Phosphorylation and inactivation of glycogen synthase kinase 3 by protein kinase A

Reagents.

cAMP analogs [8-bromo-cAMP (8-Br-cAMP) and 8-bromo-cGMP (8-Br-GMP)], forskolin, isoproterenol, 3-isobutyl-1-methylxanthine, and wortmannin were purchased from Sigma. H89 and LY294002 were obtained from Calbiochem. Insulin-like growth factor (IGF)-1 and the PKA peptide inhibitor PKI were from Upstate Biotechnology (Lake Placid, NY). Insulin was from GIBCO/BRL.

Cell Lines.

HEK293 and NIH 3T3 were obtained from American Type Culture Collection. Rat1 cells were provided by Wouter H. Moolenaar (Netherlands Cancer Institute). These cell lines were grown in DMEM supplemented with 10% FBS (Sigma). The cells were frozen at early passages and used for less than 10 weeks in continuous culture.

Plasmids.

GSK-3 expression vectors (pMT2GSK-3α and pcDNA3-HA-GSK-3β) were described previously (11). pcDNA3-PKAc containing the human PKAc cDNA was a gift of H. Wen (University of Texas M. D. Anderson Cancer Center).

Transfection.

HEK293, NIH 3T3, and Rat1 cells were plated in 60-mm dishes. One day later, at around 60% confluence, the cells were transfected (5 μg DNA/dish) by using Fugene 6 according to the protocol of the supplier (Roche Molecular Biochemicals). In cotransfection experiments, the molar ratio of the GSK-3 and PKAc plasmids was adjusted to 1:5 to ensure the PKAc was coexpressed with GSK-3α or -β.

Preparation of Cell Lysates and Western Blotting.

Cells were lysed on ice for 30 min in a lysis buffer containing 1% Triton X-100, 50 mM Hepes, pH 7.4, 150 mM NaCl, 1.5 mM MgCl₂, 1 mM EGTA, 10% glycerol, 100 mM NaF, 10 mM sodium pyrophosphate, 25 mM β-glycerol phosphate, 1 mM DTT, 1 mM sodium vanadate, 1 mM benzamidine, 0.1 μM okadaic acid, 10 μg/ml aprotinin, and 10 μg/ml leupeptin and protease inhibitor mixture (Roche Molecular Biochemicals). Equal amounts of total cellular protein were separated by SDS/PAGE, transferred to immobilon [poly(vinylidene difluoride)], and immunoblotted with antibodies following the protocols provided by manufacturers. Anti-phospho-GSK-3α is a sheep polyclonal IgG reactive with GSK-3α phosphorylated at serine 21 (Upstate Biotechnology). Antiphospho-GSK-3β is a rabbit polyclonal IgG recognizing GSK-3β phosphorylated at serine 9 (New England Biolabs). Anti-GSK-3 is a phosphorylation-independent monoclonal antibody that recognizes both GSK-3α and -β (Upstate Biotechnology). Immunocomplexes were visualized by an enhanced chemiluminescence detection kit (Amersham Pharmacia) by using horseradish peroxidase-conjugated secondary antibodies (Bio-Rad).

In Vitro Kinase Assays.

For measuring GSK-3 activity, approximately 150 μg (for GSK-3α) and 75 μg (for GSK-3β) of total cellular protein were diluted in freshly made lysis buffer and immunoprecipitated with 5 μg anti-GSK-3α (sheep anti-rat GSK-3α; Upstate Biotechnology) or with 0.75 μg anti-GSK-3β (mouse monoclonal anti-rat GSK-3β; Transduction Laboratories, Lexington, KY). After 2 h of rotation at 4°C, protein G-Sepharose was added for another 1.5 h of incubation. Immunoprecipitates were washed twice with lysis buffer and twice with kinase reaction buffer (10 mM 4-morpholinepropanesulfonic acid, pH 7.4/1 mM EDTA/10 mM MgAc/50 mM β-glycerol phosphate/1 mM sodium vanadate/0.5 mM NaF/0.1 μM okadaic acid/1 mM benzamidine/1 μg/ml aprotinin/1 mM DTT). Kinase activity of immunoprecipitated GSK-3 was assayed in a total of 40 μl of reaction buffer supplemented with 62.5 μM phosphoglycogen synthase peptide-2 (Upstate Biotechnology)/20 mM MgCl₂/125 μM cold ATP/10 μCi [γ-³²P] ATP. After 20 min of incubation at 30°C, reaction mixtures were centrifuged, and 15 μl of the supernatant was spotted onto Whatman P81 phosphocellulose paper. Filters were washed in three changes of 0.75% phosphoric acid, rinsed in acetone, dried, and counted in a liquid scintillation counter.

