GATA-4 and Serum Response Factor Regulate Transcription of the Muscle-specific Carnitine Palmitoyltransferase I b in Rat Heart*

Transcriptional regulation of nuclear encoded mitochondrial proteins is dependent on nuclear transcription factors that act on genes encoding key components of mitochondrial transcription, replication, and heme biosynthetic machinery. Cellular factors that target expression of proteins to the heart have been well characterized with respect to excitation-contraction coupling. No information currently exists that examines whether parallel transcriptional mechanisms regulate nuclear encoded expression of heart-specific mitochondrial isoforms. The muscle CPT-I b isoform in heart is a TATA-less gene that uses Sp-1 proteins to support basal expression. The rat cardiac fatty acid response element ( 2 301/ 2 289), previously characterized in the human gene, is responsive to oleic acid following serum deprivation. Deletion and mutational analysis of the 5 * -flank-ing sequence of the carnitine palmitoyltransferase I b (CPT-I b ) gene defines regulatory regions in the 2 391/ 1 80 promoter luciferase construct. When deleted or mutated constructs were individually transfected into cardiac myocytes, CPT-I/luciferase reporter gene expression was significantly depressed at sites involving a putative MEF2 sequence downstream from the fatty acid response element and a cluster of heart-specific regulatory regions flanked by two Sp1 elements. Each site demonstrated binding to cardiac nuclear proteins and competition specificity (or supershifts) with oligonucleotides and antibodies. Individual expression

Expression of nuclear and mitochondrial encoded expression of respiratory chain subunits occurs despite physical separation of transcriptional events within separate genomes. Stim-ulation and coordination of mitochondrial gene expression from these two sites is accomplished by the nuclear respiratory factors, NRF-1 and NRF-2 (1,2). Using electrical stimulation to produce hypertrophic growth of neonatal cardiac myocytes, the transcriptional activation of cytochrome c is preceded by induction of NRF-1 mRNA (3). This observation is consistent with NRF-1 induction as a prerequisite for synthesis of respiratory chain components. These basic insights into cellular factors that link nuclear events to mitochondrial gene activation are critical for adaptation to environmental stresses and the necessity for enhanced energy production.
In contrast to subcellular coordination of mitochondrial biogenesis and respiratory chain synthesis, less is known concerning tissue-specific transcriptional regulation of nuclear encoded genes involved in energy metabolism. These genes are particularly important in cardiac muscle where contractile activity must be supported by a high rate of aerobic ATP production. There are examples of muscle-or heart-specific mitochondrial proteins that contain sites for ubiquitous transcription factors as well as striated muscle-specific motifs in the proximal promoter, e.g. muscle-specific cytochrome oxidase genes (4). Increases in contractile activity induced by electrical stimulation of neonatal rat cardiac myocytes in culture results in increased mRNA levels of the striated muscle-specific, energymetabolizing enzymes, carnitine palmitoyltransferase I ␤ (CPT-I␤) 1 (3), and adenylate-succinate synthase I, a component of the purine nucleotide cycle (5).
CPT-I␤ (the muscle isoform) has been cloned and is the predominant isoform in rat heart and the sole enzyme expressed in skeletal muscle (6). CPT-I␣ is expressed in most tissues, including the heart. The latter isoform is present at very low levels in the adult cardiac myocyte and is absent in skeletal muscle (6). CPT-I is an example of an enzyme for which the isoforms are very different kinetically. The muscle isoform exhibits an affinity for carnitine that is at least an order of magnitude higher in K m and a K I for malonyl-CoA that is an order of magnitude lower than the K I measured for CPT-I␣ (6). High expression of the muscle isoform of CPT-I is characteristic of adult heart, so that this particular isoform appears adapted for efficient derivation of energy from long chain fatty acids in an active contracting myocyte.
