Phospholipase C-independent Activation of Glycogen Synthase Kinase-3β and C-terminal Src Kinase by Gαq*

It is generally thought that activation of phospholipase Cβ (PLCβ) by Gαq accounts for most of the effects of Gq-coupled receptors. Here we describe a novel effect of Gαq that is independent of the PLCβ pathway. Expression of the constitutively active Gαq mutant Gαq(Q209L) promoted an increase in glycogen synthase kinase-3β (GSK-3β) activity that was associated with increased phosphorylation of Tyr216 on GSK-3β. Gαq(Q209L)-AA, a mutant that cannot activate PLCβ, also induced GSK-3β activation and phosphorylation of Tyr216. We speculate that the protein-tyrosine kinase Csk (C-terminal Src kinase), which is also activated by Gαq(Q209L) and Gαq(Q209L)-AA, acts upstream of GSK-3β. Expression of Csk accentuated the activation of GSK-3β by Gαq(Q209L), whereas catalytically inactive Csk blocked GSK-3β activation by Gαq(Q209L). Recombinant Csk phosphorylated and activated GSK-3β in vitro, and GSK-3β coprecipitated with Csk from cell lysates. These results suggest that activation of Csk and GSK-3β by Gαq may contribute to the physiological and pathological effects of Gq-coupled receptors.

The G␣ q protein transduces signals from cell surface receptors that are activated by hormones such as angiotensin II, endothelin 1, catecholamines, and prostaglandin F 2␣ to regulate diverse physiological functions. Activation of certain G qcoupled receptors appears to induce insulin resistance, because treatment of patients with receptor antagonists can improve insulin sensitivity (1,2). In addition, persistent activation of G␣ q is involved in the development of heart failure (3), which is one of the most common causes of death in humans. The most well characterized effector of G␣ q is phospholipase C␤ (PLC␤), 1 the activation of which leads to increased hydrolysis of phosphatidylinositol 4,5-bisphosphate, release of Ca 2ϩ from intracellular stores, and activation of protein kinase C (4). It is generally thought that activation of this canonical PLC␤ pathway by G␣ q accounts for most of the pleiotropic effects of G q -coupled receptors (5).
Glycogen synthase kinase-3 (GSK-3) was discovered as a serine/threonine protein kinase that phosphorylates and inactivates glycogen synthase, but it is now known to regulate a diverse array of cellular processes (6,7). Dysregulation of GSK-3 signaling is thought to play a role in the development of type II diabetes mellitus (8), neuronal cell loss in Alzheimer disease (9), and cancer (10). The two major isoforms of GSK-3 in mammalian tissues (GSK-3␣ and -3␤) are structurally similar but not functionally equivalent. This was demonstrated upon deletion of the GSK-3␤ gene in mice, which resulted in embryonic lethality due to liver degeneration, even though GSK-3␣ was still presumably expressed at the normal level (11). GSK-3␤ activity is relatively high in resting cells, and it can be positively or negatively regulated by various stimuli. For example, insulin activation of the protein kinase Akt results in phosphorylation of GSK-3␤ at Ser 9 , thus reducing its enzymatic activity (12).
Several reports indicated that agonist stimulation of G qcoupled receptors antagonizes the signaling events initiated by insulin and other growth factors that lead to Akt activation (13)(14)(15)(16)(17)(18)(19). To directly investigate how G␣ q affects Akt activity, we used G␣ q (Q209L), a mutant that cannot hydrolyze GTP and therefore constitutively activates PLC␤. We found that G␣ q (Q209L) blocks insulin and platelet-derived growth factor activation of Akt by inhibiting its upstream regulator phosphatidylinositol 3-kinase (PI3K) (20). Active G␣ q still inhibited PI3K/Akt in cells treated with U73122, a PLC inhibitor, suggesting that the inhibitory mechanism is independent of the canonical PLC␤ pathway (20). In this study, we asked whether suppression of PI3K/Akt signaling by activated G␣ q results in increased GSK-3␤ activity. We show here that G␣ q (Q209L) does induce the activation of GSK-3␤, and use of a G␣ q (Q209L) mutant that cannot activate PLC indicates that this response is independent of the canonical PLC␤ pathway. Surprisingly, GSK-3␤ activation is not a consequence of decreased Akt activity but is instead because of activation of a tyrosine kinase. Finally, we identify Csk (C-terminal Src kinase), which has been shown to be activated by G␣ q (Q209L) (21), as a tyrosine kinase that activates GSK-3␤.
were used in cotransfections, except if otherwise noted. Cell lysates were prepared 2 days after transfection in lysis buffer containing 1% Triton X-100 as described earlier (20). Immunoprecipitation, Western blotting, and stripping of membranes were done as described elsewhere (17,20).
