Protein Kinase C Phosphorylates G12α and Inhibits Its Interaction with Gβγ

Of nine G protein α subunits examined, only α12 and αz served as substrates for phosphorylation by various isoforms of protein kinase C in vitro. A close homolog of α12, α13, was not phosphorylated. Exposure of NIH 3T3 cells that stably express α12 to phorbol 12-myristate 13-acetate also resulted in phosphorylation of the protein. Phosphorylation in vitro occurred near the amino terminus (probably Ser38), and approximately 1 mol of phosphate was incorporated per mol of α12. Although G protein heterotrimers containing either α12 or αz were poor substrates for phosphorylation, the isolated α subunits were phosphorylated equally well in their GDP- or GTPγS-bound forms. The guanine nucleotide binding properties of purified α12 and αz were unaltered by phosphorylation, as was the capacity of αz to inhibit type V adenylyl cyclase. However, phosphorylation of either protein greatly reduced its affinity for G protein βγ subunits, consistent with the newly determined crystal structure of a G protein heterotrimer. We suggest that protein kinase C regulates α12- and αz-mediated signaling pathways by preventing their association with βγ.

The two members of the ␣ 12 and ␣ 13 subfamily, discovered most recently, are expressed ubiquitously (4) and share interesting biochemical characteristics, including relatively slow guanine nucleotide exchange and hydrolysis (5,6). Although the receptors and effectors that interact with these G proteins have not yet been identified, overexpression of wild type or mutationally activated ␣ 12 or ␣ 13 transforms fibroblasts (7)(8)(9). Furthermore, overexpression of constitutively activated ␣ 12 or ␣ 13 stimulates Na ϩ /H ϩ exchange activity (10,11). Of interest, Dhanasekaran et al. (10) showed that this stimulatory effect of ␣ 12 , but not that of ␣ 13 , is lost after prolonged exposure of cells to PMA. These results suggest that ␣ 12 and ␣ 13 transduce similar regulatory signals related to cell growth or transformation and that there is further regulation of the ␣ 12 pathway by PKC.
There are other interactions between G protein-regulated pathways and PKC. Treatment of cells with PMA has a variety of often confusing effects on their capacity to synthesize cyclic AMP in response to various activators or inhibitors; certain adenylyl cyclases are activated following phosphorylation by PKC in vitro (12,13). Activation of phospholipase C by muscarinic or ␣ 1 -adrenergic agonists is blocked by treatment of astrocytoma cells or hepatocytes, respectively, with PMA (14,15). The inhibitory effects of substance P on an inward rectifier K ϩ channel appear to be mediated by a pertussis toxin-insensitive G protein and protein kinase C (16). With regard to direct effects of PKC on G protein subunits, there are descriptions of phosphorylation of G z␣ and G i␣ both in vitro and in vivo (17)(18)(19)(20), but the functional significance of such modification has been unclear. We describe here the phosphorylation of ␣ 12 by PKC in vitro and, in addition, in cells exposed to PMA. We further demonstrate that phosphorylated ␣ 12 and ␣ z have reduced affinity for G protein ␤␥ subunits compared to the unmodified ␣ subunits. Similar results with ␣ z have just been reported by Fields and Casey (21).

EXPERIMENTAL PROCEDURES
Purification of G Protein Subunits from Sf9 Cells-␣ 12 and ␣ z were purified from Sf9 cells infected with appropriate baculoviruses after their coexpression with ␤ 1 and His 6 -␥ 2 subunits as described by Kozasa and Gilman (5), with the following modifications. A novel His 6 -␥ 2encoding virus (amino acid sequence MAHHHHHHGG-␥ 2 -(3-71)) was utilized. The resulting protein binds with higher apparent affinity to Ni-NTA than does the protein without the two glycine residues inserted after the hexahistidine tag. After application of the Sf9 cell membrane extract to the Ni-NTA column (Qiagen), the resin was washed extensively with buffer containing 15 (instead of 5) mM imidazole. ␣ 12 was eluted from the Ni-NTA column with buffer containing AMF (30 M AlCl 3 , 50 mM MgCl 2 , and 10 mM NaF) and was purified further on a Mono S HR5/5 column (Pharmacia Biotech Inc.) with solutions containing 10% glycerol to prevent precipitation during concentration of peak fractions.
Trypsin Protection-Phosphorylated ␣ 12 was incubated with GDP (100 M) or GDP ϩ AMF for 10 min at 0°C prior to treatment with TPCK-treated trypsin (20% of the mass of ␣ 12 ) for 20 min at 30°C. After addition of an equal volume of 2 ϫ sample buffer, the products were analyzed by SDS-PAGE, followed by autoradiography or Western blotting with ␣ 12 antiserum J169 (5).
