GATA-4 Is a Nuclear Mediator of Mechanical Stretch-activated Hypertrophic Program*

In overloaded heart the cardiomyocytes adapt to increased mechanical and neurohumoral stress by activation of hypertrophic program, resulting in morphological changes of individual cells and specific changes in gene expression. Accumulating evidence suggests an important role for the zinc finger transcription factor GATA-4 in hypertrophic agonist-induced cardiac hypertrophy. However, its role in stretch-induced cardiomyocyte hypertrophy is not known. We employed an in vitro mechanical stretch model of cultured cardiomyocytes and used rat B-type natriuretic peptide promoter as stretch-sensitive reporter gene. Stretch transiently increased GATA-4 DNA binding activity and transcript levels, which was followed by increases in the expression of B-type natriuretic peptide as well as atrial natriuretic peptide and skeletal α-actin genes. The stretch inducibility mapped primarily to the proximal 520 bp of the B-type natriuretic peptide promoter. Mutational studies showed that the tandem GATA consensus sites of the proximal promoter in combination with an Nkx-2.5 binding element are critical for stretch-activated B-type natriuretic peptide transcription. Inhibition of GATA-4 protein production by adenovirus-mediated transfer of GATA-4 antisense cDNA blocked stretch-induced increases in B-type natriuretic peptide transcript levels and the sarcomere reorganization. The proportion of myocytes with assembled sarcomeres in control adenovirus-infected cultures increased from 14 to 59% in response to stretch, whereas the values for GATA-4 antisense-treated cells were 6 and 13%, respectively. These results show that activation of GATA-4, in cooperation with a factor binding on Nkx-2.5 binding element, is essential for mechanical stretch-induced cardiomyocyte hypertrophy.

Cardiac hypertrophy is the primary adaptive mechanism for terminally differentiated cardiac myocytes to increased hemodynamic load. In addition to mechanical stretch, a number of humoral factors, such as G-protein-coupled receptor agonists angiotensin II and endothelin-1 (ET-1) 1 as well as adrenergic receptor agonists released by the activated symphatic nervous system participate in the adaptive process and modify the growth of cardiac myocytes in vivo (for review, see Refs. 1 and 2). When the cardiac overload is sustained, this initially compensatory mechanism fails, and the contractile function is impaired, leading to congestive heart failure (1).
The hypertrophic response in cardiac myocytes is characterized by morphologic changes that include increase in cell size and protein synthesis and enhanced sarcomere reorganization as well as specific changes in cardiac gene expression (2,3). The early genetic response to hemodynamic overload is the activation of immediate early response genes such as c-fos and c-jun, components of the activator protein-1 (AP-1) transcription factor complex. This is followed by up-regulation of B-type natriuretic peptide (BNP) and reactivation of fetal genes such as atrial natriuretic peptide (ANP), ␤-myosin heavy chain, and skeletal muscle ␣-actin (sk␣A) (2, 4 -6). The mechanisms regulating the genetic reprogramming in cardiac overload has been the subject of intensive research. GATA-4 was originally shown to be a crucial regulator of cardiogenesis and cardiac-specific genes such as ANP and BNP (for review, see Refs. 7 and 8). Several lines of evidence suggest the involvement of GATA-4 in the development of cardiac hypertrophy. First, hemodynamic overload caused by aortic banding, Arg 8 vasopressin infusion, or nephrectomy in vivo (9 -11) as well as in vitro treatment of cultured neonatal rat cardiac myocytes with ␣ 1 -adrenergic agonist phenylephrine, ␤-adrenergic agonist isoprenaline, or ET-1 (10,(12)(13)(14)(15)(16)(17)(18)(19) activate GATA-4 binding on cis-acting elements of target genes. Second, hemodynamic overload in vivo induces transcription via GATA binding elements present in the regulatory region of the angiotensin type 1 A receptor, ␤-myosin heavy chain, and BNP genes (9,11,20). Similarly, hypertrophic agonists phenylephrine and isoprenaline activate GATA-dependent transcription of ET-1, BNP, and ANP in vitro (12,14 -16,21). Third, overexpression of GATA-4 in cultured neonatal cardiomyocytes by adenoviral gene transfer or in the hearts of transgenic mice was shown to be sufficient to induce cardiomyocyte hypertrophy (13,22). Finally, expression of dominant negative GATA-4 or antisense GATA-4 cDNA blocked GATA-4-directed transcriptional responses and features of cardiomyocyte hypertrophy induced by phenylephrine and ET-1 (13,22). Although these observations have implicated GATA-4 as sufficient and, in the case of hypertrophic agonist-induced hypertrophy, a necessary mediator of the hypertrophic response, it is not known whether direct mechanical stretch also activates the hypertrophic program via GATA-4-dependent mechanism(s).