DNA Constructs-Akt-HA was obtained from Richard Roth (Stanford University, Stanford, CA). The GSK-3␤ cDNA was isolated by reverse transcriptase-polymerase chain reaction using RNA from HEK 293 cells as template (19). (The RNA in Ref. 19 was erroneously reported to be from Swiss mouse 3T3 cells.) HA-GSK-3␤ and HA-GSK-3␤ S9A were described earlier (19). HA-GSK-3␤ Y216F was constructed using Pfu DNA polymerase (Stratagene) and the forward primer 5Ј-GAGGAGA-ACCCAATGTTTCGTTTATCTGTTCTCGGTACTATAG. G␣ q (Q209L) was described earlier (20). (The mouse G␣ q sequence we isolated shows two differences when compared with the published sequence for G␣ q from mouse brain (22). The C at position 84 in the coding region is G, and the G at position 85 is C. Amino acids 28 and 29 in the mouse G␣ q used here are thus Gln and Leu instead of His and Val. Others have reported a sequence identical to ours (23).) G␣ q (Q209L)-AA was constructed using Pfu DNA polymerase and the forward primer 5Ј-CATGG-AGGAGAGCAAAGCACTCTTTGCAGCAATTATCACCTACCCCTG-GTTC to change Arg 256 and Thr 257 to Ala. The GSK-3␤, G␣ q , and Akt constructs were subcloned into pcDNA3.1 (Invitrogen). The cDNA for human Csk in pcDNA3.1/GS (tagged at the C terminus with the V5 epitope and His 6 ) was obtained from Invitrogen. Csk-KD was constructed using Pfu DNA polymerase and the forward primer 5Ј-CCGAG-GGAACAAAGTCGCCGTCCGGTGCATTAAGAACGACGCCACTGC to mutate Lys 222 to Arg. The fidelity of all cDNA constructs was verified by sequencing.
Kinase Assays-GSK-3␤ activity was assayed using phospho-GS2 substrate peptide (Upstate Biotechnology) as described elsewhere (19). To assay Csk, equal amounts of protein in lysis buffer containing Triton X-100 (20) were immunoprecipitated with His 6 antibody and protein G-agarose at 4°C. The immunoprecipitates were washed twice with lysis buffer and twice with buffer A (50 mM HEPES, pH 7.4, 0.1 mM EGTA, 10 mM MgCl 2 , 1 mM MnCl 2 , and 1 mM sodium orthovanadate). Csk activity was measured in 20 l of buffer A containing 0.25 l/assay of [␥-32 P]ATP, 40 M ATP, 1 mM dithiothreitol, and 5 g/assay of poly(Glu, Tyr) (Sigma). After incubating the reactions at 30°C for 30 min, the mixtures were subjected to SDS-PAGE and autoradiography. Radioactive poly(Glu, Tyr) was cut out of the gel and counted in a scintillation counter.

G␣ q (Q209L) Effects on Akt and GSK-3␤
Are PLC-independent-In an earlier study we suggested that inhibition of PI3K/ Akt by activated G␣ q is independent of the canonical PLC␤ pathway, based on its resistance to U73122 (20). Because U73122 is a relatively poor PLC inhibitor in our experimental systems, we decided to re-examine this issue using G␣ q (Q209L)-AA, a G␣ q (Q209L) mutant that still binds to GTP but does not activate PLC (mutant 10 cited in Ref. 24). Coexpression of G␣ q (Q209L)-AA with HA-tagged Akt in HEK 293 cells caused a decrease in Akt activity that was comparable to that seen using G␣ q (Q209L) (Fig. 1A). These results support our hypothesis that the inhibitory effect of active G␣ q on Akt is independent of PLC activation.