Very little is known about the heart-specific elements that drive transcription of CPT-I␤ in heart. Recent work has dem-onstrated the presence of a fatty acid response element (FARE) in the human CPT-I␤ promoter (7). After exposing neonatal rat cardiac myocytes to serum-free conditions, addition of exogenous oleate increases expression of the human CPT-I␤ reporter gene 8 -20-fold and CPT-I␤ mRNA levels rise 4 -5-fold (7). Peroxisome proliferator-activated receptor ␣ and the retinoid X receptor act to activate CPT-I␤ through the FARE site. Peroxisome proliferator-activated receptor ␣ may also play a pivotal role in the expression of enzymes of ␤-oxidation (8). The physiological impact of fatty acid induction of CPT-I␤ in heart is less certain since cardiac CPT-I is resistant to fasting, a condition that enhances serum fatty acid concentrations (9,10).
We hypothesized that factors that increase expression of tissue specific proteins involved in contractility and energyutilizing reactions in heart would also be important in regulating the expression of an enzyme involved in transformation of its major energy substrate, long chain fatty acids. Our studies demonstrate for the first time that CPT-I␤ is regulated by the muscle-specific factor, GATA-4, and by combinatorial interactions between GATA-4 and the nuclear factor, serum response factor (SRF). Interaction between MADS box and C4 zinc finger proteins represents a novel coregulator mechanism of the cardiac actin promoter (11). SRF is especially abundant in embryonic and adult cardiac, skeletal and smooth muscle cells (12)(13)(14)(15). The recent homologous recombinant knockout of the murine SRF gene locus demonstrated that SRF is absolutely required for the appearance of mesoderm and muscle lineages during mouse embryogenesis (12). The identification of CPT-I␤ as a GATA-4-dependent gene that is coactivated by SRF is the first suggestion that this paradigm may represent an important mechanism by which expression of cardiac specific genes is synchronized between subcellular compartments.

EXPERIMENTAL PROCEDURES
Cell Culture and Transfections-Primary cultures of neonatal rat cardiac myocytes were prepared as described previously (16). Cells were plated at a density of 6 ϫ 10 5 cells/well in six-well plates (Primeria, Fisher), and maintained in Dulbecco's modified Eagle's medium (DMEM, CellGro, Fisher) with 1% penicillin/streptomycin (Life Technologies, Inc.) and 10% calf serum (HyClone Laboratories, Inc.). The cells were incubated at 37 o in the presence of 95% O 2 and 5% CO 2 for 36 h before transfection. Myocytes were transfected using calcium phosphate precipitation in the presence of serum as described previously (18). The calcium phosphate precipitate contained 1.0 g of CPT-I␤ firefly luciferase vector and 0.25 g of a CMV-driven Renilla luciferase expression vector (Promega) as a control for transfection efficiency. Six hours following transfection, the myocytes were washed twice with phosphate-buffered saline and maintained in DMEM with serum for an additional 48 h. For treatment with oleic acid, the myocytes were transfected with the CPT-I␤ (Ϫ318/ϩ80) reporter gene and maintained in DMEM ϩ serum for 12 h, washed with phosphate-buffered saline and incubated in serum-free DMEM Ϯ 0.5 mM oleate (2:1 molar ratio oleate: bovine serum albumin) for an additional 24 -30 h. CV-1 monkey kidney fibroblasts (ATCC no. CCL-70) were maintained as above but were trypsinized 24 h before transfection and plated to reach 85% confluence by the start of the experiment. CV-1 cells were transfected using Lipo-fectAMINE Plus reagent system (Life Technologies, Inc.) in serum-free medium. After 3 h, the transfection medium was replaced with fresh serum-containing medium for 48 h. Cotransfections included 1.0 g of the wild-type, truncated, or mutant CPT-I␤ firefly luciferase reporter constructs and various combinations of the following CMV expression vectors: 0.4 g of Nkx2.5, 0.4 g of GATA-4, 0.1 g of SRF, 0.1 g of SRF-⌬C, and 0.1 g of SRF-pm. Total DNA for each transfection was corrected to a final concentration of 2.0 g by addition of empty CMV vector. Total protein in each well was measured using the BCA protein assay reagent kit (Pierce). CPT-I␤ firefly luciferase was corrected for protein and normalized to that of the CMV/Renilla expression for each separate experiment.
