted in a variety of effects (8). In none of these cases is it completely clear whether palmitoylation alters the structure of the palmitoylated protein itself, whether the palmitoyl group forms or blocks part of a protein-protein interface, or whether it only increases binding to hydrophobic surfaces.
It has recently been reported that several RGS proteins are also palmitoylated in cells (12)(13)(14). RGS proteins are a family of variably selective GAPs for members of the G i and G q families of heterotrimeric G proteins (15). RGS proteins inhibit G protein signaling in fungi and roundworms (16 -18) and their overexpression can inhibit signaling in mammalian cells (see Ref. 15 for review). RGS proteins are also important in modulating the decay kinetics of G protein signaling (19 -23). The closely related proteins RGS4 and RGS16 are palmitoylated at Cys 2 and/or Cys 12 (13,14), and mutation of these residues diminished the ability of RGS16 to inhibit cellular signaling (14). Membrane-bound GAIP is also palmitoylated near its N terminus, but in a cysteine string that characterizes a separate RGS protein subfamily (12). We report here that RGS4 can autopalmitoylate stoichiometrically in vitro at two sites, one near the N terminus and the other in the conserved RGS box domain. RGS10, which lacks the N-terminal site, also autopalmitoylates at the conserved Cys residue. Autopalmitoylation of both proteins correlates well with their palmitoylation in cells. We report further that dual palmitoylation can either inhibit or potentiate GAP activity depending on the site of palmitoylation, the assay medium and on the identity of the RGS protein.
Mutagenesis of RGS4 cDNA-The Cys mutants C2A/C12A/C33A and C2A/C12A RGS4 were prepared by sequential polymerase chain reaction reactions. C33A RGS4 cDNA was first prepared by substituting codon 33 by GCG. The product and the wild-type cDNA were cut with NcoI and BamHI and ligated in-frame into a modified pQE60 (Qiagen) that encodes the sequence MGH 6 MG before the cloning site (30). In subsequent reactions, codons 2 and 12 were replaced by GCC. The final cDNA products, C2A/C12A and C2A/C12A/C33A RGS4, were cloned into the modified pQE60 vector as described above. The ⌬N57 RGS4 deletion mutation was generated by polymerase chain reaction with the primer 5Ј-GATCCATGGGCAAATGGGCTGAATCGCTGGAA. The product was cut with NcoI and BamHI and cloned into the corresponding sites of the modified pQE60 (30). C95V RGS4 was generated using the QuikChange mutagenesis kit (Stratagene) to change codon 95 to GTT. All wild-type or mutant RGS proteins contained N-terminal His 6 . The accuracy of all constructions was checked by DNA sequencing. For expression in Sf9 cells, cDNAs were removed from pQE60 with NcoI and BamHI and inserted into pVL1392 modified to include a NcoI site at the 5Ј end of the multiple cloning site. Recombinant baculoviruses were produced as described previously (27). The plasmid that encodes RGS10 box was prepared as described (30).
Expression and Palmitoylation of RGS Proteins in Sf9 Cells-For labeling with [ 3 H]palmitate and small scale preparation, Sf9 cells (4 ϫ 10 6 ) were infected with RGS4 or RGS10 virus for 32 h prior to metabolic labeling. Sodium [ 3 H]palmitate (1 mCi) was suspended in 2 ml of IPL-41 that contained 1% ethanol and 2% heat-inactivated fetal calf serum. Cells were incubated in this medium for 1 h. The cells were harvested by centrifugation, washed with 2 ml of phosphate-buffered saline, and suspended in 0.5 ml of lysis buffer (20 mM NaHepes (pH 8.0), 2 mM MgCl 2 , 1 mM 2-mercaptoethanol, 1 mM phenylmethylsulfonyl fluoride, 20 g/ml leupeptin, 10 g/ml aprotinin). Cells were lysed by 5 freeze/thaw cycles, followed 10 passes through a 25-gauge needle. The lysates were centrifuged at 100,000 ϫ g at 4°C for 20 min in a Beckman TL100.3 rotor. The supernatant (cytosolic fraction) was removed, and the pellet (crude membrane) was resuspended in lysis buffer. Both fractions could then be analyzed by SDS-PAGE. For purification of active RGS protein, the pellet was solubilized by resuspension and stirring for 30 min at 4°C in 100 mM NaCl, 50 mM NaHepes (pH 8.0), 0.5% deoxycholate, 0.5% Triton X-100, 0.1% SDS, 1 mM 2-mercaptoethanol, and the protease inhibitors listed above. After centrifugation for 30 min as described above, the supernatant was diluted 5-fold with 50 mM NaHepes (pH 8.0) and applied to nitrilotriacetic acid-Ni 2ϩ -agarose and purified as described for RGS proteins expressed in E. coli (24). RGS4 from the cytoplasmic fraction was purified similarly.