To directly address the role of GATA-4 and its potential cofactors in mechanical stretch-activated hypertrophic program, we employed an in vitro mechanical stretch model of cardiomyocytes. The model provides the advantage to dissect the mechanical component from the humoral and neural components of hemodynamic overload, which are always present in the whole animal. Stretch of cultured neonatal rat cardiomyocytes has been shown to stimulate BNP transcription (23), therefore providing a useful target gene to elucidate the mechanisms of stretch-induced gene expression in cardiac myocytes. Our present data show that GATA-4 is a critical regulator of transcriptional and morphological changes induced by mechanical stretching of cardiomyocytes.  -19) were from Santa Cruz Biotechnology (San Diego, CA), and anti-rabbit IgG antibody linked to horseradish peroxidase was from New England Biolabs Ltd. (Hertfordshire, UK). Fu-GENE™ 6 transfection reagent was from Roche Molecular Biochemicals. ECLϩPlus reagents, poly-(dI-dC)⅐(dI-dC), [␣ 32 P]dCTP, and Rediprime™II random prime labeling system were from Amersham Biosciences. The BNP peptides and antiserum as well as a 390-bp fragment of rat BNP cDNA probe (24) were gifts from Dr. Kazuwa Nakao, Kyoto University School of Medicine (Kyoto, Japan). Rat ANP cDNA probe Car-55 (25) was provided by Dr. Peter L. Davies, Queen's University (Kingston, Canada). Cell culture reagents were from Sigma-Aldrich.
Cell Culture and Gene Transfer-Neonatal rat ventricular myocytes (hereafter referred as cardiomyocytes) were prepared and cultured in conditions as previously described (17,25). Briefly, the ventricular apexes of hearts of 2-4-day-old Sprague-Dawley rats were digested with disaggregation medium (collagenase type 2, 2 g/liter and 50 M CaCl 2 in phosphate-buffered saline) at 37°C. The cells were filtered, washed twice, and suspended in a 1:1 mixture of DMEM/F-12 supplemented with 10% fetal bovine serum (DMEM/F-12-FBS). The cells were pre-plated into 100-mm culture dishes (BD Falcon™, BD Biosciences) for 45 min at 37°C in humidified air with 5% CO 2 to reduce the number of contaminating non-muscle cells. Cells not attached to the pre-plated dishes were plated on Bioflex collagen 6-well plates (Flexcell, McKeesport, PA) or 24-well culture plates (BD Biosciences Falcon™) at a density of 1800 cells/mm 2 . The next day, when designed, myocytes were transfected with FuGENE™ 6 in fresh DMEM/F-12-FBS for 6 -8 h. The total amounts of FuGENE™ 6 reagent and DNA were 6 and 3 g/1.8 ϫ 10 6 cells, respectively, unless otherwise indicated. To control the transfection efficiency, luciferase reporter plasmids were cotransfected with Rous sarcoma virus-␤-galactosidase plasmids at the ratio of 2:1. The DMEM/F-12-FBS was replaced with complete serum-free medium 24 h after plating. The composition of complete serum-free medium was similar to that described earlier (25), except that the concentration of triiodothyronine was 0.1 nM. In adenovirus experiments, infections were performed to myocytes 4 h after switching to complete serum-free medium at a multiplicity of infection of 4 by adding the appropriate recombinant adenovirus to the culture media overnight. The media were replaced every 24 h. After experiments, the cells were washed twice with phosphate-buffered saline and quickly frozen at Ϫ70°C. The reporter gene activities were measured by using the luciferase assay system (Promega) and the luminescent ␤-galactosidase detection kit II (BD Biosciences Clontech, Palo Alto, CA) with Luminoskan RS luminometer (ThermoLabsystems, Vantaa, Finland).
COS-1 cells were maintained in Dulbecco's modified Eagle's medium containing 10% FBS. Cells were plated onto 100-mm culture plates and transfected with 1 g of pEF-FLAG-Nkx2.5 or pMT2 control plasmid with FuGENE™ 6 reagent. Forty-eight hours after transfection, cells were harvested and subjected to nuclear protein extraction.
Application of Mechanical Stretch-Stretch was introduced to attached myocytes after 18 -24 h in complete serum-free medium by applying a computer-controlled vacuum suction under the flexible-bottomed Bioflex collagen plates with Flexercell Strain Unit FX-3000 (Flexcell). The vacuum varied in two-second cycles at a level sufficient to promote 10 -25% elongation of the cardiomyocytes at the point of maximal distension of the culture surface.
Immunofluorescence-The cells were washed twice with Hanks' balanced salt solution, fixed, and permeabilized with 4% paraformaldehyde and 0.2% Triton X-100 in phosphate-buffered saline for 15 min at room temperature and washed 3 times with phosphate-buffered saline. Filamentous actin was labeled with Alexa Fluor 488 phalloidin (1:20, Molecular Probes Inc, Eugene, OR). To score for reorganized cells, random fields (minimum of 10 per well) were taken by laser confocal microscope (LSM 510, Zeiss) using green fluorescence channel. The number of cardiomyocytes harboring striated actomyosin fibers extending from one extremity of the cell to the other was counted and divided by the total number of cardiomyocytes in the same field. These observations were performed blindly.
Western Blot-Protein extracts were boiled in Laemmli buffer, resolved by SDS-PAGE, and transferred to Optitran BA-S 85 nitrocellulose membranes (Schleicher & Schuell). The membranes were blocked in 5% nonfat milk and then incubated with GATA-4 antibody (H-112) at a 1:5000 dilution in 1% milk in Tris-buffered saline-Tween 20 overnight at 4°C. Antibody binding was detected with an anti-rabbit horseradish peroxidase-IgG at a 1:2000 dilution and revealed using ECLϩPlus as described by the manufacturer.