Electrophoretic Mobility Shift Assays (EMSAs)-Nuclear extracts from primary rat neonatal cardiac myocytes were prepared as described previously (17). Double-stranded DNA probes containing sequences from the rat CPT-I␤ promoter were synthesized by Operon Technolo-gies, Inc. (Alameda, CA) as shown in Table I. EMSA reaction mixtures included 2-15 g of nuclear extract, 25 mM Hepes, 100 mM KCl, 0.1% Nonidet P-40 (v/v), 1 mM dithiothreitol, 5% glycerol, and 50 ng of poly(dI⅐dC) as a nonspecific competitor in a 20-l reaction volume. After incubation for 10 min at room temperature, 0.3 ng of radiolabeled probe was added and the reaction incubated for 20 min. When included, specific antibodies (Santa Cruz) or 100-fold molar excess of cold probe was added during the first incubation. Protein-DNA complexes were separated on a 4% nondenaturing polyacrylamide gel at 25°C.
Plasmids and Constructs-Construction of the rat CPT-I␤ promoter fragment Ϫ391/ϩ80 has been reported previously (18). The expression vector pCMV-Nkx 2.5 was a gift from Dr. Janet Mar. Construction of the expression vectors for pCGN-SRF, pCGN-SRFpm, pCGN-SRF⌬C, and pCDNA3-GATA-4 has been described (15). A series of CPT-I␤ promoter constructs with the 5Ј end between Ϫ315 and Ϫ31 and the 3Ј end from Ϫ1 to ϩ80 were created by PCR amplification from the plasmid template, Ϫ361/ϩ80. Specific primers were designed with an artificial restriction site and the desired region of the rat CPT-I␤ promoter. For deletion of the MEF2 region contained in the Ϫ306/ϩ80 construct, 10 bp (Ϫ281 to Ϫ270) were deleted from the artificial primer. The PCR products were restricted and cloned into the multiple cloning site of the promoterless firefly luciferase vector, pGL3 basic (Promega, Madison, WI). The inserted CPT-I␤ fragment in each construct was confirmed by DNA sequencing. The Renilla expression vector, pRL, was purchased from Promega. Mutations were introduced to the Ϫ391/ϩ80 CPT-I construct with the QuickChange site-directed mutagenesis kit (Stratagene, La Jolla, CA). Briefly, high performance liquid chromatography-purified primers containing the desired mutation were used in concert with the high fidelity polymerase Pfu-turbo for PCR amplification of the Ϫ391/ϩ80 plasmid. Wild-type template was digested with DpnI, and the remaining mutated product was transformed into Escherichia coli and sequenced for correct insertion of the mutant primer.
Quantification of CPT-I␤ Transcripts-Specific quantitative assay of rat CPT-I␤ was performed using real time PCR (7700 Prism, PerkinElmer Life Sciences) based on hydrolysis of a specific fluorescent probe at each amplification cycle by the endonuclease activity of Tac polymerase as described previously (19). The sequence for the musclespecific CPT-I was obtained from GenBank (accession no. D43623, nucleotide numbers 869 -889 (forward primer) and 932-952 (reverse primer)). The level of transcripts for the constitutive housekeeping gene product, cyclophilin, was quantitatively measured for each sample to control for sample-to-sample differences in RNA concentrations. The PCR data are reported as the number of transcripts per number of cyclophilin mRNAs (19).
Statistics-Each experiment was performed in triplicate, and the reported values represent the mean of three to five separate cultures Ϯ standard error. The significance of the differences was determined using Student's t test for nonpaired and paired variates (SigmaPlot statistics software).

RESULTS
A region within 391 bases 5Јof the first exon is sufficient to drive the heart-specific expression of the CPT-I␤ luciferase reporter gene (18). By normalizing the Ϫ391/ϩ80 reporter construct to 1.00, the ratio of the larger genomic fragment, Ϫ1188/ ϩ80, to Ϫ391/ϩ80 is 0.93 Ϯ 0.075. Therefore, we conclude that the promoter elements that regulate CPT-I␤ expression to the heart are within 390 bp of the transcription start site. A FARE was located in the rat gene at Ϫ303 to Ϫ296. When 0.5 mM oleic acid (2:1 molar ratio with bovine serum albumin) was added to serum-free medium and neonatal cardiac myocytes cultured for 24 -30 h, CPT-I␤/luciferase expression was increased from 0.63 Ϯ 0.02 relative luciferase units (in the absence of oleate) to 1.88 Ϯ 0.02 in the presence of oleate. Partial deletion of this region (⌬ bases Ϫ301 to Ϫ293) resulted in a diminished fatty acid effect, i.e. 0.80 Ϯ 0.01 relative luciferase units (in the presence of oleate) from 0.65 Ϯ 0.06 (in the absence of oleate). Electrophoretic mobility shift assays of this region (Table I) demonstrated binding of myocyte nuclear protein to this site, and this binding was competed by 100-fold excess of cold oligonucleotides and antibodies to the binding factor COUP-TF (20), but not by mutant oligonucleotides (data not shown).