Autopalmitoylation of RGS Proteins-Autopalmitoylation of RGS proteins was performed as described previously for G␣ subunits with slight modification (4,10). Routinely, 5 M RGS protein was incubated with 100 M radioactive or non-radioactive Pal-CoA for up to 6 h at 30°C in 50 mM NaHepes (pH 8.0), 0.005% Lubrol PX, and 100 M 2-mercaptoethanol. Residual free Pal-CoA was removed by adsorbing the palmitoylated RGS protein to nitrilotriacetic acid-Ni 2ϩ -agarose followed by elution as described (24). To assess the degree of palmitoylation, the preparation was either precipitated with 10% trichloroacetic acid and isolated either on a glass fiber filter (10) or by SDS-PAGE before liquid scintillation counting. Gels were stained with Coomassie Blue to detect proteins and appropriate slices were solubilized with 30% H 2 O 2 for 16 h at 60°C. RGS concentrations were determined by Amido Black binding (31) using bovine serum albumin as standard.
Electrophoresis and Immunoblotting-SDS-PAGE was performed as described (10). Gradient gels (15-22%) were used for CNBr-cleaved samples. Samples for PAGE were prepared as described (10) to maintain palmitoylation during denaturation. Proteins were transferred to nitrocellulose as described (10) and immunoblots were developed according to instructions in the ECL kit (Amersham Pharmacia Biotech).
CNBr Cleavage of RGS4 -RGS4 (10 g/ml, 50 -100 pmol) in 70% (v/v) formic acid was mixed with 100 mg/ml CNBr and 1 mg/ml tryptophan and was incubated under argon in the dark at room temperature for 20 h. Products were diluted in water and dried under vacuum.
Mass Spectrometry-Palmitoylated or non-palmitoylated RGS4 was alkylated by incubation with 15 mM N-ethylmaleimide to block free thiol groups and digested with CNBr. The mixture was resolved with SDS-PAGE, the gels were stained by Copper Stain (Bio-Rad), and appropriate bands were cut and extracted with 17% formic acid, 33% 2-propanol (32). The molecular masses of peptides were determined by matrix-assisted laser desorption ionization/time of flight spectrometry using a matrix of 3,5-dimethoxy-4-hydroxycinnamic acid on a Voyager-DE spectrometer (PE Biosystems).
GAP Assays-GAP activity was assayed in two formats. In the simpler assay, purified G␣ is first bound to [␥ -32 P]GTP, and the rate of hydrolysis of the [␥-32 P]GTP-G␣ is measured in detergent solution in the presence and absence of the GAP. In this assay, GAP activity is defined as the increase in the first-order hydrolysis rate constant or is approximated as an increase in the initial rate of hydrolysis (26,33). Such single-turnover GAP assays, using either ϳ2 nM G␣ z -[␥-32 P]GTP at 15°C or ϳ10 nM G␣ i1 -[␥-32 P]GTP at 8°C, were performed as described (10,26). A more sensitive and presumably more physiological assay for GAP activity monitors the enhancement of agonist-stimulated steady-state GTPase activity in proteoliposomes reconstituted with receptor and heterotrimeric G protein. Reconstitution of purified m2AChR with either G i or G z and was performed as described (24). RGS proteins were usually incubated with the vesicles for 1 h at 30°C prior to assay. Conditions for measuring carbachol-stimulated GTPase activity in this system have been described (24).

RGS4 and RGS10 Autopalmitoylate in Vitro and
Are Naturally Palmitoylated in Sf9 Cells at a Highly Conserved Residue in the RGS Box-When RGS10 or RGS4 were incubated with [ 3 H]Pal-CoA under the conditions originally described by Duncan and Gilman (4) for autopalmitoylation of G␣ subunits, they incorporated 3 H through a bond that was sensitive to both NH 2 OH and DTT (Fig. 1A), presumably a palmitoyl thioester. Both RGS4 (13)  1B). In vivo [ 3 H]palmitoylation was essentially limited to the RGS protein found in the particulate fraction. Little label was found on soluble RGS protein despite the fact that about 30% of RGS4 and almost all of the RGS10 were found in the cytoplasmic fraction.
