p38 Mitogen-activated Protein Kinase Mediates the Transcriptional Induction of the Atrial Natriuretic Factor Gene through a Serum Response Element

In various cell types certain stresses can stimulate p38 mitogen-activated protein kinase (p38 MAPK), leading to the transcriptional activation of genes that contribute to appropriate compensatory responses. In this report the mechanism of p38-activated transcription was studied in cardiac myocytes where this MAPK is a key regulator of the cell growth and the cardiac-specific gene induction that occurs in response to potentially stressful stimuli. In the cardiac atrial natriuretic factor (ANF) gene, a promoter-proximal serum response element (SRE), which binds serum response factor (SRF), was shown to be critical for ANF induction in primary cardiac myocytes transfected with the selective p38 MAPK activator, MKK6 (Glu). This ANF SRE does not possess sequences typically required for the binding of the Ets-related ternary complex factors (TCFs), such as Elk-1, indicating that p38-mediated induction through this element may take place independently of such TCFs. Although p38 did not phosphorylate SRF in vitro, it efficiently phosphorylated ATF6, a newly discovered SRF-binding protein that is believed to serve as a co-activator of SRF-inducible transcription at SREs. Expression of an ATF6 antisense RNA blocked p38-mediated ANF induction through the ANF SRE. Moreover, when fused to the Gal4 DNA-binding domain, an N-terminal 273-amino acid fragment of ATF6 was sufficient to support trans-activation of Gal4/luciferase expression in response to p38 but not the other stress kinase, N-terminal Jun kinase (JNK); p38-activating cardiac growth promoters also stimulated ATF6 trans-activation. These results indicate that through ATF6, p38 can augment SRE-mediated transcription independently of Ets-related TCFs, representing a novel mechanism of SRF-dependent transcription by MAP kinases.

The activation of signal transduction pathways by growth factors stimulates transcription of certain genes (e.g. c-fos) that participate in the growth response. Among the regulatory regions in the promoter of the well studied c-fos gene that are responsible for conferring growth factor inducibility is the serum response element (SRE) 1 (1). The c-fos SRE, which is comprised of a core sequence (CC(A/T) 6 GG) and a nearby Ets motif ((C/A)(C/A)GGA(A/T)), binds serum response factor (SRF) over the A/T-rich core sequence (2), as well as the ternary complex factors (TCF), Elk-1 or Sap-1, over the Ets motif. SRF and TCFs interact with each other as well their cognate DNA binding elements to form a ternary complex (3,4). Growth factor-activated signal transduction pathways often converge on either the SRF or the TCF components of the ternary complex (1,5). For example, all three MAP kinases, extracellular signal-regulated kinase (ERK), N-terminal jun kinase (JNK), and p38, phosphorylate the TCF, Elk-1, as part of the mechanism by which c-fos transcription is activated through the SRE (6). Although SRF itself is phosphorylated as a result of the activation of certain signaling pathways (e.g. pp90 rsk and Ca/ CaMKIV; Refs. 7 and 8), and this phosphorylation leads to increased c-fos transcription, there is no evidence that any of the MAP kinases can influence transcription by directly phosphorylating SRF. Accordingly, it is widely believed that MAP kinase-stimulated transcriptional induction through the c-fos SRE is dependent on Elk-1-, Sap-1-, or as yet unidentified TCFs.
In contrast to the c-fos promoter there are a number of genes that possess critical SREs which do not have flanking-or nearby Ets motifs and therefore are not likely to bind Etsrelated TCFs (4, 9 -11); among them are several cardiac-specific genes that are induced as part of the well studied hypertrophic growth program. For example, the skeletal ␣-actin (12,13), cardiac ␣-actin (14), and atrial natriuretic factor (ANF) genes (9) all possess such SREs that are critical for induction by growth factors. Recent studies have shown that the p38 pathway plays an important role in conferring growth factor inducibility to these cardiac genes (15,16). Accordingly, induction by p38 presumably involves the SREs that are required for growth factor-stimulated transcription. However, the only mechanism presently known by which p38 can activate transcription through SREs requires phosphorylation of Elk-1 by p38 (6). Thus, it is unclear how p38 could enhance transcription through SREs in cardiac genes in an Ets/TCF-independent manner. To address this question we used the ANF gene in cardiac myocytes as a model system and asked first whether p38 pathway signals converge on the ANF SRE to confer inducibility and if so, through what mechanism.

