Mitogen-activated protein kinases p38 and ERK 1/2 mediate the wall stress-induced activation of GATA-4 binding in adult heart.

The zinc finger transcription factor GATA-4 has been implicated as a critical regulator of inducible cardiac gene expression and as a potential mediator of the hypertrophic program. However, the precise intracellular mechanisms that regulate the DNA-binding activity of GATA-4 are not fully understood. The aim of the present study was to examine the role of mitogen-activated protein kinases (p38 kinase, extracellular signal-regulated protein kinase, and c-Jun N-terminal protein kinase) in the left ventricular wall stress-induced activation of GATA-4 DNA binding in adult heart. Isolated perfused rat hearts were subjected to increased left ventricular wall stress by inflating a balloon in the ventricle. Gel mobility shift assays were used to analyze the transacting factors that interact with the GATA motifs of the B-type natriuretic peptide promoter. The left ventricular wall stress rapidly activated GATA-4 DNA binding and significantly increased the levels of phosphorylated p38 kinase, extracellular signal-regulated protein kinase, and c-Jun N-terminal protein kinase. The wall stress-induced increase in the DNA-binding activity of GATA-4 was abolished both in the presence of the p38 inhibitor SB239063 and MEK1/2 inhibitor U0126. In contrast, the inhibition of c-Jun N-terminal protein kinase by CEP11004 had no effect on the baseline or stretch-induced GATA-4 DNA binding. Moreover, GATA-4 DNA binding was up-regulated by mechanical stretch in the isolated rat atria via p38 and extracellular signal-regulated protein kinase. In conclusion, the present study demonstrates that both p38 and extracellular signal-regulated protein kinase are required for the stretch-induced GATA-4 binding in intact heart.

Myocardial hypertrophy is an adaptive process that develops as a response to hemodynamic overload and a number of different neurohumoral factors. Initially, hypertrophy is considered as a beneficial compensatory mechanism to increased workload, but when prolonged, it ultimately leads to deterio-ration of cardiac function and chronic heart failure (1). At the cellular level, the hypertrophic response is characterized by increase of the size of the myocytes, accumulation of cellular proteins and increased sarcomeric assembly of myofibrils (2,3). At the genetic level, the exposure to hypertrophic stimuli leads to the reprogramming of gene expression, i.e. to the up-regulation of immediate early genes (such as c-jun and c-myc), B-type (BNP) 1 and A-type (ANP) natriuretic peptide genes and genes encoding the structural proteins, such as skeletal ␣-actin and ␤-myosin heavy chain (2,3).
The nuclear signaling cascades that link the hypertrophic stimuli into the changes in cardiac gene expression and hypertrophic growth are not fully understood. GATA-4 is a member of the GATA family of zinc-finger transcription proteins and a critical regulator of cardiac cell growth, differentiation, and survival during the embryogenesis (4 -7). There is also considerable evidence of the involvement of GATA-4 in the hypertrophic signaling and gene expression in the heart (8,9). Indeed, a recent study indicates that overexpression of GATA-4 in transgenic mice is alone sufficient to induce cardiac hypertrophy (9). Furthermore, GATA-4 has been suggested to be a key regulator of the inducible cardiac gene expression (8 -10). It has been shown that GATA-4 regulates the transcriptional activation of BNP, endothelin-1 (ET-1), ␤-myosin heavy chain, and angiotensin II type 1 (AT 1 ) receptor genes during pressure overload and hypertrophic process (10 -14). Moreover, we have recently shown that arginine-vasopressin-induced pressure overload in vivo and direct left ventricular wall stress in vitro result in an increased DNA-binding activity of GATA-4 (15,16). However, there is no data available regarding the signaling mechanisms regulating the activation of GATA-4 DNA binding in the adult heart.
The branches of the mitogen-activated protein kinases (MAPKs), the extracellular signal-regulated protein kinase (ERK) pathway, the p38 kinase pathway and the c-Jun Nterminal protein kinase (JNK) pathway, are known to regulate a wide array of cellular events by phosphorylation in response to extracellular stimuli (17). The aim of the present study was to characterize the role of MAPK pathways in the regulation of left ventricular wall stress-induced GATA-4 binding by using isolated perfused rat hearts. Gel mobility shift assays were used to analyze the transacting factors that interact with the GATA motifs of the BNP promoter. The hearts were subjected to increased wall stress by inflating a balloon in the left ventricle. Moreover, the role of MAPK pathways in the stretchinduced GATA-4 binding was investigated in the isolated rat atria. Our results demonstrate that both p38 and ERK 1/2 play a crucial role in the regulation of wall stress-induced GATA-4 binding in intact heart.
