A Secondary Phosphorylation of CREB341 at Serf2' Is Required for the CAMP-mediated Control of Gene Expression A ROLE FOR GLYCOGEN SYNTHASE KINASE-3 IN THE CONTROL OF GENE EXPRESSION*

The CAMP-dependent protein kinase (PKA) phospho- rylates CREBS27/541 at a single serine residue, Ser""", respectively. Phosphorylation at this site creates the sequence motif sxxxS(P), a consensus site of the glycogen synthase kinase-3 (GSK-3) enzyme (Fiol, C. J., Mahren-holz, A. M., Wang,Y., Roeske, R. W., and Roach, P. J. (1987) J. Biol. Chem. 262,14042-14048). We examined the phos- phorylation of CREB at the sxxxS(P) consensus site and its role in CREB transactivation to C A M P induction. Neither isoform of the GSK-3 enzyme (GSK-3 a or p) utilizes CREB as its substrate unless CREB is already phospho~lated at Ser1*wK'89. A 13-amino acid peptide containing the sequence surrounding Ser11s/189 was phosphorylated by GSK-3, at Ser'1N12s,

/ I To whom correspondence should be addressed. Tel.: 317-494-8131; Fax: 317-494-9193; E-mail jsw@vet.vet.purdue.edu. been reported (reviewed in Ref. 2). The nuclear factor CREB interacts specifically with the cyclic AMP response element (CRE)l in genes controlled by the cyclic AMP-mediated pathways of signal t r~s d u c t i o n (3,4). Phosphorylation of CREB by cyclic AMP-dependent protein kinase was initially thought to be necessary and sufficient for the activation of transcription by CAMP (5). In either of two forms of CREB (CREB327 and CREB341) that result from alternative splicing, a mutation of the PKA phosphorylation site Ser119/'33 to an alanine or an aspartic residue prevented activation by PKA in vitro and abolished the CAMP control of transcription in vivo (6). Aphosphate group in the transactivating region is therefore essential for a transcriptionally active protein. Later, measurements in vivo of the activities of deletion and point-mutated chimeric fusion proteins of CREB327 showed that Serllg was necessary but not sufficient for the full CAMP-inducible transactivation functions of CREB (7). It was suggested that phosphorylation within a domain termed the P-box (a 46-amino acid sequence, residues 92-137) was required for the full response to CAMP. This domain includes consensus phosphorylation sites for protein kinase C, casein kinase 11, and the well established PKA site It is now recognized from the study of protein phosphorylation in general that multiple phosphorylation of proteins is the norm rather than the exception (8). Furthermore, in several instances the multiple phosphorylations do not occur independently (1, 9) but follow an obligate order in what has been termed hierarchical phosphorylation (8) and involves "primary" and "secondary" kinases. The mechanistic basis for the phenomenon is that the secondary kinases appear to require prior phospho~lation of the substrate by primary kinases. The best studied example of this synergism is the phosphorylation of the metabolic enzyme glycogen synthase (1) by GSK-3. GSK-3 preferentially phosphorylates substrates where primary kinases have produced the primary sequence motif SXXXS(P). This motif is found in other GSK-3 substrates such as ATP citrate lyase (10) and the G subunit of type I protein phosphatases (11). The role of GSK-3 as the activator of the A~-~~-d e p e n d e n t phosphatase is also well studied and involves a synergistic phosphorylation of the inhibitor 2 protein, However, in this instance the secondary site is 13 residues N-terminal to the primary CK-I1 site (12).
The cDNAs encoding two isoforms of GSK-3 termed GSK-3a (51 kDa) and GSK-3p (47 kDa) have been isolated from a rat brain library by Woodgett (13). These two proteins share 85% homology at the amino acid level. It is becoming clear that The abbreviations used are: CRE, cyclic AMP response element; PKA, cAMP-dependent protein kinase; GSK-3, glycogen synthase kinase-3; CAT, chloramphenicol acetyltransferase.