Isolation and Analysis of RNA-Total RNA was isolated using the guanidine thiocyanate-CsCl method (30). For the RNA Northern blot analyses, 0.5-6-g samples were separated by electrophoresis and transferred to nylon membrane (Osmonics, Westborough, MA). The cDNA probes complementary to rat ANP, BNP, sk␣A, and GATA-4 mRNA or ribosomal 18 S RNA were random prime-labeled with Re-diprime™ II (Amersham Biosciences). Probes for sk␣A and GATA-4 were made by reverse transcription-PCR technique and confirmed by sequencing to correspond bases 2950 -3184 of rat sk␣A and 1196 -2613 of rat GATA-4 (GenBank TM accession numbers v01218 and l22761, respectively). The membranes were hybridized and washed 3 ϫ 20 min at ϩ62°C as previously described (25). Thereafter, the membranes were exposed with PhosphorImager screens (Amersham Biosciences), which were scanned with Molecular Imager FX Pro Plus and quantitated using Quantity One software (Bio-Rad). Northern blot analysis using these cDNA probes revealed one single band of appropriate size. The hybridization signals of specific mRNAs were normalized to that of 18 S RNA in each sample.
Protein Extraction and DNA Gel Mobility Shift Assay (EMSA)-Nuclear extracts were prepared from cells as described by Schreiber et FIG. 2. The proximal ؊520 bp are sufficient to confer stretch inducibility of the rat BNP promoter. Luciferase (LUC) reporter constructs containing various lengths of the rBNP 5Ј-flanking region were co-transfected with Rous sarcoma virus-␤-galactosidase plasmid (2 and 1 g/35 mm well, respectively) into neonatal rat ventricular myocytes that were cyclically stretched for 24 h. The amplitude of stretch varied between 10 and 25% in 2-s cycles. Schematic presentation of the constructs is shown on the left. Deletion of the 5Ј-flanking region between Ϫ520 and Ϫ114 significantly decreased the inducibility of the rBNP promoter in response to stretch. This proximal region contains two GATA binding elements (GG) located at Ϫ90, consensus AP-1 binding element located at Ϫ373, and two NKE-like elements (NKE) located at Ϫ387 and Ϫ496. Open bars indicate non-stretched control, and solid bars indicate stretched cells. Basal promoter activities are shown in the middle relative to Ϫ60rBNP construct. Results are expressed as luciferase to ␤-galactosidase ratio Ϯ S.E. of stretched cells relative to that of non-stretched control cells transfected with the same construct (n ϭ 10 -18 from 3-5 independent cultures). *, p Ͻ 0.05; ***, p Ͻ 0.001 versus control (Student's t test).
al. (31). Cells were washed and scraped with Tris-buffered saline. After centrifugation, cells were resuspended in 400 l of ice-cold buffer A (10 mM HEPES, pH 7.9, 10 mM KCl, 0.1 mM EDTA, 0.1 mM EGTA, supplemented with 0.2 mM Na 3 VO 4 , 20 g/ml leupeptin, 2 g/ml pepstatin, 20 g/ml aprotinin, 0.5 mM phenylmethylsulfonyl fluoride, 50 mM NaF, 1 mM dithiothreitol, 6 g/ml L-1-tosylamido-2-phenylethyl chloromethyl ketone, and 6 g/ml 1-chloro-3-tosylamido-7-amino-2-heptanone) by gentle pipetting and allowed to swell on ice for 15 min. Cells were lysed by adding 25 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 25 l of ice-cold buffer C (20 mM HEPES, pH 7.9, 400 mM NaCl, 1 mM EDTA, and 1 mM EGTA with supplements similar to those in buffer A) and rocked for 15 min. The samples were centrifuged at 12,500 rpm for 5 min. The supernatant was saved as the nuclear fraction. The entire procedure was carried out at 4°C. Probes were prepared by Klenow fragmentmediated filling of the sticky ends of double-stranded (ds) oligonucleotides and labeled with [␣ 32 P]dCTP. Binding reactions contained 3-6 g of nuclear protein and 2 g of poly-(dI-dC)⅐(dI-dC) in a buffer containing 16 mM HEPES, pH 7.9, 120 mM NaCl, 1 mM MgCl 2 , 40 mM KCl, 1 mM dithiothreitol, 0.7 mM EDTA, 0.3 mM EGTA, 8% glycerol, 0.02% Nonidet P-40, 0.25 mM phenylmethylsulfonyl fluoride, and 1 g/ml of each aprotinin, leupeptin, pepstatin, and when appropriate, various molar excesses of competitor DNAs. When designed, supershift experiments were performed by preincubating nuclear extract with 1 g of appropriate antibody for 20 min at room temperature before performing the binding reaction. After a 10-min preincubation of extract, the labeled probe was added, and binding was allowed to proceed at room temperature for 20 min. The reaction mixes were then analyzed by electrophoresis on 5% polyacrylamide gel in 0.5ϫ Tris-borate-EDTA buffer at 4°C. After drying, the gel DNA-protein complexes were detected and quantitated as the Northern blot membranes.
