Dual regulation of Akt/protein kinase B by heterotrimeric G protein subunits.

While positive regulation of c-Akt (also known as protein kinase B) by receptor tyrosine kinases is well documented, compounds acting through G protein-coupled receptors can also activate Akt and its downstream targets. We therefore explored the role of G protein subunits in the regulation of Akt in cultured mammalian cells. In HEK-293 and COS-7 cells transiently transfected with beta(2)-adrenergic or m2 muscarinic receptors, respectively, treatment with agonist-induced phosphorylation of Akt at serine 473 as evidenced by phosphoserine-specific immunoblots. This effect was blocked by the phosphatidylinositol-3-OH kinase inhibitor LY294002 and wild-type Galpha(i1), and was not duplicated by co-transfection of the constitutively active Galpha(s)-Q227L or Galpha(i)-Q204L mutant. Co-transfection of Gbeta(1), Gbeta(2) but not Gbeta(5) together with Ggamma(2) activated the kinase when assayed in vitro following immunoprecipitation of the epitope-tagged enzyme. In contrast, constitutively activated G protein subunits representing the four Galpha subfamilies were found unable to activate Akt in either cell line. The latter results are in disagreement with a report by Murga et al. (Murga, C., Laguinge, L., Wetzker, R., Cuadrado, A., and Gutkind, J. S. (1998) J. Biol. Chem. 273, 19080-19085) that described activation of Akt in response to mutationally activated Galpha(q) and Galpha(i) transfection in COS cells. To the contrary, in our experiments Galpha(q)-Q209L inhibited Akt activation resulting from betagamma or mutationally activated H-Ras co-transfection in these cells. In HEK-293 cells Galpha(q)-Q209L transfection inhibited insulin-like growth factor-1 activation of epitope-tagged Akt. In m1 muscarinic receptor transfected HEK-293 cells, carbachol inhibited insulin-like growth factor-1 stimulated phosphorylation at Ser(473) of endogenous Akt in an atropine-reversible fashion. We conclude that G proteins can regulate Akt by two distinct and potentially opposing mechanisms: activation by Gbetagamma heterodimers in a phosphatidylinositol-3-OH kinase-dependent fashion, and inhibition mediated by Galpha(q). This work identifies Akt as a novel point of convergence between disparate signaling pathways.

Membrane bound heptahelical receptors utilize both the G protein ␣ and ␤␥ subunits to transmit extracellular signals to the inside of the cell (1). Numerous reports now conclusively show that the ␤␥ heterodimer can regulate a diverse array of effector molecules such as inwardly rectifying potassium channels, G protein-coupled receptor kinases, certain isoforms of adenylyl cyclase, phospholipase C-␤, and phosphatidylinositol-3-OH kinase (PI3K), 1 and other as yet unidentified upstream targets in the mitogen-activated protein kinase (MAPK) and c-Jun N-terminal kinase (JNK) pathways (2,3). Attenuation of ␤␥-mediated signaling is achieved by the binding of GDPbound G␣ to the ␤␥ subunits. Studies have shown that the effector interaction sites on the ␤␥ dimer partially overlap one another as well as with the G␣-binding site, and hence could explain GDP-bound G␣'s ability to blunt ␤␥ signaling potential (4).
3Ј-Phosphorylated phospholipids generated by PI3K act as regulatory cofactors to a variety of cellular components (5). One such cellular target is Akt (6). c-Akt is the cellular homologue of the viral Akt isolated from a transforming retrovirus (AKT8 which induces T-cell lymphoma in AKR mouse, hence the name Akt) (7). Since Akt shares sequence similarity with protein kinase A and protein kinase C it is also known as protein kinase B or PKB (8,9). The three isoforms (Akt-1, -2, and -3) of Akt represent a small subfamily of second messenger-regulated serine/threonine protein kinases (10). Structurally, Akt contains a pleckstrin homology domain at its amino-terminal end, a kinase domain in the central region and a carboxyl-terminal tail region. The pleckstrin homology domain of Akt has been shown to bind phosphatidylinositol (3,4,5)-triphosphate and phosphatidylinositol (3,4)-bisphosphate (11,12) and the ␤␥ subunits of G proteins (13). Two key regulatory residues, threonine 308 and serine 473, are located in the kinase and the carboxyl-terminal tail region, respectively (see below).