MATERIALS AND METHODS
Cell Culture-Primary ventricular myocytes, prepared from 1-4day-old neonatal rats as described previously (9,15,17), were transfected (see below), plated onto fibronectin-coated plastic dishes, and then maintained for about 18 h in DMEM, 10% fetal bovine serum. The cultures were then washed briefly with medium, refed with serum-free, hormone-free DMEM containing 1 mg/ml bovine serum albumin, maintained for an additional 48 h and then extracted for reporter enzyme assays.
Transfections-After isolation, myocardial cells were resuspended at a density of 30 million cells/ml minimal medium (DMEM (Sigma) containing 1 mg/ml bovine serum albumin) and transfections were carried out as described previously (9,15,17). Briefly, for each transfection, 300 l, or 9 million cells, were mixed with 15 to 30 g of a reporter construct (e.g. ANF/Luc (9) or pG5E1bLuc (1, 18), 12 to 18 g of pCH110 (SV40-␤-galactosidase), which was used for normalization, and in some experiments, 15 to 45 g of an MKK6, ATF6/Gal4, or Gal4 DNA-binding domain expression construct (see below). The levels of plasmid used in each culture within an experiment were equalized using empty vector DNA, such as pCMV6. Each 300-l aliquot was then electroporated in a Bio-Rad Gene Pulser at 500 V, 25 microfarads, 100 ⍀ in a 0.2-cm gap cuvette, a protocol that allows for the selective transfection of only cardiac myocytes (9,15,17). This procedure results in an approximate 30% viability (9); accordingly, the 3 million viable cells were plated into fibronectin-coated 35-mm wells.
Reporter Enzyme Constructs-The ANF/luciferase reporter constructs used in this study include ANF 134 and ANF 65, which are composed of ANF (Ϫ134 to ϩ65) or ANF (Ϫ65 to ϩ65) driving luciferase expression, respectively. ANF 134 is inducible, while ANF 65 displays mostly basal promoter activity; both constructs display myocardial cellspecific expression (9). The structures of ANF 65 (c-fos SRE), ANF 65 (ANF SRE), ANF 65 (M c-fos SRE), and ANF 65 (M ANF SRE) are described in Fig. 1. In each case the relevant synthetic oligonucleotide pairs (see below) were hybridized, ligated to form trimers, and cloned upstream of the minimal, cardiac-specific ANF promoter (ANF 65), as described previously (9). Reporter Enzyme Assays-Transfected cells were maintained in DMEM supplemented with 10% fetal bovine serum for approximately 16 h after electroporation. The cells were then washed thoroughly and the medium was replaced with minimal medium. Unless otherwise stated, 24 to 48 h later the cultures were extracted for reporter enzyme assay. Luciferase and ␤-galactosidase assays were performed as described (9). Luciferase activity was measured for 30 s on a Bio Orbit 1251 Luminometer (Pharmacia Biotech Inc., Piscataway, NJ). Data are expressed as "Relative Luciferase (Rel Luc)" ϭ arbitrary integrated luciferase units/␤-gal units, representative of at least three independent experiments performed with two different plasmid preparations, and represent the mean and S.E. of triplicate cultures.
Preparation of Recombinant ATF6 and SRF-The full-length cDNAs for human ATF6 (18) and SRF, the latter of which was obtained from E. Olson, University of Texas, were subcloned into pcDNA3.1 and used as templates for in vitro transcription/translation using TNT ® T7 Quick Coupled Transcription/Translation System (Promega) Ϯ [ 35 S]Met, as per the manufacturer's instructions. The identities of the products were confirmed by observing the mobilities of the [ 35 S]Met-labeled products after SDS-PAGE and by immunoprecipitating 35 S-Met-labeled products with the anti-ATF6 (19) or anti-SRF (20) antisera followed by SDS-PAGE.