Assay of Immunoreactive ANP and BNP-Radioimmunoassays of ANP and BNP were done as previously described (5,32). The sensitivities of the BNP and ANP assays were 2 and 1 fmol/tube, respectively. 50% displacements of the respective standard curve occurred at 16 and 25 fmol/tube. The intra-assay and inter-assay variations were 10 and 15%, respectively.
Statistical Analysis-The results are expressed as the mean Ϯ S.E. For multiple comparisons, data were analyzed with one-way analysis of variance followed by a least significant difference post hoc test. For comparison of two groups at each time point Student's t test for unpaired data was used. A value of p Ͻ 0.05 was considered statistically significant.

Mechanical Stretch Induces a Hypertrophic
Phenotype-To validate the experimental system we first analyzed the effect of mechanical stretch on the gene expression of BNP, ANP, and sk␣A, which represents genetic hallmarks of the stretch-induced cardiac hypertrophic program (23,(33)(34)(35). Cyclic mechanical stretch elevated BNP mRNA levels significantly after 4 h, peaking at 24 h (2.1-fold, Fig. 1A). Similarly, mechanical stretch caused significant increases in ANP and sk␣A mRNA levels, respectively (Fig. 1, B and C). Elevated natriuretic peptide mRNA levels were accompanied by increased peptide secretion from cardiomyocytes into the culture medium (Fig. 1, D  and E). Conclusively, because of rapid response at the level of mRNA and the peptide released, BNP provides a sensitive target gene to further study the stretch-inducible signaling mechanisms.
Proximal 520 bp of rBNP 5Ј-Flanking Sequences Is Sufficient to Confer Stretch Inducibility-To identify the rat BNP promoter region that mediates the transcriptional activation by mechanical stretch, we transfected cardiomyocytes with luciferase reporter constructs driven by various lengths of the rBNP 5Ј-flanking sequence. Removal of about 1.2 kilobases of the Ϫ2200 rBNP 5Ј-flanking sequence down to Ϫ940 resulted in about a 30% decrease in basal reporter activity (Fig. 2). Further deletion down to Ϫ520 increased basal promoter activity, in agreement with the suggested presence of repressor elements distally from Ϫ520 (21). Interestingly, deletion of the rBNP 5Ј-flanking region from Ϫ2200 to Ϫ520 did not significantly change the inducibility of the rBNP promoter in response to mechanical stretch. Yet truncation down to Ϫ114 reduced the basal as well as mechanical stretch-induced promoter activity by 70 -80% compared with Ϫ520rBNP. Further deletion to Ϫ60 completely abolished the stretch response. The Ϫ60 rBNP possessed only 3% of original basal promoter activity.
GATA-4 and AP-1 Are Activated by Mechanical Stretch-The tandem GATA elements within the Ϫ520-bp BNP promoter are well conserved between mammalian species (rat, human, and dog, Ref. 26). Moreover, these elements have been shown to mediate the activation of rBNP promoter in response to hemodynamic volume overload in vivo (11), suggesting an important role in stretch-activated BNP transcription. Therefore, we used 30-bp ds oligonucleotide probe containing the Ϫ90 GATA sites of rBNP (rBNP-90 GATA) as the probe in EMSA to analyze whether mechanical stretch activates nuclear protein binding on these sites. As shown in Fig. 3A, stretch transiently increased GATA DNA binding activity peaking at 1 h. Octamer-1 binding activity remained unaffected by stretch excluding unspecific effect on nuclear protein-DNA interaction (Fig. 3B). To determine the specificity of GATA binding activity, competition analyses were performed (Fig. 3D). The formation of complexes with the rBNP-90 GATA probe was dose-dependently inhibited by the unlabeled self DNA, but not by the mutated BNP GATA site, confirming the specific protein-DNA interaction. Supershift analysis clearly showed complete antibody-induced supershift of the GATA-4 but not GATA-5 or -6 complexes (Fig. 3E).
To study the effect of mechanical stretch on GATA-4 gene expression, Northern blot analysis using rat GATA-4 cDNA probe was performed. The activation of GATA-4 binding was accompanied by a transient 1.6-fold increase in GATA-4 mRNA levels peaking at 4 h (Fig. 3F). Western blot analysis showed that the levels of GATA-4 protein were almost undetectable in the cytosolic fraction. In the nuclear extracts of stretched cardiomyocytes GATA-4 protein accumulation appeared to increase; however, this change was not statistically significant (Fig. 3G).
