Glycogen synthase kinase 3beta regulates GATA4 in cardiac myocytes.

Inactivation of glycogen synthase kinase 3beta (GSK3beta) is critical for transcription of atrial natriuretic factor (ANF) by beta-adrenergic receptors in cardiac myocytes. We examined the mechanism by which GSK3beta regulates ANF transcription. Stimulation of beta-adrenergic receptors induced nuclear accumulation of GATA4, whereas beta-adrenergic ANF transcription was suppressed by dominant negative GATA4, suggesting that GATA4 plays an important role in beta-adrenergic ANF transcription. Interestingly, GATA4-mediated transcription was markedly attenuated by GSK3beta. GSK3beta physically associates with GATA4 and phosphorylates GATA4 in vitro. Overexpression of GSK3beta suppressed both basal and beta-adrenergic increases in nuclear expression of GATA4, whereas inhibition of GSK3beta by LiCl caused nuclear accumulation of GATA4, suggesting that GSK3beta negatively regulates nuclear expression of GATA4. The nuclear exportin Crm1 reduced nuclear expression of GATA4, and the reduction was enhanced by GSK3beta but not by kinase-inactive GSK3beta. Leptomycin B, an inhibitor for Crm1, increased basal nuclear GATA4 and suppressed GSK3beta-induced decreases in nuclear GATA4. These results suggest that GSK3beta negatively regulates nuclear expression of GATA4 by stimulating Crm1-dependent nuclear export. Inhibition of GSK3beta by beta-adrenergic stimulation abrogates GSK3beta-induced nuclear export of GATA4, causing nuclear accumulation of GATA4, which may represent an important signaling mechanism mediating cardiac hypertrophy.

Cardiac hypertrophy is a fundamental adaptive process employed by postnatal terminally differentiated cardiac myocytes (1). Understanding of the signaling mechanism of cardiac hypertrophy is important, because the continued presence of hypertrophy with structural alteration in the myocardium (cardiac remodeling) leads to development of congestive heart failure (2). A series of phosphorylation or de-phosphorylation events in myocardial signaling molecules mediated by protein kinases or protein phosphatases play an important role in cardiac hypertrophy by many stimuli, including mechanical forces, myocardial ischemia, catecholamines, and autocrine/ paracrine factors (reviewed in Refs. [3][4][5]. Cardiac hypertrophy is characterized not only by increases in the cell size but also by changes in gene expression, including reactivation of the fetal gene program, as well as by cytoskeletal reorganization (6). Changes in gene expression seen in cardiac hypertrophy are in many cases mediated by changes in nuclear transcription. However, only few studies have shown thus far that the protein kinases or phosphatases activated by hypertrophic stimuli regulate activities of specific nuclear transcription factors (7)(8)(9).
Although the cardiac phenotype of hypertrophy by ␤-adrenergic receptor (␤AR) 1 stimulation is similar to that caused by other hypertrophic stimuli, such as mechanical stretch and agonists for G q protein-coupled receptors (angiotensin II, phenylephrine, and endothelin-1), the signaling mechanism of hypertrophy by ␤ARs differs from that by other hypertrophic stimuli (10 -13). We have previously shown that, in ␤-adrenergic cardiac hypertrophy, the Akt-GSK3␤ pathway, rather than the mitogen-activated protein kinase pathway, plays an essential role in transcription of atrial natriuretic factor (ANF), a "fetal type" gene that is commonly activated in many types of cardiac hypertrophy (13,14). It is unknown, however, which transcription factor is regulated by the Akt-GSK3␤ pathway.
Members of the GATA family zinc finger transcription factor regulate critical steps of cellular differentiation during vertebrate development (reviewed in Ref. 15). Among the GATA family, GATA-4, -5, and -6 are expressed in the heart and gut. GATA-4 is essential for survival of cardioblasts and terminal differentiation of cardiac myocytes (15)(16)(17)(18)(19). Functional GATAbinding sites have been identified within the promoters of multiple cardiac-specific genes, including ANF (20), slow/cardiac muscle-specific troponin C (21), and ␣-myosin heavy chain (22), and GATA4 plays an important role in cardiac musclespecific gene expression (15,22). It has been suggested that GATA transcription factors also mediate cardiac gene transcription in response to hypertrophic stimuli, including pressure overload, leukemia inhibitory factor, phenylephrine, endothelin-1, and electrical stimulation (see  reviewed in Ref. 28). A number of genes, whose expression is altered during cardiac hypertrophy, including ANF, brain na-triuretic peptide (BNP), ␣and ␤-myosin heavy chain, cardiac troponin I, platelet-derived growth factor receptor ␤, angiotensin II type 1a receptor, and Na ϩ -Ca 2ϩ exchanger, are critically regulated by GATA4 and other GATA-binding proteins (20,(23)(24)(25)(26)29). However, the mechanism as to how transcriptional activities of GATA4 are regulated by hypertrophic stimuli has not been fully understood.
The present studies were conducted to elucidate the mechanism by which the Akt-GSK3␤ pathway regulates nuclear transcriptional events in cardiac myocytes, leading to expression of ANF. We have identified that GATA4 plays an important role in ␤AR-regulated ANF transcription and that the Akt-GSK3␤ pathway regulates transcription through GATA4 in cardiac myocytes. Furthermore, we have obtained evidence that GSK3␤ physically interacts with GATA4 and decreases nuclear expression of GATA4 possibly through direct phosphorylation and subsequent stimulation of nuclear export via nuclear exportin Crm1. We propose that GATA4 is an important downstream effector of GSK3␤ and inhibition of GSK3␤ by hypertrophic stimuli mediates cardiac hypertrophy at least in part by removing its negative constraint upon GATA4.

EXPERIMENTAL PROCEDURES
Materials-Lithium chloride and leptomycin B were purchased from Sigma.