32187
The Role of a Secondary P~o~p h o~l~~~o n of CRE€3341 at Ser129 there is a family of related GSK-3-like enzymes. A third mammalian form, DM,-1, has been identified by polymerase chain reaction.' Two yeast enzymes, MCK-I and MDS-1, appear to recognize the SXXXS(P) sequence motif. MCK-I has 45% sequence identity with GSK-3 within the catalytic domain, and MDS-1 has 57% sequence identity over a 296-amino acid overlap with GSK-3a from rats (14). Zeste white/Shaggy has been shown to be the Drosophila melanogaster homologue of GSK-3p, with 88% homology within the catalytic domains and similar specificity for the known GSK-3 protein substrates. It is therefore reasonable to expect that Shaggy will participate in hierarchical phosphorylations, although it has not been directly demonstrated. In transgenic flies, it has been shown that GSK-3p can substitute for Shaggy (15). Shaggy plays a crucial role in embryogenesis, and it is speculated that it may be involved in the regulation of transcription factors involved in the expression of genes leading to embryonic segmentation.  (19) demonstrated that the CAMP refractory nature of the UF9 cells is not due to the absence or reduced levels of known positive-acting factors, such as CREB or PKA. In the present study, we employed the undiff~rentiated F9 cell line as a model system to examine if the GSK-38 kinase is an additional positive factor in the CAMP transduction pathway.
EXPERIMENTAL PROCEDURES Peptides-CREB peptide was synthesized using the solid phase method with an Applied Biosystems Model 431A machine run with the small scale (0.1 mmol) t-Boc program. Purification of the crude peptide was by reverse phase chromatography using Aquapore C8 semi-preparative (1 x 10 cm) cartridge columns (Brownlee Laboratories). Sequences were confirmed using a Porton Instruments model 2090 microsequencer.
Peptide Phosphorylation-The CREB peptide (100 p~) was incubated with PKA (3.6 pg/ml), GSK-3a (3 pg/ml) or GSK-3P (8 pg/ml), or the sequential combination of PKA and either isoform of GSK-3 in a 50-pl reaction volume containing C-Y-~~PJATP (500 cp~pmol). Other reaction conditions were as described in Ref. 1. Peptide phosphorylation was analyzed by isoelectric focusing with a modified, more basic gradient (pH 3.5-9.5) as described in Ref. 1. The peptides were localized by autoradiography; for quantitation, the phosphopeptides were excised from the gel, and the radioactivity was measured. The locations of phosphorylated residues were identified according to the method described in Ref. 20. phosphorylated peptide was purified with SEPAK C18 cartridges and reacted (1-nmol samples) with ethanethiol for 1 h at 50 "C. This reaction converts the phosphoserines to S-ethylcysteines, which could then be identified by sequencing using a Porton Instruments microsequencer model 2090.
In Vitro P~osphory~tion of CREB Proteins-CREB proteins were phosphorylated by PKA (3.6 pg/mlf, GSK-30 (3 pg/ml), or GSK-38 (8 pglmlf, i n~v i d u a l~y or in combination. The reaction mixture contained 42 m~ Tris-HC1, pH 7.5, 1.4 m M [y32PlATP (1000 cpndpmol), 8.6 m M Mg(CH,COOf,. The labeled proteins were located by autoradiography of SDS-polyacrylamide gel separations. Quantitation was carried out by excising the bands and dissolving the gel in 30% hydrogen peroxide at 60 "C followed by liquid scintillation counting. Enzymes-Rabbit skeletal muscle GSK-3a was purified as described in Ref. 21. Using antibodies raised to a synthetic peptide common to both isoforms (a gift from Dr. John C. Lawrence, Washington University), we identified a single immunoreactive species of 57 kDa. We conclude that the enzyme purified from rabbit skeletal muscle was predominantly GSK-3a Rabbit skeletal muscle GSK-3P was expressed and purified from Escherichia coli (221. Homogenous bovine cyclic Ah4Pdependent protein kinase catalytic subunit was the gift of Dr. Edwin G. Krebs (U~versity of Washington). Expression and Ptlrification of R e c o~b i n a~t CREB-Wild type or mutant forms of CREB127'3*1 proteins were expressed in bacteria via the T,4121 Lys S(DE3) system (23,241. Purification of the recombinant protein was carried out as described in Colbran et al. (251. Construction of Ser -3 Ala CREB Mutants-Site-directed Ser --f Ala substitutions were constructed at the GSK-3 consensus phosphorylation sites, Ser"' and SerlZ9 of CREBSZ7 and CREB341, respectively, employing the site-directed mutagenesis protocol of Olsen and Eckstein (26). The introduced mutation was verified by dideoxy sequencing.