Electrophoretic Mobility Shift Assays-EMSA was performed as described (9), with minor modifications. ANF SRE and MEF2C probes were prepared by T4 kinase end labeling of the relevant oligonucleotide pairs (see above). A typical binding assay contained 20,000 cpm of double-stranded probe and 5 to 10 g of protein in 1 ϫ binding buffer (30 mM NaCl, 0.1 mM EDTA, 8 mM Tris-HCl (pH 8.2), 8% glycerol, 1 nM dithiothreitol, 0.2 mM ZnCl 2 ). After a 10-min preincubation of extract and 0.2 g of nonspecific competitor (poly(dI-dC), Pharmacia Biotech Inc.) and test competitor oligonucleotide pairs, as indicated, the probe was added. Binding was allowed to proceed at room temperature for 30 min prior to separation of bound and free probe on a 4% native polyacrylamide (PA) gel (29:1 bis/acrylamide) high ionic strength buffer (50 mM Tris (pH 8.5), 380 mM glycine, 2.1 mM EDTA) at room temperature. DNA-protein complexes were detected by autoradiography. The autoradiograms of some gels in this article were scanned using a Molecular Dynamics Personal Densitometer and the resulting image was imported to Adobe Photoshop and Claris MacDraw Pro II for final figure preparation.
In Vitro Phosphorylation by p38 and MAPKAP Kinase-3-In vitro phosphorylation reactions with p38 and MAPKAP kinase-3 were carried out essentially as described previously (21)(22)(23). Briefly, approximately 1 g of either the active form of p38 (obtained from Dr. J. Han) or the active form of MAPKAP kinase-3 (obtained from Dr. Stephan Ludwig) were mixed with recombinant ATF6, MEF2C, or HSP27 (StressGen), in 30 l of kinase buffer (21) and [␥-32 P]ATP and allowed to incubate at 30°C for 30 min. Approximately 1 l of antisera specific for the protein in question was then added to each reaction mixture and allowed to incubate for 4 h at 4°C after which protein A-Sepharose beads were used to collect immune complexes, as described previously (15). Immune complexes were then eluted from the beads using Laemmli sample buffer and the eluted material analyzed by SDS-PAGE followed by PhosphorImager analysis, as described in the legend to Fig.  3. In some cases, the recombinant proteins were prepared in the presence of [ 35 S]methionine allowing them to be used as markers on parallel lanes of a gel.
ERK, JNK, and p38 Assays-For ERK and JNK assays, cultures were treated for 30 min with control medium, or phenylephrine (10 M), serum (10%), phorbol ester diburyate (100 nM), or endothelin-1 (10 nM), then extracted in 10 mM Tris (pH 7.6), 1% Triton X-100, 0.05 M NaCl, 5 mM EDTA, 2 mM sodium o-vanadate, and 20 g/ml aprotinin. After brief centrifugation, extracts to be tested for ERK activity were incubated for 2 h at 4°C with anti-ERK (raised against the C-terminal 16 amino acids of ERK-1; Santa Cruz SC-093) bound to Protein A-Sepharose (Pharmacia) and immune-complex kinase assays were carried out using the appropriate substrates, as described previously (8,15,24). Briefly, reactions were initiated by the addition of 1 g of myelin basic protein and 6 M [␥-32 P]ATP (5000 Ci/mmol) in a final volume of 30 l of kinase buffer (20 mM HEPES (pH 7.4), 20 mM MgCl 2 , 20 mM ␤-glycerophosphate, 2 mM dithiothrietol, 20 M ATP). After 30 min at 25°C, the reactions were terminated by the addition of Laemmli sample buffer and the phosphorylation level of substrate proteins was evaluated by SDS-PAGE followed by autoradiography and PhosphorImager analyses.
p38 assays were carried out as described previously (15). Briefly, after treating with test agent, myocardial cells (1.5 ϫ 10 6 cells/35-mm culture well) were extracted in 80 l of Laemmli buffer containing 1 mM p-nitrophenyl phosphate, 100 M sodium o-vanadate, 10 g/ml aprotinin, 10 g/ml leupeptin, and 1 mM phenylmethylsulfonyl fluoride. Sixty-l aliquots of each extract were fractionated on a 10% SDS-polyacrylamide gel followed by Western blotting using a p38 antibody specific for the phosphorylated/activated form of the kinase (specific for Thr(P) 180 / Tyr(P) 182 ; New England Biolabs, Inc., Beverly MA, catalog number NE 92115). In each experiment 3 identically treated cultures (1.5 ϫ 10 6 cells/35 mm dish) were used for each treatment and following densitometric analyses of the exposed PhosphorImager plates, values for each treatment were averaged.