Primary Culture of Neonatal Rat Ventricular Myocytes-Primary cultures of cardiac ventricular myocytes from 1-day-old Crl:(WI)BR-Wistar rats (Charles River Laboratories) were prepared as described previously (30). In brief, ventricular myocytes were dispersed from the ventricles by digestion with collagenase type IV (Sigma), 0.1% trypsin (Life Technologies, Inc.), and 15 g/ml DNase I (Sigma). Cell suspensions were applied on a discontinuous Percoll gradient as described (10). Cells were cultured in the cardiac myocyte culture medium containing Dulbecco's modified Eagle's medium (DMEM)/F-12 supplemented with 5% horse serum, 4 g/ml transferrin, 0.7 ng/ml sodium selenite (Life Technologies, Inc.), 2 g/liter bovine serum albumin (fraction V), 3 mM pyruvic acid, 15 mM HEPES, 100 M ascorbic acid, 100 g/ml ampicillin, 5 g/ml linoleic acid and 100 M 5-bromo-2Ј-deoxyuridine (Sigma). We obtained myocyte cultures in which more than 95% were myocytes, as assessed by immunofluorescence staining with a monoclonal antibody against sarcomeric myosin (MF20). Culture media were changed to serum-free at 24 h. Myocytes were cultured under serum-free conditions for 48 h before experiments.
Transient Transfection and Reporter Gene Assays-For transient transfection, myocytes were plated at a density of 1 ϫ 10 6 per well in 6-well plates. Twenty four hours after plating, the medium was changed to DMEM/F-12 without supplement. Transfections were carried out using 10 l/ml of LipofectAMINE (Life Technologies, Inc.) in 1 ml/well DMEM. To determine the activity of the ANF promoter, a plasmid (1 g/ml) containing a 638-or 3003-base pair fragment of the rat ANF promoter linked to firefly luciferase (ANF-Luc-(638), courtesy of Dr. K. R. Chien, University of California, San Diego (10); ANF-Luc-(3003) (13)) was used. In addition, a series of 5Ј truncations of the ANF promoter were generated by PCR and subsequently subcloned into pGL3 (Promega). Construction and characterization of the ANF-Luc-(638) reporter genes (GM1 and GM2), which have mutations at the GATA4-binding sites, have been previously described (34). The GATA4 reporter gene (GATA4-Luc) in the pGL2-Promoter (Promega) vector is composed of the duplicate GATA4 target sequence from the ␣MHC gene (22) linked to the SV40 promoter and fused to the luciferase gene (34).
To determine the activity of the transcription activation domains of GATA4, a plasmid containing one or two GATA4 activation domains fused to the DNA binding domain of yeast transcription factor GAL4 (pFA-GATA4-(2-116) and pFA-GATA4-(2-205)) and another plasmid containing 5 copies of the GAL4-binding sequence fused to the firefly luciferase gene (pFR-Luc, Stratagene) were co-transfected. To evaluate the transcriptional activity of NF-ATs, the NF-AT-dependent reporter gene (NF-AT-Luc, Stratagene), which is composed of four NF-AT-binding sites upstream of the TATA box and the luciferase gene, was used. In some experiments, myocytes were co-transfected with various doses of expression plasmid encoding signaling molecules. Total amounts of DNA were adjusted to 2 g/ml by adding appropriate empty vectors. Twenty four h after transfection, media were changed to the cardiac myocyte culture media without serum. Myocytes were further cultured in the presence or absence of isoproterenol (ISO, 10 M), an agonist for the ␤ARs, or phenylephrine (10 M), an agonist for the ␣ 1 -adrenergic receptors, for an additional 24 h. Myocytes were then lysed with the Reporter lysis buffer (Promega), and luciferase activities were measured. An SV40 promoter-driven ␤-galactosidase construct (SV40-␤-galactosidase, 0.5 g/ml) was co-transfected, and ␤-galactosidase activities were determined using Lumi-Gal 530 (Lumigen). The luciferase Preparation of Nuclear Extracts-Nuclear proteins were extracted as described (38). Cells were washed and scraped with Tris-buffered saline. After centrifugation, cells were resuspended in 200 l of ice-cold buffer A (10 mM HEPES, pH 7.9, 10 mM KCl, 0.1 mM EDTA, 0.1 mM EGTA, 0.1 mM Na 3 VO 4 , 1 mM dithiothreitol, 0.5 mM AEBSF) by gentle pipetting and allowed to swell on ice for 15 min. Cells were lysed by adding 12.5 l of 10% Nonidet P-40 and vortexing vigorously for 10 s, followed by centrifugation (12,000 rpm) for 30 s. Supernatant was saved as the cytosolic fraction. Pellets were resuspended in 50 l of ice-cold buffer B (20 mM HEPES, pH 7.9, 400 mM NaCl, 1 mM EDTA, 1 mM EGTA, 0.1 mM Na 3 VO 4 , 1 mM dithiothreitol, 0.5 mM AEBSF) and rocked for 15 min. The samples were centrifuged at 14,000 rpm for 5 min. The supernatant was saved as the nuclear fraction. The entire procedure was carried out at 4°C. The samples were stored at Ϫ85°C.
Gel Mobility Shift Assays-The gel mobility shift assays were performed as described previously (34,39). Double-stranded synthetic oligonucleotides containing the DNA sequence around the GATA consensus binding site in the rat proximal ANF promoter and the mouse BNP promoter or their mutants (Table I) were labeled by [␥-32 P]ATP and polynucleotide kinase. The labeled probe was purified by electrophoresis on 20% polyacrylamide gels. Samples of 15 l containing 5 g of nuclear extract were incubated with 3 g of poly(dI-dC) and 1 ng of probe (50,000 cpm) in the presence or absence of competitor oligonucleotides for 20 min at room temperature. Some samples were incubated with anti-GATA4 antibody (Santa Cruz Biotechnology) for 15 min at room temperature before adding probe. Binding reactions were subjected to electrophoresis on a 4 -6% non-denaturing polyacrylamide gel in 0.5ϫ TBE and visualized by autoradiography. To determine if the equal amount of nuclear extracts are quantitated for GATA binding, the same nuclear extracts were subjected to another gel shift assay, in which radiolabeled, double-stranded oligonucleotide probes containing the SP-1 and NF-B consensus binding sequences (Promega and Santa Cruz Biotechnology, respectively, Table I) were use as probes for COS7 cell and cardiac myocyte nuclear extracts, respectively.
