Characterization of a Cardiac-specific Enhancer, Which Directs α-Cardiac Actin Gene Transcription in the Mouse Adult Heart*

Expression of the mouse α-cardiac actin gene in skeletal and cardiac muscle is regulated by enhancers lying 5′ to the proximal promoter. Here we report the characterization of a cardiac-specific enhancer located within -2.354/-1.36 kbp of the gene, which is active in cardiocytes but not in C2 skeletal muscle cells. In vivo it directs reporter gene expression to the adult heart, where the proximal promoter alone is inactive. An 85-bp region within the enhancer is highly conserved between human and mouse and contains a central AT-rich site, which is essential for enhancer activity. This site binds myocyte enhancer factor (MEF)2 factors, principally MEF2D and MEF2A in cardiocyte nuclear extracts. These results are discussed in the context of MEF2 activity and of the regulation of the α-cardiac actin locus.

Expression of the mouse ␣-cardiac actin gene in skeletal and cardiac muscle is regulated by enhancers lying 5 to the proximal promoter. Here we report the characterization of a cardiac-specific enhancer located within ؊2.354/؊1.36 kbp of the gene, which is active in cardiocytes but not in C2 skeletal muscle cells. In vivo it directs reporter gene expression to the adult heart, where the proximal promoter alone is inactive. An 85-bp region within the enhancer is highly conserved between human and mouse and contains a central AT-rich site, which is essential for enhancer activity. This site binds myocyte enhancer factor (MEF)2 factors, principally MEF2D and MEF2A in cardiocyte nuclear extracts. These results are discussed in the context of MEF2 activity and of the regulation of the ␣-cardiac actin locus.
␣-Cardiac actin is a member of a highly conserved multigene family, found in all eukaryotic cells, of which the isoforms are differentially expressed in a tissue-specific and developmentally regulated manner (1). In mouse, ␣-cardiac actin is expressed in the early myocardium and in somites from embryonic day 8.5. as the skeletal muscle of the myotome begins to form (2). After birth, the ␣-cardiac actin gene remains expressed at a high level in the heart and is down-regulated in skeletal muscles at the same time as ␣-skeletal actin remains expressed in skeletal muscle and is down-regulated in the heart (3)(4)(5).
Several lines of genetic evidence suggest that ␣-cardiac actin is essential for normal structure and function of adult cardiac myocytes. In BALB/c mice, a 5Ј-partial duplication of the ␣-cardiac actin gene is associated with the down-regulation of the ␣-cardiac actin mRNA and protein (6), leading to enhanced expression of ␣-skeletal actin mRNA and protein after birth through a compensatory mechanism (5). However, the hearts of adult BALB/c mice present functional alterations with increased contractility (7). When a null mutation is introduced into the ␣-cardiac actin gene, the mice either do not survive to term or die within 2 weeks of birth because of extensive loss of thin filaments and sarcomere disorganization leading to heart failure (8). Attempts to rescue the deficient cardiocytes by transgenic expression of a non-cardiac actin in such mutant mice causes heart enlargement and dysfunction (8) resembling human idiopathic dilated cardiomyopathy. In humans, mis-sense mutations of the ␣-cardiac actin gene predispose affected cardiocytes to mechanical injury leading to idiopathic dilated cardiomyopathy (9) and eventual cell death. Taken together, these results indicate that both the type of actin expressed in the adult heart and the levels at which it is expressed are important for correct contractile function.
Several regulatory sequences involved in the expression of the ␣-cardiac actin gene during muscle cell differentiation in vitro and the development of mammalian striated muscles in vivo have been identified. Transcriptional activation in differentiating skeletal muscle cells depends on critical sites in the proximal promoter of the human and mouse genes, notably a CArG-box binding serum response factor (10,11), an E-box binding myogenic regulatory factors (10 -13), and an SpI site (10,13). In cardiocyte cultures, the same sites seem to be involved (14,15), although the role of the E-box is controversial (12). Interestingly, it has been shown that the multiple CArGboxes found in the ␣-cardiac actin promoters of several species (16,17) are sites of serum response factor and Nkx-2.5 interaction (18) acting to promote high levels of activity (19). However, previous observations on the BALB/c mouse line suggested that there might be other levels of regulation operating on this gene (6).