Mapping p38-inducible Elements in the ANF Promoter-To
begin determining how the p38 MAPK pathway participates in cardiac gene induction, p38-inducible elements in the rat ANF promoter were mapped. Primary myocardial cells were cotransfected with ANF promoter/luciferase reporter plasmids (Fig. 1A) and a construct encoding MKK6 (Glu), an activated form a p38 MAP kinase kinase (18,21) which, among the MAPKs, activates only p38 in cardiac myocytes (15,16). In previous studies it has been shown that nearly all the information for optimal hormonal induction of ANF transcription resides in the 134-bp region just 5Ј of the transcriptional start site (9). Consistent with those results, transcription from a reporter construct comprised of the region spanning from Ϫ134 to ϩ65 of the rat ANF 5Ј-flanking sequence was strongly induced (approximately 6-fold) by MKK6 (Glu) (Fig. 1B, Construct 1), similar to levels previously seen using reporters comprised of 638 bp of ANF 5Ј-flanking sequence (15). The induction of ANF 134 was completely blocked by SB 203580, a cell permeable compound shown to selectively inhibit p38 (26). Truncating ANF 134 by just 70 nt resulted in a construct, ANF 65, which had completely lost p38 inducibility (Fig. 1B, Construct 3), indicating that p38-inducible sequences must lie between Ϫ134 and Ϫ65 in the ANF 5Ј-flanking sequence.
An A/T-rich SRE, the core of which resides between Ϫ117 and Ϫ108 in ANF (Fig. 1A), is highly conserved and found in the ANF genes of all species studied to date. This SRE was previously shown to mediate ANF induction in response to myocardial cell growth factors, such as ␣ 1 -adrenergic agonists (9). To evaluate whether the ANF SRE participated in p38mediated ANF 134 induction, a cluster mutation known to distrupt SRF binding to the ANF SRE was tested. This mutated reporter construct, ANF 134 (M ANF SRE), was completely incapable of displaying inducibility by MKK6 (Glu) (Fig.  1B, Construct 2), indicating the importance of the ANF SRE in p38-stimulated ANF transcription. When the same cluster mutation was tested in the context of ANF 638, p38-stimulated reporter production was nearly completely inhibited compared with the native form of ANF 638 (not shown).
To further characterize whether and how this region of the ANF gene conferred induction, ANF sequences spanning from Ϫ125 to Ϫ102, which contains the 10-nt core ANF SRE and 8 and 6 nt of 5Ј-and 3Ј-flanking sequence, respectively, were cloned into ANF 65. In this context, the ANF SRE alone resulted in the recovery of nearly 100% of p38-inducibility to the otherwise, uninducible ANF 65 (Fig. 1B, compare Constructs 3 and 4). Unlike the well characterized SRE present in the c-fos gene, which serves as the prototype of this element, the ANF SRE does not possess an Ets motif ((C/A)(C/A)GGA(A/T)) and, therefore, is not known to bind the Ets-related ternary complex factors (e.g. Elk-1 or Sap-1) (4, 9, 10) through which many signaling pathways confer transcriptional induction at SREs (1,(27)(28)(29). When the c-fos SRE (10-nt core ϩ8 and 6 nt of c-fos 5Ј-and 3Ј-flanking sequence, respectively) was cloned 5Ј of ANF 65, full responsiveness to MKK6 (Glu) was regained (Fig.  1B, Construct 5), even when the Ets-binding site was mutated in a manner known to distrupt TCF binding (Fig. 1B, Construct  6).
Thus, it is apparent that the A/T-rich region between Ϫ117 and Ϫ108 of the rat ANF 5Ј-flanking sequence is critical for p38 inducibility, and that SRF, or a different protein capable of binding to A/T-rich sequences, is required for this inducibility, but probably does not require the presence of TCFs that typically bind to the Ets domain of the c-fos SRE.
SRF and MEF2C Electrophoretic Mobility Shift Analysis-An obvious candidate for another A/T-rich binding protein that could mediate p38 inducibility is MEF2C. MEF2C, an SRF-related transcription factor (30,31), is found in high quantities in the heart, binds to A/T-rich sequences in cardiac genes (32,33), and is required for proper cardiac development and gene expression (32)(33)(34). Furthermore, MEF2C is phosphorylated by p38 (22) in a manner that confers transcriptional inducibility in the cardiac context (15).