Because previous studies suggest functional cooperation between GATA-4 and AP-1 in pressure overload-induced hypertrophy (20), we also studied nuclear protein binding on the AP-1 site by using ds oligonucleotide probe containing the rat BNP-373 AP-1 site (rBNP-373 AP-1). Stretch increased AP-1 binding activity, peaking at 1 h (2.5-fold increase) and sustaining for at least 24 h (Fig. 4A). Competition experiments demonstrated specificity of the complex formation, which was effectively inhibited by the unlabeled rBNP-373 AP-1 probe (self) (Fig. 4B), but not by the mutated BNP AP-1 site DNA. To identify the components of the AP-1 complex we performed supershift experiments using antibodies against different Fos and Jun family members. As shown in Fig. 4C, moderate supershifts were observed with JunD-specific antibody and unspecific antibody against Fos family proteins. Instead, when using c-Fos-specific antibody, no supershift complex was seen, suggesting the presence of either Fos B, Fos-related antigen-1 or -2 but not c-Fos in the complex. Weak antibody-induced supershifts were observed also with c-Jun-and JunB-specific antibodies. Mechanical stretch did not change the supershift profile compared with nuclear extracts obtained of nonstretched cardiomyocytes (Fig. 4C).
Stretch-induced BNP Promoter Activity Is Inhibited by Mutation of Proximal GATA Sites-Next we studied whether the activation of GATA-4 had functional consequences on BNP transcription. We introduced site-directed mutations to the reporter construct containing the proximal 534-bp rBNP promoter that is sufficient to confer full stretch inducibility (Ϫ534rBNP, Fig. 5A). Mutation of the tandem Ϫ90 GATA sites resulted in 37% decrease in the stretch-induced reporter activity compared with intact Ϫ534rBNP promoter, which was activated by 2.2-fold (Fig. 5B). To study the possibility that stretch-induced rBNP transcription could be mediated by cooperation of GATA-4 with AP-1 as reported previously in pressure overload-activated angiotensin type 1 A receptor transcription in vivo (20), we mutated the Ϫ373 AP-1 site of rBNP. Despite increased complex formation of cardiac nuclear proteins with rBNP-373 AP-1 (Fig. 4A), mutation of this site alone or in combination with the tandem GATA sites did not modify stretch-induced transcription of rBNP (Fig. 5B).
Ϫ387 NKE-like Element Cooperates with GATA-4 in Stretchinduced BNP Transcription-In addition to AP-1, GATA-4 has been shown to cooperate with several other transcription factors including Nkx-2.5 in regulation of cardiac-specific gene expression (36 -38). Therefore, we utilized a computer-based search for cis-acting elements in a Ϫ534-bp rBNP promoter and identified two sequences with high homology to NKE located at Ϫ387 and Ϫ496. To confirm whether these NKE-like elements are bona fide binding sites for Nkx-2.5, we prepared nuclear extracts from COS-1 cells transfected with Nkx-2.5 expression vector. The nuclear extracts were incubated with labeled ds DNA probes corresponding to Ϫ387 and Ϫ496NKE-like sites of BNP and a previously described high affinity duplicate NKE from the rat ANP gene (ANP NKE) (37,38). As shown in Fig. 6 (see Table I), gel shift analysis demonstrated that the Ϫ387 NKE-like and the ANP NKE probes formed complexes that were specifically supershifted by an anti-Nkx-2.5 antibody; the affinity of the Ϫ387 NKE-like element for Nkx-2.5 was significantly weaker than that of the ANP NKE. Under the same conditions, the Ϫ496NKE-like site of BNP was not able to bind Nkx-2.5, in agreement with our recent study demonstrating that this site specifically binds the ETS transcription factor Elk-1 (39).
Given that the Ϫ387 NKE-like element proved to be the true binding target for Nkx-2.5 (hereafter referred as Ϫ387 NKE), we studied whether it cooperates with GATA element in stretch-induced transcription of BNP in cardiomyocytes. As Fig. 7 shows, mutation of Ϫ387 NKE alone had no effect on stretch-induced rBNP promoter activity. Interestingly, the inducibility by stretch was almost completely abolished when mutation of the Ϫ387 NKE was combined with the mutation of the tandem Ϫ90 GATA sites. Therefore, the requirement of Ϫ387 NKE in the stretch-induced BNP transcription was unmasked when the tandem Ϫ90 GATA sites were mutated, suggesting that Nkx-2.5 acts in cooperation with GATA-4. In support of this notion, mutation of Ϫ496NKE-like site, which is unable to bind Nkx-2.5 although it shares similar core sequence to Ϫ387 NKE, did not modify stretch-induced transcription of Ϫ534rBNP with intact or mutated GATA sites (Fig. 7). into neonatal rat ventricular myocytes that were cyclically stretched for 24 h. The mutation of Ϫ90 GATA sites (mGATA) inhibited stretchinduced rBNP transcription, whereas mutation of Ϫ373 AP-1 site did not modify stretch responsiveness of BNP promoter with either intact (mAP-1) or mutated GATA-sites (mGATA/AP-1). Results are expressed as the luciferase to ␤-galactosidase ratio of stretched cells relative to that of non-stretched control cells transfected with the same construct (mean Ϯ S.E., n ϭ 12-18 from four to five independent cultures). p Ͻ 0.01 (**) and p Ͻ 0.001 (***) are versus control cells transfected with the same construct; p Ͻ 0.01 (#) is versus stretched cells transfected with Ϫ534rBNP (one way analysis of variance followed by least significant difference post hoc test).