Two DNase I hypersensitive sites (HSp 1 and HSd) were identified upstream of the mouse ␣-cardiac actin gene in the C2 skeletal muscle cell line. When these sequences where tested for activity in different cell lines with homologous and heterologous promoters they were shown to act as striated musclespecific enhancers (20). Analysis of the more distal sequence, at ϳϪ7 kbp from the gene, showed that its activity depends on an E-box, a target of myogenic regulatory factors, and also on an AT-rich 3Ј sequence, which binds the homeodomain protein Emb in interaction with MEF2D and p300 (21). This complex potentially plays a role in opening chromatin at the ␣-cardiac actin locus, rendering the E-box and downstream regulatory regions accessible; in transgenic mice when this sequence is present robust expression is seen (22). The proximal promoter itself is a weak regulatory element directing expression to embryonic skeletal muscle and to the embryonic heart in a few transgenic lines. The distal sequence HSd acts as a skeletal muscle enhancer in vivo. Expression in the adult heart is seen with transgenes containing ϳϪ5 kbp of ␣-cardiac actin DNA, which includes the proximal DNase I hypersensitive site, con-sistent with a role for this proximal enhancer region in the transcription of the gene in cardiac muscle.
In this study, we have examined in more detail this aspect of ␣-cardiac actin expression. We report on a third enhancer located within Ϫ2.354 kbp of the gene that directs transgene expression in the adult heart but not in skeletal muscle. In the embryo, it directs expression in the skeletal muscle of the somites, as well as the heart. However, unlike the previously described distal and proximal enhancers, it is not active in the C2 skeletal muscle cell line derived from postnatal muscle but only in cardiocytes. Its activity depends on a MEF2 site.

EXPERIMENTAL PROCEDURES
Reporter Constructs and Site-directed Mutagenesis-A 2.473-kbp BamHI fragment derived from the clone IG 10 of the murine ␣-cardiac actin gene (6) containing the Ϫ2.354, ϩ0.119-kbp region of the gene was inserted into the BamHI site of the promoterless plasmid, pBLCAT6 (23) generating the construct EnC proximal promoter region (POX). Deletions in this sequence were generated either by using convenient restriction sites in the 5Ј region or by using partial digestion. Fragments were also cloned into a CAT construct (20) in which the POX (Ϫ0.667, ϩ 0.119 kbp) of the ␣-cardiac actin gene had been cloned into the promoterless plasmid pBLCAT6. Mutations of the MEF2 site were introduced separately, by site directed mutagenesis (transformer sitedirected mutagenesis kit or QuikChange site-directed mutagenesis, Stratagene), in the Ϫ2.3to Ϫ1.4-kbp region cloned in BlueScript. The two mutations substituted for the native MEF2 site were, respectively, TTTTCTACCCGGGACTGGT (m1) and TTTTCGATTCTTAACTGGT (m2). L. Bold characters indicate mutated nucleotides. Mutated DNA fragments were subsequently sequenced and transferred into POX (Ϫ0.669 kbp) or TkCAT. Multimerized sequences of the native or mutated ␣-cardiac actin MEF2 box were cloned into TkLacZ at a SfiI site introduced previously (provided by F. Relaix), and confirmed by sequencing.