Thus, since the ANF SRE nearly fits the MEF2C consensus binding site ((C/T)T(A/T)(A/T)AAATA(A/G)), it seemed possible that MEF2C could be responsible for conferring p38-inducible ANF transcription. This hypothesis was investigated initially using electrophoretic mobility shift assays (EMSAs) and recombinant MEF2C or, as a control, recombinant SRF. However, when MEF2C was incubated with radiolabled oligonucleotides, there was no apparent binding to the ANF SRE probe, while the positive control, a muscle creatine kinase probe, bound MEF2C, as expected (34) (Fig. 2A). The competition profile showed that only the MCK MEF2C oligonucleotide was able to compete for MEF2C binding, with no competition obtained with either the c-fos SRE or the ANF SRE. The same result was obtained using ventricular nuclear extracts (not shown). The converse experiment showed that while recombinant SRF could bind specifically to the ANF SRE probe, and while the appropriate competition profile was obtained, SRF did not bind to the MCK MEF2C probe under these conditions (Fig. 2B).
Phosphorylation Analysis-Since MEF2C cannot bind to the ANF SRE, the most likely candidate for conferring p38-mediated induction of ANF through this element is SRF. This suggests an as yet uncharacterized mechanism where SRF itself, and/or some other non-Ets-related accessory protein, serves as the target for p38 and/or p38-activated kinases. Accordingly, experiments were carried out to evaluate whether SRF itself could serve as a downstream target for p38. When recombinant SRF was incubated with p38 in the presence of [␥-32 P]ATP no apparent phosphorylation was observed (Fig. 3A, lanes 2 and  3). Accordingly, the possibility that p38, or a downstream kinase, might phosphorylate a non-Ets-related SRF-binding partner was considered. One such protein is ATF6, a 670-amino acid member of the b-ZIP, ATF/CREB family recently shown to bind SRF, but not SREs, and to enhance SRF-mediated transcription (19). Interestingly, when ATF6 was incubated with p38 it was phosphorylated (Fig. 3A, lanes 6 and 8 -10), consistent with a role for ATF6 in p38-mediated transcriptional induction.
Further experiments were carried out to determine whether a kinase downstream of p38 could be involved with either SRF or ATF6 phosphorylation. The MAP kinase-activated protein kinases (MAPKAP-K) are members of a kinase family known to be activated specifically by p38 (35,36). The most recently identified MAPKAP kinase family member, MAPKAP kinase-3, also known as 3pK, has been shown to be phosphorylated and activated by p38 (23, 37). However, when either ATF6 or SRF were incubated with a constitutively active form of MAPKAP kinase-3, MAPKAP kinase-3 (Glu), no phosphoryla-tion was apparent, even though heat shock protein-27, a well characterized substrate of MAPKAP kinase-3 (23), was clearly phosphorylated (Fig. 3B). These results are consistent with the possibility that ATF6 may serve as a substrate for p38.
ATF6 Antisense-To test whether ATF6 is involved in p38activated transcription through the ANF SRE, an expression construct encoding ATF6 in the antisense orientation was employed. This construct has previously been used in HeLa cells to demonstrate the dependence of serum activation of SRF on ATF6 (19). Myocardial cells were co-transfected with MKK6 (Glu), various levels of the ATF6 antisense construct, and ANF 65 (ANF SRE). Interestingly, the ATF6 antisense construct (ATF6-AS) displayed a dose-dependent ability to inhibit FIG. 1. Mapping p38-inducible elements in the ANF promoter. Panel A, diagram of luciferase reporter constructs: the luciferase reporter constructs used to map the p38-inducible element(s) in the ANF 5Ј-flanking sequence are shown. Regions depicted as white bars represent native rat ANF promoter sequences and the SRE-like core region (SRE consensus ϭ CC(A/T) 6 GG) is single underlined in each construct. 1) ANF 134 is comprised of rat ANF 5Ј-flanking sequences extending to Ϫ134 bp from the ANF transcriptional start site, defined as ϩ1, fused to luciferase. In construct 2, ANF 134 (M ANF SRE), the positions shown in bold and lowercase were mutated to distrupt the SRE core. Construct 3, ANF 65, is a truncated version of ANF 134, comprised of rat ANF promoter sequences extending to Ϫ65 bp from the ANF transcriptional start site fused to luciferase (9). Construct 4, ANF 65 (ANF SRE), possesses ANF sequences from Ϫ125 to Ϫ102 (see construct 1), which contains the core ANF SRE and a small amount of flanking sequence, cloned as a 3 times concatomer in front of ANF 65. Construct 5, ANF 65 (c-fos SRE), has the c-fos SRE core and the same number of flanking nucleotides as ANF65 (ANF SRE) cloned as a 3 ϫ concatomer in front of ANF 65; the Ets domain is double underlined. In construct 6, ANF 65 (M c-fos SRE), the Ets domain in construct 5 has been mutated (lowercase bold) so Elk will not bind, and in construct 7, ANF 65 (M ANF SRE), the SRE core in ANF 65 (ANF SRE) has been mutated to distrupt SRF binding. Constructs 4 -7 all contain 3 ϫ concatomers of the SRE-related sequences shown cloned in front of ANF 65/luciferase. At the bottom the ANF-and c-fos SREs are aligned for comparison purposes. Panel B, effects of MKK6 (Glu): the constructs diagrammed in Panel A, and SV40-␤-galactosidase, were each co-transfected into primary myocardial cells Ϯ MKK6 (Glu). After the 18-h cell plating period, the media were replaced with serum-free media and the MKK6 (Glu)-transfected cells were treated Ϯ SB 203580 (5 M) for 48 h, extracted and luciferase and ␤-galactosidase reporter assays carried out as described (9,15,17). Luciferase values were normalized with ␤-galactosidase values and plotted as relative luciferase compared with the control for each construct. Each value is the mean (n ϭ 3) Ϯ S.E. This experiment is representative of at least three identical replicate experiments.