RT-PCR-To detect GATA4 transcript in cardiac myocytes, a reverse transcription-polymerase chain reaction (RT-PCR) was carried out as described previously (40). Total RNA was isolated and subjected to reverse transcription with a first strand cDNA synthesis kit (Stratagene). The PCR primers were designed on the basis of published cDNA sequence for mouse GATA4 and rat GAPDH as follows: sense for GATA4, AAGACGCCAGCAGGTCCTGCTGGT; antisense for GATA4, CGCGGTGATTATGTCCCCATGACT; sense for GAPDH, GGAAAGCT-GTGGCGTGATGG; antisense for GAPDH, GTAGGCCATGAGGTC-CACCA. The following conditions were chosen for the PCRs in a volume of 50 l: reverse transcription products from 1 g of RNA, 2.5 units of Taq polymerase, and 33 cycles of amplification for GATA4 or 20 cycles for GAPDH, and 250 nM each of sense and antisense primers. The amplification was carried out as follows: denaturation, 30 s at 94°C; annealing, 30 s at 54°C; extension, 30 s at 72°C. The PCR products were subjected to electrophoresis on a 2% agarose gel and stained with ethidium bromide.
Immunoblotting of the Nuclear and Cytosolic Extracts-Nuclear and cytosolic extracts were prepared as described above. The samples were subjected to SDS-PAGE and transferred onto polyvinylidene difluoride membranes. The membranes were probed with anti-GATA4 or anti-GSK3␤ antibody (Santa Cruz Biotechnology). Horseradish peroxidaseconjugated goat anti-rabbit IgG antibody was used as secondary antibody. The bound secondary antibody was detected by enhanced chemiluminescence (ECL) or ECL Plus (Amersham Pharmacia Biotech).
In Vitro Phosphorylation of GST-GATA4 by GSK3␤-cDNA encoding the amino acid residues 2-116, 2-205, and 206 -440 of GATA4 were generated by PCR and subcloned into pGEX-4T-3 to generate pGEX-GATA4-(2-116), pGEX-GATA4- , and pGEX-GATA4-(206 -440). Note that PCR primers for GATA4 were made by using the updated mouse GATA4 sequence (GenBank TM accession number NM008092). Bacterially expressed GST-GATA4 or GST alone was purified on glutathione-Sepharose and used at 1 g of recombinant protein per 10 l of bead slurry. The fusion proteins were pre-phosphorylated on Sepharose by addition of 5 units of cAMP-dependent protein kinase (New England Biolabs) per g of fusion protein at 30°C in kinase buffer (20 mM Tris, pH 7.5, 10 mM MgCl 2 , and 1 mM dithiothreitol) with 1 mM ATP for 2 h and then thoroughly washed (with kinase buffer without ATP) to remove cAMP-dependent protein kinase and ATP. For kinase assays, fusion protein (1 g) on glutathione-Sepharose was incubated with 100 M ATP and 5 Ci of [␥-32 P]ATP (6000 Ci/mmol) in 50 l of kinase buffer for 30 min at 30°C. Beads were incubated with 2.5 units of purified GSK3 (New England Biolabs). Kinase reactions were terminated by washing the Sepharose beads twice with 1 ml of TEN (50 mM Tris, pH 7.5, 1 mM EDTA, 150 mM NaCl, and 0.5% Nonidet P-40) to remove free ATP and enzyme. Phosphorylated GATA4 was eluted by boiling with 1ϫ SDS-PAGE loading buffer for 10 min, fractionated on SDS-PAGE, and stained with Coomassie to ensure that the substrate was not degraded. The gels were dried and subjected to autoradiography.
Immunoprecipitation-For immunoprecipitation of GSK3␤, total cell extracts (1 ml) were prepared by using the cell lysis buffer C, containing 20 mM Tris, pH 7.5, 150 mM NaCl, 1 mM EDTA, 1 mM EGTA, 1% Triton X-100, 2.5 mM sodium pyrophosphate, 1 mM ␤ glycerophosphate, 1 mM Na 3 VO 4 , 0.5 g/ml leupeptin, 0.5 mM AEBSF. The samples were incubated with 1 g of anti-GSK3␤ antibody (Transduction Laboratory) for 2 h and then with 30 l of protein A/G-Sepharose slurry (Santa Cruz Biotechnology) at 4°C for 1 h. Immunoprecipitates were washed three times with the lysis buffer and subjected to SDS-PAGE.
Immunostaining-Cardiac myocytes were grown on glass coverslips. After stimulation, myocytes were fixed with methanol at Ϫ20°C for 10 min followed by a brief dip in acetone at Ϫ20°C. For staining of GATA4 and GSK3␤, anti-GATA4 antibody (Santa Cruz Biotechnology, 1 g/ml) and anti-GSK3␤ antibody (Transduction Laboratory, 1 g/ml) were used, respectively. Fluorescein isothiocyanate-conjugated anti-rabbit or anti-mouse IgG antibody (Jackson ImmunoResearch) was used as secondary antibody at 1:200 dilution in phosphate-buffered saline. Methods of immunostaining and subsequent light microscopic observations have been described (30). To confirm nuclear localization of GATA4, confocal microscopic analyses were performed in some experiments as described previously (30).
Statistics-Data are given as mean Ϯ S.E. Statistical analyses were performed using the analysis of variance. The post-test comparison was performed by the method of Tukey. Significance was accepted at p Ͻ 0.05 levels.