CREB-dependent Assay System-The indicator plasmid contains the somatostatin promoter (-750 to +50) fused to the chloramphenicol acetyltransferase (CAT) reporter gene (27). The CRE site at nucleotide position -43 has been replaced by the insertion of five Gal4 DNA binding sites. Specifically, the PstI site of the pBxSSTb4 construct, described in Andrisani et al. (27), was used to insert the synthetic Gal4 DNA binding site. The resulting indicator plasmid is pBxSS~Gal4)CAT. The CREB expressor vector used is described by Berkowitz and Gilman (28) and was kindly provided by Dr. M. Gilman (Cold Spring Harbor Laboratory). In this vector, the Ga141"47 DNA binding domain is fused at the NH,-terminal end of CREB. The resulting Gal4-CREB fusion proteins are functional, as demonstrated in earlier studies (281. Dansfection of PC12 Cells-Expressor plasmid (5 pg) was transfected with indicator plasmid (10 pg) in PC12 cells by the Ca(PO,), coprecipitation method employing the Life Technologies, Inc. transfection kit. PC12 cells were grown in Dulbecco's modified Eagle's medium containing heat-inactivated horse serum t 10%) and heat-inactivated fetal bovine serum (5%) on 100-mm tissue culture dishes coated with rat collagen. PC12 cells were transfected at passage 18. 48 h following introduction of the DNA, the cells were harvested, and extracts were prepared in the lysis buffer described in Ref. 29, which yields higher CAT extract activity. Cellular extract (25 pg) was assayed for CAT enzyme activity for 30 min at 37 "C as previously described (301. In uiuo metabolic labeling of transfected PC12 cells, with 135Slmethionine, was carried out by employing the protocol described by Lee et al. (7). Immunoprecipitation reactions with Gal4 antibodies were carried out as described (7). The Gal4 antibody was kindly provided by Dr. M. Ptashne.
Zkansfection of Undifferentiated F9 Cells-Undifferentiated F9 cells were maintained as monolayer cultures on 100-mm dishes, coated with 0.1% gelatin, in Dulbecco's modified Eagle's medium supplemented with 15% heat-inactivated fetal bovine serum. Subconfluent (90%) monolayers were passaged 1:30 the day before the transfection. Transfections were carried out via the CaPO, coprecipitation method, employing the Life Technologies, Inc. transfection kit. The Gal4 programmed transfection assay system utilized 8 pg of indicator plasmid pBxSSTfGal4)-CAT, 5 pg of expressor Ga14-CREB32"~1, and 10 pg of CMV4-GSK-38 plasmid DNA. The amount of total DNA transfected was kept constant by addition of CMV4 vector DNA. The cells were harvested 48 h posttransfection following 20 p~ forskolin stimulation. Cell extracts were prepared in the lysis buffer described in Ref. 29 and assayed for CAT enzyme activity for 2 h at 37 "C as described earlier (30).
The pBxSST transfections employing the endogenous CREB were carried out with 5-8 pg of pBxSST indicator plasmid, 1 pg of CbfY4-PKA (catalytic subunit), and 10 pg of CMV4-GSK-3P expressor plasmid DNA. Cells were harvested 24 h post-infection. Cellular extracts were prepared and assayed as described above.