Growth factor-mediated activation (via tyrosine kinase receptors) of Akt is abrogated by pharmacological inhibitors of PI3K such as wortmannin (14 -16) and LY294002 (17). Evidence is growing to support the following mode of activation of Akt. PI3K-generated phospholipids phosphatidylinositol (3,4,5)-triphosphate and phosphatidylinositol (3,4)-bisphosphate bind to the pleckstrin homology domain of Akt and lead to the translocation of the kinase to the plasma membrane (18). Two distinct 3Ј-phosphoinositide-dependent kinases, PDK1 and a yet to be identified PDK2 phosphorylate Akt on the regulatory threonine 308 and serine 473, respectively (19,20). Much less is known regarding Akt activation by wortmannininsensitive pathways in response to cAMP and its analogues (21,22) as well as cellular stresses such as heat shock and hyperosmolarity (23). Both class I A PI3K and class I B PI3K can activate Akt (24). Class I A PI3K are heterodimers consisting of a catalytic subunit (␣, ␤, or ␦ isoforms) bound to a p85 or analogous adapter molecule and mediate tyrosine kinase receptor pathways connecting to Akt. Class I B PI3K contain the ␥ catalytic isoform, does not bind p85 (or analogous adapter molecules), and is stimulated by G protein ␤␥ subunits (25). A novel PI3K␥specific adapter was recently cloned and may mediate co-activation by G protein ␤␥ (26), although the role of p101 as an essential adapter has been questioned (27). Although both classes of PI3K contain a Ras-binding domain in their aminoterminal region and bind Ras-GTP, only class I A isoforms are effectively stimulated by Ras (28). While class I A PI3K is stimulated by tyrosine-phosphorylated peptides, class I B subclass is not stimulated by such peptides. There is no evidence to suggest that the PI3K␥ participate in the tyrosine kinase receptormediated pathways. On the contrary, recent reports indicate that ␤␥ subunits may interact with both class I A and I B isoforms of PI3K (29,30). ␤␥ Subunits have also been shown to mediate activation of MAPK in a PI3K␥but not PI3K␣-dependent manner (31). Physiological substrates of Akt include glycogen-synthase kinase 3 (GSK3) (32), 6-phosphofructo-2kinase (33), and the Bcl2/BclXL-associated death factor (BAD) (34,35). Apart from mediating insulin-stimulated metabolic pathways (36), activated Akt plays a critical role in promoting cell survival by opposing apoptotic pathways in cerebellar neurons and other cell types (37)(38)(39).
Regulation of Akt by G protein-coupled receptors is poorly understood. Initial studies done in rat-1 fibroblasts reported that the mitogen lysophosphatidic acid which binds to a G i /G qcoupled GPCR did not activate Akt (15). However, recent data indicate that Akt can be transiently activated in a variety of cells in response to GPCR ligands such as thrombin (14), isoproterenol (22), fMLP, interleukin-8 (40), and carbachol (41). Furthermore, forskolin, which stimulates adenylyl cyclase and thus increases cyclic AMP, as well as the non-hydrolyzable analogs of cAMP have also been shown to activate Akt (21). What is unclear is the fact that such GPCR-mediated activation of Akt has been shown to involve both wortmannin-sensitive (12,40,41) and -insensitive (21,22) pathways. To better understand the immediate downstream partners involved in GPCR-mediated activation of Akt we focused our attention on the G protein subunits ␣ and ␤␥. We demonstrate here that in HEK-293 cells, isoproterenol acting through ␤ 2 -adrenergic receptors can induce phosphorylation of Akt, an effect blocked by inhibitors of PI3K, but not duplicated by mutationally activated G␣ s . We further show that only ␤␥, not G␣ are capable of activating an epitope-tagged transiently transfected Akt. We present additional evidence that c-Akt can be dually regulated by G proteins in a highly specific fashion in as much as constitutively active G␣ i and G␣ q interfere with Akt activation at a step downstream from ␤␥-effector interactions. This sets the stage for exploration of possible downstream effects of GPCR, mediated by Akt, such as attenuation in programmed cell death or metabolic or gene regulating effects mediated by GSK3␤.