PKB was immunoprecipitated with a sheep polyclonal anti-rat PKB (Upstate Biotechnology). After immunoprecipitation, PKB kinase activity was determined with a PKB kinase assay kit by using crosstide–paramyosin fusion protein as substrate (New England Biolabs). Phosphorylation of the crosstide–paramyosin substrate at a serine site corresponding to serine 21/9 of GSK-3 was revealed by immunoblotting with a GSK-3 phospho-specific antibody, provided with the kit, that recognizes both GSK-3α phosphorylated at serine 21 and GSK-3β phosphorylated at serine 9.

Analyses of Physical Association Between PKA and GSK-3.

GSK-3 antibodies used for immunoprecipitation are the same as those used in the GSK-3 kinase assays. The anti-GSK-3β antibody crossreacts with human, rat, and mouse GSK-3β, whereas the anti-GSK-3α antibody recognizes human and rat but not murine GSK-3α. Rabbit polyclonal antibodies against PKAc or the PKA regulatory subunit RII (PKA RII) (Santa Cruz Biotechnology) were used for PKAc or PKA RII immunoprecipitation and Western blotting. Immunoprecipitates were washed three times with lysis buffer, twice with a washing buffer (8 mM 4-morpholinepropanesulfonic acid, pH 7.4/0.2 mM EDTA/10 mM MgAc) and analyzed by Western blotting.

PKA Phosphorylation of GSK-3 in Vitro.

PKA phosphorylation of GSK-3 was assessed by using a PKA assay kit according to the protocol provided by the manufacturer (Upstate Biotechnology), except that immunoprecipitated GSK-3α, immunoprecipitated HA-GSK-3β, purified GSK-3α (0.1 unit/reaction) (Upstate Biotechnology), or recombinant GSK-3β (5 units/reaction) (New England Biolabs), instead of Kemptide, were used as substrates. HA-GSK-3β was immunoprecipitated with anti-HA antibody from HEK293 cells transfected with HA-GSK-3β (approximately 180 μg total protein for each reaction), and GSK-3α was immunoprecipitated from the same amount of protein of untransfected HEK293 cells by using the GSK-3α-specific antibody. In both cases, HEK293 cells were starved in serum-free medium for at least 12 h before lysing to decrease the background phosphorylation level of GSK-3. Immunoprecipitates were washed twice with lysis buffer and twice with washing buffer (8 mM 4-morpholinepropanesulfonic acid, pH 7.4/0.2 mM EDTA/10 mM MgAc). PKA activity toward each of these substrates was assayed in a total volume of 50 μl of reaction buffer (20 mM 4-morpholinepropanesulfonic acid, pH 7.2/25 mM glycerol phosphate/5 mM EGTA/1 mM sodium vanadate/1 mM DTT/15 μM MgCl2/100 μM ATP) without or with PKAc (50 ng) (Upstate Biotechnology) or with PKAc in the presence of the PKA peptide inhibitor PKI (1.2 μM) (Upstate Biotechnology). After 20 min at 30°C, the reaction was stopped by adding an equal volume of 2 × SDS sample buffer. PKA-mediated phosphorylation of GSK-3 was analyzed by immunoblotting with GSK-3α and GSK-3β phospho-specific antibodies.

cAMP Induces PKA-Dependent Phosphorylation and Inactivation of GSK-3.

In Rat1, NIH 3T3, and HEK293 cells, increases in the intracellular levels of cAMP stimulate phosphorylation of GSK-3α and -β, as demonstrated by immunoblotting with GSK-3 phosphorylation-specific antibodies. As shown in Fig. 1, the cell-permeable cAMP analogue 8-Br-cAMP induced a marked increase in phosphorylation of GSK-3α and -β at serine 21 and 9, respectively, whereas the structurally related cGMP analogue 8-Br-cGMP had little effect. Forskolin, which activates adenyl cyclase thus raising intracellular cAMP levels (15), triggered a similar elevation in GSK-3 phosphorylation at serine 21 and 9. In Rat1 cells, isoproterenol, which activates the β-adrenergic receptor stimulating adenylate cyclase and increasing endogenous cAMP levels (16, 17), also efficiently stimulated GSK-3 phosphorylation at these serine sites. In NIH 3T3 cells that contain few of the β-adrenergic receptors, stimulation with other G-protein-coupled receptor agonists, such as lysophosphatidic acid. also led to GSK-3 phosphorylation (data not shown).

GSK-3 phosphorylation induced by 8-Br-cAMP, forskolin, and isoproterenol was abrogated in the presence of the PKA inhibitor H89 (18), whereas IGF-1-mediated phosphorylation of GSK-3 was insensitive to H89 (Fig. 1), indicating an essential and specific role for PKA in cAMP-initiated phosphorylation of GSK-3 at serine 21 and 9. Consistent with the inhibitory effect of phosphorylation at serine 21 and 9 on GSK-3 activation (9–12), 8-Br-cAMP, forskolin, or isoproterenol dramatically inhibited GSK-3 kinase activity as shown in in vitro kinase assays in Rat1 (Fig. 1A) and HEK293 cells (data not shown).