In Vitro P~~~p h o~~u t~o n
of CREB-To examine if the transcription factor CREB is a substrate for the GSK-3 protein kinase, recombinant CREB327 protein was used in in vitro enzymatic reactions employing PKA and either of the two GSK-3 isoforms, GSK-3a or GSK-3P. The CREB327 proteins were phosphorylated in vitro by PKA to a stoichiometry of 0.6 ( Fig. 1, A  and B, lanes 3 ) . A substitution of Ser1Ig by an alanine in CREB abolished PKA phosphorylation, confirming that Ser1Ig is the only phosphorylation site for CAMP-dependent protein kinase (Fig. 1, A and B, lanes 5). GSK-3a (Fig. lA, lane 2 ) or GSK-3P (Fig. lB, lane 2 ) alone did not phosphorylate CREB. However, if CREB was previously phosphorylated by PKA at Ser1Ig, GSK-3a or GSK-3P could stoichiometrically introduce another phosphate (Fig. 1, A and B, respectively, lanes 4 ) . The S e P g mutant was unable to act as a substrate for either isoform of GSK-3 (lane 6 ) . Thus, the in vitro phosphorylation reactions shown in Fig. 1 demonstrate that CREB can be phosphorylated by either isoform of GSK-3 but only after primary phosphorylation by PKA.
Mapping of Phosphorylation Sites in CREB-To map the GSK-3 phosphorylation sites in CREB, a peptide was synthesized with sequence as shown in Fig. 2. The CREB peptide was stoichiometrically phosphorylated by cyclic AMP-dependent protein kinase (lane 3 ) , but it was not phosphorylated by GSK-3 (lane 2). Once the CREB peptide had been phosphorylated by PKA, it became a substrate for both GSK-3a and Gsk-3p reproducing the synergistic phosphorylation of the CREB protein consistent with previous results (16). Stoichiometric phosphorylation by GSK-3a of the monophosphopeptide produced by PKA is shown in lane 4  phosphorylated by PKA alone or in combination with GSK-3 was reacted with ethanethiol and subjected to sequence analysis. The sequencing pattern shown in Fig. 3 is consistent with a phosphate being present only on Ser119/133 after phosphorylation with PKA (panel B ) and with phosphate also present at were confident that we had mapped the GSK-3 site to Ser"5/'29 of CREB.
To confirm the identification of the phosphorylation site in the intact native CREB protein, we carried out site-directed mutagenesis of Ser""'29 to Ala, as shown in Table I. Ser"'/'29 is the putative phosphorylation site for GSK-3, based on the peptide phosphorylation studies. The mutant proteins were expressed in bacteria, purified, and utilized for in vitro phosphorylation reactions (Fig. 4). Controls for these experiments include the wild type CREB327 protein (lanes 1 4 ) and the CREB proteins that contain Ser"' --f Ala substitutions at the PKA phosphorylation site (lanes 5 and 6 ) . Lane 7 in panel A shows that the mutant CREB327 with the Ala"' substitution is a substrate for PKA with a stoichiometry essentially equal to that of wild type CREB327 shown in lane 3. Unlike the native CREB327 (lane 4 ) , mutant CREB did not become a substrate for either GSK -3a (panel A, lune 8) or GSK-3P (data not shown) after phosphorylation by PKA. This result confirms that Ser"' was the site for GSK-3 phosphorylation in vitro. The same results were obtained with mutant CREB341 protein containing SerIz9 4 Ala substitution, employing GSK -3P (panel B, lane 8) or GSK-3a (data not shown).
I)-ansactivation Properties of GSK-3 Site Mutant CREB327'34' Proteins-"he role of this secondary phosphorylation reaction of CREB in its transactivation response to CAMP was examined by in vivo functional assays in the well documented CAMPresponsive PC12 cell line (17). The transcriptional activity of the wild type CREB327/341 protein was compared with the activity of the CREB protein mutants containing Ser -+ Ala substitutions at the GSK-3 phosphorylation site.