EXPERIMENTAL PROCEDURES
cDNA Constructs-Expression plasmid pAU5 was generated by modification of pcDNA3 (Invitrogen). Between the HindIII and XbaI sites of pcDNA3, complementary synthetic oligonucleotides were ligated encoding a nonapeptide of the sequence MATDFYLKGS, in which the starting Met is in a context favorable for translation initiation (42) and the Gly and Ser codons comprise a BamHI site GGATTC, followed by XhoI and EcoRI restriction sites. The peptide sequence TDFYLK is the target of the monoclonal antibody AU5 (BabCo). The full-length coding region of murine Akt-1 (GenBank TM accession number X65687) was obtained by amplification of mouse brain cDNA (CLONTECH) with specific primers (sequences of primers employed available upon request) by the polymerase chain reaction (PCR). The PCR reactions employed Pyrococcus woesei (Roche Molecular Biochemicals) thermostable DNA polymerase. Sequences encoding the restriction endonuclease sites for BamHI and EcoRI were designed into the 5Ј-sense and 3Ј-antisense primers, respectively. The PCR-amplified product corresponding to codons 2 to 481 of wild-type Akt-1 was digested with BamHI and EcoRI, then subcloned in-frame into the corresponding restriction sites of the pAU5 vector. Seven silent mutations were noted relative to the Gen-Bank TM sequence corresponding to the wild-type codons as follows: Ala 107 -GCa, Lys 154 -AAg, Lys 163 -AAg, Arg 243 -CGt, Asp 302 -GAc, Tyr 340 -TAc, Ser 396 -TCc. Constructs in pcDNA3 encoding G␤ 1 , ϪG␤ 2 , -G␤ 5 , -G␥ 1 , -G␥ 2 , -G␥ 2 * (C68S) were described previously (43)(44)(45)(46). pCDPS-G␣ i1 was a gift from Dr. M. Y. Degtyarev. pCDPS-G␣ il -QL, in which glutamine 204 was mutated to leucine, was prepared from this construct using the QuikChange (Stratagene) kit according to manufacturer's protocol.
pCEFL-G␣ 12 and -G␣ 12 -Q229L and pcDNA1-G␣ q and -G␣ q -Q209L were a kind gift of S. J. Gutkind. The coding regions of the latter two constructs were subcloned between the BamHI and NotI sites in pcDNA3. The constructs pCDNA1.1-G␣ s and -G␣ s -Q227L were purchased from American Type Tissue Collection (ATCC). pCIS2-H-Ras-Q61L, a constitutively active form of human H-Ras, was a kind gift of Dr. Michael Quon (NHLBI, National Institutes of Health). cDNAs encoding the human m1 mAchR, m2 mAchR, and ␤ 2 -adrenergic receptor (␤ 2 AR) in pCD-PS were kindly provided by Dr. Jurgen Wess (NIDDK, National Institutes of Health). Constructs were amplified in Escherichia coli XL1Blue (Stratagene) and purified using either a plasmid DNA isolation kit (Qiagen Maxi Prep) or by equilibrium sedimentation in a cesium chloride density gradient. The DNA sequence of all inserts generated or modified by PCR was verified by the dideoxy chain termination employing ABI Prism Dye Terminator Cycle Sequencing Ready Reaction Kit (PerkinElmer Life Sciences).

Transient Transfection of Expression Plasmids into Cultured Mammalian Cells-Human embryonic kidney 293 (HEK-293) and COS-7
African green monkey kidney cells were maintained in Dulbecco's modified Eagle's medium supplemented with glutamine, penicillin, streptomycin, and 10% fetal bovine serum (termed "complete medium") and were passaged every 2-3 days. Purified plasmids were introduced into the cells (70 -80% confluent) using LipofectAMINE (Life Technologies Inc.) or Superfect (Qiagen) reagent according to manufacturer's recommendations. Typical transfection reactions consisted of a total of 16 g of DNA per 75-cm 2 flask and 40 l of the transfection reagent. The negative control flasks were transfected with the equivalent amount of vector DNA. 24 h after transfection, the cells were serum starved for 16 h before harvesting. Isoproterenol and carbachol (without or with atropine at 100 nM) treatments were performed by the addition of prewarmed, drug-containing serum-free medium to the adherent cells and incubating them at 37°C in a tissue culture incubator for 30 or 10 min, respectively. IGF stimulation was performed by adding serum-free medium containing 1% bovine serum albumin and IGF-1 (Life Technologies, Inc.) at the concentration indicated in the figure legends to the adherent cells and incubating them at 37°C for 5-10 min. Pretreatment of cells with LY294002 or wortmannin was done by incubating the adherent cells at 37°C for 1 h in serum-free medium containing 10 M LY294002 (Sigma, in ethanol) or 100 nM wortmannin (Sigma) prior to stimulation. The inhibitor was included during the stimulation step.