MKK6-activated reporter expression from ANF 65 (ANF SRE) (Fig. 4), however, it had no inhibitory effect on the ability of MKK6 to activate ANF through an isolated Sp1-like element (not shown). These results indicate that overexpression of the ATF6 antisense transcript can inhibit p38-inducible transcription specifically through the ANF SRE, implying a role for ATF6 in this induction process.
ATF6/Gal4 Trans-activation-Further studies were carried out to assess whether ATF6 can serve as a p38-inducible transcriptional activator. ATF family members investigated to date commonly possess an N-terminal trans-activation domain, however, such a domain has not been identified in ATF6. Moreover, there is very little homology between ATF6 and other ATF family members in the N terminus of the protein. Accordingly, we tested whether ATF6 could confer transcriptional activation to a DNA-binding protein using a Gal4 trans-activation system (38). An expression construct was prepared wherein the Gal4 DNA-binding domain was fused to either the full-length, 670-amino acid ATF6, creating Gal4/ATF6 (670), or to an N-terminal fragment of ATF6 comprised of amino acids 1-273, creating Gal4/ATF6 (273) (Fig. 5A). Myocardial cells were then co-transfected with Gal4/ATF6 (670), or Gal4/ATF6 (273) and pG5E1bLuc, a luciferase reporter construct possessing five Gal4 upstream activating sequences driving luciferase expression (38), and either an empty vector control (CMV6) or MKK6 (Glu). When fused to the Gal4 DNA-binding domain, both the full-length and truncated forms of ATF6 were able to confer significant enhancement of trans-activation to the relatively inactive Gal4 DNA-binding protein (Fig. 5B, left panel). For the experiments shown in Panels A and B, protein⅐DNA complexes were separated by PAGE as described previously; the resulting gels were dried, exposed to autoradiographic film, and the developed film scanned and printed from Claris Draw Pro, and as described previously (9). The identification of MEF2C⅐DNA and SRF⅐DNA complexes on the gels was established by the competition pattern and by using anti-MEF2C or anti-SRF antisera in supershift analyses (see Ref. 9). The bands that appear beneath the SRF⅐SRE complexes in Panel B are the results of nonspecific binding of proteins from the in vitro transcription/translation reagent mixture. This was verified by carrying out EMSAs using proteins derived from an in vitro transcription/translation reaction performed using a DNA template without an insert.

FIG. 4. Effects of antisense ATF6 on ANF promoter activity.
Primary myocardial cells were transfected with ANF 65 (ANF SRE), SV40-␤-galactosidase, Ϯ MKK6 (Glu), Ϯ various levels of ATF6-AS, the latter of which codes for an antisense form of the ATF6 (19). Cultures were maintained for 48 h in serum-free media and then the cells were extracted, analyzed for reporter enzyme activity, and the results plotted as described in the legend to Fig. 1.
This result verified that ATF6 operates as a transcription factor in the cardiac context and that it is capable of responding to activation of the p38 MAPK pathway. Consistent with this hypothesis, recombinant ATF6 (273) was phosphorylated by p38 in vitro (Fig. 5B, left panel inset) and the relative levels of phosphorylation of ATF6 (273) and ATF6 (670) correlate with their relative abilities to trans-activate.