Cell Culture, DNA Transfection, and Assays-The C2/7 line, a subclone of the C2 cell line derived from the skeletal muscle of adult C3H mice, was grown as reported previously (24). For DNA transfection, the cells were cultured in 6-cm diameter dishes with 20% fetal calf serum and allowed to differentiate in 2% fetal calf serum. The embryonic mouse fibroblast cell line C3H10T1/2 was grown in 10% fetal calf serum (24). Ventricular cardiocytes were prepared from hearts of 18 embryonic day Wistar rat fetuses according to the protocol of (25). Hearts were trimmed of the atria and outflow tract, and the ventricular cells were dissociated under gentle shaking in 1ϫ trypsin (T4674 Sigma) in ADS medium complemented with glucose. Trypsin was inhibited by adding decomplemented newborn calf serum (Invitrogen). Cells were recovered by centrifugation and kept at 37°C in newborn calf serum. Fresh enzyme solution was added to the tissue until complete dissociation was reached. Cells were plated in 6-cm dishes in plating medium and grown in 5% CO 2 .
Cardiocytes and myogenic cells were transfected by the calcium phosphate transfection technique (26). The embryonic fibroblast cell line, C3H10T1/2, was transfected with superfect (Qiagen) under the conditions described by the manufacturer. 9 g of CAT reporter construct and 0.5 g of a reporter gene containing the Rous sarcoma long terminal repeat linked to the luciferase gene (RSVluc) were added to each dish. Myotubes were rinsed in phosphate-buffered saline and collected after being maintained 48 h in Dulbecco's modified Eagle's medium supplemented with 2% fetal calf serum as reported elsewhere (24). Cardiocytes were rinsed with phosphate-buffered saline for 60 h following transfection, collected in 40 mM Tris-HCl (pH 7.5), 150 mM NaCl, and 1 mM EDTA, and lysed by three freeze-thaw cycles.
Co-transfection was carried out with 15.5 g of total DNA consisting of 0.5 g of RSVLuc, 10 g of a TkLacZ construct regulated by a concatamerized MEF2 site (x4) from the mouse muscle creatine kinase (MCK) or ␣-cardiac actin genes and 5 g of MEF2A, MEF2C, or MEF2D expression vectors (27). As a control an empty expression vector was used.
CAT activity was measured by the simple phase extraction procedure (28), and luciferase activity was measured as described previously (20). ␤-Galactosidase activity was measured with a galactolight kit (Tropix). The resulting CAT activity of the different constructs was normalized to the corresponding luciferase activity. All transfections were carried out in duplicate, and three different DNA preparations of each construct were tested.

Preparation of Nuclear Extracts and Gel Mobility Shift Assays-
Nuclear extracts from ventricles of adult rat hearts were prepared according to the procedure of Mar and Ordahl (29) with slight modifications. Hearts were trimmed of the atria and outflow tract, washed in sterile phosphate-buffered saline, and homogenized in a Tissue Mizer for 12 s at half-power in homogenization buffer (10 mM Hepes, pH 7.6, 25 mM KCl, 1 mM EDTA, 0.15 mM spermine, 0.5 mM spermidine, 0.4 mM phenylmethylsulfonyl fluoride, 1.8 M sucrose, 5% glycerol), with added protease inhibitor mixture (Roche Applied Science, number 1697498). Liberation of nuclei was checked with trypan blue dye, and the homogenate was centrifuged on a pad of homogenization buffer at 25K for 60 min at 4°C. Nuclei were transferred to a Dounce homogenizer in lysis buffer (10 mM Hepes, pH 7.6, 100 mM KCl, 0.1 mM EDTA, 3 mM MgCl 2 , 1 mM dithiothreitol, 0.1 mM phenylmethylsulfonyl fluoride, 10% glycerol), with added protease inhibitor mixture. Following 6 strokes with an A pestle, NaCl was added to a 0.5 M final volume. The homogenate was left for 30 min on a slow rotating table and then spun for 60 min at 35,000 rpm. The supernatant was dialyzed against 25 mM Hepes, pH 7.6, 100 mM KCl, 0.1 mM EDTA, 1 mM dithiothreitol, 0.1 mM phenylmethylsulfonyl fluoride, and 10% glycerol, with added protease inhibitor mixture, and stored in liquid nitrogen. Protein concentration was determined using the Bio-Rad protein assay.