Expression, Labeling, and Immunoprecipitation of ␣ 12 -NIH 3T3 cells that stably express ␣ 12 (NIH 3T3-G12) were obtained by transfection with plasmids pCMV␣ 12 and pSV2Neo and selection in medium containing 600 g/ml G418 (Life Technologies, Inc.). A mixture of G418resistant colonies was collected 20 days after transfection. Expression of ␣ 12 was confirmed by immunoblotting of cell membrane extracts with antiserum J169.
For immunoprecipitation, 25 l of membrane extract was incubated with 2.5 l of 10% fixed Staphylococcus aureus (Pansorbin; Calbiochem) on ice for 30 min. After centrifugation at 15,000 ϫ g for 5 min, supernatants were incubated overnight at 4°C with 7.5 g of anti-␣ 12 IgG or control rabbit IgG. Pansorbin (5 l; 10%) was added for an additional 30 min prior to collection of immunoprecipitates by centrifugation and suspension in 100 l of RIPA buffer. The suspension was layered over 1 ml of RIPA buffer containing 20% sucrose (w/v) and centrifuged at 15,000 ϫ g for 5 min. Pellets were extracted with SDS-PAGE sample buffer, heated (90°C; 3 min), and subjected to SDS-PAGE followed by autoradiography. Gels containing [ 35 S]methionine-labeled proteins were treated with EN 3 HANCE (DuPont NEN).
Miscellaneous Procedures-SDS-PAGE was performed as described by Laemmli (27); gels were stained with silver according to Wray et al. (28). Protein concentrations were estimated by staining with Amido Black (25) or by the method of Bradford (29). Immunoblotting was performed using the ECL chemiluminescence detection system (Amersham). GTP␥S binding and adenylyl cyclase assays were performed as described (5). An IgG fraction of antiserum J169 was prepared by chromatography on a Mono Q HR10/10 column using a gradient of NaCl (0 -200 mM). Membranes from Sf9 cells expressing type V adenylyl cyclase were generously provided by Dr. Christiane Kleuss (this laboratory), while the recombinant baculovirus encoding PKC␣ was a gift from Dr. Shigeo Ohno (Yokohama City University).

RESULTS
Phosphorylation of ␣ 12 and ␣ z by PKC-The results of Dhanasekaran et al. (10), described above, prompted examination of possible phosphorylation of ␣ 12 by PKC. Of the nine G protein ␣ subunits tested, only ␣ 12 and ␣ z were phosphorylated in vitro by PKC (Fig. 1A); the effect on ␣ z was anticipated based on the work of Lounsbury et al. (17). The related ␣ subunit, ␣ 13 , is not a substrate for PKC, also consistent with Dhanasekaran et al. (10). Phosphorylation of ␣ 12 was dependent on Ca 2ϩ and phosphatidylserine, characteristic of PKC (Fig. 1B). Of the several types of PKC tested (PKC␣, -␦, -⑀, and -), all showed the same pattern, phosphorylating only ␣ 12 and ␣ z (data not shown).
We examined NIH 3T3 cells that had been stably transfected with an expression plasmid encoding ␣ 12 to test phosphorylation of the protein in vivo. Immunoblotting of membranes from these cells (NIH 3T3-G12) demonstrates significant expression of ␣ 12 ( Fig. 2A); we could not detect the protein in these cells prior to transfection (using antiserum J169). This antiserum could be used to immunoprecipitate ␣ 12 from a membrane extract of [ 35 S]methionine-labeled NIH 3T3-G12 cells (Fig. 2B), and phosphorylated ␣ 12 was immunoprecipitated from cells FIG. 1. Phosphorylation of G protein ␣ subunits by PKC. A, G ␣ subunits (2.5 pmol) were incubated with 2.5 milliunits of PKC for 20 min. The products were separated by SDS-PAGE, stained with silver, and subjected to autoradiography. Upper panel, silver staining of ␣ subunits; lower panel, autoradiography. ␣ 12 , ␣ 13 , ␣ i1 , ␣ i2 , ␣ i3 , ␣ z , and ␣ q were purified from Sf9 cells as described under "Experimental Procedures." ␣ o and ␣ s were purified from bovine brain and E. coli, respectively. B, ␣ 12 (2.5 pmol) was incubated with 2.5 milliunits of recombinant PKC␣ for 20 min in the presence or absence of 5 M PMA, 10 g/ml phosphatidylserine, or 125 M CaCl 2 as indicated. Proteins were resolved by SDS-PAGE and subjected to autoradiography. labeled with [ 32 P]P i after exposure to PMA (Fig. 2C). Thus, ␣ 12 appears to be phosphorylated in vivo after PKC is activated by phorbol esters.