Isolated Perfused Rat Heart Preparation-Male Sprague-Dawley rats (n ϭ 131), weighing 250 -320 g, from the Center for Experimental Animals at the University of Oulu, were used. The experimental design was approved by the Animal Use and Care Committee of the University of Oulu, and it was similar to that described previously in detail (16, 18 -21). Briefly, the rats were decapitated, the aorta was cannulated above the aortic valve, and the hearts were arranged for retrograde perfusion by the Langendorff technique. The hearts were perfused with modified Krebs-Henseleit bicarbonate buffer (pH 7.40) equilibrated with 95% O 2 -5% CO 2 at 37°C. The composition of the buffer was (in millimolar) 113.8 NaCl, 22.0 NaHCO 3 , 4.7 KCl, 1.2 KH 2 PO 4 , 1.1 MgSO 4 , 2.5 CaCl 2 , and 11.0 glucose. The coronary flow rate was set to 15 ml/min, and the heart rate was increased 15-20% above the spontaneous beating frequency. Perfusion pressure, reflecting coronary vascular resistance, was measured by a pressure transducer (Isotec, Hugo Sachs Elektronik, Germany) situated on a side arm of the aortic cannula. A fluid-filled balloon connected to the pressure transducer (Isotec) was inserted through the mitral valve to the left ventricle to measure the ventricular pressure. Initially, the hearts were unloaded by opening the valve of the cannula of the intraventricular balloon against the air so that the balloon was inflated just enough to obtain a pressure signal to monitor preparation stability, and perfused under these conditions for 20 min. After the 20-min period of equilibration the hearts were exposed within 5 min to a peak systolic pressure of between 130 and 150 mmHg by closing the valve and inflating the balloon for 10 min, 30 min, or 2 h. The end-diastolic pressure was set to a level of between 20 and 25 mmHg to maximize the active pressure generation and systolic pressure. In the control groups, the perfusion was continued further under the unloaded conditions. The drugs or the vehicle were added into the aortic cannula as a continuous infusion in the absence and presence of increased wall stress. At the end of the experiments, the left ventricles were immersed in liquid nitrogen and stored in Ϫ70°C until assayed.
Calculation of Wall Stress-Peak systolic circumferential wall stress () was derived from ventricular pressure measurements, intraventricular balloon volume (V B ) and the weight of the left ventricle, as described by Strömer et al. (22). Briefly, a spherical model was assumed in which the radius of the left ventricle (R i ) can be calculated by the cubic formula, R i ϭ 3 ͌V B /[(4/3)⅐]. Total left ventricular volume is the sum of V B and volume of the left ventricular wall (V Wall ϭ left ventricular weight/1.05 (specific gravity of myocardium)). Therefore, Isolated Rat Atria Preparation-In another set of experiments, the isolated rat atrial appendix, prepared as described previously, was used (23). Briefly, the rats were decapitated and their hearts were rapidly removed and placed in oxygenated cool (25°C) buffer solution consist-ing of (in mM) NaCl 137, KCl 5.6, CaCl 2 2.2, HEPES 5.0, MgCl 2 1.2, and glucose 2.5 (pH 7.4), which was also used for superfusion of the atrial preparation. An X-branch polyethylene adapter was inserted into the lumen of the left auricle, and the tissue was placed in a constant temperature (37°C) organ bath. Another tube with smaller diameter was inserted inside the adapter to carry perfusate inflow into the lumen of the auricle. The outflow from the lumen came from one cross-branch of the X-cannula. The stretch of the atrium was produced by changing intra-atrial pressure. Pressure inside the atrium was increased by increasing the height of the outflow tube. The other cross-branch of the X-cannula was connected to a pressure transducer (TCB 100, Millar Instruments, Inc.) so that the pressure in the lumen of the auricle could be recorded. Inflow and outflow (3 ml/min) both to the auricle lumen and to the organ bath with constant temperature were controlled by a peristaltic pump (7553-85, Cole-Parmer Instrument Co.).
Double-stranded synthetic oligonucleotides containing GATA (5Ј-TGTGTCTGATAAATCAGAGATAACCCCACC-3Ј) motifs of the rat BNP promoter were labeled with [␣-32 P]dCTP. Binding reactions consisted of 40 g of protein extract and 2 g of poly(dI-dC)(dI-dC) in a buffer containing 10 mM HEPES, pH 7.9, 1 mM MgCl 2 , 50 mM KCl, 1 mM/DTT, 0.1 mM EDTA, 10% glycerol, 0.025% Nonidet P-40, 0.25 mM PMSF, and 1 M each of aprotinin, leupeptin, and pepstatin, and when appropriate, various molar excesses of competitor DNAs. Supershift experiments were performed by preincubating the nuclear extract with 1 g of appropriate antibody for 20 min at room temperature before the binding reaction. The binding reactions were carried out at room temperature for 20 min. Protein-DNA complexes were separated by electrophoresis on 5% polyacrylamide gel in 0.5 ϫ TBE (Tris borate-EDTA buffer) at 4°C. The gels were dried and exposed with PhosphorImager screens (Molecular Dynamics, Sunnyvale, CA), which were scanned by using a Molecular Imager (Bio-Rad Laboratories). All results were quantitated by using Quantity One software (Bio-Rad).