To map the regions in the CPT-I␤ gene that influence expression of the luciferase reporter construct, Ϫ391/ϩ80 was progressively deleted from the 5Ј end and analyzed for luciferase expression following transfection into neonatal cardiac myocytes (Fig. 1). Compared with the full-length Ϫ391/ϩ80, deletion of 85 bp to Ϫ306/ϩ80 produced minimal changes in luciferase expression. Deletion of the MEF2 site within the Ϫ306/ϩ80 reporter gene resulted in an overall decrease in luciferase expression of 34%. With further deletion of the 5Јflanking sequence to Ϫ270/ϩ80, luciferase expression progressively decreased by 60%. Subsequent deletion of a consensus E box (Ϫ252/Ϫ247) led to a significant (85%) enhancement of luciferase expression. In the presence of neonatal cardiac myocyte nuclear protein, EMSA analysis of the Ϫ252 E box demonstrated a shift ( Table I) that was competed by 100ϫ wildtype oligonucleotide, but not by 100ϫ mutant oligonucleotide (Fig. 2). The shift was supershifted by antibodies to either USF1 or USF2 and supershifted to a higher molecular weight complex in the presence of both antibodies ( Fig. 2A). The data suggest that the Ϫ252 E box interacts with USF1 and USF2 and may function as a suppressor of CPT-I␤. Deletion of 165 bp of flanking sequence diminishes full-length CPT-I␤ expression by 50% (Ϫ126/ϩ80). This truncation removed one Sp1 site at Ϫ136/Ϫ131 and a GATA site at Ϫ129/Ϫ126; deletion of 15 more bp containing a CA box (Ϫ117/Ϫ112) resulted in a dramatic decrease in reporter gene expression by 90%. Further deletion resulted in partial restoration of activity and an extended deletion produced a minimal promoter containing an Sp-1 site at Ϫ74/Ϫ68 (Fig. 1).
To examine potential regulatory elements in the context of the intact promoter, mutations were targeted to known sequences contained in the deletion constructs that appear to serve as regulators of CPT-I␤ expression. The mutated constructs were transfected into neonatal cardiac myocytes for measurement of luciferase expression. In agreement with the deletion data, mutation of the MEF2 site (Ϫ280/Ϫ271) produced a 46% decrease in reporter gene activity (Fig. 3). Gel mobility-shift assays were conducted to examine nuclear protein binding to double-stranded oligonucleotides corresponding to Ϫ280/Ϫ271 in the CPT-I␤ gene. Two bands were shifted and were identical in mobility to TNT-produced proteins corresponding to MEF2C and MEF2A (Fig. 2B). Both bands were competed by 100ϫ wild-type but not 100ϫ mutant oligomers (Fig. 2B), and both were supershifted by antibodies to MEF2. These data suggest that cardiac specificity of CPT-I␤ expression is partly conferred by the presence of a MEF2 site at the distal end of the promoter construct.
Although mutation of the Sp1 (Ϫ137/Ϫ131) did reduce CPT-I␤ expression by 60%, mutations in the CA box (CACCC) at Ϫ117/Ϫ112 and the Sp1 site (Ϫ74/Ϫ68) produced strong repression of luciferase expression, i.e. 75% and 76%, respectively (Fig. 3). These data suggest that factors binding to these sites are most likely to be responsible for basal transcription of the CPT-1␤ gene. Electrophoretic mobility shift assay of the oligomer containing the Ϫ74/Ϫ68 region demonstrated two bands, both of which were competed by wild-type oligonucleotide but not by mutant oligomers (Fig. 2C and Table I). Similarly, antibodies to Sp1 and Sp3 abolished formation of the protein/DNA complex (Fig. 2C), suggesting that Sp1 and Sp3 proteins bind to the proximal promoter at Ϫ74/Ϫ68 and may also play a role in binding to the CA box at Ϫ117/Ϫ112 (21).