GATA-4 Is Essential for Stretch-induced
Sarcomere Reorganization-In addition to changes in gene expression, the hypertrophic phenotype of cultured cardiomyocytes is characterized by morphologic changes including increased reorganization of sarcomeres (3). To test whether the activation of GATA-4 is involved in the morphologic changes associated with stretchinduced hypertrophy, we specifically inhibited the production of GATA-4 protein by employing an adenovirus expressing antisense GATA-4 cDNA (GATA-4as). In agreement with previous studies (22,28), GATA-4 protein levels decreased significantly in cells infected with GATA-4as compared with cells infected with control adenovirus expressing a nuclear localization signal-lacZ cDNA (LacZ) (Fig. 8A). Down-regulation of GATA-4 in cardiomyocytes blocked a stretch-induced increase in BNP mRNA levels (Fig. 8B) and inhibited sarcomere reorganization (Fig. 8C). Quantification of reorganized cardiomyocytes showed that stretch increased the number of cardiomyocytes that underwent sarcomere reorganization in LacZ-infected cultures by 4.1-fold (p Ͻ 0.01) and that this effect was significantly smaller (2.0-fold) in GATA-4as-treated cells (Fig. 8D). GATA-4as treatment also decreased the basal sarcomere assembly in non-stretched cells compared with cells infected with LacZ. Collectively, the data demonstrate an essential role for GATA-4 in the genetic and morphologic response of cardiomyocytes to mechanical stretch. DISCUSSION Development of the in vitro stretch model for cultured neonatal rat cardiocytes has provided an established method for the study of the molecular mechanisms in stretch-activated changes in cardiomyocyte hypertrophy. These changes closely resemble those of cardiac overload-induced hypertrophy in vivo (for review, see Ref. 35). Experiments employing this model have led to increasing knowledge of cytosolic signal transduction pathways mediating the cellular responses to stretch. In the present study, we show for the first time that mechanical stretch per se activates GATA-4 binding on the well conserved proximal GATA element of BNP promoter. This element to-FIG. 6. Nkx-2.5 binds to ؊387NKE. COS-1 cells were transfected with expression plasmid coding Nkx-2.5 or empty pMT2 control plasmid. Forty-eight hours after transfection cells were harvested, and the nuclear proteins were extracted. Binding reactions containing 3 g of extracts were incubated with labeled ds DNA probes corresponding to intact or mutated (mut) Ϫ496 or Ϫ387NKE-like sites of rBNP or a previously described high affinity Nkx-2.5 binding element of rat ANP gene (ANP NKE) before EMSA. The probe sequences are described in Table I. A, intact Ϫ387NKE-like and ANP NKE probes form complexes with overexpressed Nkx-2.5 (indicated with asterisks). Note that there is no difference in binding on the Ϫ496NKE-like site of rBNP or any of the mutated probes between extracts with or without overexpressed Nkx-2.5. B, supershift reactions were performed by incubation of reaction mixtures with 1 g of anti-Nkx-2.5 antibody (ab). Position of supershifts are indicated (SS). NS indicates nonspecific band that was inconstantly formed (see A). gether with Ϫ387 NKE mediates the transcriptional activation of the gene. We also provide evidence that GATA-4 mediates mechanical stretch-activated enhanced sarcomere reorganization, a major hallmark of cardiac myocyte hypertrophy in vitro (3).
Little has been known about nuclear signaling mechanisms that activate the genetic reprogramming in response to stretch. Recently, it was reported that activation of human BNP promoter by mechanical stretch was mediated partly via nuclear factor-B binding on the shear stress response elements present in the human BNP promoter (40). Although the result is not mutually exclusive with our results, it should be pointed out that shear stress response elements are not conserved in the rat BNP promoter, whereas the GATA element is conserved across species (26). The activity of GATA-4 is controlled by specific interactions with several other transcription factors that cooperate with GATA-4 in transcriptional regulation of cardiac-specific gene expression. These factors include Nkx-2.5 (36 -38), nuclear factor of activated T-cells 3 (NF-AT3) (41), GATA-6 (28), friend of GATA 2 (42), serum response factor (43), myocyte enhancer factor 2 family members (44), yin yang 1 (45), p300 (46,47), and dHAND (47). Two of these interactions, GATA-4/NF-AT3 and GATA-4/serum response factor, have been associated with genetic responses to some forms of cardiac hypertrophy (19,41,48), but the exact role of other GATA-4 cofactors in GATA-4-dependent hypertrophic responses has not been explored.
Interestingly, in the present study inhibition of GATA-4 production alone by GATA-4as adenovirus was sufficient to completely block the stretch-induced increase in BNP mRNA levels. On the contrary, the mutation of the Ϫ90 GATA sites in the context of Ϫ534rBNP resulted in about a 40% decrease in the stretch response (Fig. 5B), whereas the stretch response of Ϫ114 rBNP, still containing intact GATA sites, was minimal (Fig. 2). Because intact Ϫ534rBNP was sufficient to confer full stretch inducibility, these data suggest that other factor(s), which bind between Ϫ534 bp and Ϫ114 bp of rBNP promoter, are involved together with GATA in stretch-activated transcription. Also, the apparently conflicting results regarding necessity/sufficiency of GATA-4 in stretch response between transfection experiments and GATA-4as adenovirus experiments may be contributed to the interaction of GATA-4 with other cofactors. Down-regulation of GATA-4 protein may limit the complex formation between GATA-4 and its cofactors, whereas inhibition of GATA-4 DNA binding does not necessarily have an effect on the interaction with other factors that may recruit the complex on the promoter.