Kinase Assays-p38 MAPK activity was determined with immunocomplex kinase assay by using Elk-1 as substrate. Cardiac tissue homogenates were clarified at 14,000 ϫ g for 10 min, and the supernatants were immunoprecipitated for 4 h with immobilized phospho-p38 antibody. The pellets were washed twice with a buffer, which consisted of 20 mM Tris (pH 7.5), 10 mM NaCl, 0.1 mM EDTA, 0.1 mM EGTA, supplemented with 1 mM ␤-glycerophosphate, 2 mM dithiothreitol, 1 mM Na 3 VO 4 , 10 g/ml leupeptin, 10 g/ml aprotinin, 2 g/ml pepstatin, 2 mM benzamidine, 1 mM phenylmethylsulfonyl fluoride and 50 mM NaF. Then the pellets were washed once with kinase buffer consisting of 25 mM Tris (pH 7.5), 5 mM ␤-glycerophosphate, 2 mM DTT, 0.1 mM Na 3 VO 4 , and 10 mM MgCl 2 . Finally, the pellets were suspended in 40 l of kinase buffer, including 200 M ATP and 2 g of Elk-1 fusion protein (Cell Signaling Technology, Beverly, MA). The kinase reaction was conducted at ϩ30°C for 30 min and stopped by placing the samples on ice and adding 20 l of 3 ϫ SDS. Next the samples were boiled, microcentrifuged, and analyzed by Western blotting for phospho-Elk-1.
An assay for measuring JNK activation was performed according to the manufacturer's instructions (Cell Signaling Technology). Briefly, c-Jun fusion protein was used to pull down active JNK from clarified cardiac tissue homogenates. Pellet was washed twice with lysis buffer and twice with kinase buffer consisting of 25 mM Tris (pH 7.5), 5 mM ␤-glycerophosphate, 2 mM DTT, 0.1 mM Na 3 VO 4 , and 10 mM MgCl 2 . Finally, the pellet was suspended in 40 l of kinase buffer, including 100 M ATP, and the samples were incubated in 30°C for 30 min. The kinase reaction was stopped by placing the samples on ice and adding 20 l of 3 ϫ SDS. Next the samples were boiled, microcentrifuged, and analyzed by Western blotting for phospho-c-Jun.

RNA Extraction and Northern Blot Analysis-
The RNA extraction and Northern blot analysis were performed as previously (8,15,16). The RNA was extracted from the left ventricles using the guanidinethiocyanate-CsCl method. For the Northern blot analyses, 20-g samples of RNA were separated by electrophoresis and transferred to nylon membranes (Osmonics, Westborough, MA). The cDNA probes complementary to rat GATA-4, BNP, c-fos, or ribosomal 18 S RNA were random prime-labeled, and the membranes were hybridized and washed 3 ϫ 20Ј at 62°C. The results were detected and quantitated as the gel mobility shift assays.
Statistical Analysis-The results are presented as mean Ϯ S.E. The hemodynamic data were analyzed with one-way analysis of variance (ANOVA) followed by Student-Newman-Keuls post hoc test. The statistical significance between two groups was determined by using Student's t test. A protocol of one-way ANOVA followed by a least significant difference post hoc test were used for multivariate analysis. Correlation coefficients were calculated by linear regression analysis. A p value of Ͻ0.05 was considered statistically significant.

Hemodynamics and Gene Expression-To examine the stretch-induced activation of MAPKs and transcription factors
in the left ventricle of adult heart, we applied an in vitro stretch model in isolated perfused rat hearts (18 -21). In this experimental system, wall stress is imposed on isovolumetrically beating rat hearts by distension of a fluid filled balloon in the left ventricle, whereas the control hearts are perfused with flaccid balloons to minimize the left ventricular pressure generation. In the present experiments, the hearts were initially unloaded by opening the valve of the cannula of the intraventricular balloon against the air so the flaccid balloon was filled just enough to obtain a pressure signal (Ϸ10 mmHg systolic pressure at an end-diastolic pressure of Ϸ0 mmHg) to monitor preparation stability. After the 20-min equilibration period the valve of the cannula was closed and the volume of the left ventricular balloon was increased within 5 min to reach the maximal point on the Frank-Starling curve, corresponding to a left ventricular systolic pressure of 135-150 mmHg at an enddiastolic pressure of 20 -25 mmHg. Thereafter the hearts were perfused under these conditions for 10, 30, and 120 min. In the control groups, the perfusion was continued further under the unloaded conditions. The hemodynamic data are shown in Table I. The coronary perfusion pressure remained constant during the experimental period in all experimental groups. No significant differences were observed in the perfusion pressure, left ventricular systolic pressure, and end-diastolic pressure between the vehicle-infused group and drug-infused groups (Table I).