RESULTS
The GATA-binding Site Is Essential for ISO-induced Increases in ANF Transcription-We have shown previously that ISO, an agonist for ␤ARs, stimulates transcription of ANF in cultured neonatal rat cardiac myocytes in vitro and adult heart in vivo (14). In order to identify the cis-acting element that is critical for ␤AR-stimulated ANF transcription, we transfected ANF-Luc reporter gene constructs, in which the 5Ј region of the rat ANF promoter is progressively deleted. Commonly used ANF promoter luciferase constructs having Ϫ3003 and Ϫ638 bp upstream of the transcription start site responded to ISO stimulation by 3-4-fold, consistent with our previous observations (13). Deletion to Ϫ294 as well as that to Ϫ134 caused a modest decrease in ISO responsiveness (2.3-and 2.4-fold induction versus respective controls) compared with ANF-Luc-(Ϫ3003), whereas deletion to Ϫ109 abolished ISO-induced increases in the promoter activities ( Fig. 1). Since the ANF-Luc reporter construct deleted to Ϫ109 remained inducible to stimulation by phenylephrine, an agonist for ␣ 1 -adrenergic receptor and a well established hypertrophic stimulus, the loss of inducibility in this construct was not nonspecific. These results suggest that sequences Ϫ134/Ϫ109 in the rat ANF promoter play a critical role in ANF transcription by ␤AR stimulation.
The rat ANF promoter sequences Ϫ134/Ϫ109 contain the functional serum-response element (SRE) and the proximal GATA element (20,29,34). To examine if these cis-acting regulatory elements mediate ISO-induced increases in ANF transcription, we mutated these elements in the context of the Ϫ638-bp ANF promoter. When the SRE was mutated, although the basal activity of ANF-Luc was significantly attenuated, ISO significantly stimulated ANF transcription (dark gray column, Fig. 2A). Thus, additional elements may be responsible for the ISO responsiveness.
The proximal ANF promoter contains two conserved GATA elements, and GATA4 dose-dependently stimulated the ANF-Luc-(Ϫ638) activity ( Fig. 2B), consistent with previous observations (20,29,34). To examine the role of the GATA elements in ISO-induced increases in ANF transcription, we used ANF-Luc-(Ϫ638) with mutations in these GATA elements (GM1 with mutation at the distal (Ϫ270) GATA site and GM2 at the proximal (Ϫ122) site (34)). Basal luciferase activities were slightly reduced in both GM1 and GM2 compared with those of wild type ANF-Luc-(Ϫ638), and GATA4-induced increases in ANF transcription were markedly attenuated in GM1 and GM2 (Fig. 2B), suggesting that these GATA elements mediate GATA4-induced transcription in cardiac myocytes. ISO-induced increases in ANF transcription were modestly reduced in GM1 and significantly reduced in GM2 ( Fig. 2A). Interestingly, responsiveness to phenylephrine was not significantly affected in GM1 or GM2 (data not shown).
To confirm further the essential role of GATA4 in ␤-adrenergic ANF transcription, we examined the effect of dominant negative GATA4, in which amino-terminal activation domains of GATA4 were deleted (GATA4-(199 -440)) (34,35). Although expression of GATA4-(199 -440) did not affect basal activities of ANF-Luc-(Ϫ638), it significantly reduced ISO-induced increases in ANF transcription (Fig. 2C). These results suggest that the GATA elements and GATA4 play an important role in induction of ANF transcription by ␤AR stimulation.
ISO Increases GATA4 Binding to the ANF Promoter-We next examined whether or not stimulation of the ␤AR affects GATA4 binding to the GATA element. Cardiac myocytes were stimulated with ISO, and nuclear extracts were incubated with a radiolabeled, double-stranded DNA probe containing the rat proximal ANF GATA element (ANF-GATA probe), and the gel mobility shift assays were performed (Fig. 3A, left). One retarded band found in ISO-treated samples (indicated by an thick arrow) represented GATA sequence-specific binding because it was competed by an excess of the unlabeled ANF-GATA probe (lane 9) but not by an excess of the unlabeled ANF-GATA oligonucleotide containing point mutations (ANF-GATA mutant oligonucleotide) (data not shown). Furthermore, the intensity of the band was attenuated, and the faint supershifted band (indicated by thin arrow) was observed (lanes 4, 6, and 8) when anti-GATA4-specific antibody was included in the binding reaction, suggesting that GATA4 is a component of the GATA complex. ISO treatment increased the intensity of the retarded band at 30 min, and the intensity of the retarded band was still elevated at 6 h. Control gel mobility shift assays were performed by using the same samples and a radiolabeled oligonucleotide containing a consensus SP-1-binding sequence.
In these experiments, a retarded band with a similar intensity was observed in each sample (Fig. 3A, right), suggesting that each sample contains almost equal amounts of nuclear protein.
The intensity of the retarded bands found in the gel mobility shift assays with the GATA and SP-1 site probes was quantitated by densitometric analyses, and the intensity of the band representing GATA4 was normalized by that representing SP-1. Mean of the results obtained from five independent experiments showed that intensity of the GATA4-containing band was significantly increased within 30 min of ISO stimulation, and it was significantly elevated at 6 h (Fig. 3B).
ISO Causes Rapid Nuclear Accumulation and Subsequent Up-regulation of GATA4 -Since ISO stimulates GATA4 binding to the ANF promoter, we next examined the effect of ISO on the subcellular localization of GATA4 by immunostaining. In control myocytes, weak staining of GATA4 was observed in the nucleus, but the boundary between the nucleus and cytoplasm was not prominent (Fig. 4A). ISO increased the intensity of the GATA4 staining, which was clearly localized at the nucleus within 1 h. Nuclear localization of GATA4 in myocytes stimulated with ISO for 30 min was also confirmed by confocal microscopic analyses (data not shown, see Fig. 10B).
To examine if ISO treatment causes changes in subcellular localization of GATA4 in myocytes, both nuclear and cytoplasmic fractions were prepared and immunoblot analyses performed. ISO stimulation caused increases in nuclear expression of GATA4 within 1 h, consistent with the result of the immunostaining. A small level of GATA4 expression was observed in the cytoplasm, and the level of GATA4 in the cytoplasm was slightly decreased after ISO stimulation (Fig. 4B). However, due to very low signal intensities of GATA4 in the cytoplasmic fraction, changes in the amount of GATA4 in the cytoplasm were only modest. Densitometric analyses of the immunoblots, in which cytoplasmic and nuclear fractions were quantitated on the same filter, indicated that the intensity of GATA4 band in the nucleus was 7.7 Ϯ 2.7 (n ϭ 4)-fold higher than that in the cytoplasm.