Xia et al. (48) report that GATA, NF-AT and myocyte enhancer factor elements were required for activation of adenylosuccinate synthetase 1 gene in electrical pacing-induced hypertrophy of cardiomyocytes (48). The previously described NF-AT binding element responsible for GATA-4/NF-AT3 synergy in activation of human BNP is located far distally (Ϫ927) in the promoter (41) and does not appear to be required for stretch response. Moreover, we were not able to detect specific binding activity on NF-AT consensus-like sites at Ϫ320 and Ϫ340 of rBNP 5Ј-flanking sequence (data not shown). In addition, 534 bp of 5Ј-flanking region of rat BNP gene contains no potential myocyte enhancer factor nor serum response factor binding elements, and the yin yang 1 element locates more proximally (between Ϫ80 to Ϫ60) of the rBNP 5Ј-flanking sequence (45), suggesting that these factors do not cooperate with GATA-4 in stretch-activated BNP transcription. The inducible expression of angiotensin type 1 A receptor by pressure overload in vivo has been reported to be regulated by possible functional cooperation of GATA and AP-1 in adult rat heart (20). However, the inability of mutation of Ϫ373 AP-1 to modify stretch-induced reporter activity of Ϫ534rBNP with mutated tandem GATA sites indicates that AP-1 does not cooperate with GATA-4 or that the possible cooperation does not require AP-1 binding on the Ϫ373 AP-1 site in mechanical stretch-induced activation of rBNP.
A major finding of this study was that mutation of the tandem Ϫ90 GATA sites in combination with mutation of the NKE at Ϫ387 almost completely abolished the stretch inducibility of rBNP promoter, whereas mutation of Ϫ387 NKE alone had no effect on it. This suggests that GATA-4/Nkx-2.5 interaction is targeted by stretch-induced signaling. Furthermore, GATA-4 may recruit Nkx-2.5 on rBNP promoter via intact Ϫ90 GATA sites when Nkx-2.5 binding on its cognate binding site is disrupted by mutation, thereby enabling normal stretch response. Therefore, the involvement of Nkx-2.5 in stretch-induced rBNP transcription becomes evident only when Ϫ387 NKE is mutated and its recruitment on rBNP promoter is blocked by mutation of the tandem Ϫ90 GATA sites, resulting in further decrease in stretch inducibility compared with mutation in GATA sites only. Previously, the GATA-4/Nkx-2.5 interaction has been shown to be targeted by bone morphogenetic protein-2 and -4 signaling during the early stages of cardiogenesis (49,50). Our results extend the relevance of the GATA-4/Nkx-2.5 interaction to postnatal cardiomyocytes and implicate it in the response of cardiomyocytes to mechanical stretch. FIG. 7. Functional ؊387NKE cooperates with ؊90 GATA sites in mechanical stretch-induced rBNP transcription. Luciferase reporter constructs were cotransfected with Rous sarcoma virus-␤-galactosidase plasmid (2 and 1 g/35-mm well, respectively) into neonatal rat ventricular myocytes that were cyclically stretched for 24 h. Combining mutation of Ϫ387 NKE with mutation of Ϫ90 GATA (mGATA/NKE387) almost completely abolished the stretch responsiveness of the BNP promoter, whereas the inhibitory effect of mutation of Ϫ90 GATA sites (mGATA) on stretch inducibility was not further potentiated by combination with mutation of Ϫ496 NKE (mGATA/ NKE496). Mutation of either Ϫ387 NKE (mNKE387) or Ϫ496 NKE-like (mNKE496) elements alone had no effect on stretch-induced BNP transcription. Results are expressed as luciferase to ␤-galactosidase ratio relative to induction of intact Ϫ534 rBNP promoter (Ϫ534rBNP) (mean Ϯ S.E. n ϭ 6 -14 from 3-5 independent cultures); p Ͻ 0.05 (*), p Ͻ 0.01 (**), p Ͻ 0.001 (***) are versus stretched cells transfected with Ϫ534rBNP; p Ͻ 0.05 (#) is versus stretched cells transfected with the Ϫ534rBNP-containing mutation in Ϫ90 GATA sites (one way analysis of variance followed by least significant difference post hoc test).