Where indicated, antibodies or competing oligonucleotides (20 -100fold molar excess) were added to the incubation 10 min before adding the probe. Polyclonal preimmune and anti-MEF2 antibody were obtained from Santa Cruz Biotechnology. Polyclonal antibodies to the three MEF2 isoforms (MEF2A, MEF2C, and MEF2D) (30) were used at a 1:20 final dilution. The mixture was directly loaded on a 5% polyacrylamide gel in 0.5ϫ Tris-buffered EDTA, prerun for 1 h. Then the gels were fixed, dried in a vacuum desiccator, and exposed to x-ray film overnight.
Generation of Transgenic Mice, Screening of the Transgenic Lines, and Histochemical Staining for ␤-Galactosidase-The 2.4-kbp BamHI fragment was inserted at the BamHI site of pAC1SDKnLacZ (n represents a nuclear localization sequence) (22). The insert was extracted by cutting with SacII and XhoI, eluted, purified, resuspended, and injected as described previously (22). A 0.5-cm tail biopsy was digested, and DNA was prepared as reported before (22). In the case of founder animals, ϳ10 g of DNA were digested with EcoRI, separated in an 0.8% agarose gel and transferred onto a Hybond Nϩ membrane, which was treated according to standard procedures (31). Probes (recognizing the ␤-galactosidase sequence or 5Ј regions of the ␣-cardiac actin gene) were labeled using Amersham Biosciences Megaprime RPN 1606 and purified on a Chromaspin 100 column. Filters were hybridized in 50 mM Na 2 HPO 4 , pH 7.6, 5ϫ SSC, 5ϫ Denhardt's solution, 0.1% SDS, 100 g/ml salmon sperm DNA, and washed in 0.1ϫ SSC, 0.1% SDS at 65°C. Transgenic litters were detected by amplification of a 0.9-kbp fragment encompassing the 3Ј-terminal of the ␣-cardiac actin promoter and the 5Ј-end of the nLacZ gene using the primers ␣-cardiac actin, 5Ј-GCT-GCTCCAACTGACCCCGTCCATCAGAGAG and nLacZ, 5Ј-CGCATCG-TAACCGTGCATCTGCCAGTTTGAG. The reaction was carried out for 25 cycles (94°30 s, 57°1 min, 72°45 s) followed by 1 cycle of 72°for 10 min in 30 l of 1ϫ PCR buffer, 2.5 mM MgCl 2 , 0.2 mM dNTP mix, 0.5 M primer mix, 1 unit of Taq DNA polymerase (Invitrogen).
Embryos or organs were dissected, fixed (22), and stained (32) for various lengths of time. Thee copy number was estimated by Southern blotting (22). The transgenic line shown here had a copy number of 2-3.
Comparison of Mouse with Human Genomic Sequences-A 150-kbp contig of human genomic sequence was extracted from the human genome data base using the promoter sequence of the human ␣-cardiac actin gene (16). Different alignments of parts of this sequences with the 2-kbp cognate sequence of the mouse gene were performed. Two methods were used with the DNA Strider TM 1.3f8 program. DNA matrix analysis was performed with variable windows and stringencies. DNA alignment was performed by the blocks method with variable mismatch and gap penalties.