The assay system used has been previously reported by Berkowitz and Gilman (281, Lee et al. (7) and Sheng et al. (31). The DNAbinding domain of CREB is reprogrammed, via fusion to the heterologous Ga141"47 DNA binding domain. A schematic diagram of the vector is shown in Fig. 5A. A diagram of the SerllS/129 after phosphorylation with GSK-3 (panel C). Thus, we

Mutagenesis of CREB327'34' Proteins
The constructed site-directed Ser + Ala mutants are listed below.

GSK-3 site mutants
PKA site mutants indicator plasmid used in the above CREB-dependent assay system is also shown in Fig. 5A. The indicator plasmid contains the rat somatostatin promoter, spanning the sequences between nucleotide position -750 to +50. The CRE site at position -43 of the promoter is replaced by the Gal4 DNA binding site. Thus, this Gal4-CAT indicator plasmid maintains the native arrangement of the somatostatin promoter as opposed to a minimal GalCdriven promoter.
The wild type CREB327'341 and the Ser +. Ala mutants shown in Table I were also cloned into the vector shown in Fig. 5A. Following transfection in PC12 cells and forskolin stimulation, the transcriptional activity of the GSK-3 site mutants of CREB327/341 was compared with the activity of the wild type CREB327'341 proteins. In addition, the inactive PKAsite mutants of CREB327/341 were also transfected in parallel and used as a negative control. An additional negative control in the transfection assays included an expression vector encoding only the Ga141-147 DNA binding domain and lacking an activation domain. The results of the transient transfection CAT assays are shown in Fig. 5B. The  The Ser + Ala substitution within the GSK-3 phosphorylation site in CREB341 (lane 5) also renders the CREB protein transcriptionally inactive. However, in CREB327, the Ser"' -+ Ala mutation at the GSK-3 site reduces its transcriptional response to CAMP induction, on average, to 30% of its wild type activity. The histogram in Fig. 5C shows the quantitation of three independent transfection assays in PC12 cells. The in vivo synthesis and stability of the transfected Ga14-CREB proteins was confirmed by in vivo metabolic labeling studies, employing [35S]methionine as shown in Fig. 6, A and B. The radiolabeled proteins were immunoprecipitated with Ga14specific antibodies. Because we have encountered difficulties in transfecting the PC12 cell line (Fig. 6A), we carried out additional in vivo labeling assays in HeLa cells (Fig. 6B). The results of the immunoprecipitation reactions confirm the expression of the CREB341 proteins as shown in lanes 1 and 2. The control immunoprecipitation reaction shown in Fig. 6B, lane 3, employing extracts of 3sS-labeled cells transfected with vector lacking the Gal4-CREB cDNA insert, confirms the specificity of the Gal4 antibody and supports that the immunoprecipitated proteins shown in Fig. 6 correspond to the Gal4-CREB fusion.

Dansfection of GSK-3P Encoding Plasmids in UF9 Cells-
The undifferentiated F9 cell line is refractory to CAMP induction of the CRE-dependent promoters of somatostatin and vasoactive intestinal peptide (19), although the known positive effectors CREB and PKA are present and functional (19). Overexpression of exogenous CREB and PKA compensates for the C, histograms show the quantitation of three independent transfection assays in PC12 cells. Percent CAT activity is plotted against each expressor plasmid tested. Each bur is one independent assay. Groups [1][2][3][4][5][6][7] correspond to the expressor plasmid tested as follows: 1, wild type Ga14-CREB?"; 2, wild type Ga14-CREB327; 3, PKA mutant Gal4-CREB3'I; 4, PKA mutant Gal4-CREB"'; 5, GSK-3 mutant Ga14-CREB3"; 6, GSK-3 mutant Ga14-CREB"'; and 7, Gall-"' DNA binding domain. lack of CRE-dependent induction by the endogenous cAMP effector molecules (6). This observation has been interpreted to mean (19) that negative regulators block the activity of the We have employed the UF9 cell line to examine if GSK-3 kinase activity is an additional positive factor required for the CAMP transduction pathway. We have carried out transfection assays of the mammalian expression vector CMV4, encoding the GSK-3P kinase, to assess the effect on transcription directed from the CREKREB-dependent promoter of the rat somatostatin gene.