To further validate the experimental system, we analyzed the effects of increased left ventricular wall stress on the expression of BNP and c-fos genes, which are known to be upregulated during the hypertrophic process and during acute hemodynamic overload (2,3). Previously, we have reported that a 4-h period of stretch in the cultured neonatal cardiomyocytes produced a 1.6-fold increase in the BNP gene expression, and infusion of angiotensin II in conscious animals resulted in a 2.2-fold increase in BNP mRNA levels at 2 h (8,25). In agreement with these studies, the increased ventricular wall stress for 2 h resulted in a 1.6-fold increase in BNP gene expression (Fig. 1A, p Ͻ 0.01 versus control, n ϭ 6), whereas a 1-h period of stretch resulted in a 1.2-fold increase in the BNP gene expression (data not shown). We also observed a 3.0-fold increase in the c-fos gene expression (data not shown) at 1 h and and a 2.1-fold increase in c-fos gene expression (Fig. 1B, p Ͻ 0.01 versus control, n ϭ 6) at 2 h. Consistently with these results, Cornelius et al. (20) have reported a 3.5-fold increase and Thienelt et al. (19) a 2.1-fold increase in the c-fos gene expression after a 1-h wall stretch period by using similar experimental setup. Previously, it has also been reported that an acute increment of left ventricular wall stress activates c-fos and c-jun gene expression independently of passive diastolic stretch (18). In agreement with these findings, the left ventricular BNP mRNA levels correlated positively with the levels of developed pressure (r ϭ 0.73, n ϭ 10, p Ͻ 0.05), whereas there was no correlation between the BNP mRNA levels and levels of the left ventricular end-diastolic pressure (r ϭ Ϫ0.18, n ϭ 10, p ϭ NS). Taken together with the results of previous studies, the active generation of an acute increment in left ventricular systolic force generation is the major stimulus leading to the activation of gene expression and signal transduction pathways under these experimental conditions. Interestingly, Yamamoto et al. (26) have shown that strain imposed on cultured myocytes during systolic phase stimulate ERK 1/2 phosphorylation and BNP gene expression more than the strain imposed during diastolic phase.
Activation of p38 Kinase, ERK, and JNK by Increased Left Ventricular Wall Stress-The mitogen-activated protein kinases are known to be activated in the heart in response to numerous stimuli. To establish the wall stretch-induced acti- vation of MAPKs under these experimental conditions, the hearts were exposed to increased left ventricular wall stress for 10 min, and Western blotting using phospho-p38-, phospho-p44/p42-, and phospho-JNK-specific antibodies was performed.
In agreement with a previous study in isolated guinea pig hearts (27), the left ventricular wall stretch for 10 min increased phospho-p38 levels ( Fig. 2A, 2.3-fold versus control, p Ͻ 0.001) and phospho-ERK1/2 levels (Fig. 2B, 1.6-fold versus control, p Ͻ 0.05). Also the phospho-JNK levels were markedly increased (Fig. 2C, 1.9-fold versus control, p Ͻ 0.05), in contrast to the previous study (27). The reason for this discrepant finding may be related to species, differences in the experimental protocols or to the higher level of end-diastolic pressure used in the previous study. In contrast to phospho-p38, phospho-ERK, and phospho-JNK, total p38, ERK, and JNK protein levels remained unchanged (Fig. 2, A-C). Effect of SB239063, U0126, and CEP11004 on MAPK Activities-SB239063 is a novel p38 inhibitor that exhibits improved kinase selectivity and in vivo activity compared with the other p38 inhibitors (28). Because p38 inhibitors, including SB239063, are known to affect the catalytic activity of p38 kinase rather than the levels of phosphorylated p38, a kinase assay that uses Elk-1 as a substrate was performed to confirm the inhibition of the p38 kinase. As shown in Fig. 3A, the p38 kinase activity in the left ventricle was clearly increased by the wall stretch and the increase was abolished in the presence of 10 M SB239063. To influence on ERK 1/2 activity, we used U0126, which is a potent inhibitor of MEK1/2, an upstream regulator of the phosphorylation of ERK 1/2 (29). As described in Fig. 3B, administration of 1 M U0126 significantly reduced the levels of phospho-ERK both in the stretched and unstretched left ventricles. On the other hand, treatment with 1 M U0126 had no effect on the levels of phosphorylated p38 kinase and, similarly, the administration of SB239063 did not affect the levels of phospho-ERK 1/2 (data not shown). Finally, we infused CEP11004, which is a novel inhibitor of MLK1, MLK2, and MLK3, which activate JNK through MKK4 and MKK7 (30). A kinase assay using c-Jun as substrate was performed. The wall stretch increased the phosphorylation of c-Jun, and the increase was markedly attenuated by treatment with 100 nM CEP11004 (Fig. 3C). To further confirm the inhibition of JNK, we performed Western blotting from the same samples using anti-phospho-JNK antibody. As shown in Fig.  3C, the levels for phospho-JNK were reduced by CEP11004. Effect of Endothelin-1 and Angiotensin II Receptor Antagonism on p38 and ERK Activation-Because wall stretch is known to activate the hypertrophic response in part by increased local production or release of angiotensin II and ET-1 (1, 2, 15, 16), we next studied whether p38 and ERK 1/2 are activated directly by increased left ventricular wall stress or indirectly via angiotensin II or ET-1. The hearts were perfused in the presence of 1 M of mixed ET A /ET B receptor antagonist bosentan and 100 nM of AT 1 receptor antagonist CV-11974 for 10 min. These concentrations of bosentan and CV-11974 have previously been shown to be effective under these experimental conditions (31). As shown in Fig. 4, bosentan and CV-11974 had no effect on wall stress-induced increase in phospho-p38 and phospho-ERK 1/2 levels, thereby suggesting that ET-1 and angiotensin II are not involved in the stretch-induced activation of MAPKs.