Immunoblotting-Phosphorylation of Akt was estimated by subjecting the cell lysates (see below) to SDS-PAGE under reducing conditions and immunoblotting using phospho-Akt (Ser 473 ) antibody (1:1000 dilution) (New England Biolabs), which detects Akt only when phosphorylated at serine 473. In addition, anti-Akt polyclonal antibody (New England Biolabs), which recognizes Akt-1 irrespective of its activation state, was used to evaluate the expression of Akt under various transfection conditions. The mouse monoclonal antibody AU5 in ascites fluid (BabCo) was employed for blotting of AU5 epitope-tagged Akt. Expression of G␣ q , G␤ 1 , and H-Ras was estimated in immunoblots employing antibodies anti-G q / 11␣ C-19 (Santa Cruz), RA (47), and anti-Ras (Transduction Laboratories), respectively. The bound primary antibodies were visualized either by the enhanced chemiluminescence detection (ECL) method with appropriate secondary antibody (Roche Molecular Biochemicals), or by incubation with 125 I-labeled Protein A (DuPont) and autoradiography on a storage phosphor screen (Molecular Dynamics) when quantitation was desired.
Cell Lysis, Immunoprecipitation of Akt, and in Vitro Kinase Assay-The cells were scraped off the flask using 1 ml of lysis buffer per 75-cm 2 flask. Lysis buffer consists of 20 mM Tris, pH 7.4, 300 mM NaCl, 3.5 mM EDTA, 1.0 mM EGTA, 1.0% Triton X-100, 2.5 mM sodium pyrophos-phate, 40 mM ␤-glycerophosphate, 1.0 mM sodium vanadate, 100 nM okadaic acid, 2.0 g/ml leupeptin, 2.0 g/ml aprotinin, and 1.0 mM Pefablock (Roche Molecular Biochemicals). The whole cell lysates were centrifuged at 16,000 ϫ g for 30 min at 4 o C and 800 l of the clear supernatant was incubated for 2 h with 3 l of AU5 monoclonal antibody. At the end of the incubation 30 l of GammaBind TM G-Sepharose (Amersham Pharmacia Biotech) beads (50% slurry) were added and incubated on a rocking platform at 4 o C for a further 30 -60 min. The beads were sequentially washed with ice-cold lysis buffer, water, and kinase buffer for a total of 5 times. The washed immunoprecipitates were then incubated with 50 l of kinase reaction buffer consisting of 25 mM Tris, pH 7.5, 10 mM MgCl 2 , 10 mM MnCl 2 , 100 M ATP, 2 mM dithiothreitol, 5 mM ␤-glycerophosphate, 0.1 mM sodium vanadate, and 30 M cross-tide peptide (32) (Upstate Biotechnology) or histone H2B (Roche Molecular Biochemicals) and 4 Ci of [␥-32 P]ATP or [␥-33 P]ATP (DuPont), at 30°C for 15 min. At the end of the kinase reaction, 15 l of the reaction solution was spotted on to a phosphocellulose SpinZyme tube (Pierce). The filter was washed with 0.75% phosphoric acid solution, air-dried, and the peptide specific radioactivity was measured using a liquid scintillation counter.

RESULTS AND DISCUSSION
Isoproterenol-mediated Akt Phosphorylation Not Mimicked by Activated G␣ s -The agonist isoproterenol acting through ␤ 3 -adrenergic GPCR has been shown to activate Akt and thereby inactivate GSK3 in rat epididymal fat cells (22). It is not clear, however, whether signals generated by activated G␣ s or freed G␤␥ subunits or both are involved. This same group showed that forskolin, a direct activator of the Gs ␣ -effector adenylyl cyclase, was unable to inhibit GSK3 activity (22), implying that forskolin did not activate Akt. On the contrary, the eicosanoid prostaglandin E1, which binds to and activates a G s -coupled GPCR, as well as forskolin have been shown to activate Akt in HEK-293-EBNA cells (21). Moreover, wortmannin, a fungal metabolite and a known inhibitor of PI3K had no effect on the activation of Akt in either of these studies. In order to address the question of which G protein subunits participate in Akt activation we used a transient transfection assay system where we tested the ability of overexpressed signaling proteins to regulate Akt activation.