The extent of in vitro autopalmitoylation of RGS10 was approximately 1 mol of palmitate/mol (Fig. 1A). A truncated RGS10 consisting only of the conserved RGS box was also labeled to about 1 mol/mol with approximately similar kinetics. Because Cys 66 is the only Cys residue in the RGS10 box, its palmitoylation accounts for that observed in full-length RGS10. Intact RGS4 incorporated 2 mol of palmitate/mol, but an Nterminal truncation mutant of RGS4 was labeled to only 1 mol/mol. This finding suggests that one palmitoylation site lies in the RGS4 box and that the other lies in the N-terminal region. Because Cys 66 in RGS10 is the only highly conserved Cys residue in the RGS protein family, its palmitoylation suggests that the corresponding Cys residue in RGS4, Cys 95 , might be the site of palmitoylation in the RGS4 box. Both Cys 2 and Cys 12 are candidate sites for the N-terminal palmitoylation site (13,14), and we did not try to distinguish between them in this study.
We analyzed the palmitoylation sites of RGS4 by a combination of CNBr peptide mapping and mutagenesis. As shown in Fig. 2A, autopalmitoylated RGS4 contained [ 3 H]palmitate in two CNBr-generated peptides, one of about 15,000 Da and the other of about 2000 Da. Both peptides were also labeled in RGS4 that had been palmitoylated in Sf9 cells. The palmitoylated peptides were identified both by Edman sequencing and mass spectrometry as Lys 20 -Met 141 , which includes most of the RGS4 box, and Cys 2 -Met 19 . Comparison of the masses of these peptides in samples prepared from palmitoylated and nonpalmitoylated samples indicated that each incorporated 1 palmitoyl group. Mutation of Cys 95 decreased palmitoylation of intact RGS4 that was labeled either in vitro or in vivo, and essentially eliminated incorporation of palmitate into the ϳ15-kDa Lys 20 -Met 141 peptide (Fig. 2B). This finding, coupled with the unique palmitoylation of the homologous Cys 66 residue in RGS10, indicates that Cys 95 is the site of palmitoylation in the RGS4 box. In autopalmitoylated RGS4, Cys 95 and the more N-terminal site were each labeled to approximately 1 mol/mol. In Sf9 cells, however, more [ 3 H]palmitate was incorporated into the N-terminal site than into Cys 95 under our standard conditions for Sf9 cell growth. This difference might reflect either relatively more incorporation of palmitate into the Nterminal site or more complete turnover at that site, although net incorporation of label into RGS4 was essentially complete within 1 h.
To study the interdependence of palmitoylation of the Nterminal region and the RGS box, we mutated Cys 2 and Cys 12 , which are probable sites of cellular palmitoylation in both RGS4 and RGS16 (13,14). Surprisingly, mutation of these sites eliminated all RGS4 autopalmitoylation, including autopalmitoylation of Cys 95 (Fig. 3B). To determine whether N-terminal palmitoylation precedes Cys 95 autopalmitoylation in wild-type RGS4, we monitored autopalmitoylation of both sites in vitro. As shown in Fig. 3C, autopalmitoylation at Cys 95 lagged somewhat behind autopalmitoylation at the N terminus. While the lag was not great, it was reproducible in three similar experiments. These findings indicate that palmitoylation of the Nterminal site is kinetically favored over Cys 95 palmitoylation. Consistent with this idea, the ⌬N57 mutant of RGS4 and RGS10 (which is not N-terminally palmitoylated) both autopalmitoylate very slowly (t 1/2 ϳ 2 h, compared with a t 1/2 of 30 min for wild-type RGS4). These results suggest that initial N-terminal palmitoylation promotes subsequent palmitoylation in the RGS4 box. N-terminal palmitoylation is not absolutely required for autopalmitoylation within the RGS box because both Cys 95 in the RGS4 box construct and Cys 66 in RGS10 do slowly autopalmitoylate. Coupling of palmitoylation in the RGS box with that in the N-terminal region also appears to hold true in cells, where the C2A/C12A mutation eliminates all RGS4 palmitoylation (13).