To examine if changes in nuclear expression of GATA4 are due to accumulation of GATA4 mRNA, total RNA was prepared from control and ISO-stimulated myocytes, and the equal amount of RNA was subjected to RT-PCR. ISO increased mRNA expression of GATA4 after 6 h (lane 6) of ISO stimulation, and increases in GATA4 expression persisted for at least 12 h (lane 7) (Fig. 4C). It should be noted that although increases in nuclear staining and DNA binding of GATA4 were observed within 1 h of ISO stimulation, changes in mRNA expression of GATA4 were not observed during the initial 3 h (Fig. 4B, left panel). These results suggest that accumulation of GATA4 mRNA does not mediate ISO-induced increases in GATA4 expression in the nucleus during the initial few hours, although it may contribute to increases in nuclear expression of GATA4 after 6 h of ISO stimulation.
The Akt-GSK3␤ Pathway Regulates GATA4 Transcription-We have recently shown that ISO-induced increases in ANF transcription are predominantly mediated by the Akt-GSK3␤ pathway (13). GSK3␤ is catalytically active and inhibits ANF transcription in unstimulated myocytes, whereas ISOinduced activation of Akt phosphorylates/inactivates GSK3␤ (13). ISO-induced inactivation of GSK3␤ removes the negative constraint of GSK3␤ on ANF transcription, thereby stimulating ANF transcription (13). To determine if GATA4 is regulated by the Akt-GSK3␤ pathway, we examined the effect of GSK3␤ on GATA4-induced increases in ANF transcription. Expression of GATA4 increased both basal and ISO-induced increases in ANF-Luc activities (Fig. 5A). Co-expression of wild type GSK3␤ significantly attenuated GATA4-induced increases in ANF-Luc activities in both control and ISO-stimulated cardiac myocytes although ISO-induced increases in ANF transcription were partially preserved (Fig. 5A). Co-expression of GSK3␤(S9A), an Akt-insensitive mutant in which the Akt phosphorylation site (Ser-9) was mutated, inhibited both basal and ISO-induced increases in ANF-Luc activities (Fig. 5A). These results are consistent with the notion that GSK3␤ negatively regulates GATA4 transcription, and Ser-9 phosphorylation of GSK3␤ by FIG. 3. ISO stimulates DNA binding of GATA4. Cardiac myocytes were stimulated with isoproterenol (10 M) for various periods. Nuclear extracts were probed with a radiolabeled, doublestranded oligonucleotide containing the ANF proximal GATA site (see Table I)  Akt reverses GSK3␤-induced inhibition of GATA4 in cardiac myocytes.
To examine further the effect of the Akt-GSK3␤ pathway on transcriptional activities of endogenous GATA, we used the GATA-Luc reporter gene composed of the duplicated GATA4 target sequence from the ␣MHC gene (22) in front of the SV40 promoter (34). ISO did not affect the activity of luciferase in pGL2-promoter without GATA sequences, suggesting that ISO does not affect the SV40 promoter. ISO stimulation significantly increased GATA-Luc activities in cardiac myocytes, suggesting that ISO increases transcriptional activities of GATA4. Co-transfection of GSK3␤ (S9A) significantly suppressed the ISO-induced increases in the GATA-Luc activity. Kinase-inactive GSK3␤ significantly stimulated basal GATA4-Luc activities (Fig. 5B). Co-transfection of constitutively active Akt, which phosphorylates and inhibits endogenous GSK3␤, significantly stimulated GATA-Luc (Fig. 5B). These results suggest that the Akt-GSK3␤ pathway regulates transcription through GATA4.
NF-AT Does Not Mediate ISO-induced Activation of GATA4 Transcription-GATA4 has been shown to associate with NF-AT3, thereby enhancing transcriptional activities of NF-AT3 (9). By inference, such interaction between GATA4 and NF-AT3 may also enhance transcriptional activities of GATA4. It has been shown that GSK3 directly phosphorylates NF-ATc and causes nuclear exit of NF-ATc in COS cells and cardiac myocytes (41,42). Thus, it is possible that the effect of GSK3␤ on GATA4 is mediated by NF-ATs. To examine if ISO activates transcription of NF-AT in cardiac myocytes, we used the NF-AT-Luc reporter gene. Although co-expression of NF-AT3 significantly stimulated the activity of NF-AT-Luc, ISO failed to stimulate it in cardiac myocytes. Co-transfection with wild type or kinase-inactive GSK3␤ also failed to affect NF-AT-Luc activities (Fig. 6). These results suggest that the transcriptional activity of NF-AT is neither stimulated by ISO nor significantly affected by GSK3␤ in cardiac myocytes. Thus, it is unlikely that the effect of GSK3␤ on GATA4 is mediated by NF-AT3.