Stimulation of cells transfected with ␤ 2 -adrenergic receptor with 200 M isoproterenol markedly increased the phosphorylation of a key regulatory serine residue in the carboxyl-terminal tail of endogenous Akt in HEK-293 cells, as evidenced by immunoblotting with a phospho-specific antibody (Fig. 1, panels A-C, lane 3). Prior incubation of cells with the ␤-adrenergic receptor antagonist propranolol completely abolished the isoproterenol-induced Akt phosphorylation (not shown). A positive control, mutationally activated H-Ras (RasQL) known to activate endogenous PI3K, also induced Akt phosphorylation. The extent of ␤ 2 -adrenergic receptor-mediated phosphorylation of Akt is comparable to that of Ras-QL (Fig. 1, panels A-C, cf.  lanes 3 and 7). The compound LY294002, a reversible inhibitor of PI3K, which does not discriminate among the PI3K isoforms, completely inhibited isoproterenol-induced Akt phosphorylation (Fig. 1, panels A-C, lane 4). Our results are consistent with the published reports in which wortmannin or LY294002 inhibited Akt activation (12,40,41), and we conclude that isoproterenol-mediated Akt phosphorylation in transfected HEK-293 cells, and by extension activation, involves PI3K.
Mutant G␣ subunits, in which a conserved glutamine residue (Gln 227 in the long form of G␣ s ) is mutated to leucine, lack GTPase activity, and are hence persistently active (48). Cotransfection of mutationally activated G␣ s , G s -QL, did not stimulate Akt phosphorylation (Fig. 1, panels A-C, lane 6), suggesting that ␤ 2 -adrenergic receptor-mediated Akt signaling does not involve G␣ s subunit. HEK-293 cells transfected with both ␤ 2 -adrenergic receptor and wild-type G␣ i1 (G␣ i ) did not respond to isoproterenol (Fig. 1, panels A-C, lane 5). Since wild-type G␣ i1 acts as one of the natural binding partners of G␤␥ subunits, it is possible that the transfected G␣ i was able to sequester free ␤␥ complex that is released upon ␤ 2 -adrenergic receptor stimulation. Overexpression of these signaling molecules did not alter the steady-state expression of endogenous Akt (Fig. 1, panel B). Taken together, these results lead to the hypothesis that ␤ 2 -adrenergic receptor mediated activation of Akt in HEK-293 cells involves ␤␥ subunits and an LY294002sensitive PI3K.
m2 Muscarinic Receptor-mediated Akt Phosphorylation Not Mimicked by Activated G␣ i -To probe the generality of the above findings, a similar approach was applied to COS-7 cells transfected with m2 mAchR, a GPCR known to couple to G iregulated pathways. Because of the lower level of endogenous Akt in these cells compared with HEK-293 cells, they were co-transfected with a cDNA encoding an AU5 epitope-tagged form of Akt-1. The muscarinic agonist carbachol induced phosphorylation of Akt-Ser 473 in m2 co-transfected cells as evidenced by phospho-specific immunoblots (Fig. 1, panels D-F,  lane 3), an effect blocked by pretreatment with LY294002 ( Fig.  1, panels D-F, lane 4). In COS cells lacking transfected m2 receptors, carbachol induced no such Akt phosphorylation (not shown). These results are consistent with a recent report documenting carbachol-induced Akt activation in COS cells cotransfected with m2 mAchR and hemagglutinin-tagged Akt reporter enzyme (41). As seen with the isoproterenol response in ␤ 2 -adrenergic receptor-transfected HEK-293 cells (Fig. 1, panels A-C), coexpression of wild-type G␣ i blunted Akt-Ser 473 phosphorylation in response to carbachol in m2 mAchR-transfected COS cells (Fig. 1, panels D-F, lane 5). Co-transfection of COS cells with ␤ 1 ␥ 2 or Ras-QL, but not mutationally activated G␣ i subunit, G␣ i -QL, induced Akt-Ser 473 phosphorylation (Fig.  1, panels D-F, lanes 6 -8). None of the co-transfections significantly affected expression of the AU5-tagged Akt reporter (Fig.  1, panel E). These data are consistent with a model in which m2 mAchR-mediated agonist activation of Akt involves G␤␥ subunits acting through a PI3K-dependent pathway, but would appear at odds with reports of G␣ i -QL-mediated Akt activation in COS cells (41). The possible basis of this apparent discrepancy in the G␣ data was explored (see below), after further characterization of the G␤␥ effects.