Initially, we employed the Gal4-CREB-dependent assay system described earlier in Fig. 5. The pBxSST(Gal4)CAT indicator plasmid (Fig. 5A) and the Gal4-CREB expressor (Fig. 5 4 ) were cotransfected in the presence of a CMV4 expressor vector encoding the GSK-3P isoform. The transfection assays were carried out in the presence of 20 PM forskolin. The histogram shown in Fig. 7 demonstrates that the transcriptional activity of CREB341 is induced between 5-25-fold in the presence of the cotransfected GSK-3P kinase. Similarly, CREB327 activity is induced approximately 10-fold by the cotransfected GSK-3P kinase. In contrast, the Ser*15/129 --.f Ala substitution at the GSK-3 site of CREB327R41 displayed an approximately 2-fold activation in the presence of cotransfected GSK-3P plasmid, suggesting the importance of the GSK-3 phosphorylation on CREB-dependent transcription and the CAMP transduction pathway. The histograms of the independent transfection assays depicted in Fig. 7 show some variability in the level of transcriptional induction of CREB341 by the cotransfected GSK-3P kinase; this variation is most likely due to differences in the transfection efficiency of the UF9 cell line. However the results of these experiments demonstrate the overall inducible effect of the GSK-3P kinase on CREB transactivation.
To further demonstrate that GSK-3P is a positive effector in the CAMP transduction pathway, we examined in UF9 cells whether exogenously transfected GSK-3P kinase could compensate for the overexpression of exogenous CREB and PKA required to obtain transcription from the CRE-dependent somatostatin promoter. We carried out transient expression assays employing the pBxSST reporter plasmid (271, which contains the rat somatostatin promoter (+50 to -750) in front of the CAT gene. This CRE/CREB-dependent reporter plasmid was cotransfected with the expression plasmid encoding the GSK-3P kinase. In the results shown in Fig. 8, the CREB protein mediating the transcriptional response is the endogenous CREB; the catalytic PKA activity was obtained either by transfection of an expression vector encoding the catalytic subunit of the PKA enzyme (Fig. 8 A ) or via forskolin stimulation of UF9 cells (Fig. 8B). An approximately 60-fold induction in the transcriptional activity of the somatostatin promoter mediated via the endogenous CREB is observed in the presence of cotransfected GSK-3P kinase. The cotransfected GSK-3 kinase requires the presence of the catalytic subunit of PKA for CREB transactivation in agreement with the well documented mode of action of this class of enzymes. In agreement with earlier studies (6,191, overexpression of the catalytic subunit of PKA without overexpression of exogenous CREB is not sufficient for CRE-dependent transcriptional induction in UF9 cells. The results shown in Fig. 8 support the positive effector role of GSK-3 type kinases in the CAMP transduction pathway.

DISCUSSION
This study provides evidence in vitro and in vivo to support a role for hierarchical phosphorylation in the regulation of the transactivation properties of CREB. The amino acid residues 92-137 are critical for the transcriptional activation of CREB in response to CAMP induction (7). This domain of CREB is serine-rich and has been shown to be multiply phosphorylated in addition to the phosphorylation by PKA. We therefore tested the possibility of hierarchical phosphorylation reactions occurring within CREB327'341 in the regulation of its transcriptional response to CAMP. A synthetic CREB peptide and site-directed CREB mutants were used to map the secondary CAMP-dependent phosphorylation to Ser*15'129 within the activation domain of CREB327'341. The in vitro data strongly suggested that the primary kinase is PKA and the secondary kinase is one or both isoforms of GSK-3, though it is not certain which kinase(s) carry out the sequential phosphorylations in vivo. Our results from the in vivo transfection assays in PC12 cells are consistent with the conclusion that phosphorylation by PKA is essential for the activation of CREB327'341. However, the PKA phosphorylation is not sufficient since we demonstrated that ablation of the secondary phosphorylation site of CREB327"41, Ser115'129, impaired the transactivating function of the protein. Furthermore, analysis of the UF9 cell system directly demonstrates that GSK-3P is a positive effector of the CAMP transduction pathway. Transfection of the GSK-3P expression vector results in a 60-fold induction in transcription of the CRE-dependent promoter, employing the available endogenous CREB. This transcriptional induction requires the presence of the catalytic PKA subunit. The results of these two types of in vivo experiments in PC12 cells (Fig. 5) and UF9 cells (Fig. 8) are interpreted to mean that two phosphorylations must occur to generate a fully activated form of CREB. This is consistent with the proposal that PKA phosphorylation is not sufficient for full CREB activation. Overexpression of exogenous CREB and PKA is required for detectable CRE-dependent transcription in UF9 cells, which compensate for the partial activation state of CREB.