Increased Wall Stress Rapidly Up-regulates the DNA Binding of GATA-4 -Increased left ventricular wall stress for 30 min resulted in a 1.7-fold increase in DNA-binding activity of nuclear extracts and 30-bp double-stranded oligonucleotide probe containing the Ϫ90-bp BNP GATA sites (Fig. 5A, p Ͻ 0.001). The extent of the increase was similar to that we have previously reported (16). We also noted an increase in the AP-1 DNA binding, whereas the DNA-binding activities of nuclear factor-B (NF-B) and nuclear factor of activated T cells (NFATc) remained unchanged (data not shown). No changes were observed in the Octamer-1 DNA binding, either. To confirm the specificity of BNP GATA-4 binding, supershift assays using GATA-4, GATA-5, and GATA-6 antibodies were performed. An antibody-induced supershift was seen for GATA-4 but not for GATA-5 or GATA-6 complexes (Fig. 5A). To make certain that the formation of DNA-protein complexes was the result of a specific interaction, competition analysis was performed. The binding was markedly attenuated by an excess of unlabeled rBNP Ϫ90-bp GATA DNA but remained unchanged by an excess of non-related Oct-1 competitor DNA or rBNP Ϫ90-bp DNA with mutated GATA site (Fig. 5B).

GATA-4 Protein and mRNA Levels in Loaded Left
Ventricles-To test whether the increased DNA-binding activity is associated with increased nuclear accumulation of GATA-4 protein, Western blotting using nuclear protein extract and anti-GATA-4 specific antibody was performed. As shown in Fig.  6A, no significant differences were observed in the levels of GATA-4 protein between the stretched and unstretched hearts, suggesting that the increase in the DNA-binding activity does FIG. 3. Inhibition of p38 by SB239063, ERK by U0126, and JNK by CEP11004 in the isolated rat heart. The hearts were exposed to increased wall stress for 30 min, the cytoplasmic protein was extracted and Western blotting, and kinase assays were performed. A, SB239063, a novel and selective inhibitor of p38 kinase, was infused as a continuous infusion via the aortic cannula during the perfusion. p38 MAPK activity was determined with immunocomplex kinase assay by using Elk-1 as substrate. Cardiac tissue homogenates were clarified, and the supernatants were immunoprecipitated with immobilized phospho-p38 antibody. The pellets were washed and suspended in kinase buffer, including Elk-1 fusion protein, the kinase reaction was conducted for 30 min, and finally the samples were boiled, microcentrifuged, and analyzed by Western blotting for phospho-Elk-1. A representative Western blot analysis showing that inhibition of p38 by 10 M SB239063 resulted in a marked attenuation of the stretch-induced activation of the p38 kinase. B, a representative Western blot analysis showing the inhibition of ERK1/2 by a potent MEK 1/2 inhibitor U0126. 30-g samples of protein were boiled, resolved by SDS-PAGE, and blotted on nitrocellulose membranes. The blot was incubated with phospho-p44/ p42-and p44/42-specific antibodies and detected by enhanced chemiluminescence. The administration of 1 M U0126 during the perfusion resulted in a decrease in both baseline and stretch-induced phospho-ERK 1/2 levels, whereas the levels of total ERK 1/2 remained unchanged in both unstretched and stretched hearts; C, the hearts were perfused in the presence of 100 nM CEP11004, a novel MLK inhibitor. c-Jun fusion protein was used to pull down active JNK from clarified cardiac tissue homogenates. Pellet was washed and suspended in kinase buffer, the samples were incubated for 30 min, boiled, microcentrifuged, and analyzed by Western blotting for phospho-c-Jun. A representative Western blot showing that the inhibition of JNK by 100 nM CEP11004 decreases the phosphorylation of c-Jun. To further confirm the inhibition of JNK by CEP11004, Western blotting for phospho-JNK was performed using the same protein extracts. A representative Western blot showing the effect of CEP11004 on the levels of phospho-p46and phospho-p54JNK.