Palmitoylation in the RGS Box Inhibits G Protein GAP Activity in Solution-based Assays-Autopalmitoylation of RGS4 inhibited its GAP activity essentially completely when activity was measured in a single-turnover assay in detergent solution (Fig. 3A). Upon incubation with Pal-CoA, GAP activity declined with a time course similar to that of incorporation of palmitate (data not shown; see Fig. 4 for similar data for RGS10). Incubation of autopalmitoylated RGS4 with DTT largely reversed inhibition (Fig. 3A), and incubation of RGS4 without Pal-CoA was without effect. The GAP assays shown in Fig. 3A used G␣ z -GTP as substrate, but similar results were obtained in similar experiments that used G␣ i1 -GTP as substrate and Triton X-100 instead of Lubrol PX.
Mutation of Cys 95 in RGS4 blocked inhibition of GAP activity by incubation with Pal-CoA, suggesting that inhibition is me- ). As would be predicted by its inability to incorporate palmitate at Cys 95 , Both C2A/C12A and C2A/C12A/C33A RGS4 were also not inhibited by incubation with Pal-CoA (Fig. 3B).
Palmitoylation also inhibited the GAP activity of RGS10 when activity was measured in the solution-based assay (Fig.  4). Again, fractional inhibition of GAP activity paralleled fractional palmitoylation and suggested that stoichiometrically palmitoylated RGS10 is inhibited 80 -90% at this substrate concentration. Control incubations without Pal-CoA did not cause inhibition (Fig. 4), and inhibition was reversed by incubation with DTT (not shown). Because RGS10 is palmitoylated only on Cys 66 , these data combine with those of Fig. 3 to indicate that palmitoylation of the conserved Cys residue in the RGS box blocks the GAP activity of these RGS proteins in the single-turnover assay.
Palmitoylation of RGS10 Potentiates GAP Activity in a Vesicle-based Assay System-In contrast to the inhibition described above, palmitoylation of RGS10 markedly stimulated its GAP activity as measured during receptor-stimulated, steady-state GTP hydrolysis. The agonist-stimulated GTPase activity of unilamellar phospholipid vesicles that contained heterotrimeric G i and m2AChR was measured in the presence of increasing concentrations of RGS10 (Fig. 5). When agonistbound receptor drives GDP/GTP exchange in these vesicles, hydrolysis of G i -bound GTP becomes rate-limiting and a GAP increases steady-state hydrolysis until the overall reaction again approaches the rate of receptor-catalyzed GDP/GTP exchange (29,34,35). In m2AChR-G i vesicles, RGS10 increases agonist-stimulated GTPase activity about 10-fold at 5 M, the highest concentration tested (Fig. 5). Palmitoylation at Cys 66 markedly potentiated the GAP activity of RGS10. Although we were not able to demonstrate saturation with RGS10, which displays relatively low potency in this assay, the activity of palmitoylated RGS10 was equal to that of about 20-fold more non-palmitoylated RGS10 (Fig. 5).
N-terminal Palmitoylation of RGS4 Enhances Its GAP Activity in Phospholipid Vesicles-In contrast to RGS10, initial experiments indicated that complete palmitoylation of wild-type RGS4 (2 mol of palmitate/mol) inhibited its GAP activity in the vesicle-based steady-state assay (Fig. 6). In contrast, N-terminal palmitoylation of C95V RGS4 increased its GAP activity

FIG. 3. Palmitoylation of RGS4 at Cys 95 blocks its GAP activity.