PKA-Dependent Phosphorylation and Inactivation of GSK-3 Does Not Correlate with Activation of the PI3K–PKB Cascade.

PI3K-dependent activation of PKB is an intermediary in the phosphorylation and inactivation of GSK-3 induced by insulin and other growth factors (10, 12, 14). Insulin and IGF-1 efficiently activate the PI3K pathway in NIH 3T3 and Rat1 cells, respectively. GSK-3 phosphorylation induced by insulin and IGF-1 were eliminated by pretreatment with the PI3K inhibitors wortmannin or LY294002. In contrast, the effects of 8-Br-cAMP or forskolin on GSK-3 phosphorylation in these cell lines were not altered by inhibition of PI3K activity with wortmannin or LY294002 (Fig. 2A), excluding the possibility that PKA induces GSK-3 phosphorylation through activation of PI3K. In some systems, cAMP or PKA has been suggested to induce a PI3K-independent activation of PKB (19). To determine whether PKA-induced PKB activation accounts for GSK-3 phosphorylation after treatment with 8-Br-cAMP, forskolin, or isoproterenol, PKB phosphorylation at serine 473 and threonine 308, which is crucial for activation of PKB (20), was assessed. As expected, insulin and IGF-1 strongly stimulated phosphorylation of PKB at serine 473 in NIH 3T3 or Rat1 cells (Fig. 2B). In contrast, 8-Br-cAMP, forskolin, and isoproterenol did not significantly alter PKB phosphorylation at serine 473 (Fig. 2B). Similar results were obtained with a threonine 308 phosphorylation-specific antibody (data not shown). Furthermore, PKB activity was not increased by treatment of cells with 8-Br-cAMP at concentrations that strongly induced phosphorylation and inactivation of GSK-3 (Fig. 2C). In fact, incubation with 8-Br-cAMP modestly decreased PKB activity because of an as yet undefined mechanism. Forskolin increased PKB activity weakly in Rat1 and NIH 3T3 cells. A slightly stronger stimulation of PKB activity was seen in Rat1 cells treated with isoproterenol. However, activation of PKB by forskolin or isoproterenol was much less than that induced by IGF-1 or insulin (Fig. 2C). As compared with the basal activity present in the unstimulated cells, isoproterenol or forskolin-induced PKB activity was only one-third or less of that stimulated by IGF-1 or insulin. As demonstrated in Fig. 2C, however, 8-Br-cAMP, forskolin, or isoproterenol stimulated phosphorylation of GSK-3α more efficiently than IGF-1 or insulin and increased phosphorylation of GSK-3β to levels comparable to those induced by IGF-1 or insulin. The lack of correlation between PKB activation and GSK phosphorylation in cells after elevation of cAMP levels argues in favor of a PKB-independent route for GSK-3 phosphorylation and inactivation.

PKA Physically Complexes with GSK-3.

PKA and PKB, a known GSK-3 kinase, are structurally related, exhibiting a high homology especially within their catalytic domains (21–23). The amino acid sequences around serine 21 (SGRARTSSFA) in GSK-3α and serine 9 in GSK-3β (SGRPRTTSFA) comply with the “high-probability” PKA consensus phosphorylation site RXXS/T (24, 25). Further, both GSK-3 and PKA are present in cytosol (1, 25). To explore the possibility that PKA may function as a GSK-3 kinase that directly phosphorylates GSK-3, we examined whether PKA physically interacts with GSK-3 in intact cells. As demonstrated in Fig. 3A, in Rat1 and HEK293, GSK-3α forms a stable complex with the α catalytic subunit of PKA (PKAc) (24), as reflected by the presence of PKAc in GSK-3α immunoprecipitates. The amount of PKAc that coprecipitated with GSK-3α was enhanced by overexpression of GSK-3α in transiently transfected cells, indicating that the ability to coprecipitate PKAc depended on GSK-3α protein levels. The GSK-3α antibody used for immunoprecipitation does not react with murine GSK-3α, precluding assessment of association between GSK-3α and PKAc in NIH 3T3 cells (data not shown). Similar to GSK-3α, GSK-3β immunoprecipitates also contained PKAc (Fig. 3A). Transient transfection with HA-tagged GSK-3β (HA-GSK-3β) increased the amount of PKAc that was coprecipitated with GSK-3β. When the transfected GSK-3β was immunoprecipitated by using an anti-HA epitope tag antibody, PKAc was coprecipitated from HA-GSK-3β-transfected cells but not from control vector-transfected cells (Fig. 3B), indicating a specific association between GSK-3β and PKA. In the reciprocal experiment, both endogenous GSK-3β and transfected HA-GSK-3β were readily detected in PKAc precipitates (Fig. 3B). We were unable to assess whether GSK-3α was present in PKAc precipitates, because GSK-3α (≈52 K_d) (1) colocalizes with the Ig heavy chain. To determine whether the association between GSK-3 and PKAc is regulated by cAMP levels, GSK-3β was immunoprecipitated from HEK293 and NIH 3T3 cells treated with or without forskolin. The level of PKAc in GSK-3β immunoprecipitates was markedly reduced in forskolin-treated cells compared with that in unstimulated cells (Fig. 3C), suggesting that dissociation of the GSK-3β-PKAc complex occurs after PKA activation or GSK-3 phosphorylation. Because GSK-3 also coimmunoprecipitated with the PKA regulatory subunit RII and the association was not affected by forskolin (Fig. 3C), it seems that the association of PKAc with GSK-3 is indirect, mediated by PKA regulatory subunits or A-kinase-anchoring proteins.