FIG. 4. The left ventricular wall stretch-induced activation of MAPKs is independent of ET-1 and angiotensin II.
To establish the role of endothelin-1 and angiotensin II in the stretch-induced activation of MAPKs, the hearts were perfused for 10 min in the presence and absence of wall stretch, and mixed ET A /ET B receptor antagonist bosentan and angiotensin II receptor antagonist CV-11974 were infused through the aortic cannula during the perfusion. The cytoplasmic protein was extracted and Western blotting was performed. The stretch-induced increase in phospho-p38 and phospho-ERK 1/2 levels was not influenced by administration of either bosentan or CV-11974. Representative Western blots showing the effect of bosentan and CV-11974 on the levels of phospho-p38 (A) and phospho-ERK 1/2 (B). not result from increased production or increased nuclear accumulation of GATA-4 protein. This finding agrees with our recent results by using cultured neonatal cardiomyocytes (8). Furthermore, increased left ventricular wall stress for 30 min had no effect on GATA-4 mRNA levels (Fig. 6B, 1.1-fold versus  control, p ϭ NS) showing that the increase in GATA-4 DNAbinding activity is not related to increased transcription of GATA-4 gene. Taken together, these findings support the hypothesis that the activation of GATA-4 DNA binding results from changes in the protein phosphorylation.
Wall Stress-induced Activation of GATA-4 Binding Is Abolished by p38 and ERK Inhibition But Not by JNK Inhibition-To evaluate the role of p38 kinase-mediated phosphorylation in the wall stress-induced activation of GATA-4 binding, the hearts were perfused in the presence of 10 M SB239063. The wall stress-induced increase in the GATA-4 DNA binding was completely abolished by 10 M SB239063 (Fig. 7A, p Ͻ 0.01  versus vehicle-infused stretch). Similarly, the stretch-induced increase in the DNA binding of GATA-4 was significantly attenuated in the presence of the selective MEK1/2 inhibitor U0126 (Fig. 7B, p Ͻ 0.01 versus vehicle-infused stretch). In contrast to ERK and p38 inhibition, JNK inhibition by CEP11004 (100 nM) had no significant effect on stretch-induced GATA-4 DNA-binding activity (Fig. 7C, p ϭ NS versus vehicleinfused stretch, p Ͻ 0.05 versus vehicle-infused control). Infusion of either SB239063 (Fig. 7A), U0126 (Fig. 7B), or CEP11004 (Fig. 7C) had no effect on baseline GATA-4 binding.
The vehicle (0.02% dimethyl sulfoxide) alone had no effect on GATA-4 DNA binding (data not shown).
Mechanical Stretch Activates GATA-4 Binding via p38 and ERK in the Isolated Atria-To further examine the role of MAPK pathways in the stretch-induced GATA-4 binding, the effects of p38 and ERK 1/2 inhibition were studied by using the isolated atria preparation. The isolated left auricles were perfused for 30 min without stretch and thereafter stretched at 8 cm of H 2 O for 30 min. The 30-min stretch produced a 2.0-fold increase in the binding activity of nuclear extracts and the Ϫ90-bp BNP GATA oligonucleotide probe (Fig. 8, p Ͻ 0.01 versus control). p38 inhibition by 5 M SB203580 significantly attenuated the stretch-induced GATA-4 DNA binding (Fig. 8,  p Ͻ 0.01 versus vehicle-treated stretch). Similarly, the stretchinduced GATA-4 DNA binding was decreased in the presence of MEK1 inhibitor PD98059 (5 M) (Fig. 8, p Ͻ 0.05 versus vehicle-treated stretch). These concentrations of SB203580 and PD98059 were shown to effectively inhibit p38 kinase and ERK 1/2, respectively, under these experimental conditions (data not shown).