A, inhibition by autopalmitoylation in vitro. RGS4 (5 M) was incubated at 30°C either alone (OE, ‚) or with (q, E) 100 M Pal-CoA. At the times shown, 5-l aliquots (25 pmol) were diluted and assayed for GAP activity toward G␣ z -GTP. At 120 min, half of each of the remaining two samples were treated with 50 mM DTT for 60 min at 30°C (‚, E). DTT-treated and untreated samples were then assayed for GAP activity. B, palmitoylation of Cys 95 accounts for RGS4 inhibition and can be blocked by mutation of either Cys 95 itself or of the N-terminal Cys residues. Samples of mutant or wild-type RGS4 (5 M) were incubated at 30°C with 100 M Pal-CoA for 2 h. G z GAP activity was assayed as described (lower panel). Data are shown as the percentage of activity in a sample of the same protein (mutant or wild type) that was incubated similarly but without Pal-CoA. Control activities were 60 milliunits for wild-type and C95A RGS4 and 35 milliunits for C2A/C12A (AAC) and C2A/C12A/C33A (AAA) RGS4. In a parallel experiment, the same RGS proteins (25 pmol  severalfold in this assay. Both effects were reversed by depalmitoylation with DTT, and the intrinsic GAP activities of non-palmitoylated C95V RGS4 is nearly the same as that of wild type (Ͼ80%; Fig. 6, Table I). As shown in Fig. 7B, Nterminal palmitoylation increased the potency of C95V RGS4 about 6-fold, whereas palmitoylation of wild-type RGS4 at both sites decreased potency more than 75%. Maximal GTPase activity was the same for each protein (Fig. 7B), presumably because GDP/GTP exchange became rate-limiting (35). Selective palmitoylation of RGS4 near its N terminus thus stimulates its GAP activity in vesicles as long as Cys 95 is not also palmitoylated. Conversely, palmitoylation at Cys 95 accounts for the inhibition of wild-type RGS4. Palmitoylation at the N terminus of C95V RGS4 had no effect on its GAP activity in solution, but dual palmitoylation of wild-type RGS4 inhibited by more than 99% (Fig. 3B and 7A; residual activity probably reflects incomplete palmitoylation). The activity of the C2A/ C12A mutant, which does not autopalmitoylate at either site (Fig. 3), was predictably unaltered by treatment with Pal-CoA or DTT (Fig. 6).
To analyze the effect of palmitoylating only the N-terminal site in wild type RGS4, we took advantage of the likelihood that N-terminal palmitoylation in Sf9 cells precedes or is preferred to palmitoylation of Cys 95 and the fact that cytoplasmic RGS4 is essentially not palmitoylated (Figs. 1B and 3). As shown in Fig. 8A, wild-type RGS4 purified from Sf9 cytosol can autopalmitoylate to approximately 2 mol of palmitate/mol, as was the case for RGS4 purified from E. coli (Fig. 1A). In contrast, RGS4 purified from Sf9 cell membranes could only incorporate about 1.0 -1.2 mol of palmitate/mol, suggesting that labeling is blocked because one or both sites are naturally palmitoylated in cells to a total of 0.8 -1.0 mol of palmitate/mol of RGS4. Based on these data and on the ratio of [ 3 H]palmitate labeling at the two sites (about 2; Fig. 2), we estimate that RGS4 bound to Sf9 cell membranes is about 60% palmitoylated at the Nterminal site and 30% palmitoylated at Cys 95 to produce an average total palmitoylation of about 1 mol of palmitate/mol of RGS4.
We then tested the GAP activities of samples of RGS4 purified from either the membrane or cytoplasmic fraction of Sf9 cells, both without treatment and after complete in vitro autopalmitoylation. In the solution-based single-turnover assay, non-palmitoylated RGS4 purified from Sf9 cytosol was about 50% more active as a GAP than was the partially palmitoylated RGS4 purified from the membrane fraction (Fig. 8B). Partial palmitoylation at Cys 95 is thus inhibitory in solution as expected, and partial palmitoylation near the N terminus had little if any effect. Autopalmitoylation at both sites completely inhibited GAP activity, consistent with the experiment shown in Figs. 3 and 7A. When GAP activity was measured during steady-state GTP hydrolysis by m2AChR-G z vesicles, the opposite order was observed. Partially palmitoylated RGS4 from the membrane fraction was nearly twice as active as RGS4 from cytoplasm, consistent with significant stimulation by palmitoylation in the N-terminal region (Fig. 8C). Again, complete palmitoylation of both RGS4 preparations inhibited their activities to the same low level. These data, taken together, suggest that RGS4 palmitoylated near its N terminus is more active as a GAP when assayed at a membrane surface and that a second palmitoyl group added in the RGS box inhibits activity regardless of the palmitoylation state of the N-terminal site. DISCUSSION These data establish that multi-site palmitoylation of an RGS protein, within the RGS box and near the N terminus, can either potentiate or inhibit its G protein GAP activity. We were able to determine that these distinct sites in RGS4 are both FIG. 5. Palmitoylation of RGS10 potentiates GAP activity in receptor-G i proteoliposomes. The steady-state GTPase activity of proteoliposomes reconstituted with m2AChR, and heterotrimeric G i was assayed in the presence of 1 mM carbachol (q, OE) or 10 M atropine (E, ‚). Detergent-free preparations of RGS10 (q, E) or in vitro palmitoylated RGS10 (OE, ‚) were added to the vesicles to yield the concentrations shown and incubated for 60 min before assay. Vesicles used for each assay contained 0.45 nM G i and 0.13 nM receptor. The palmitoylated RGS10 was prepared by incubating RGS10 with Pal-CoA for 6 h at 30°C, followed by removal of free Pal-CoA by nitrilotriacetic acid-Ni 2ϩ chromatography.