Characterization of G␤␥-mediated Akt Activation-The results of Akt-Ser 473 phospho-specific immunoblotting experiments in both HEK-293 and COS cells described above were consistent with two classes of GPCR acting through G␤␥ to activate Akt in a PI3K-dependent fashion. Since G␤␥ subunits have been shown to stimulate PI3K isoforms (26,27,30) and PI3K is placed upstream of Akt activation (49), we tested the ability of transfected combinations of ␤ and ␥ isoforms and point mutants to activate Akt in a direct immunoprecipitation assay. To increase the ratio of signal to noise we used the AU5-epitope-tagged version of Akt. Akt activity was measured in immunoprecipitates containing the epitope-tagged Akt by an in vitro kinase assay employing a synthetic peptide derived from GSK3 ("cross-tide") (32). Co-transfection of ␤ 1 ␥ 2 with epitope-tagged Akt led to a consistent 2-3-fold activation of Akt. In HEK-293 cells, under the conditions tested, singular co-transfection of ␤ 1 , ␤ 2 , and ␥ 2 (Fig. 2), or ␤ 5 (not shown) did not stimulate Akt. As shown in Fig. 2, ␤ 1 ␥ 2 -mediated Akt activation is completely inhibited by the irreversible PI3K inhibitor wortmannin (lane 6). Our results are in concurrence with three of the four studies performed in a variety of cell lines in which PI3K inhibitors were tested against G protein-coupled Akt activation, in which the inhibitors markedly decreased Akt activity and/or function (12,40,41).
As has been previously shown by Murga et al. (41), membrane targeting of ␤␥ subunits is critical inasmuch as the non-isoprenylated ␥ 2 point mutant, ␥ 2 -C68S (␥2*) which targets the ␤ subunit to the cytosol did not support Akt activation when co-transfected with ␤ 1 or ␤ 2 (Fig. 2, lanes 8 and 11). The specificity of coupling between ␤ and ␥ subunits is highlighted by the fact that co-transfection of ␤ 2 /␥ 1 , a pair of subunits which does not form functional dimers (50), fails to support Akt activation (Fig. 2, lane 10). As shown in Fig. 2, ␤ 5 ␥ 2 failed to activate Akt (lane 12). It has been shown that the brain specific ␤ 5 which can efficiently dimerize with ␥ 2 can activate phospholipase C-␤ 2 (51), but not the MAPK or JNK pathways (46). Recent reports have implicated PI3K as a mediator of ␤␥ signaling to MAPK and JNK (31,52). Taken together, the inability of G␤ 5 to activate the MAPK/JNK cascades and Akt likely represents the failure to interact with a common intermediate such as PI3K␥ (53).
The role of G␣ and G␤␥ Subunits in Akt Activation-Certain G protein-regulated effectors such as adenylyl cyclase and phospholipase C can be independently regulated by G␣ and G␤␥ subunits (1). In light of this, and the apparent inconsistencies between the results of Murga et al. (41) and our own Akt-Ser 473 phospho-specific immunoblots, with respect to the GTPase-deficient G␣ i -QL mutant noted above (Fig. 1), we further investigated the ability of G␣ subunits to activate epitopetagged Akt in a direct immunoprecipitation assay. We employed the GTPase-deficient mutants of G i1 , G s , G q , and G 12 (G i -QL, G s -QL, G q -QL, G 12 -QL) as representatives of the four known G␣ subfamilies (1).