Wall Stress-induced Activation of GATA-4 DNA Binding Is Attenuated by Inhibition of the Rho Kinase But Not by PKC Inhibition-Because the activation of p38 and ERK were independent of endothelin-1 and angiotensin II, we characterized further the upstream pathways responsible for the activation of GATA-4 DNA binding. Because a recent study suggested that Rho/ROCK pathway may be a proximal activator of the phenylephrine-induced activation of GATA-4 DNA binding in the cultured cardiomyocytes (32), we examined the effects of Rho kinase inhibition on the wall stress-induced GATA-4 DNA were perfused for 30 min, the nuclear proteins were extracted, and gel mobility shift assays were performed as described in more detail under "Experimental Procedures." A, the nuclear extracts from ventricular tissue were incubated with rBNP Ϫ90-bp oligonucleotide probe. The ventricular wall stretch for 30 min increased the DNA-binding activity in nuclear extracts with rBNP Ϫ90-bp probe (lanes 1 and 2). binding by using Y-27632 at a concentration of 3 M. Y-27632 has been shown to specifically and potently inhibit Rho-dependent kinases (K i ϭ 0.14 M for p160ROCK), and has been used at a concentration of 3 M in cultured cardiomyocytes (33,32). The wall stress-induced activation of GATA-4 DNA binding was significantly attenuated by treatment with Y-27632 (Fig. 9A, p Ͻ 0.05 versus vehicle-infused stretch), thereby suggesting that Rho kinase is involved in the stretch-induced activation of GATA-4 DNA binding. Treatment with Y-27632 did not significantly affect the baseline GATA-4 DNA binding (Fig. 9A, p ϭ NS versus vehicle-infused control). The hemodynamic parameters in Y-27632 or bisindolylmaleimide-infused hearts parameters did not significantly differ from the vehicleinfused hearts (data not shown).
Protein kinase C has been suggested to be involved in the stretch-induced activation of MAPKs in the isolated heart (27).
To study the role of PKC in the stretch-induced GATA-4 binding, the hearts were perfused in the presence on 90 nM of bisindolylmaleimide I (GF-109203X), a potent PKC inhibitor. We have previously shown that this dose of bisindolylmaleimide I is effective under these experimental conditions (34). The increased wall stress resulted in a 2.1-fold increase in GATA-4 DNA binding (Fig. 9B, p Ͻ 0.05 versus vehicle-infused control) in the presence of bisindolylmaleimide I.

DISCUSSION
The precise intracellular signaling pathways that link the hypertrophic stimuli into the changes in cardiac gene expression, changes in nuclear transcription, and hypertrophic growth have not been fully established. Yet, reversible protein phosphorylation and dephosphorylation have been suggested to play an essential role in these processes (2). Accumulating evidence also indicates that GATA-4 DNA-binding activity is regulated by protein phosphorylation, whereas GATA-4 protein expression remains stable in response to various hypertrophic stimuli (7)(8)(9)(10). Indeed, at least seven potential sites for serine phosphorylation by MAPKs have been identified in the GATA-4 protein (35). Moreover, in vitro studies support an important role for MAPKs in the regulation of GATA-4 in G-proteincoupled receptor agonist-induced cardiomyocyte hypertrophy (24,32,36). The functional significance of GATA-4 in the regulation of several genes, including the BNP gene, has been well characterized in vivo (11), but the current knowledge concerning the intracellular signaling pathways controlling GATA-4 is restricted only to studies performed in cultured neonatal cardiomyocytes. Furthermore, the mechanisms of stretch-induced activation of GATA-4 binding are unknown. Thus, we tested the hypothesis whether the up-regulation of GATA-4 DNA binding by increased wall stress is mediated by the MAPK pathways in the adult heart. Our present results show that both p38 kinase and ERK 1/2 are essential for the wall stretchinduced activation of GATA-4 binding in the normal adult ventricle and atria.
A major finding of the present study was that mechanical stretch activated p38 MAPK and that the activation of the kinase was necessary for wall stretch-induced GATA-4 binding to BNP promoter both in the left ventricle and in the atria. Previously, we have reported that in the cultured cardiomyocytes ET-1-induced increase in the GATA-4 binding was abolished by inhibition of p38 kinase (24). Interestingly, ET A /ET B and AT 1 receptor antagonists did not influence the left ventricular wall stretch-induced activation of p38 kinase, although both ET-1 and angiotensin II are involved in mediating stretchinduced GATA-4 binding (16). This suggests that mechanical stress per se, without paracrine mechanisms, activates p38, which directly affects the DNA-binding activity of GATA-4. In cultured cardiomyocytes, blockade of the p38 pathway leads to decreased phosphorylation of serine residues in the GATA-4 protein (24). Whether p38 kinase directly phosphorylates the serine residues of the GATA-4 protein in the intact heart remains to be studied in the future experiments.