Identification of a Cardiac Muscle Enhancer-
The 5Ј-flanking region of the mouse ␣-cardiac actin gene was re-examined for cardiac as well as skeletal muscle regulatory elements by transfection experiments in primary cultures of cardiac muscle cells (cardiocytes), as well as the C2 skeletal muscle cell line, previously used to identify muscle enhancers (20). The POX of Ϫ770 bp is active in both cell types but is a weak regulatory element in vivo where it does not direct transgene expression to the adult heart. The activity of the CAT reporter gene driven by POX was taken as 1 to compare the effect of upstream sequences, cloned in front of the POX transgene (Fig. 1). The distal enhancer region (HSd), which is active in differentiated skeletal muscle myotubes, shows very low activity in cardiocytes. The proximal enhancer, HSp, is active in cardiocytes as well as in C2/7 myotubes. In this deletion analysis a new enhancer region (EnC), which is active only in cardiocytes, is now identified in the Ϫ2.354to Ϫ0.669-kbp region immediately 5Ј of the proximal promoter.
We had previously shown that ϳϪ5.7 kbp of ␣-cardiac actin upstream sequence (construction T3, (22)) contained the necessary regulatory elements to direct nlacZ transgene expression to the adult heart. We how tested whether Ϫ2.354 kbp was sufficient for this transcriptional activity in vivo. Of four founder mice, one had a rearranged transgene, as shown by Southern blotting, and one was infertile, the two others gave rise to transgenic lines, which expressed the nlacZ transgene in the adult heart. In one case ␤-galactosidase activity was patchy in both atria and the ventricles, with prior expression in the embryonic heart also (results not shown); this expression was lost in subsequent generations. In the other line, all aspects of transgene expression were stable over the four generations examined. Between embryos/mice within a single litter, the pattern was also maintained, with minor variations in the extent of labeling in the right ventricle. Strong expression was seen in the embryonic (Fig. 2, a and b) and adult (Fig. 2, c and  d) heart. In the adult both atria show intense X-gal staining.
Most of the left ventricular myocardium is labeled. Part of the right ventricle is positive, with negative areas of myocardium ventrally. This was also the case at birth and during postnatal development (3 weeks, 3 months, results not shown). This is similar to the expression profile of the T3 transgenic lines, where Ϫ5.746 kbp, which included the HSp and EnC enhancers, was driving the transgene. The results shown here demonstrate that the EnC with the POX promoter is sufficient for expression in the adult heart. Expression in somites, in the skeletal muscle of the myotome, is also seen with this transgene in the strongly expressing line (Fig. 2, a and b). This is probably because of the proximal promoter, which can direct expression in embryonic muscle but not in the adult heart (22). No expression was detected in postnatal (3 weeks) or adult skeletal muscles (results not shown).
Sequence Conservation in the 5Ј-Flanking Region of the ␣-Cardiac Actin Gene-Comparison of DNA genomic sequences located 8.6 kbp 5Ј to the CAP site of the mouse and human ␣-cardiac actin genes was carried out using the DNA matrix program (Fig. 3a). A genomic contig of 150 kbp was identified in which the human gene was located. The sequence comparison shows regions of homology at ϳϪ8 to Ϫ7 kbp and Ϫ5 to Ϫ4 kbp, where the HSd and HSp sites are located, respectively. It also identifies a conserved region lying within Ϫ2 kbp from the CAP site potentially corresponding to the new enhancer region (EnC), in addition to sequences in the POX, which have been shown previously to play a role in its activity (16). The mouse Ϫ2 kbp fragment does not display significant homology with unrelated fragments of the human contig, using the matrix program (results not shown). Similar results were obtained using a larger (15 kbp) fragment of human genomic DNA (results not shown). We then focused on the conserved region at about Ϫ1.4 kbp from the CAP site. This relative location is conserved between human and mouse, and the homology is maintained at high stringency. Using the blocks method of Martinez (51), DNA alignment shows that the conserved sequence is located between Ϫ1.432 and Ϫ1.350 kbp from the CAP site, with 85% identity and two gaps, and with most mismatches located at the extremities of the region (Fig. 3b).