The G␣ mutants were tested in both mammalian cell types used in the Akt-Ser 473 phospho-specific immunoblot experiments described above, in kinase assays involving immunoprecipitated AU5-Akt and the GSK3-derived synthetic peptide substrate. In COS-7 cells, co-transfection of neither wild-type G␣ (not shown) nor mutationally activated G␣ subunits stimulated Akt activity, under conditions in which Akt stimulation by Ras-QL and ␤ 1 ␥ 2 co-transfection was readily demonstrable (Fig. 3A). In COS-7 cells, the co-transfection of the mutationally activated G␣ subunits had little effect on the expression of the AU5-tagged reporter enzyme, with the exception of G s -QL which enhanced its expression, an effect not reflected in the corresponding activity measurements (Fig. 3A). In HEK-293 cells, similar results were seen: ␤ 1 ␥ 2 but not the four mutation- FIG. 2. Stimulation of Akt activity by G␤␥ subunits. HEK-293 cells were transiently transfected with pAU5-Akt in combination with either vector alone (C) or plasmids encoding the G␤ and G␥ isoforms alone or together as indicated. G␥ 2 * is the non-isoprenylated C68S mutant of G␥ 2 (43). Cells were incubated with wortmannin (100 nM) prior to harvesting where indicated. Akt enzyme activity was estimated in immunoprecipitates of AU5-Akt by an in vitro kinase assay using cross-tide peptide as a substrate as described under "Experimental Procedures." Akt expression in each flask was determined by SDS-PAGE of cell lysates followed by immunoblotting with AU5 monoclonal antibody (lower panel). A comparable pattern of results was obtained in two additional independent experiments. ally activated G␣ subunits consistently activated Akt (Fig. 3B). These findings were at odds with previously published results in COS cells in which both transfected G i -QL and G q -QL were found to activate epitope-tagged Akt (41).
Methodologic differences between Murga et al. (41) and the present work were considered as a source of the discrepant findings. Beside the use of different NH 2 -terminal Akt epitope tags (hemagglutinin versus AU5) and corresponding immunoprecipitating antibodies, a possibly important difference between the studies was the use of different peptide substrates to assess Akt activity in the kinase assays. To directly examine this question, immunoprecipitates from lysates of COS-7 cells co-transfected with AU5-Akt and ␤ 1 ␥ 2 , G␣ q -QL, or Ras-QL were split and Akt activity assayed in parallel with either histone 2B (H2B) (41) or GSK3-derived peptide as substrate. As seen in Fig. 4, using kinase substrate, ␤ 1 ␥ 2 , and Ras-QL, but not G␣ q -QL was found to activate Akt. Specific immunoblots from this experiment document that the failure of G␣ q -QL to activate Akt was not due to lack of expression of either the reporter enzyme or the G␣ q -QL construct (Fig. 4, lower panels). Thus the choice of different peptide substrates in the kinase assay is unlikely to account for the differences between Murga et al. (41) and the present study.
Inhibitory Effects of G␣ q -QL and G q -linked Receptors on Akt Activation-As noted above, no stimulatory effects of G␣ i -QL and G␣ q -QL on Akt were noted in our studies. Moreover, during the course of the functional survey of the mutationally activated G␣ subunits, we consistently found inhibitory effects due to G␣ q -QL transfection in both COS-7 and HEK-293 cells under certain conditions, as illustrated in Figs. 5 and 6. As noted above, immunoblotting with Akt-Ser 473 phospho-specific antibodies of lysates from COS cells co-transfected with AU5tagged Akt revealed ϳ4-fold stimulation by ␤ 1 ␥ 2 (Fig. 5A, lanes  3 and 4). This ␤ 1 ␥ 2 -dependent increase in Akt-Ser 473 phosphorylation was abolished by co-transfection with G␣ q -QL (Fig. 5A,  lanes 5 and 6). Parallel immunoblots for AU5-Akt, G␤ 1 , and G␣ q documented their expression under conditions in which Akt-Ser 473 phosphorylation was abolished (Fig. 5A, lower panels), excluding the possibility that the inhibitory effect was due to impaired expression of the former two constructs. Functional assays of AU5-Akt reporter enzyme activity demonstrated the inhibitory effects of G␣ q -QL in COS cells as well when tested against either G␤ 1 ␥ 2 and Ras-QL stimulation of Akt (Fig. 5B). The possibility that the inhibition of Akt function by G␣ q -QL was due to reduced expression of AU5-tagged reporter, G␤ 1 , or mutant H-Ras was excluded by analysis of protein expression with specific antibodies in parallel immunoblots (Fig. 5B, lower  panels).