The in vivo data support the concept that a secondary phosphorylation at Ser12' is an important component of the trans-activation response of CREB and suggest a mechanism for integration of signals from different signal transduction pathways. Evidence for CREB being involved in the cross-talk between different signaling pathways has appeared (31). There are reports of CREB341 functioning as a Ca2+-regulated transcription factor through the phosphorylation of Ser'33 (31). It is also possible that a secondary phosphorylation, dependent on primary phosphorylation of Ser'33, mediates the response to Ca2+ signal since mutation of Ser133 would destroy the phosphorylation at the secondary site as well.
At the moment, it is not clear what may regulate the activity of GSK-3 in the cell. One possibility is that GSK-3 activity is constant and dependent on the level of substrate maintained by primary kinases responding to several different signal transduction pathways. There is evidence that suggests that GSK-3 must be phosphorylated on a tyrosine residue to be in a n active state (32). Another report suggests that one form of GSK-3, GSK-3P, may be a target for protein kinase C phosphorylation resulting in down-regulation (33). It has been proposed that a down-regulation of GSK-3 activity by protein kinase C in response to phorbol esters results in a reduced level of c-Jun phosphorylation and subsequent stimulation of c-Jun binding to DNA. Though there is no evidence for inactivation of CREB upon 12-0-tetradecanoylphorbol-13-acetate treatment, such down-regulation of GSK-3P activity would also imply a possible role for protein kinase C inactivation of CREB341 via a decrease in phosphorylation of SerlZ9 by GSK-3. Based on these models, GSK-3 would appear to have opposing roles in the expression of CRE or TRE containing genes. It is important to note that presently it is not known which cellular isoform of GSK-3 carries out the phosphorylation of Ser'29, and GSK-3a appears not to be down-regulated by protein kinase C phosphorylation. This is consistent with the observation that 12-0-tetradecanoylphorbol-13-acetate treatment does not influence CRE-mediated stimulation of transcription (7). More recent reports (34,35) have shown that phosphorylation of GSK-3a or GSK-3P by mitogen-activated protein kinase-activated protein kinase-1 (34) and p70 S6 kinase (35) results in almost complete inactivation. Thus, GSK-3 has been implied as a target for inactivation through the insulin response pathway. This inactivation by insulin would provide a mechanism to antagonize CAMP-dependent gene expression. A secondary phosphorylation of CREB341 at SerlZ9 potentiated by the primary phosphorylation of Ser133 by PKA appears to be necessary to fully evoke the change in transcriptional function of CREB341. Another example in which GSK-3 has been shown to act synergistically with CAMP-dependent protein kinase is in the phosphorylation of two sites in the glycogen binding subunit of type 1 protein phosphatase (11). Judging from these two examples, one can postulate that the sequence SRR(G/P)S is a consensus for a coupled pair of P W G S K -3 sites. At present, the only enzymes known to have the appropriate specificity, for the sequence motif sxxXS(P) are enzymes designated GSK-3 as previously discussed. This work strongly suggests a role for GSK-3 or a related family of acidotropic kinases in the control of transcriptional events through hierarchical protein phosphorylations, in particular the expression of genes responding to changes in CAMP levels.