A region flanked by the two Sp1 sites at Ϫ137/Ϫ131 and Ϫ74/Ϫ68 contains a cluster of potential cardiac regulatory elements including two GATA binding sites, an SRE, and one potential Nkx site (Ϫ94/Ϫ88) that partially overlaps with a second GATA binding motif at Ϫ96/Ϫ93. Mutation of the SRE (Ϫ112/Ϫ104) results in a 56% decrease in CPT-I␤ expression in cardiac myocytes (Fig. 3). Serum withdrawal for 44 h dramatically decreases the mRNA concentration of CPT-I␤ in cardiac myocytes (Fig. 4). This decrease in mRNA content is reversed by replacement of serum-free DMEM with media containing 10% bovine calf serum (Fig. 4), supporting a role for serumcontaining factors in the up-regulation of CPT-I␤ transcription. Mutation of either GATA site results in a 43-38% decrement in gene expression (Ϫ129/Ϫ126 and Ϫ 96/Ϫ93, respectively), whereas mutation of both sites (double GATA mutant) causes a slightly greater fall in luciferase activity (55%). The binding of SRF to the SRE (Ϫ112/Ϫ104) and GATA-4 to the two potential GATA sites at Ϫ96/Ϫ93 and Ϫ129/Ϫ126 was confirmed by EMSA (Fig. 2, D and E). Mutation of the Nkx site (Ϫ94/Ϫ88) has no effect on reporter gene expression in cardiac myocytes (Fig. 3).
Although homeodomain factors achieve specificity via protein binding to DNA, Nkx, SRF, and GATA-4 have been demonstrated to exert regulatory control over cardiac gene expression via protein-protein interactions (11,22,23). To study biological activity of these proteins independent of DNA binding, Nkx, GATA-4, and/or SRF expression vectors were cotransfected into CV-1 cells with the full-length CPT-I␤ reporter gene construct. Transfection of Nkx 2.5 or SRF alone produces modest expression of CPT-I␤/luciferase in CV-1 cells (ϳ4-fold activation, Fig. 5). Cotransfection of Nkx and SRF with the promoter gene construct produces additive effects on luciferase expression (Fig. 5A), suggesting that both factors contribute independently to CPT-I␤ gene activation. GATA-4 transfection alone stimulated CPT-I␤ 36-fold, making GATA-4 the most potent tissue-specific regulatory element thus far described for this gene. Cotransfection of GATA-4 and Nkx 2.5 with the CPT-I␤ reporter construct diminished the response of the CPT-I␤ gene to this combination of factors (Fig. 5B). An additive effect would be predicted if both factors were acting independently. It is possible that Nkx/GATA4 protein-protein interactions (22-24) decreased the amount of GATA4 (and Nkx) available for DNA binding to GATA elements on the CPT-I␤

Regulation of Carnitine Palmitoyltransferase I␤ Transcription
gene (25). In contrast to the GATA-4/Nkx cotransfections, coexpression of GATA-4 and SRF produced a synergistic response of the CPT-I␤ gene (Fig. 5C). This combination of factors resulted in a synergism that was significantly different from the predicted value if the actions of these two proteins were independent. These data provide the first example of a nuclear encoded, cardiac mitochondrial gene where SRF and GATA-4 act as mutual coregulators (11).
To confirm the role of the predicted regulatory domains and protein/DNA interactions in reporter gene expression, point mutations in SRE, the two GATA-4 sites (double mutation) and Nkx2.5 in the Ϫ391/ϩ80 CPT-I␤ construct were cotransfected with SRF or GATA-4 alone or in combination. Mutation of the SRE (and the Nkx), but not of the GATA sites, significantly reduced CPT-I␤ induction by SRF (Fig. 6A). The induction of CPT-I␤ reporter expression by GATA-4 alone was significantly FIG. 1. Deletion analysis of the CPT-1␤ ؊391/؉80 promoter fragment. Serial deletions of the Ϫ391/ϩ80 promoter fragment ligated to the luciferase reporter were transfected into neonatal cardiac myocytes. Transfection reactions included 1.0 g of reporter and 250 ng of CMV-driven Renilla luciferase expression vector as a control for transfection efficiency. Following transfection, cells were maintained in DMEM containing 10% fetal calf serum for 48 h. Firefly luciferase was normalized to Renilla and is expressed as a ratio multiplied by 100. Bars represent the means Ϯ S.E. of two to three experiments performed in triplicate.