The functional characteristics of the conserved 82-bp sequence were tested by transfection experiments in primary cardiocyte cultures, in myotubes of the skeletal muscle C2/7 cell line and in the C3H 10T1/2 embryonic fibroblast line (Fig.   4). Upstream sequences were introduced in front of the POX directing a CAT reporter sequence. The activity of POX in the different cell cultures was taken as 1. As reported in Fig. 1 the complete Ϫ2.354-kbp sequence upstream of the ␣-cardiac actin CAP site was active in cardiocytes only. A series of deletions demonstrate that activity in cardiocytes depends on sequences lying 5Ј to the SpeI site at Ϫ1.35 kbp from the gene. The BamHI/SpeI fragment (Ϫ2.354 to Ϫ1.364 kbp) was cloned in front of the Tk promoter driving the CAT reporter and shown to be active in cardiocytes, with some activity also in C2/7 myotubes (Table I). Activity in cardiac muscle cells was retained in the reverse orientation, and we therefore concluded that the sequence acts as an enhancer. Full activity is retained with a subfragment, which extends from the SpeI site to the HincII site a further 500 bp upstream (Fig. 4). The highly conserved 85-bp region alone does not exert more than about 2-fold activity, with incremental rises in activity as sequences extending to the HincII site are included. Between the HincII and the XbaI site, there is an Ebox, and its mutation results in some loss of activity, but we have not identified any single site that is essential for activity in contrast to the MEF2 site, which is essential for activity in the context of the larger fragment (BamHI/SpeI fragment). There are potential sites for cardiac regulators, some of which had a minor effect on activity.

Activity of the Cardiac Enhancer Depends on a Critical MEF2
Site-We focused our attention on the highly conserved region, which contains an AT-rich sequence, CTATTTTTAAC, which is a potential binding site for the MEF2 family of transcription factors (consensus CTA(A/T)4 TAG/A (33)). When this sequence is mutated (m1, m2) in the context of the BamHI/SpeI fragment all enhancer activity is lost, both with the POX (Fig. 4) and Tk (Table  1) promoters. The low level activity seen with the conserved XbaI/SpeI fragment is also eliminated when the central MEF2 site is mutated (Fig. 4). These results demonstrated that the activity of the cardiac enhancer depends on this critical MEF2 site. The functional role of this site was further illustrated by experiments in which it was cloned as a multimer (4 ϫ 24 bp of site and flanking sequences) in front of the Tk promoter driving a lacZ reporter. This resulted in a 7.8-fold increase in activity over the TkLacZ transgene alone, specifically in cardiocytes (Table 1). The fact that this multimer is not active in C2/7 myotubes suggests that a cardiac-specific complex is involved. A larger fragment with the Tk promoter shows some activity in C2 myotubes, which is lost when the site is mutated, consistent with a role for MEF2 in skeletal muscle cells.
The binding of MEF2 isoforms to the sequence is shown in Fig. 5. The same 25-bp sequence, radioactively labeled, was tested by gel mobility shift assay. Nuclear extracts from adult cardiac muscle form a complex, which is competed by excess cold oligomer and by the standard MEF2 site present in an enhancer of the MCK gene but not by the mutated ␣-cardiac actin oligomer (m2) (Fig. 5A). When different antibodies to MEF2 isoforms, mHox or Oct1, which also bind AT-rich sequences, are added to the nuclear extract, the presence of MEF2 in the complex is demonstrated (Fig. 5B). A complete band shift is seen with a MEF2 antibody and also to a large extent with an antibody to the MEF2D isoform. A MEF2A antibody also shifts part of the complex. MEF2B does not appear to be involved. The extent of the shift with the MEF2D (and 2A) antibody suggests that if MEF2C, for which specific antibodies are not available, is also part of the complex, it is probably a minor component. DISCUSSION We have identified a novel enhancer (EnC) situated in the Ϫ2.354 to Ϫ1.364-kbp region upstream of the mouse ␣-cardiac actin gene. In cell culture experiments this sequence is only active in cardiocytes, not in differentiated