GSK3␤ Interacts with GATA4 and Phosphorylates GATA4 -We next tested the possibility that GSK3␤ physically interacts with GATA4, thereby directly modulating GATA4mediated transcription. We co-transfected GSK3␤ and GATA4 into COS7 cells, and GSK3␤ was immunoprecipitated from the total cell extracts. The immunoprecipitates were subsequently immunoblotted with anti-GATA4 antibody. GATA4 was found in the GSK3␤ immunoprecipitates (lanes 2-4) but not in control (lane 5) (Fig. 7A). The band corresponding to GATA4 was not found in the GSK3␤ immunoprecipitates prepared from COS7 cells transfected with GSK3␤ alone (lane 1). The result FIG. 4. ISO stimulates nuclear accumulation of GATA4. A, cardiac myocytes were grown on gelatin-coated glass coverslips and cultured under serum-free conditions for 48 h before experiments. Myocytes were then stimulated with or without isoproterenol (10 M) for 1 h and then fixed. Immunofluorescent staining with anti-GATA4 antibody was performed. Panel a, control; panel b, 1-h stimulation with ISO. B, cardiac myocytes were stimulated with ISO for various periods. Cytosolic (50 g) and nuclear (50 g) fractions were subjected to immunoblot analyses by using anti-GATA4 antibody. Note that the immunoblot of the cytosolic fraction and that of the nuclear fraction were conducted separately. ECL-Plus was used for the immunoblot of the cytosolic fraction. C, cardiac myocytes were stimulated with ISO for various periods. In some experiments (lane 4), cardiac myocytes were transduced with adenovirus harboring GSK3␤ (100 multiplicity of infection) and harvested at 48 h. One g of total RNA/per each time point was subjected to RT-PCR using the primers specific for GATA4. As an internal control, GAPDH was amplified. Ethidium bromide staining of the RT-PCR products is shown. Note that samples in lanes 1-4 and  (Fig. 7B), suggesting that GSK3␤ directly phosphorylates the amino terminus of GATA4 in vitro.
GSK3␤ Decreases Nuclear Expression of GATA4 -Since GSK3␤ failed to change the activity of the GATA4 activation domains, we next examined the effect of GSK3␤ on nuclear expression of GATA4. We transfected COS7 cells with GATA4 together with control vector, wild type GSK3␤, or kinase-inactive GSK3␤. Nuclear extracts were prepared, and DNA binding of GATA4 was determined (Fig. 9A). In the gel mobility shift assays with the oligonucleotide probe containing the GATA4binding sequence, the band containing the nuclear GATA4 (shown by thick arrow) was identified by the supershift assay, using anti-GATA4 antibody (lane 7). Co-expression of GSK3␤ reduced DNA binding of GATA4 (lane 5) whereas that of kinase-inactive GSK3␤ did not significantly affect them (lane 6). To show that other proteins in the nuclear extract are not similarly affected, additional gel shift assays were performed by using the same nuclear extracts and an oligonucleotide probe containing the specific NF-B-binding sequence (Fig. 9A,  right). Although the DNA binding to the NF-B site was increased by GATA4 (lane 8 versus lanes 9 -11), almost equal levels of binding to the NF-B site were observed for the samples used for lanes 4 -6 (lanes 9 -11 in the right panel of Fig. 9A).
The effect of GSK3␤ on subcellular localization of GATA4 was determined by immunoblot analyses. Co-expression of GSK3␤ significantly reduced nuclear expression of GATA4 (Fig. 9B). Although GATA4 was predominantly found in the nucleus, small expression of GATA4 in the cytosol was also reduced in the presence of GSK3␤. These results suggest that chronic overexpression of GSK3␤ decreases the total amount of GATA4 in COS7 cells.
Since ␤-adrenergic stimulation by ISO inactivates GSK3␤ in cardiac myocytes (13), we examined the effect of LiCl, a well established inhibitor of GSK3␤ (44), on nuclear expression of GATA4 in cardiac myocytes. Immunostaining with anti-GATA4 antibody and confocal microscopic analyses showed that 1 h of treatment with LiCl (20 mM) caused significant increases in nuclear staining of GATA4 similar to ISO-induced nuclear accumulation of GATA4 (Fig. 10B). These results indicate that inhibition of GSK3␤ by either LiCl or ISO causes nuclear accumulation of GATA4 in cardiac myocytes.
Nuclear Expression of GSK3␤ Is Reduced in Response to ISO Stimulation-We examined subcellular localization of GSK3␤ in cardiac myocytes before and after ISO stimulation and determined if there is any correlation regarding the intracellular localization between GSK3␤ and GATA4. In the control state, GSK3␤ was localized both in the cytoplasm and in the nucleus (Fig. 11A). After 1 h of stimulation with ISO, staining of GSK3␤ was predominantly localized in the cytoplasm, and the nuclear staining was decreased (Fig. 11A). Changes in subcellular localization of GSK3␤ were confirmed by immunoblot analyses. In the control state, GSK3␤ was localized both in the nucleus and in the cytoplasm (Fig. 11B). ISO stimulation decreased nuclear expression of GSK3␤, whereas it increased GSK3␤ in the cytoplasm (Fig. 11B). Thus, active GSK3␤ and GATA4 may co-exist in the nucleus in unstimulated myocytes, and ISO stimulation not only inactivates GSK3␤ (13) but also causes nuclear exit (and predominant cytosolic localization) of GSK3␤.
GSK3␤ Stimulates Nuclear Export of GATA4 -The results presented thus far indicate that active GSK3␤ decreases nuclear expression of GATA4, whereas inhibition of GSK3␤ causes nuclear accumulation of GATA4. Since cytoplasmic expression of GATA4 is low in both cardiac myocytes and GATA4transfected COS7 cells and because chronic overexpression of GSK3␤ in COS7 cells failed to induce cytoplasmic accumulation of GATA4, it is unlikely that nuclear accumulation of GATA4 upon inhibition of GSK3␤ is caused by translocation of Inactivation of GSK3␤ by Akt removes the inhibitory effect of GSK3␤, thereby stimulating transcription through GATA4. We propose that transcription through GATA4 is regulated by the phosphorylation status of GATA4. PKB, protein kinase B; prox GATA, proximal GATA element. Lower, nuclear events controlling nuclear expression of GATA4 in unstimulated myocytes (left) and myocytes stimulated with agonists for the ␤-adrenergic receptors (right). In control myocytes, active GSK3␤ exists in the nucleus, phosphorylates GATA4, and stimulates nuclear export of GATA4 with nuclear exportin Crm1. After ␤-adrenergic stimulation, GSK3␤ is inactivated and predominantly localized in the cytoplasm. Nuclear export of GATA4 is not stimulated, which allows GATA4 to accumulate in the nucleus.