To establish a connection between known myocardial cell growth factors and ATF6 trans-activation, the ability of Gal4/ ATF6 (273) to respond to several test agents that activate p38 in myocardial cells was assessed. The ␣ 1 -adrenergic receptor agonist, phenylephrine (PE), serum (Ser) and the peptide hormone, endothelin-1 (ET-1), three well known myocardial cell growth factors, each activated p38 by 3-6-fold (Fig. 6A) and stimulated Gal4/ATF6 (273) trans-activation (Fig. 6B). The stimulatory effects of each growth factor on ATF6 trans-activation were effectively inhibited by SB 203580 indicating that the p38 pathway plays a major role in this process.
Further experiments were carried out in order to evaluate  1 and 2). Unlabeled ATF6 (273) was incubated Ϯ p38 ϩ [␥-32 P]ATP (lanes 3 and 4, respectively), as was unlabeled ATF6 (670) (lanes 5 and 6, respectively). Right panel, triplicate cultures of primary myocardial cells were transfected with the constructs shown. Following 48 h in culture the cells were extracted and assayed for luciferase and ␤-galactosidase as described under "Materials and Methods." The results are expressed as fold of control, as described above. Each value is the mean (n ϭ 3) Ϯ S.E. This experiment is representative of at least three identical replicate experiments. possible roles for ERK or JNK on ATF6 trans-activation. The constitutively active form of Raf, Raf-BXB, resulted in very little stimulation of ATF6 trans-activation, suggesting that the ERK pathway is unlikely to be a major activator (Fig. 6C). Like MKK6 (Glu), ⌬MEKK4, which has been shown in HEK293 cells to activate JNK (40), but not ERK or p38, resulted in a significant stimulation of ATF6 trans-activation, however, the SB 203580 sensitivity suggested that this effect was mediated through the p38 pathway in myocardial cells. We have found that in myocardial cells, ⌬MEKK4 activates JNK by about 100-fold, as expected, but it unexpectedly activates p38 by about 10 -15-fold (not shown). Thus, the complete inhibition of ⌬MEKK4-stimulated ATF6 by SB 203580 indicates that in myocardial cells JNK is unlikely to be a major activator. Consistent with this hypothesis was the finding that constitutively active forms of Rac or cdc42, both of which activate JNK and not ERK or p38 in myocardial cells (15), had no effect on ATF6 trans-activation. A control using AP-1/ANF 65-driven luciferase verified that Rac could serve as a potent stimulator of transcription through AP-1-binding sites (Fig. 6D) which are elements through which JNK typically activates transcription. Taken together, these results support the hypothesis that in myocardial cells, among the MAP kinases, p38 is the major stimulator of ATF6 trans-activation. DISCUSSION In response to various stimuli, cardiac myocytes undergo an unusual growth program typified by increased cell size, enhanced sarcomeric organization, and augmented cardiac-specific gene expression, but no cell division. Recently, it was shown that many compounds that promote hypertrohic cardiac myocyte growth stimulate p38 and that the overexpression of MKK6 (Glu), which specifically activates p38 in cardiac myocytes, leads to the replication of all three features of this novel, hypertrophic growth program, including a robust activation of cardiac-specific transcription (15). In the present study it was shown that p38-mediated transcriptional induction of one such cardiac gene (ANF) requires a promoter proximal SRE. Since p38 had recently been demonstrated to phosphorylate and activate the cardiac myocyte-enriched transcription factor MEF2C (22), and since MEF2C was also shown to be activated by p38 in cardiac myocytes (15) and could conceivably bind to the ANF SRE, we first investigated the possibility that it was through MEF2C that p38 enhanced ANF transcription. However, the inability of MEF2C to bind to the ANF SRE led to the alternate hypothesis that p38-mediated trans-activation was SRF-dependent but occurred through an as yet unidentified mechanism.
The recent discovery that a new ATF family member, ATF6, which is present in the heart, binds to SRF and participates in SRF/SRE-enhanced transcription in response to serum (19), led us to evaluate whether ATF6 might serve as a convergence point for p38. When studied in vitro, ATF6 was phosphorylated by p38, while SRF was not. However, MAPKAP kinase-3 (23, 24), a kinase downstream of p38 (23) which is known to phosphorylate another ATF family member, cAMP response element-binding protein, in response to p38 activation (42), was incapable of phosphorylating ATF6 in vitro. Accordingly, it appears as though p38 might augment SRE-mediated ANF induction by directly phosphorylating ATF6. Our preliminary experiments have shown that ATF6 can bind to SRF but does not appear to bind to SREs, indicating that it is probable that SRF serves as an anchoring protein with which ATF6 can partner to confer p38-inducibility to ANF transcription.