In the previous studies hypertrophic response has been associated with a transient increase in GATA-4 DNA binding activity without significant change in total cellular GATA-4 protein levels, suggesting that GATA-4 activity is mainly regulated by posttranslational mechanisms (12,13,15,20). Indeed, there is evidence that GATA-4-mediated changes in gene expression and cell morphology are controlled not only by interaction with other cofactors but also by phosphorylation of GATA-4 by specific protein kinases (16,22,51). ET-1 and phenylephrine activate small GTPase family member RhoA previously shown to activate sarcomere reorganization in cardiac myocytes (52). RhoA in turn potentiates GATA-4 via p38 mitogen-activated protein kinase-dependent phosphorylation on Ser-105, leading to genetic reprogramming and induction of sarcomere reorganization (18,22). In addition to p38 mitogenactivated protein kinase, extracellular signal-regulated kinases have been shown to phosphorylate GATA-4 followed by both enhanced binding and transactivation activity of GATA-4 (12, 51). Interestingly, mechanical stretch has been shown to activate both of these mitogen-activated protein kinase pathways (35,40,53). Moreover, Rho family members have been shown to play a critical role in mechanical stretch-induced activation of extracellular signal-regulated kinase and p38 mitogen-activated protein kinase in cultured cardiomyocytes (54,55), thereby providing a potential link between mechanical stretch and the GATA-4-dependent enhanced sarcomere assembly as well as BNP transcription observed in the present study.
Recently, Morisco et al. (16) reported that stimulation of cardiac myocytes with ␤-adrenergic agonists increased nuclear GATA-4 protein levels (16). The data suggest that activation of protein kinase B/Akt by isoprenaline led to inhibition of glycogen synthase kinase 3␤, which phosphorylates the amino-terminal part of GATA-4, resulting to increased export from the nucleus. Previously, Akt has been shown to be activated by mechanical stresses in the ventricles of rats subjected to aortocaval shunt (56). Therefore, the possibility that increased GATA-4 protein accumulation in the nuclear extracts of stretched cells may partially result from decreased export via activation of the Akt-glycogen synthase kinase-3␤ pathway remains to be studied.
Previous studies have shown that autocrine/paracrine fac- Cardiomyocytes were infected at a 4 multiplicity of infection with adenoviruses expressing a nuclear localization signal-lacZ gene for control or antisense regions directed specifically toward GATA-4 (GATA-4as) as described under "Experimental Procedures." The cells were subjected to mechanical stretch for 48 h. A, Western blot analysis confirming that the GATA-4as specifically decreased nuclear GATA-4 protein levels. 10 g of nuclear proteins were boiled in Laemmli buffer, resolved by SDS-PAGE, and blotted on nitrocellulose membranes. The blots were incubated with GATA-4 antibody (H-112) at a 1:5000 dilution. Antibody was detected with an anti-rabbit horseradish peroxidase-IgG at a 1:2000 dilution and revealed by chemiluminescence. B, after extraction, 4 g of total RNA was separated on a Northern gel. The blots were sequentially hybridized with specific cDNA probes complementary to rat BNP and ribosomal 18 S RNA. The results are expressed as the ratio of specific mRNA to ribosomal 18 S RNA as determined by Northern blot analysis. Open bars indicate unstretched control (n ϭ 6), and solid bars indicate stretched cells (n ϭ 6). Results are the mean Ϯ S.E. from two independent cultures. *, p Ͻ 0.001 versus LacZ control; # , p Ͻ 0.001 versus stretched GATA-4as cells (one way analysis of variance followed by least significant difference post hoc test). For immunofluorescence microscopy cardiomyocytes were fixed, and actin filaments were stained using Alexa Fluor 488 phalloidin. C, representative images were taken with laser confocal microscope (LSM 510). D, quantification of the percentage of reorganized cardiomyocytes. Open bars indicate unstretched control, and solid bars indicate stretched cells mean Ϯ S.E. *, p Ͻ 0.05 versus LacZ control (Student's t test). tors play a significant role in the development of load-induced cardiac hypertrophy. Mechanical stretch is coupled with the cellular release of angiotensin II and ET-1, which act as chemical mediators of stretch-induced myocyte hypertrophy in cultured rat cardiomyocytes (35,57). Liang and Gardner (58) propose that stretch-triggered release of angiotensin II sequentially stimulates secretion of ET-1, which accounts for 50% of the stretch-induced human BNP gene transcription (58). In addition, ET-1 stimulates GATA-4 binding activity (10,(17)(18)(19), thus raising the question of whether mutation of the tandem Ϫ90 GATA sites decreased the activation of rBNP promoter through endothelin-dependent component of stretch. Recently, we have shown that mutation of the tandem GATA sites at Ϫ90 of the rBNP promoter does not inhibit transcriptional activation by either ET-1 or angiotensin II (17). Furthermore, mutation of the Ϫ496 NKE-like site that is a functional target for Elk-1 was shown to specifically inhibit ET-1-induced BNP transcription (39), whereas it had no effect on stretch-induced promoter activity in the present study. Taken together these observations suggest that the role of GATA-4 in the stretchinduced response is ET-1-independent.
In summary, our results indicate that myocyte stretch, a major component of hemodynamic overload, activates a hypertrophic response in cultured neonatal cardiac myocytes, as demonstrated by enhanced sarcomere organization and activation of BNP transcription. Both responses occur via GATA-4dependent mechanisms, indicating that GATA-4 is a nuclear mediator of mechanical stretch-activated hypertrophic program. Moreover, our results suggests that the GATA-4/Nkx-2.5 interaction is targeted by stretch-induced signaling, therefore extending the requirement for GATA-4/Nkx-2.5 interaction to postnatal cardiomyocytes and implicating its role in the response of cardiomyocytes to external stimuli.