GATA4 from the cytoplasm to the nucleus. To examine if nuclear localization of GATA4 is regulated by the nuclear export, we transfected Crm1, a nuclear exportin, and GATA4 into COS7 cells with or without activation of GSK3␤. Expression of Crm1 decreased nuclear expression of GATA4 (Fig. 12A, lane  10). Concomitant overexpression of GSK3␤ (lane 11), but not kinase-inactive GSK3␤ (lane 12), further decreased nuclear expression of GATA4. Modest expression of GATA4 in the cytoplasm was slightly decreased by overexpression of Crm1 plus GSK3␤ (lane 4). Decreases in nuclear expression of GATA4 by Crm1 plus GSK3␤ (lane 11) were completely reversed in the presence of the specific inhibitor of nuclear exportin, leptomycin B (33) (Fig. 12A, lane 13). Although cytoplasmic GATA4 slightly decreased after 1 h of treatment with leptomycin B (Fig. 12A, lane 6 versus lane 4), this change was modest due to low expression levels of GATA4 in the cytoplasm. These results suggest that nuclear expression of GATA4 may be at least in part regulated by GSK3␤ and the nuclear exportin Crm1.
To examine if Crm1 regulates nuclear expression of GATA4 in cardiac myocytes, we treated cardiac myocyte with leptomycin B and performed immunostaining with anti-GATA4 antibody. Leptomycin B treatment significantly increased nuclear staining of GATA4 in cardiac myocytes (Fig. 12B), indicating that nuclear localization of GATA4 is regulated by the nuclear exportin in cardiac myocytes. DISCUSSION In this study, we have demonstrated that GATA4 plays an important role in ␤AR-stimulated ANF transcription. We have also provided evidence that GSK3␤ regulates GATA4; GSK3␤ decreases nuclear GATA4 at least in part by stimulating nuclear export of GATA4, thereby inhibiting transcription through GATA4. Together with our previous observation that the Akt-GSK3␤ pathway mediates ␤AR-induced ANF transcription (13), these results suggest that inactivation of GSK3␤ and subsequent nuclear accumulation of GATA4 play an important role in ␤AR-induced ANF transcription (Fig. 13).
GATA4 Is an Essential Factor Mediating ANF Transcription by ␤AR Stimulation-Our results suggest that sequences between Ϫ134 and Ϫ109 in the rat ANF promoter, containing the SRE (45) and the GATA element (20,29,34), are required for the ISO responsiveness in the context of the reporter genes used. Although a mutation in the SRE reduced basal activities of ANF-Luc, significant induction of ANF-Luc was observed by ISO stimulation. This suggests that other elements may suffice to mediate ISO response. By contrast, ISO-induced increases in ANF transcription were significantly attenuated when the proximal GATA element was mutated. Thus, the proximal GATA element is required for the ISO responsiveness of the ANF promoter. It should be noted, however, that since mutations at the GATA element did not completely abolish ISO responsiveness, other elements may be also required to confer full ISO responsiveness to the ANF promoter.
Several lines of evidence support the notion that transcription through GATA4 is activated by ␤AR stimulation in cardiac myocytes. ISO stimulation increases DNA binding of GATA4 to the GATA element. ISO stimulates transcription from the promoter, which consists of tandem GATA-binding sequences. ISO also causes striking nuclear accumulation of GATA4. To our knowledge, agonist-stimulated accumulation of GATA4 in the nucleus has not been reported. Since ISO-induced increases in ANF transcription were significantly inhibited in the presence of dominant negative GATA4, our results indicate that GATA4 critically mediates ANF transcription, an important phenotype in ␤-adrenergic cardiac hypertrophy.
The Akt-GSK3␤ Pathway Regulates GATA4 -Several lines of evidence support the notion that GATA4 is a nuclear target of the Akt-GSK3␤ pathway. First, overexpression of GSK3␤ inhibits GATA4-mediated activation of GATA4-Luc and ANF-Luc. Second, both overexpression of kinase-inactive GSK3␤ and inhibition of GSK3␤ by constitutively active Akt stimulate transcription through GATA4. Third, GSK3␤ inhibits both nuclear expression of GATA4 and binding of GATA4 to the GATA element. Fourth, inhibition of GSK3␤ by LiCl or ISO causes nuclear accumulation of GATA4. These results suggest that GSK3␤ negatively regulates transcription through GATA4. Although a possibility remains that Akt may also directly affect GATA4 (46), such a mechanism may not be a major component, since ␤AR-induced ANF transcription was completely abolished in the presence of active GSK3␤ (13). Regulation of GATA4 by GSK3␤ has not been previously reported. Since the kinase activity of GSK3␤ is inhibited by several cardiac hypertrophic stimuli (13,42), we predict that the removal of the GSK3␤'s negative constraint upon GATA4 may be an important signaling mechanism to promote cardiac hypertrophy by many hypertrophic stimuli. It should be noted, however, since GSK3␤ also regulates other transcription factors (41,(47)(48)(49)(50)(51), GSK3␤ may affect development of cardiac hypertrophy by regulating multiple transcription factors in a coordinated fashion. Phosphorylation of GATA4 Regulates GATA4-mediated Transcription-Several lines of evidence suggest that GSK3␤ regulates GATA4 through direct phosphorylation. First, GSK3␤ effectively phosphorylates GST-GATA4-(2-116) and GST-GATA4-(2-205) in vitro. Second, GSK3␤ and GATA4 physically interact with one another in COS7 cells. Third, both catalytically active GSK3␤ and GATA4 exist in the nucleus, suggesting that they may interact each other in unstimulated cardiac myocytes. Mouse GATA4 has consensus amino acid sequences for GSK3␤ phosphorylation (-SXXXS(P)-) (52) in the amino terminus between two activation domains as well as in the carboxyl terminus adjacent to the DNA binding domain. Both GATA4-(2-116) and GATA4-(2-205) contain two aminoterminal consensus GSK3␤ phosphorylation sites. Identifying the in vivo GSK3␤ phosphorylation site will be important to confirm that GSK3␤ is the relevant GATA4 kinase and to elucidate the role of phosphorylation in transcription through GATA4. It should be noted that our results do not rule out the possibility that GSK3␤ phosphorylates GATA4-associating transcription factors, including AP-1 (25), GATA6 (29), Nkx 2.5 (34,(53)(54)(55), FOG-2 (56,57), and MEF2 (58), to suppress GATA4.