The time course and stoichiometry of phosphorylation of ␣ 12 in vitro are shown in Fig. 3A. Since the substrate is over 90% pure (based on silver staining; Fig. 1A) and other phosphorylated proteins do not appear in the reaction mixtures (Fig. 1B), we estimated stoichiometry by filtration. When 7 pmol of ␣ 12 was included in the assay, the maximal incorporation of phosphate was about 3 pmol. Since the stoichiometry of binding of GTP␥S to the ␣ 12 used here was about 50% (based on the protein assay), we believe that 1 mol of phosphate is incorporated per mol of ␣ 12 . (␣ 12 is not phosphorylated when denatured; data not shown.) Of interest, ␣ 12 is phosphorylated very poorly after incubation with a 2-fold excess of ␤ 1 ␥ 2 (Fig. 3A); the reaction is almost completely suppressed when ␣ 12 and ␤ 1 ␥ 2 are present at equimolar concentrations (Fig. 3B). Similar results were obtained with ␣ z (Fig. 3C). Nonprenylated ␤␥ subunit complexes have reduced affinity for at least certain G ␣ subunits (22); appropriately, the ␤␥ complex comprised of ␤ 1 and the nonprenylated Cys 68 3 Ser ␥ 2 mutant was a less potent inhibitor of ␣ 12 phosphorylation (Fig. 3B). Since ␤ 1 ␥ 2 did not inhibit the activity of PKC when a specific substrate peptide from myelin basic protein (MBP 4 -14 ) was utilized (data not shown), we conclude that ␣ 12 and ␣ z are not substrates for PKC when associated with ␤␥ in the G protein heterotrimer.
Both the GDP-bound and the GTP␥S-bound forms of ␣ 12 and ␣ z are phosphorylated almost equally well by PKC (Fig. 4A); there was no significant difference in the time course of phos-phorylation of both forms of both proteins (data not shown). Lounsbury et al. (17) reported that the GDP-bound form of ␣ z was phosphorylated more efficiently than the GTP␥S-activated species. The discrepancy may be explained by the fact that the ␣ subunits used in this work were purified from Sf9 cells and thus myristoylated at their amino termini; the protein used by Lounsbury et al. (17) was synthesized in Escherichia coli and was not so modified. Myristoylation of the amino terminus may alter the conformation of this domain, which is the site of phosphorylation (see below).
Since ␤␥ inhibits the phosphorylation of ␣ 12 and ␣ z , we assessed the dependence of this effect on ␤␥ concentration (using both GDP-and GTP␥S-bound forms of ␣ z at the lowest possible concentrations (0.5 nM)) in an attempt to estimate the affinity of ␤␥ for the protein (Fig. 4B). Efforts to measure these affinities have been thwarted in the past by the very high affinity of ␣-GDP for ␤␥ and resultant difficulty in detection of an effect of ␤␥ on ␣ at appropriately low concentrations. However, phosphorylation of ␣ by PKC offers a very sensitive signal. The concentrations of ␤ 1 ␥ 2 required to inhibit (by 50%) phosphorylation of ␣ z -GDP and ␣ z -GTP␥S were 0.5 and 50 nM, respectively. Since the effect of ␤ 1 ␥ 2 on ␣ z -GDP was still close to stoichiometric, there exists at least a 100-fold difference in apparent affinity of ␤ 1 ␥ 2 for ␣ z -GDP and ␣ z -GTP␥S.
The Site of Phosphorylation of ␣ 12 -Phosphorylated ␣ 12 was digested with trypsin either in the presence of GDP or GDP ϩ AMF. Activation of ␣ 12 by AMF protects the bulk of the protein from digestion, and a 40-kDa fragment accumulates (Fig. 5). This fragment is recognized by antiserum J169, which was generated using a peptide corresponding to the carboxyl terminus of ␣ 12 (Fig. 5, lane 2). However, this fragment is no longer phosphorylated (Fig. 5, lane 5). Thus, phosphorylated ␣ 12 can still be activated by AMF, and phosphorylation by PKC occurs near the amino terminus. Similar results were obtained with ␣ z , in which Ser 16 and Ser 27 both appear to be phosphorylated by PKC (17).