FIG. 2. Identification of specific transcription factors binding to the CPT-1␤ promoter.
Protein binding to sequences in the CPT promoter was characterized by gel-mobility shift assay. Oligomers containing binding sites from the CPT-1 promoter were end-labeled with [␥-32 P]ATP and incubated with neonatal rat heart-myocyte, nuclear-protein extracts (see "Experimental Procedures"). Unlabeled competitor oligomers or antibodies were included as indicated above each lane. A, labeled oligomers corresponding to Ebox Ϫ252 incubated with rat myocyte nuclear extract (RMNE) and 2 l of either USF1 or USF2 antibodies or 1 l of each for a total of 2 l. B, sequences representing MEF2 Ϫ280 incubated with 4 g of RMNE and MEF2A antibody (2 l) or with TNT proteins (1 l of final TNT reaction). C, labeled probes for the Sp1 Ϫ74 binding site incubated with 10 g of RMNE and 2 l of either Sp1 or Sp3 antibody. D, Labeled oligomers representing GATA-129 and GATA-96 sites incubated with 10 and 15 g of RMNE, respectively; GATA4 antibodies added as indicated. E, labeled probes for the SRE-112 binding site incubated with 5 g of RMNE; SRF antibody (2 l) added as shown.
reduced by ϳ50% by all three point mutations (Fig. 6B). The point mutation in the Nkx site (Ϫ94/Ϫ88) adjacent and overlapping the GATA site likely represents interruption of GATA-4 binding at Ϫ96/Ϫ93. The remaining reporter gene activity in the presence of the GATA double mutation may represent physical association of the transfected GATA-4 with endogenous factors or basal transcriptional complexes in the CV-1 fibroblasts (11,25). The combinatorial effects of SRF and GATA4 on the CPT-I␤ reporter gene were abolished by mutation of either the SRE or the two GATA sites (Fig. 6C). These results reinforce the SRE dependence of these coregulators for synergism (11). Moreover, the results again demonstrate that dramatic induction in CPT-I␤ gene expression is also regulated in large part by GATA-4 interactions alone, some of which appear independent of binding to traditional GATA sites. The drop in synergism due to the point mutation in the Nkx site (Fig. 6C) appears to reflect a requirement for appropriate flanking sequences to facilitate GATA-4 interaction with the DNA at Ϫ96/Ϫ93.
Transfection of a DNA binding mutant SRFpm abolishes the synergistic response between GATA-4 and SRF (Fig. 7). This finding is consistent with requirement for SRF binding to the SRE to produce cardiac ␣ actin promoter coactivation (11). Induction of the GATA-dominated CPT-I␤ reporter gene expression remains elevated in the presence of transfected GATA-4 and SRFpm. A small reduction in the normal GATA-4 up-regulation may reflect altered affinities of the gene for the SRFpm/GATA-4 complex, including direct binding of the complex to the GATA site(s). A deletion of the C-terminal activation domain of SRF (SRF⌬C) inhibits the combinatorial action of SRF and GATA-4 and dramatically reduces the induction of gene expression due to GATA-4 (Fig. 7). The promoter activation that remains (13-fold) is on the order of induction suggestive of independent GATA binding to the DNA/protein complexes (Fig. 6B, 15-fold induction). The data confirm coactivation of CPT-I␤ by GATA-4 and SRF that augments the enhancement of gene expression seen in the presence of GATA-4 alone.