To study ATF6 trans-activation in myocardial cells we developed a reporter system that takes advantage of the fact that DNA-binding domains and trans-activation domains of transcription factors can often times be interchanged. Accordingly, expression constructs were prepared which encoded chimeric proteins which possessed the Gal4/DNA-binding domain fused to either the full-length or an N-terminal fragment of ATF6. When these fusion proteins were expressed in cardiac myocytes, along with a luciferase reporter driven by the Gal4 upstream activating sequence, which binds the Gal4/DNA-binding protein, trans-activation by the overexpressed ATF6 could be studied in isolation from endogenous ATF6. It was found that either the full-length (670 amino acids) or the N-terminal 273 amino acids of ATF6 were capable of conferring transactivation in response to MKK6 (Glu), p38, or several myocardial cell growth factors. This ATF6 trans-activation required kinase-active MKK6 and p38, but was unaffected by constitu-FIG. 6. Effects of hypertrophic stimuli and upstream MAP kinase pathway activators on p38 and ATF6 trans-activation. Panel A, effects of hypertrophic stimuli on p38: primary myocardial cells were treated for 30 min with either no additions (Con), 10 M phenylephrine (PE), 10% fetal bovine serum (Ser), or 10 nM endothelin-1 (ET-1). The cultures were then extracted and analyzed by Western blotting for p38 activation state using a phospho-p38-specific antiserum; blots were also stripped and re-probed for total p38 levels. After densitometeric analyses, the values for phosphorylated-p38 were normalized to those obtained for total p38 and the results presented as fold of control. These methods have been described elsewhere (15). Each value is the mean (n ϭ 3 cultures) Ϯ S.E. Panel B, effects of hypertrophic stimuli on ATF6 trans-activation: myocardial cells were transfected with Gal4/ATF6(273), pG5E1b/Luc, and SV40-␤-galactosidase and then treated as described in the legend to Fig.  1 except the cultures were incubated for 6 h with either no additions (Con), 10 M phenylephrine (PE), 10% serum (Ser), or 1 nM endothelin-1 (ET-1) Ϯ SB 203580. After incubation with these agents the cultures were extracted and assayed for luciferase and ␤-galactosidase enzyme activities as described under "Materials and Methods." Results are shown as luciferase/␤-galactosidase (Rel Luc). Each value is the mean (n ϭ 3) Ϯ S.E. Panel C, effects of upstream MAP kinase pathway activators on ATF6 trans-activation: myocardial cells were transfected with Gal4/ATF6(273), pG5E1b/Luc, SV40-␤-galactosidase and expression constructs for CMV6 (Con), Raf BXB (Raf), ⌬MEKK4 (M4), MKK6-Glu (M6), Rac Val-12 (Rac), and cdc42 Leu-61 (Cdc) and incubated Ϯ SB203580, as described in the legend to Fig. 1. Luciferase values were normalized with ␤-galactosidase values and plotted as relative luciferase compared with the control for each construct. Each value is the mean (n ϭ 3) Ϯ S.E. Panel D, effects of Rac Val-12 on AP-1-enhanced trans-activation: myocardial cells were transfected with AP-1/ANF-Luc (made from ANF 65 (see Fig. 1) with 5 AP-1 consensus sequences cloned upstream), SV40-␤-galactosidase either CMV6 (control construct) or Rac Val-12 (test construct). All culturing, incubation times, and reporter assays were performed as in Panel B. Each value is the mean (n ϭ 3) Ϯ S.E. tively active MAPKAP kinase-3. Interestingly, ATF6 transactivation was strongly stimulated by upstream activators of the p38 pathway, but was poorly stimulated by the ERK or JNK pathways.
In summary, this study has shown that ATF6 likely plays a key role in the induction of cardiac-specific genes, like ANF, during the hypertrophic growth program. Future studies devoted to determining the roles of ATF6 in the induction of other cardiac genes involved in hypertrophic growth will reveal a great deal about this novel SRF-mediated transcription mechanism. Additionally, it will be of interest to evaluate the roles of ATF6 in non-cardiac cell types in response to growth stimulation, as well as determining how ATF6 and other TCFs, such as Elk-1 or Sap-1, collaborate to regulate transcriptional induction through SREs that possess flanking Ets motifs.