GSK-3␤ activity is negatively regulated by Akt-mediated phosphorylation of Ser 9 (12,25). We therefore expected that suppression of PI3K/Akt by G␣ q (Q209L) would result in an increase in GSK-3␤ activity. This hypothesis was first tested using a stable cell line that expresses G␣ q (Q209L) under the control of a doxycycline-inducible promoter (Flp-In T-REx/293 cells) (20). These cells were transiently transfected with HAtagged GSK-3␤ and then treated overnight with or without 1 M doxycycline to induce expression of G␣ q (Q209L). As expected, GSK-3␤ activity from cells treated with doxycycline was 1.8 times higher than the activity from vehicle-treated cells (p Ͻ 0.01, t test, n ϭ 4). Doxycycline treatment of Flp-In T-REx/293 cells that do not express G␣ q (Q209L) did not affect GSK-3␤ activity. 2 Similarly, cotransfection of HEK 293 cells with HA-GSK-3␤ and G␣ q (Q209L) caused a 2.7-fold increase in GSK-3␤ activity as compared with the control cells (Fig. 1B). G␣ q (Q209L)-AA stimulated GSK-3␤ to a comparable extent (Fig. 1B), indicating that activation of GSK-3␤ by G␣ q (Q209L) is independent of the canonical PLC␤ pathway. Control experiments confirmed that G␣ q (Q209L)-AA does not activate PLC, as measured by accumulation of inositol phosphates, whereas G␣ q (Q209L) causes a robust activation (Fig. 1C).
G␣ q (Q209L) Activates GSK-3␤ via Phosphorylation of Tyr 216 -Using a phosphospecific antibody, we tested whether G␣ q (Q209L) activates GSK-3␤ by causing a decrease in phosphorylation of Ser 9 . Cells were cotransfected with HA-GSK-3␤ and the G␣ q constructs, and the phosphorylation state of GSK-3␤ was examined after precipitation with HA antibody. To our surprise, the phosphorylation state of Ser 9 did not change in the presence of either of the G␣ q constructs ( Fig. 2A,  top panel). Although many agonists, including insulin, regulate GSK-3␤ via changes in phosphorylation of Ser 9 (12,19,25), other stimuli have been shown to activate the enzyme by causing an increase in phosphorylation of Tyr 216 (26,27). Using a phosphospecific antibody that recognizes the latter site, we found that the basal phosphorylation of Tyr 216 in control cells was increased upon expression of either G␣ q (Q209L) or G␣ q (Q209L)-AA ( Fig. 2A, middle panel). Expression of the G␣ q proteins did not affect the amount of HA-GSK-3␤, as demon-2 G. Fan, L. M. Ballou, and R. Z. Lin, unpublished data.

FIG. 1. G␣ q (Q209L) effects on Akt and GSK-3␤ are independent of PLC.
A, cells were cotransfected with Akt-HA and the indicated constructs. Akt activity was assayed in HA immunoprecipitates as described previously (20). A Western blot probed with HA antibody shows that the amount of Akt in each cell lysate was the same (inset). The numbered lanes correspond to the bars on the graph. B, cells were cotransfected with HA-GSK-3␤ and the indicated constructs. GSK-3␤ activity was assayed in HA immunoprecipitates. C, cells were transfected with the indicated constructs. Twenty-four hours later, they were placed into growth medium containing 3 Ci/ml myo-[ 3 H]inositol. The next day, the level of inositol phosphates was assayed following methods described earlier (20). Results are means Ϯ S.E. from three (A and B) or four (C) experiments. strated by Western blotting of cell lysates using the HA antibody ( Fig. 2A, bottom panel).
To further examine the relationship between increased phosphorylation of Tyr 216 and GSK-3␤ activation, we tested whether G␣ q (Q209L) can still activate a GSK-3␤ mutant in which Tyr 216 is changed to Phe (HA-GSK-3␤ Y216F). The enzymatic activity of HA-GSK-3␤ Y216F was ϳ20% that of wild type HA-GSK-3␤ (28), and cotransfection with G␣ q (Q209L) did not increase its activity (Fig. 2B). In contrast, a GSK-3␤ mutant with Ala substituted for Ser 9 (HA-GSK-3␤ S9A) had high basal activity as compared with wild type HA-GSK-3␤ (19), and the presence of G␣ q (Q209L) caused an additional 20% increase in its activity (Fig. 2B). These results support the hypothesis that G␣ q (Q209L)-dependent activation of GSK-3␤ is mediated by phosphorylation of Tyr 216 .
Csk Activates GSK-3␤ in Vivo-It was recently reported that expression of G␣ q (Q209L) causes an increase in the activity of Csk (21), a ubiquitously expressed protein-tyrosine kinase that phosphorylates Src family members to negatively regulate their activity (29,30). Therefore, we wondered if Csk might mediate G␣ q (Q209L)-induced phosphorylation of GSK-3␤. We confirmed that Csk is activated in HEK 293 cells upon coexpression of G␣ q (Q209L) (Fig. 3). G␣ q (Q209L)-AA was as effective as G␣ q (Q209L) in activating Csk (3.3-fold stimulation), indicating that this response is also independent of PLC activation (Fig. 3).