In HEK-293 cells, analysis of Akt function in kinase assays employing immunoprecipitated AU5 epitope-tagged reporter enzyme stimulated by G␤ 1 ␥ 2 or Ras-QL revealed similar inhibition by G␣ q -QL (not shown). In other experiments in HEK-293 cells, stimulation of epitope-tagged Akt by IGF, mediated by endogenous IGF-I receptors (54), was readily demonstrable FIG. 4. Comparison of peptide substrates for assessment of Akt kinase activity. COS7 cells were transfected with AU5-Akt and either vector alone (Control) or ␤ 1 and ␥ 2 , G␣ q -QL, or Ras-QL cDNAs as indicated. Cell lysates were harvested 48 h post-transfection, divided into four aliquots, and kinase assays were performed on duplicate lysates using either cross-tide (32) (A) or histone H2B (41) (B) as substrates as described under "Experimental Procedures." Expression of AU5-Akt and G␣ q in each flask was ascertained by immunoblotting of the total lysates (lower panels). This experiment was repeated twice with the same result. (Fig. 6, lane 2), consistent with the reports of others (55). Coupling of IGF-I receptors to activation of Akt is wortmanninsensitive (55) and is mediated by interaction of the IGF-I receptor with the p85 adaptor subunit of Class I A PI3K (56). When tested against IGF stimulation, G␣ q -QL, but not the other mutationally activated G␣ subunits, reproducibly inhib-ited AU5-Akt activity in kinase assays of immunoprecipitated enzyme without affecting reporter expression (Fig. 6).
The ability of G␣ q -QL to inhibit IGF-stimulated Akt activity suggested the potential of cross-talk between GPCR and receptor-tyrosine kinase-regulated signaling pathways. To test this possibility, HEK-293 cells were transfected with m1 mAchR and the regulation of endogenous Akt phosphorylation at Ser 473 was assessed by phospho-specific immunoblotting of the cellular lysates. m1 mAchR activate downstream pathways by coupling to G q/11 (57). The stimulatory effect of IGF on Akt-Ser 473 phosphorylation was inhibited by treatment of cells with HEK-293 cells were transiently transfected with AU5-Akt, and either vector alone (lanes 1 and 2) or in combination with G␣ i -Q204L, G␣ s -Q227L, G␣ q -Q209L, or G␣ 12 -Q229L, as indicated. IGF treatment (100 ng/ml) and kinase assays on immunoprecipitated enzyme were performed as described under "Experimental Procedures." SDS-PAGE and immunoblotting (lower panel) confirmed expression of AU5-Akt. Similar results were obtained in three other experiments. the muscarinic agonist carbachol (Fig. 7, cf. lanes 2 and 3). The effect of carbachol was receptor-mediated inasmuch as the muscarinic antagonist atropine blocked the inhibition (Fig. 7,  lane 4). Treatment with IGF, carbachol, or atropine had no effect on the expression level of endogenous Akt in these cells (Fig. 7, lower panel). Because activation of m1 mAchR would be expected to release free G␤␥, which has been shown to be stimulatory to Akt in these cells (see above), these results imply the effect of activated G␣ q on Akt is dominant.
In summary these results demonstrate dual regulation of Akt by G protein subunits and the potential for opposing regulation of Akt by signals emanating from GPCR and receptortyrosine kinases. The demonstration of Akt stimulation by G␤␥ confirms the earlier report of Murga et al. (41), but the basis of discrepancies between the laboratories with respect to the G␣-QL data remains unclear. Our failure to demonstrate Akt activation in response to G␣ i -QL and G␣ q -QL transfection was consistent in two cell lines and using two different kinase assay substrates, ruling out such differences as the basis for the discrepancies.
We found instead that activated G␣ q had a prominent inhibitory effect seen in both COS and HEK-393 cells on Akt stimulated by G␤␥, Ras-QL, or IGF treatment. It is possible since G␣ q activates phospholipase C-␤ that protein kinase C may play a role in such Akt inhibition. Support for such a link comes from recent reports in which phorbol ester activators of protein kinase C inhibited insulin-stimulated Akt activation in 3T3-L1 adipocytes (58) and IGF-stimulated Akt activation in PC-12 cells (59). The inhibition of Akt by G␣ q may well contribute to the apoptosis induced by transfection of mutationally activated G␣ q in Chinese hamster ovary and COS-7 cells (60). These findings set the stage for further exploration of the potential role of GPCR on cellular processes dually regulated by receptor-tyrosine kinases and mediated by Akt such as programmed cell death and metabolic and transcriptional pathways governed by GSK3␤.