Characterization of Phosphorylated ␣ 12 and ␣ z -␣ 12 and ␣ z were phosphorylated and repurified as described under "Experimental Procedures." The stoichiometry of phosphorylation of ␣ 12 was approximately 0.5 based on total protein concentration (presumed stoichiometry approximately 1), while that for ␣ z was 1-1.5; it is possible that ␣ z is phosphorylated at more than one site (17). Phosphorylation did not change the time course of GTP␥S binding (and thus of GDP dissociation) for either ␣ 12 or ␣ z (Fig. 6).
The rate of binding of GTP␥S to nonphosphorylated ␣ 12 or ␣ z is inhibited by ␤␥ (Fig. 7). This reflects the well-known capacity  3. Inhibition of phosphorylation of ␣ 12 and ␣ z by ␤␥. A, ␣ 12 (70 nM) was incubated on ice for 10 min with or without ␤ 1 ␥ 2 (140 nM) and then phosphorylated with PKC. Aliquots (100 l) were withdrawn at the indicated times, filtered, and counted as described under "Experimental Procedures." B, ␣ 12 (50 nM) was incubated with the indicated concentration of ␤ 1 ␥ 2 or ␤ 1 ␥ 2 C68S on ice and then phosphorylated with PKC at 30°C for 20 min. Aliquots were then filtered and counted. C, ␣ 12 or ␣ z (50 nM) was incubated with the indicated concentration of ␤ 1 ␥ 2 on ice for 10 min and then phosphorylated with PKC for 20 min at 30°C. Data are expressed as percent phosphorylation relative to that observed in the absence of ␤ 1 ␥ 2 . In A, B, and C, data are the average of duplicate determinations from a single experiment that is representative of three such experiments.
of ␤␥ to stabilize the GDP-bound form of G protein ␣ subunits. However, the rate of GTP␥S binding to phosphorylated ␣ 12 (Fig. 7A) or phosphorylated ␣ z (Fig. 7B) is not inhibited substantially by a 10-fold molar excess of ␤ 1 ␥ 2 , and the modest effects seen could reflect the presence of small amounts of nonphosphorylated protein in the preparations. The effect of ␤␥ on GTP␥S binding to mock-treated proteins (PKC in the absence of ATP) was the same as that on the nonphosphorylated proteins (data not shown).
We also examined the interaction of ␣ 12 and ␤␥ by gel filtration. The peaks of both phosphorylated and nonphosphorylated ␣ 12 were in fractions 38 -40 (Fig. 8), corresponding to molecular weights of about 45,000. Addition of ␤ 1 ␥ 2 to nonphosphorylated  3 and 6) on ice for 10 min. TPCK-treated trypsin (20% of the ␣ 12 mass) was then added and incubation was continued at 30°C for 20 min. The products were resolved by SDS-PAGE, followed by immunoblotting with antiserum J169 (lanes 1-3) or autoradiography (lanes 4 -6). Lanes 1 and 4 show the sample before digestion with trypsin. to form oligomers with the G protein ␤␥ subunit complex.
Finally, we examined the effect of phosphorylation of ␣ z on its ability to inhibit the activity of type V adenylyl cyclase (5), since this is the only assay available for interaction of ␣ z or ␣ 12 with an effector (Fig. 9). Okadaic acid (1 M) was included in the assay to inhibit phosphatases that might be present in the Sf9 cell membranes utilized as the source of adenylyl cyclase. Phosphorylation of ␣ z had little or no effect on its inhibitory interactions with adenylyl cyclase. DISCUSSION We have demonstrated that ␣ 12 is phosphorylated by PKC both in vitro and in vivo; the homologous subfamily member ␣ 13 is not a substrate. Among the large number of G protein ␣ subunits tested, the only other efficient substrate for phosphorylation by various isoforms of PKC was ␣ z . The stoichiometry of phosphorylation of ␣ 12 was equal to that for GTP␥S binding and is thus assumed to be 1.
Phosphorylation of ␣ 12 occurs within the amino-terminal domain that is removed by trypsin selectively from activated G protein ␣ subunits ( Table I). Examination of corresponding sites of proteolysis in other G ␣ subunits indicates that trypsin probably removes the first 49 or 50 residues from ␣ 12 . There are three serine residues (2, 9, and 38) and one threonine (7) within the relevant sequence. Although Ser 9 and Ser 38 are both candidates for phosphorylation by PKC (30), Ser 38 is surrounded by basic residues (RRRSR) and corresponds to one of the phosphorylated serine residues in ␣ z (Ser 16 ; RRSRR). There is no equivalent of ␣ 12 residues Ser 2 , Thr 7 , or Ser 9 in ␣ z , and there is no equivalent of ␣ z residue Ser 27 (the other phosphorylation site) in ␣ 12 . Although these arguments appear to implicate Ser 38 in ␣ 12 as the site of phosphorylation, Ser 9 cannot be ruled out. Of interest, both Ser 9 and Ser 38 have homologs in ␣ 13 , which is not phosphorylated.