FIG. 6. N-terminal palmitoylation of RGS4 potentiates its GAP activity in receptor-G protein proteoliposomes. Wild-type or mutant RGS4 was incubated with Pal-CoA for 2 h, and half of each sample was further treated with 50 mM DTT for 1 h to remove palmitate. The samples were then pre-incubated for 60 min at 30°C with proteoliposomes that contained m2AChR (0.27 nM m2AChR) and trimeric G z (0.95 nM). Steady-state GTPase activity was then assayed in the presence of 1 mM carbachol. Final concentrations of RGS proteins were 10 nM wild-type RGS4, 25 nM C2A/C12A RGS4, or 12 nM C95V RGS4 to compensate for their different GAP activities (Table I).

TABLE I
Effect of Cys mutations on RGS4 GAP activity Hydrolysis rate constants were determined at 15°C for G␣ z -GTP or 8°C for G␣ i1 , as described under "Experimental Procedures," in the presence or absence of purified wild-type and mutant RGS4 (1.0 nM RGS4 in G␣ z assays, 3 nM RGS4 in G␣ i1 assays). Concentrations of substrate were 1.8 nM G␣ z -GTP and 9 nM G␣ i1 -GTP. Data are the average of duplicate determinations in three separate assays (n ϭ 6). palmitoylated in cells because both sites in RGS4 are efficiently autopalmitoylated in vitro, apparently in the same order. RGS10 also autopalmitoylated essentially quantitatively at a single site in the RGS box. Near quantitative in vitro labeling allowed identification by peptide mapping and mass spectrometry of a conserved Cys residue in the RGS box, Cys 95 in RGS4 and Cys 66 in RGS10, as the major site of both autopalmitoylation and cellular palmitoylation within the RGS box. This is the only Cys residue in the RGS10 box, and palmitoylation within the RGS4 box was eliminated by C95V mutation. This Cys residue is conserved in all mammalian RGS proteins (including axin) except RGS6 and RGS7, in which the corresponding residue is Val. The related p115 rho GEFs have two conserved Cys residues one helical turn removed from this site. RGS proteins are the most recently recognized group of cellularly palmitoylated proteins that also autopalmitoylate at the correct Cys residues in vitro. G␣ subunits were the first (4), GAP-43 autopalmitoylates at a regulatory site (36), and G protein-coupled receptors probably autopalmitoylate as well (5,37,38). RGS4 autopalmitoylated fairly quickly, such that palmitoylation was complete in about 30 min. The other RGS constructs autopalmitoylated with t 1/2 ϳ 2 h. The specificity of autopalmitoylation for specific Cys residues and the general correlation between the rates for in vitro autopalmitoylation and cellular palmitoylation resurrect the question of whether autopalmitoylation is a significant physiological mechanism for protein palmitoylation (1, 4). Although Dunphy et al. (3) have described a cellular activity that enhances the rate of G␣ palmitoylation (see also Ref. 38), no protein palmitoyltransferase has been identified at either the protein or DNA level. Druey et al. (14,39) also required a fraction from liver to palmitoylate RGS16 in vitro, but that activity might be the acyl-CoA synthetase needed to form Pal-CoA from labeled palmitate. Although it may seem intuitively unlikely that autopalmitoylation accounts totally for protein palmitoylation in cells, it remains plausible that autopalmitoylation is adequate in a local environment rich in Pal-CoA.