Csk Phosphorylates and Activates GSK-3␤ in Vitro-The results above raised the possibility that Csk might directly activate GSK-3␤ by phosphorylating Tyr 216 . To test whether Csk can phosphorylate GSK-3␤ in vitro, purified recombinant Csk was incubated with purified recombinant GSK-3␤ in the presence of [␥Ϫ 32 P]ATP and LiCl to inhibit GSK-3␤ autophosphorylation. We found that GSK-3␤ was phosphorylated by Csk  3. G␣ q (Q209L) and G␣ q (Q209L)-AA activate Csk. Cells were cotransfected with His 6 -tagged Csk and the indicated G␣ q (Q209L) constructs. Csk activity was assayed in His 6 immunoprecipitates. A Western blot probed with His 6 antibody shows that the amount of Csk in each cell lysate was the same (inset). The numbered lanes correspond to the bars on the graph. G␣ q (Q209L) and G␣ q (Q209L)-AA activated Csk in three experiments. (Fig. 5A). To determine whether phosphorylation by Csk changes GSK-3␤ activity, the two enzymes were incubated separately or together in the presence of ATP prior to assaying GSK-3␤ activity. GSK-3␤ activity was increased 3-fold after incubation with Csk (Fig. 5B). Finally, we tested whether GSK-3␤ and Csk interact in vivo. HA-GSK-3␤ and His 6 -tagged Csk were expressed in cells, and Csk was pulled down using a Ni 2ϩ affinity resin. HA-GSK-3␤ coprecipitated with Csk from lysates of cells expressing both proteins (Fig. 5C). These data support the hypothesis that GSK-3␤ is an in vivo substrate of Csk. DISCUSSION Accumulating evidence indicates that PLC␤ is not the only effector of G␣ q . Direct binding of G␣ q to Bruton's tyrosine kinase (Btk) has been shown to increase Btk kinase activity (31). We found that activated G␣ q inhibits PI3K/Akt signaling via a mechanism that is independent of PLC activation and that might involve an inhibitory interaction between G␣ q and p110␣ PI3K (Fig. 1A and Ref. 20). The results presented here suggest that the activation of Csk and subsequent tyrosine phosphorylation and activation of GSK-3␤ represents another novel effector pathway for G␣ q that functions independently of PLC. An effector pathway with similar characteristics was described in 3T3L1 adipocytes, where G␣ q (Q209L) stimulated the translocation of the GLUT4 glucose transporter in a manner that was independent of PLC activation but blocked by tyrosine kinase inhibitors (32). G␣ q (Q209L)-AA, in which Arg 256 and Thr 257 are changed to Ala, was originally made to identify residues in G␣ q that are required for interaction with PLC (24). We show here that although G␣ q (Q209L)-AA is almost completely defective in the ability to activate PLC, it is indistinguishable from G␣ q (Q209L) in the ability to inhibit Akt and activate GSK-3␤ and Csk. The simplest interpretation of these results is that these effects of G␣ q (Q209L) are independent of PLC activation. In contrast to our results, it was recently reported that the ability of G␣ q (Q209L) to depress PI3K/Akt activity in cardiomyocytes requires PLC activation (33). This conclusion was based on the use of a G␣ q (Q209L) mutant with three Ala substitutions that cannot activate PLC (mutant 7 cited in Ref. 24). An alternative explanation for the inability of this mutant to inhibit Akt is that it lacks the ability to bind to and inhibit p110␣ PI3K.
In light of the inhibitory effect of G␣ q (Q209L) on Akt, we were surprised to find that G␣ q (Q209L) activates GSK-3␤ by causing an increase in Tyr 216 phosphorylation, instead of a decrease in Ser 9 phosphorylation. Indeed, treatment of cells with LY 294002, a PI3K inhibitor, did not activate HA-GSK-3␤ or HA-GSK-3␤ S9A, indicating that GSK-3␤ activation by G␣ q (Q209L) is not simply a consequence of reduced PI3K/Akt signaling. 2 GSK-3␤ from unstimulated cells is phosphorylated on Tyr 216 , which is located in the activation loop of the catalytic domain, and mutation of Tyr 216 to Phe or dephosphorylation with a protein-tyrosine phosphatase in vitro reduces the kinase activity (28). Some stimuli such as lysophosphatidic acid (LPA) activate the enzyme by causing an increase in Tyr 216 phosphorylation (27,34). LPA signals through G protein-coupled receptors that can couple to G␣ i , G␣ q , and G␣ 12/13 . Sayas et al. (27) showed that expression of activated mutants of G␣ 12 and G␣ 13 mimicked the effect of LPA by inducing GSK-3 activation. G␣ q was not tested because treatment of cells with a PLC inhibitor did not block LPA activation of the kinase, apparently ruling out G␣ q as a possible mediator of LPA-induced GSK-3 activation (27). Our finding that G␣ q (Q209L) activates GSK-3␤ independently of PLC raises the possibility that G␣ q might contribute to the activating effect of LPA on GSK-3.