Phosphorylation of ␣ 12 and ␣ z does not appear to change their basic guanine nucleotide binding properties, nor the interactions of ␣ z with type V adenylyl cyclase. However, the affinity of both ␣ subunits for ␤␥ is clearly reduced by phosphorylation, and, reciprocally, their phosphorylation is inhibited by prior interaction with ␤␥. Similar results with ␣ z were just reported by Fields and Casey (21). This effect suggests that phosphorylation of these proteins could play a role in desensitization of the relevant signaling pathways if PKC was stimulated simultaneously. Activation of the G protein causes dissociation of ␣ from ␤␥, and PKC-mediated phosphorylation would thus be favored. Subsequent inhibition of oligomerization as a result of phosphorylation of ␣ would presumably attenuate signaling because of the requirement for ␤␥ for receptor-mediated activation of ␣.
Although the signaling pathway that is regulated by ␣ 12 is unknown, expression of constitutively activated ␣ 12 activates Na ϩ /H ϩ exchange in a PKC-dependent manner (10). Perhaps ␣ 12 activates certain isoforms of PKC either directly or indirectly to stimulate Na ϩ /H ϩ exchange, while PKC attenuates the activity of ␣ 12 in a classic feedback loop.
The crystal structure of ␣ i1 has been determined in its GTP␥S-, free GDP-, and GDP/␤␥-bound forms (31)(32)(33). The conformation of the amino terminus of the ␣ subunit is a particularly dynamic aspect of the nucleotide-and ␤␥-induced structural changes that have been observed. The amino terminus is disordered when ␣ is activated by GTP␥S; it forms a compact subdomain with the carboxyl terminus of ␣ in the free GDP-bound form; it is extended in a long ␣ helix that forms extensive contacts with the ␤ subunit in the heterotrimer. The serine residue in ␣ i1 (Ser 16 ) that is analogous to Ser 38 in ␣ 12 and Ser 16 in ␣ z is part of this interface and is hydrogen-bonded to Lys 89 in ␤ 1 , consistent with the effect of phosphorylation at this site on interactions of ␣ with ␤␥.
The compact subdomain formed by the amino and carboxyl termini of ␣ i1 in the free GDP-bound state is also of interest. This domain appears to be stabilized by interactions between arginine residues at positions 15, 21, and 32 and a sulfate ion contributed by the crystallization solution. Sulfate ions are capable of binding at sites that normally interact with phosphate or phosphoserine (34), and the arginine residues involved are close to the sites of phosphorylation of ␣ 12 and ␣ z . It will be interesting to determine if phosphorylation alters the structure of this microdomain. The specificity of phosphorylation of G protein ␣ subunits by PKC seems problematic, particularly if phosphorylation regulates a property as fundamental as ␣ subunit oligomerization. Although ␣ i2 is apparently phosphorylated following activation of PKC in hepatocytes or the promyelocytic cell line U937 (19,20) and phosphorylation in vitro of a mixture of isoforms of ␣ i by PKC was also described (18), we were not able to demonstrate phosphorylation of specific isoforms of ␣ i with the preparations of PKC used in this study. Perhaps phosphorylation of G protein ␣ subunits is a more general phenomenon than suspected and the appropriate kinases have not yet been identified. FIG. 9. Inhibition of type V adenylyl cyclase by phosphorylated ␣ z . The indicated concentrations of ␣ z were mixed with 20 g of membranes from Sf9 cells expressing type V adenylyl cyclase in the presence of 50 nM GTP␥S-␣ s . Adenylyl cyclase activity was assayed as described under "Experimental Procedures." ␣ subunits were nonphosphorylated ␣ z -GDP (f), nonphosphorylated ␣ z -GTP␥S (q), phosphorylated ␣ z -GDP (ç), phosphorylated ␣ z -GTP␥S (å). The concentrations of GTP␥S-activated ␣ subunits were estimated from [ 35 S]GTP␥S binding. Data shown are the average of duplicate determinations from a single experiment that is representative of three such experiments.