Peptide mapping of RGS4 and RGS10 that was labeled by [ 3 H]palmitate in Sf9 cells allowed us to demonstrate that both sites, Cys 2 or Cys 12 in RGS4 and Cys 95/66 in the RGS box, are palmitoylated naturally. Palmitoylation in the RGS box was missed previously because for intact RGS4 and, presumably, for RGS16 (14), it essentially depends on the prior palmitoylation of the more N-terminal site. Thus, mutation of Cys 2 and Cys 12 to Ala eliminated both their own palmitoylation and that of Cys 95 (Refs. 13 and 14 and Fig. 3B). Removal of the Nterminal domain of RGS4 also slowed palmitoylation, and the absence of an N-terminal palmitoylation site in RGS10 probably accounts in part for its slow palmitoylation in vitro and modest palmitoylation in cells. The same pattern is apparently true for RGS16, a close homolog of RGS4, when it is expressed in COS cells (14). Sequential palmitoylation may also occur in the RGSZ family (24), in which an N-terminal cysteine string (9 Cys residues out of 13) is multiply palmitoylated (12). FIG. 7. Independent effects of N-terminal and Cys 95 palmitoylation of RGS4 in soluble and vesicle-based assays. A, the GAP activities of untreated wild-type RGS4 (q), palmitoylated wild-type RGS4 (E), and palmitoylated C95V RGS4 (‚) were assayed in detergent solution using the single-turnover format, with 1.9 nM G␣ z -[␥-32 P]GTP as substrate. Assay times (40 s to 4 min) were adjusted as described (26) to account for different GAP activities. B, GAP activities of the same proteins were measured during a carbachol-stimulated, steady-state GTPase catalyzed by m2AChR-G i1 proteoliposomes. The completeness of palmitoylation (3 h at 30°C) was monitored in a parallel [ 3 H]PalCoA labeling reaction and found to be 1.9 Ϯ 0.1 mol of palmitate/mol of wild-type RGS4 and 1.0 Ϯ 0.1 mol of palmitate/mol of C95V RGS4.

FIG. 8. Differential palmitoylation and GAP activities of RGS4 purified from the cytoplasmic or membrane fractions of Sf9 cells.
Wild-type RGS4 was purified from either the soluble (S) or particulate (P) fraction of Sf9 cells that had been infected with recombinant baculovirus 33 h previously. A, a portion of each sample was tested for its capacity to autopalmitoylate by incubation with 100 M [ 3 H]Pal-CoA at 30°C for 2 h. RGS4 purified from the particulate fraction incorporated 1.1 Ϯ 0.1 mol of palmitate/mol of protein and RGS4 from the cytoplasmic fraction incorporated 1.8 Ϯ 0.2 mol/mol of protein. B, the G z GAP activity of each preparation was assayed in a solution-based single-turnover assay after incubation with or without Pal-CoA. C, the GAP activity of each preparation (15 nM each), with and without in vitro palmitoylation, was measured according to stimulation of steady-state carbachol-stimulated GTPase activity of vesicles that contained 0.6 nM G z and 0.2 nM m2AChR. The control was assayed in the absence of RGS4.
There are at least three plausible explanations for why both autopalmitoylation and cellular palmitoylation in the RGS box are accelerated by initial palmitoylation near the N terminus. First, initial N-terminal palmitoylation may help secure the RGS protein or the RGS box domain to a membrane or micelle that contains Pal-CoA (3). More likely, N-terminal palmitoylation actively facilitates palmitoylation in the box, probably by altering the conformation of the RGS protein such that the Cys 95 thiol group is more exposed, more reactive or both. This thiol group is relatively inaccessible according to the structure of the RGS4-G␣ i1 complex described by Tesmer et al. (40). Finally, it is possible that the N-terminal palmitoyl-cysteine thioester acts catalytically and directly transfers its palmitoyl group to the Cys residue in the box, after which it can be subsequently repalmitoylated. Interaction of the cysteine residues at the N terminus with Cys 95 is at least consistent with the fact that ⌬N57 RGS4 palmitoylates faster than does the C2A/C12A mutant, in which the N-terminal domain may unproductively occlude Cys 95 . Unfortunately, the N terminus of RGS4 is not visible in the only available crystal structure of RGS4 (40), and the spatial relationship between the two Cys residues is unknown.