DISCUSSION
Basal expression of the CPT-I␤ gene is analogous to the CPT-I␣ gene in that it contains no TATA box and uses Sp proteins 1 and 3 to drive expression (26). The factors that identify and separate the tissue location of these two isoforms are of particular interest to cardiac muscle where the muscle isoform is dominant, but the liver isoform (CPT-I␣) is also expressed (6). We have identified a region within Ϫ391 base pairs of the transcription start site of the muscle isoform that is sufficient to drive CPT-I␤ transcription and that contains MEF2 and E box elements as well as a cluster of other heartspecific sites. This cluster of CPT-I␤ regulatory elements is located between Ϫ137 and Ϫ68 bp of the proximal promoter and includes GATA, Nkx2.5, and SRF DNA-binding sites. Identification of these important cis-trans regulatory domains suggests a multiplicity of pathways that may be expressed and are potentially interactive depending on the CPT-I␤ gene context and the developmental environment of the myocyte.
The GATA family of transcription factors, represented by heart-specific GATA-4, are potent transactivators of several cardiac promoters. These include atrial natriuretic factor (27), cardiac sarcolemmal Na ϩ ,Ca ϩ exchanger (28), B-type natri- FIG. 4. Quantitative RT-PCR of CPT-1␤ transcripts in response to serum deprivation and restoration. Total RNA from rat neonatal cardiac myocytes was isolated and analyzed by quantitative RT-PCR. छ, freshly isolated myocytes were maintained in DMEM with 10% serum for 12 h. The medium was then removed and replaced with serum-free DMEM for an additional 8, 32, or 44 h. ࡗ, freshly isolated myocytes were placed in serum-free medium immediately after plating for 12, 20, or 44 h. Fresh DMEM with 10% serum was added, and the cells were maintained for an additional 44, 36, or 12 h for a total of 56 h/treatment. Results represent CPT-I␤ transcripts normalized to cyclophilin mRNA and are presented as mean Ϯ S.E. of triplicate determinations. uretic peptide (29), and cardiac troponin I (30). GATA-4 (and GATA-6) colocalize in postnatal cardiomyocytes and are believed to act in the differential control of various cellular processes (29). The present results are the first to identify a heart-specific protein, carnitine palmitoyltransferase I␤, as a downstream mitochondrial target of GATA-4. Two GATA binding sites are present in this gene although transcriptional activation by GATA does not always require DNA binding (25). Together with Nkx 2.5, GATA-4 has been demonstrated to be a powerful transcriptional coactivator of the ANF promoter (31) and the cardiac ␣ actin promoter (23). In the context of the CPT-I␤ gene, however, coexistence of GATA-4 and Nkx 2.5 decreases reporter gene expression in CV-1 cells. Compared with a significant, but small, transactivation by Nkx 2.5, GATA-4 expression in CPT-I␤-transfected CV-1 cells produces a greater than 30-fold increase in CPT-I␤ gene reporter activity, indicating GATA domination of the CPT-I␤ promoter. Nkx 2.5 may act to sequester GATA factors away from GATAbinding sites in GATA-dependent promoters (25). A DNA-independent interaction between GATA and Nkx 2.5 could therefore remove GATA-4 from its interaction sites in CPT-I␤, resulting in the decreased reporter gene expression observed. FIG. 6. Induction and synergistic activation of the CPT-1 promoter by SRF and GATA4 is reduced with DNA binding site mutations. Subconfluent CV-1 cells were transfected with 1 g of wild-type or mutant reporter constructs with expression vectors for SRF and GATA4 alone or in combination. Mutated promoters are schematically represented to the left. Total DNA concentrations were adjusted to 2 g with empty pcDNA3 vector. Cells were harvested 48 h after transfection, and luciferase activity was recorded and normalized to protein content and Renilla expression. Bars represent mean Ϯ S.E. of three to five experiments in triplicate. Except for the SRF induction of the CPT-I␤-GATA double mutant, all results are significantly decreased compared with wild-type reporter activity (0.01 Ͼ p Ͼ 10 Ϫ5 ).