GSK3␤ Promotes Nuclear Export of GATA4 -GATA transcription factors, including GATA-1, -2 and -4, are phosphorylated by mitogen-activated protein kinases (59 -61). In GATA-1, 6 serine residues in the amino-terminal activation domain are constitutively phosphorylated, whereas 1 serine residue in the carboxyl terminus is phosphorylated in response to dimethyl sulfoxide (59). Although inducible phosphorylation of GATA-1 and -2 does not change their transcriptional activity (59,60), that of GATA4 by endothelin-1 increases DNA binding activity of GATA4 and stimulates transcription (61). In the latter case, it is unknown if increased DNA binding is accompanied by nuclear accumulation of GATA4. In any case, the functional consequence of GATA4 phosphorylation by GSK3␤ differs from that of aforementioned phosphorylation of the GATA transcription factors, since phosphorylation by GSK3␤ rather decreases nuclear expression of GATA4. Interestingly, in Saccharomyces cerevisiae, a GATA transcription factor GLN3 is phosphorylated by TOR kinases, which causes cytoplasmic retention and consequent suppression of transcription through GLN3 (62).
Our results suggest that GSK3␤ negatively regulates the level of GATA4 in the nucleus while inhibition of GSK3␤ allows nuclear accumulation of GATA4. GATA4 is localized predominantly in the nucleus, and chronic overexpression of GSK3␤ also decreased cytoplasmic expression of GATA4 in COS7 cells. Thus, chronic overexpression of GSK3␤ decreases total GATA4 expression in cells. Although the short term ISO treatment showed slight decreases in cytoplasmic expression of GATA4, the amount of cytoplasmic expression of GATA4 was found to be very low regardless of ISO stimulation. Thus, it is unlikely that nuclear accumulation of GATA4 by ISO is due to translocation of GATA4 from the cytosol to the nucleus. Does GSK3␤ suppress synthesis or promote degradation of GATA4? It is unlikely that GSK3␤ transcriptionally regulates production of GATA4, since overexpression of GSK3␤ did not affect GATA4 mRNA levels in cardiac myocytes. At present, we cannot exclude the possibility that GSK3␤ selectively inhibits translation of GATA4. GSK3␤ phosphorylates eIF2B⑀ and negatively regulates initiation of protein translation. However, this mechanism may not explain selective inhibition of GATA4 translation. By contrast, the results of the present investigation strongly suggests that GSK3␤ promotes degradation of GATA4 by stimulating nuclear exit through the nuclear exportin (63). Nuclear expression of GATA4 was decreased in the presence of Crm1 in COS7 cells. Crm1-induced decreases in nuclear expression was enhanced by GSK3␤ but not by kinaseinactive GSK3␤. Decreased nuclear expression of GATA4 by Crm1 plus GSK3␤ was reversed in the presence of leptomycin B, a selective inhibitor of Crm1 (33). Short term treatment with leptomycin B was sufficient to stimulate nuclear accumulation of GATA4 in cardiac myocytes. These results indicate that stimulation of nuclear export by Crm1 and GSK3␤ maintains low levels of expression in the nucleus, whereas inhibition of nuclear export causes nuclear accumulation of GATA4. It has been shown recently that phosphorylation by GSK3␤ exposes the nuclear export signal in the nuclear factors, such as cyclin D1 and Dictyostelium STAT (Dd-STATa), thereby stimulating their interaction with Crm1 (64,65). Such interaction causes nuclear exit of the nuclear factors. A potential nuclear export signal sequence was found in the GATA4 sequence (LGLSYL) at positions 49 -54. Further studies are required to determine if GSK3␤-induced phosphorylation of GATA4 stimulates interaction between GATA4 and Crm1. Since GATA4 is predominantly found in the nucleus (35), we predict that GATA4 exported from the nucleus is subjected to degradation. This may explain why we do not see cytoplasmic accumulation of GATA4 in our GSK3␤ and Crm1 transfected COS7 cells at 48 h after transfection. Further investigations are required to elucidate the fate of the GATA4 exported from the nucleus and the effect of GSK3␤ on this process.
Active GSK3␤ exists both in the nucleus and in the cytosol in unstimulated myocytes, whereas it is excluded from the nucleus by ␤AR stimulation. Stimulus-dependent translocation of GSK3␤ between the nucleus and cytoplasm has been shown in NIH 3T3 fibroblasts, where GSK3␤ accumulates primarily in the cytoplasm, whereas the subpopulation of GSK3␤ becomes nuclear during S phase, the interval of the cell cycle in which cyclin D1, a substrate of GSK3␤, leaves nucleus and enters the cytoplasm (66). In our experiments, the subpopulation of GSK3␤ is nuclear only in the control state. We speculate that GSK3␤ in the nucleus interacts with GATA4 and causes nuclear export of GATA4, thereby limiting nuclear localization of GATA4. Alternatively, GSK3␤ in the cytosol may continuously prevent nuclear import of GATA4, if any GATA4 escaped from degradation exists in the cytosol. The fact that GSK3␤ is excluded from the nucleus by ISO stimulation suggests that, in addition to inactivation of GSK3␤, nuclear exit of GSK3␤ should ensure attaining the reduced level of phosphorylation and increased nuclear accumulation of GATA4 in response to ␤AR stimulation.
In summary, our results have established a novel linkage between the Akt-GSK3␤ pathway and GATA4 in cardiac myocytes. Furthermore, we have provided evidence that GSK3␤ decreases nuclear expression of GATA4 through possible phosphorylation and subsequent nuclear export of GATA4. The fact that the Akt-GSK3␤ pathway critically regulates a major cardiac transcription factor GATA4 has confirmed that the Akt-GSK3␤ pathway represents an important signaling mechanism mediating cardiac hypertrophy and warranted further studies regarding the molecular mechanism of transcriptional regulation by the Akt-GSK3␤ pathway.