The cellular palmitoylation state of the RGS proteins is regulated both by palmitoylation and depalmitoylation. The combined data of Figs. 1B, 2, and 8A suggest that the palmitate groups on RGS proteins are in constant turnover in Sf9 cells, and the same appears true in COS cells (14). At steady state, RGS4 was 50 -60% palmitoylated in Sf9 cell membranes according the capacity of purified RGS4 to add more [ 3 H]palmitate when exposed to [ 3 H]Pal-CoA in vitro (Fig. 8). CNBr peptide mapping of steady-state palmitoylated RGS4 (Fig. 2) suggests further that palmitoyl groups are found at both the N-terminal position and at Cys 95 , indicating that palmitate turns over at both sites. Labeling at both sites reached steady state in about 1 h, suggesting that removal of palmitate is essentially continuous and is required for addition of new [ 3 H]palmitate label. We do not know how depalmitoylation of each site may be regulated, however, and rapid hydrolysis of palmitate at the N-terminal site could lead to the creation of significant RGS4 that is only palmitoylated at Cys 95 .
Palmitoylation of RGS4 and RGS10 either inhibited or stimulated GAP activity depending on which assay system was used and, for RGS4, which of the two sites was palmitoylated. Palmitoylation of Cys 95/66 in RGS4 and RGS10 uniformly inhibited their GAP activities in the single-turnover assay in detergent solution. Inhibition was sufficiently complete that we could not determine whether it was caused by a decrease in k cat , an increase in K m , or both. The extent of inhibition was similar for both G i and G z substrates and was observed with either Triton X-100 or Lubrol PX. Because Cys 95/66 lies in the middle of helix 4 and is oriented inward toward helix 5 (40), it seems likely that palmitoylation inhibits GAP activity by shifting the packing of the central four-helix bundle to alter the structure of the site that binds the G␣-GTP substrate. The cysteine thiol group is itself not required for GAP activity because RGS6, RGS7, and the active C95A mutant all have a Val residue at this site. Inhibition also does not appear to result simply from a net increase in hydrophobicity because palmitoylation at the N terminus had no effect on intrinsic GAP activity (Fig. 7A).
In contrast to the inhibition observed in the single turnover assay, palmitoylation at the N terminus of RGS4 or at Cys 66 of RGS10 potentiated GAP activity measured in the vesiclebased, steady-state assay. GTPase activity in this assay depends on the coordinated stimulation of GDP/GTP exchange by receptor and of GTP hydrolysis by the GAP (29, 33). Potentia-tion in this assay probably reflects either increased hydrophobicity that attaches the RGS protein more firmly to the vesicles or enhanced interaction with some other component of the system (receptor or G␤␥). Interaction of RGS proteins with receptors has been proposed based on receptor-selective action of RGS4 in pancreatic acinar cells (41,42) and interaction with G␤␥ has been supported by the effects of RGS proteins on the regulation of G␤␥ -gated K ϩ channels (20,21,43).
It is likely that palmitoylation enhances GAP activity at least in part by increasing affinity for membranes. For RGS10, which is quite hydrophilic and does not bind appreciably to cell membranes (Fig. 1B) or to phospholipid bilayers (data not shown), palmitoylation would provide localization at the membrane surface. Increased local concentration at the site where G protein and receptor interact could account for the enhanced potency shown in Fig. 5. RGS4 is intrinsically more hydrophobic because of its N-terminal domain (Ref. 44, and data not shown). The differential hydrophobic effect of RGS4 palmitoylation would therefore not be as great as with RGS10 and would potentiate GAP activity less. Thus, only palmitoylation at the N terminus could cause net stimulation, and this effect would not overcome the inhibition by palmitoylation at Cys 95 (Fig. 7). In this context it is interesting that Druey et al. (14) found that mutation of Cys 12 and, to a lesser extent, Cys 2 in RGS16 diminished its ability to inhibit signaling by G i or G q , suggesting that N-terminal palmitoylation may contribute to GAP activity in cells. It will be important to determine whether Cys 97 in RGS16, homologous to Cys 95 in RGS4, is palmitoylated under these conditions. Alternatively, cellular effects of mutating N-terminal Cys residues may not reflect only the absence of palmitoylation. For example, a synthetic peptide that corresponds to residues 1-33 of RGS4 inhibited signaling through an apparently GAP-independent mechanism, and substitution of Cys residues in this peptide eliminated this inhibition (41). Thus, while it is clear that palmitoylation of RGS proteins occurs and can alter their activities in vitro, it will be important to find which Cys residues are palmitoylated in cells, how palmitoylation is controlled and what effects palmitoylation of specific Cys residues has on RGS protein function.