In cardiac myocytes, the double GATA site mutation reduces reporter gene expression by greater than 50%, supporting the importance of DNA binding in the cumulative GATA effects. Cotransfection of the reporter gene with the double GATA mutant into CV-1 cells also reduces reporter gene expression by GATA-4 by 43%, but significant gene induction is still retained. These data again suggest that GATA-4 may also influence CPT-I␤ gene expression by protein-protein interactions, interacting with endogenous levels of factors that affect basal transcription (25). Among these factors, SRF has recently been shown to be a mutual coregulator with GATA-4 in numerous myogenic SRE-dependent promoters (11). In the absence of GATA-DNA binding, SRF binds to the CArG box sequence and physically associates with GATA-4 through the MADS box of SRF and the second zinc finger domain of GATA-4 (11). In the CPT-I␤ promoter, the presence of serum has dramatic effects on CPT-I␤ mRNA content in the neonatal myocyte cultures. SRF binds to one SRE (Ϫ112/Ϫ104), and mutation of this site also dramatically decreases CPT-I␤ reporter gene expression. Although cotransfection of SRF with CPT-I␤ has small inductive effects comparable to those seen with Nkx 2.5, the presence of GATA-4 factors synergistically activates CPT-I␤ to levels that are significantly greater than expression of GATA-4 and SRF independently. The data suggest that even in a background of high GATA-4-induced gene expression, SRF binding to SRE and association with GATA-4 is a quantitatively important mechanism to up-regulate further expression of the muscle isoform of CPT-I. It is possible that the postnatal expression of CPT-I␤ is dependent on the presence of serum factors and GATA-4 interacting physically and functionally to increase the myocardial content of this isoform to adult levels (6). Supporting a possible physiological importance of this mechanism, we have demonstrated that electrical stimulation of neonatal cardiac myocytes in culture induces CPT-I␤ mRNA accumulation (3) subsequent to up-regulation of GATA-4 gene expression (5).
A requirement for MEF2 in CPT-I␤ gene expression is consistent in the known role for the MEF2 proteins in the differentiation of muscle cell lineages. MEF2 has also been reported to mediate hypertrophic signaling as well as synergistic transcriptional responses (32). These pathways gain additional significance in the heart where cardiac hypertrophy and failure is linked to down-regulation of the enzymes of the ␤-oxidation pathway (33), including CPT-I␤ (34). The CPT-I␤ promoter contains a MEF2/DNA binding site, and mutational analysis reveals a 46% depression in CPT-I␤ gene expression when the MEF2 site is mutated. Nuclear protein extracts of the neonatal cardiac myocytes demonstrate binding of both MEF2A and MEF2C in electrophoretic mobility shift assays. The ability of the MEF2 proteins to activate transcription in vivo depends on the dimer composition of the binding complex and the cellular context.
Twenty base pairs downstream of the MEF2 binding element, we have identified an E box that acts as a suppressor of CPT-I␤ in the promoter deletion analysis. The consensus E box binds basic helix-loop-helix regulatory proteins and are contained in the regulatory regions of most developmentally controlled, muscle-specific genes. In the CPT-I␤ gene, we have identified the E box-binding proteins as the upstream stimulatory factors, USF1 and USF2. Although a role for rat USF1 has been suggested in contractile-mediated activation of ␣ myosin heavy chain gene (35), USF can either positively or negatively regulate promoter activity via independent cis regulatory elements (36). E boxes are also frequently associated with adjacent MEF2 sites with a spacing that promotes protein-protein interaction between E box and MEF2 basic helix-loop-helix factors bound to the DNA. The interaction of this E box site with MEF2 and adjacent sites in the promoter is currently under investigation.
Finally, the presence of a FARE site in the rat CPT-I␤ promoter has been confirmed by these studies. The activity of this site in our hands produces a small induction of CPT-I␤ gene expression (2-3-fold) at high physiological concentrations of oleate (2:1 molar ratio) in the cell medium. This fold induction is less than previously reported at a 7:1 oleate to albumin molar ratio (8 -20-fold) (7). Although a role for fatty acids in the induction of CPT-I␤ gene transcription is an attractive regulatory mechanism physiologically, other studies have not been able to demonstrate a role for elevated serum fatty acids in altering the cardiac content of CPT-I␤ mRNA and protein (9,10). These studies suggest that, like other heart-specific proteins, CPT-I␤ contains the same pattern of muscle-specific control regions. Its expression in the heart is likely GATA-4dominated and is subject to protein-protein interactions that can regulate its expression in a manner that is context-dependent. Interestingly, these studies also verify that gene expression of a major controlling enzyme in mitochondrial oxidative metabolism is coordinate with expression of proteins important in contractile function and energy consumption.