Our present results in intact adult rat heart differ in part from those that have previously been obtained in cultured cardiomyocytes. Similarly to p38 kinase, wall stretch activated ERK 1/2 and administration of a potent MEK1/2 inhibitor U0126 inhibited the wall stretch-induced DNA-binding activity of GATA-4. Furthermore, the basal GATA-4 binding activity in unloaded hearts remained unchanged with administration of U0126, whereas in cultured myocytes inhibition of MEK1 by PD98059 significantly reduced basal binding activity and had no effect on ET-1-induced GATA-4 binding (24). In agreement with the important role of ERK 1/2 in the regulation of GATA-4, phenylephrine-induced up-regulation of ET-1 GATA-4 binding in cardiomyocytes has been reported to be sensitive to MEK1 inhibition by PD98059 (12). Furthermore, serine 105 of GATA-4 has been demonstrated to be phosphorylated directly by ERK 1/2 as a response to phenylephrine stimulation and Rho/ROCK pathway has been shown to be linked to GATA-4 activation via the ERK pathway (32,36). In addition to p38 and ERK 1/2, also the third member of the MAPK family, JNK, has been suggested to be involved in the hypertrophic program, yet its role has remained to some extent controversial (37,38). We found an increase in phospho-JNK activity in response to increased left ventricular wall stress, but the MLK inhibitor CEP11004 had no effect on the increase in GATA-4 DNA binding. Thus, it is unlikely that JNK would be involved in the regulation of wall stress-induced GATA-4 binding.
The finding that both p38 kinase and ERK 1/2 were needed for the augmented binding activity of GATA-4 in normal adult heart may be related to the possibility that ERK 1/2 and p38 MAPK phosphorylate different phosphorylation sites of the GATA-4 protein or that they phosphorylate the same sites, but neither active p38 nor ERK alone is sufficient for the required amount of phosphorylation. On the other hand, ERK and p38 activation occur partly in response to same stimuli and cross-talk and interaction between different MAPK pathways is known to exist at several levels. For example, the same MKKK can activate several MKKs, which then specifically activate different MAPKs (17). Finally, it should be noted that the signaling mechanisms activated by mechanical stress in the adult heart likely are far more complex than those activated by merely the administration of ET-1 or phenylephrine in neonatal cardiomyocytes.
Based on the present results, it appears that the Rho kinase pathway may be an upstream effector of the p38-and ERKmediated GATA-4 DNA binding, because the Rho kinase inhibition significantly reduced the wall stress-induced GATA-4 DNA-binding activity. This is consistent with the previous study in the cultured cardiomyocytes, which showed that phenylephrine-induced GATA-4 DNA binding was abrogated by treatment with Y-27632 (32). On the other hand, the PKC inhibition did not markedly affect the stretch-induced GATA-4 binding. In the previous study, it was suggested that the regulation of actin polymerization could provide a convergence point for Rho kinase and GATA-4 signaling (32). How the Rho kinase pathway, p38, and ERK and stretch-induced GATA-4 DNA binding are linked to each other remains to be studied in future experiments in detail. Moreover, the present results do not exclude the role of other upstream pathways and mechanisms other than MAPK mediated phosphorylation in the regulation of GATA-4 in the intact heart. For example, in a recent study it was demonstrated that glycogen synthase kinase 3␤ negatively regulates the activity of GATA-4 (39). Furthermore, GATA-4 activity has been shown to be subject to interactions with nuclear cofactors and other transcription factors. Nkx-2.5 and GATA-4 have been demonstrated to synergistically activate the ANP gene (40), and interactions between GATA-4 and myocyte enhancer factor 2 (MEF-2) (41), serum response factor (42), NFAT (43), peroxisome proliferator activator receptorbinding protein (44), and friend of GATA-2 (45) have been described.
Interestingly, recent studies suggest a partially differential role for p38 kinase and ERK 1/2 in the hypertrophic program. In cultured cardiomyocytes, the ERK signaling pathway has been shown to be necessary for phenylephrine-induced hypertrophy as well as for sarcomeric organization induced by different hypertrophic agonists (46). Studies in transgenic mice overexpressing MEK1 further support the role of ERK1/2 in the development of compensated and concentric cardiac hypertrophy (47). In contrast to ERK1/2, p38 may be more important in the development of dilated cardiomyopathy and hypertensive end-organ damage. Indeed, Liao et al. (48) have reported that transgenic mice overexpressing the upstream activators of p38 kinase (MKK3bE and MKK6bE) develop restrictive cardiomyopathy characterized by hypertrophic marker gene expression, loss of contractility, and restrictive diastolic abnormalities. It has also been shown that hypertensive end-organ damage and hypertensive cardiac hypertrophy are attenuated by long term treatment with p38 inhibitor SB239063 in a rat model (49). Taken together with the role of GATA-4 as a hypertrophic integrator and regulator of the stretch-induced hypertrophic program, one may assume that left ventricular wall stretchinduced activation of GATA-4 via p38 and ERK1/2 could be one of the very first steps in the development of both concentric hypertrophy and dilated cardiomyopathy.
In summary, we characterize for the first time the involvement of MAPKs in the regulation of GATA-4 DNA binding in normal adult heart. Our results indicate that both p38 kinase and ERK 1/2 are required for the wall stress-induced activation of GATA-4 DNA binding in the left ventricle and atria. We show that JNK is activated in response to increased left ven- tricular wall stress but is not essential for the activation of GATA-4 in normal adult heart. Finally, we show that the activation of GATA-4 DNA binding is dependent on Rho kinase but not protein kinase C.