We considered several candidate tyrosine kinases that might activate GSK-3␤ in response to G␣ q (Q209L). The most well characterized GSK-3 tyrosine kinase is Zak1, which controls cell fate specification in Dictyostelium (35,36). Binding of cAMP to the CAR3 receptor activates Zak1, which phosphorylates GSK-3 on tyrosine residues (including the Tyr 216 equivalent) to increase its activity. A mammalian homolog of Zak1 has not been identified. The Src family tyrosine kinase Fyn has been shown to phosphorylate recombinant GSK-3␤ in vitro (37). However, we found that treatment of cells with PP2, an inhibitor of Src family kinases, did not block the activation of GSK-3␤ by G␣ q (Q209L). 2 The Ca 2ϩ -dependent tyrosine kinase PYK2 (proline-rich protein kinase 2) has also been shown to phosphorylate GSK-3␤ in vitro and has been proposed to mediate increased tyrosine phosphorylation of GSK-3␤ in response to transient increases in intracellular Ca 2ϩ (38). Our finding that G␣ q (Q209L)-AA activates GSK-3␤ without activating PLC implies that an increase in intracellular Ca 2ϩ and PYK2 activation are not required for this response. Finally, as mentioned above, active G␣ q binds to Btk and stimulates its kinase activity (31). Because Btk expression is limited to a subset of hematopoietic cells, we did not consider this kinase as a likely candidate to activate GSK-3␤ in HEK 293 cells.
It was reported that Csk activity is increased upon expression of G␤␥ heterodimers and activated forms of G␣ q , G␣ 12 , and G␣ 13 (21). Interestingly, as mentioned above, activated mutants of G␣ 12 and G␣ 13 also activate GSK-3 (27). These observations led us to explore whether G␣ q (Q209L), Csk, and GSK-3␤ act within the same signaling pathway. Our data are consistent with the hypothesis that Csk acts between G␣ q (Q209L) and GSK-3␤. Both kinases are activated by G␣ q (Q209L)-AA as well as by G␣ q (Q209L), expression of Csk activates GSK-3␤, expression of G␣ q (Q209L) augments the activation of GSK-3␤ by Csk, and Csk-KD inhibits G␣ q (Q209L)mediated GSK-3␤ activation. In addition, our results showing that Csk phosphorylates and activates GSK-3␤ in vitro and that the two kinases interact with each other in cell lysates suggest that GSK-3␤ is a direct target of Csk. We speculate that activation of GSK-3 by G␣ 12 and G␣ 13 is also mediated by Csk (21,27). The mechanism for Csk activation by G␣ q (Q209L) is not understood. G␤␥, but not the activated mutants of G␣ q , G␣ 12 , or G␣ 13 , bound directly to Csk and increased its tyrosine kinase activity in vitro (21). Expression of G␤␥ and the activated G␣ subunits caused Csk to translocate from the cytoplasm to the membrane, but it is not known how this might result in Csk activation (21).
Recent studies suggest that activation of GSK-3␤ by tyrosine phosphorylation is an important mechanism by which apoptotic stimuli can lead to cell death. Nerve growth factor withdrawal or staurosporine treatment of cultured neuronal cells caused an increase in GSK-3␤ activity, Tyr 216 phosphorylation, and cell death (26). Similar observations were made using in vivo models of neuronal apoptosis induced by ischemia or blockade of N-methyl-D-aspartate receptors (26,39). Activated G␣ q also causes apoptosis in some cell types (40). In addition, trans-genic mice overexpressing G␣ q die of heart failure, and cardiomyocytes from these animals show an increased rate of apoptosis (41). The apoptotic effect of G␣ q might be mediated in part by activation of GSK-3␤. It is also well established that GSK-3 affects glucose metabolism by phosphorylating and inactivating glycogen synthase (6). In the liver, hormone stimulation of some G q -coupled receptors also inhibits glycogen synthase (42). This phenomenon could be partly because of activation of GSK-3␤ by G␣ q . Further studies are needed to elucidate how G␣ q activation of Csk and GSK-3␤ might contribute to the regulation of apoptosis, glucose metabolism, and other cellular effects of G q -coupled receptors.