PKA Phosphorylates GSK-3 in Vivo and in Vitro.

We next assessed the ability of PKA to phosphorylate GSK-3 in vivo and in vitro. Overexpression of PKAc by transient transfection results in constitutive activity in resting cells because of an excess of the catalytic subunit compared with the endogenous regulatory subunits (25). Transfection of PKAc, but not empty vector, enhanced phosphorylation of cotransfected GSK-3α and -β at serine 21 and 9 in HEK293 cells (Fig. 4A). The transfected GSK-3α (human) and cellular GSK-3α in HEK293 migrated at the same location and were not distinguishable. In NIH 3T3 cells, however, the murine GSK-3a was larger and thus separable from the transfected human GSK-3α. PKAc-mediated phosphorylation of coexpressed GSK-3α or GSK-3β was particularly high when compared with the basal phosphorylation levels of endogenous GSK-3, which were low in starved unstimulated cells.

Consistent with this in vivo activity, PKAc enzyme efficiently phosphorylated GSK-3α and -β at serine 21 and 9 in vitro (Fig. 4B), which was blocked by addition of the PKA peptide inhibitor (PKI) (26). Similar results were obtained when immunoprecipitated GSK-3α, immunoprecipitated HA-GSK-3β, purified GSK-3α, or recombinant GSK-3β was used as substrate (Fig. 4B). The crosstide–paramyosin fusion protein, which is a substrate for PKB in in vitro kinase assays, was also highly phosphorylated by PKA at the same serine site that is phosphorylated by PKB (data not shown). In contrast, other serine/threonine kinases, such as MEK-1 and Erk-2, did not phosphorylate the crosstide–paramyosin fusion protein or GSK-3α or -β at serine 21 or 9 in vitro (data not shown).

These in vivo and in vitro observations establish that PKA can directly phosphorylate and inactivate GSK-3. The result is different from the observations in Dictyostelium that cAMP up-regulates GSK-3 activity via cAMP receptor and subsequent activation of the tyrosine kinase ZAK1 (27, 28). The identification of GSK-3 as a downstream target of PKA in mammalian cells suggests an effector role for GSK-3 in cellular responses to cAMP or PKA activation. For example, cAMP has multiple diverse effects on cellular processes such as gene expression, cell cycle control, and cell survival-death decision (25, 29–31). In some systems, elevation of intracellular cAMP levels stimulates cell proliferation and survival through undefined mechanisms (29–31). Our results herein imply that PKA-mediated inhibition of GSK-3 may underlie or contribute to these effects of cAMP considering the prominent established role of GSK-3 activity in the control of cell proliferation and survival (1, 13, 14).

The convergence of the PKA and PKB signaling pathways at GSK-3 suggests that, depending on the environmental context, the activity of GSK-3 can be regulated either by growth factors (e.g., insulin and IGF-1) working through the PI3K-PKB cascade or by hormonal stimulation (e.g., adrenaline) of G protein-coupled receptors that link to changes in intracellular cAMP concentrations. Analogous to GSK-3, other functionally important proteins such as BAD, a proapoptotic member of the Bcl-2 family, and CREB, a nuclear target of PKA, also appear to represent convergence points of multiple signaling pathways (32–37). BAD was originally demonstrated to be phosphorylated at serine 136 by PKB (32, 33). However, BAD can also be inactivated through phosphorylation of serine 112 by Rsk, which is located downstream of mitogen-activated protein kinases (MAPK/Erk) (34–36), or alternatively by mitochondria-anchored PKA (37). Additionally, Rsk can phosphorylate CREB at the same serine 133 site as PKA, leading to the expression of cAMP-responsive genes (34). Our present work underscores the multiple levels of crosstalk among signaling cascades that regulate functionally critical molecules such as GSK-3.