skeletal muscle or non-muscle cells. In this respect it differs from the previously identified enhancers at ϳϪ7 and Ϫ5 kbp from the gene, which are both active in skeletal myotubes. The latter also has some activity in cardiocytes and may contribute to the expression of ␣-cardiac actin in the heart. However we now show that a transgene containing only the proximal enhancer (Ϫ2.354 kbp of 5Ј-flanking sequence) will direct expression in the adult heart, which the promoter alone (Ϫ669 bp) has never been shown to do (22). Regulation of the ␣-cardiac actin gene is therefore orchestrated by at least three enhancers, with different striated muscle specificities. Two sets of results suggest that the HSd may exercise control over the accessibility of the locus (see Ref. 21), possibly in conjunction with the HSp and EnC regions. The complete Ϫ8.6-kbp transgene gives robust, reproducible expression and is not subject to silencing in vivo. In BALB/c mice, a partial duplication of the gene (6) separates this distal enhancer from more proximal regulatory sequences, including the EnC, which is not modified. 2 In these mice, expression of the ␣-cardiac actin gene is observed in the adult heart but at a reduced level (6). The combination of enhancers with different muscle specificities is a feature of muscle gene regulation. The desmin (34) or MCK (35) genes are controlled by skeletal and cardiac enhancers, whereas the myosin MLC 1F/3F gene is regulated by distinct skeletal muscle specific enhancers active at embryonic and fetal stages of development (36). An extreme example of the fine control exercised by many distinct regulatory elements is provided by Myf5, where multiple sequences, extending over 100 kbp upstream of this key myogenic regulatory gene, control its spatiotemporal expression and hence the onset of skeletal muscle formation at different sites in the embryo (see Ref. 37).
The EnC is active in primary cultures from perinatal ventricular muscle and directs expression of the nlacZ transgene in the adult heart. Although the atria are strongly labeled by X-gal staining, not all cardiocytes in the ventricles are ␤-galactosidase positive. This is particularly evident in the adult right ventricle. A similar profile was seen with Ϫ5.746 kbp of 5Ј-flanking sequence (22) and is therefore not because of the absence of the HSp region. In the presence of the full Ϫ8.6kbp flanking sequence more complete cardiac expression is seen, suggesting that in the absence of the HSd element, shorter transgenes are more susceptible to this effect. Other muscle transgenes also show regionalized transcription in the heart, potentially reflecting the modular nature of cardiac gene regulation (see Ref. 38). In the case of a myosin light chain, MLC 3F , transgene expression in the left ventricle, but not the right, appears to reflect a phenomenon of in vivo chromatin silencing not seen in transitory transfection experiments (25).
The EnC extends over 1 kbp and contains potential regulatory sites some of which, when mutated, reduced enhancer activity (results not shown). However the main region of sequence conservation between man and mouse localizes to the Ϫ1.432 to Ϫ1.350-kbp region that surrounds the MEF2 site. Mutation of this site confirmed that it is critically important for enhancer activity. Concatemers of this site with its flanking  35 Kb) was isolated including the (B/S) fragment of the gene plus a linker of 6 bp. This (S-S) fragment was cloned at a Xba1 site in front of the Tk promoter in sense ((S)/B-S) and antisense (S-B(S)) orientation. The mutated fragment ((B)/m2/S mutation m2, 5Ј-TTTTCGATTCTTAACTGGT-3Ј) was also cloned in front of the Tk promoter. Constructs were transiently transfected into primary cultures of cardiocytes derived from dissected ventricles, and into C2/7 myotubes.
Oligonucleotides containing the native or mutated (mutation m2) MEF2 site were concatemerized and cloned at an SfiI site introduced in front of the Tk promoter driving the LacZ gene. Constructs were transfected into primary cultures of cardiocytes derived from dissected ventricles and into C2/7 myotubes and into NIH3T3 cells. Native oligonucleotide, 5Ј-CAGTTTTCTATTTTTAACTGGTGTG-3Ј; mutated oligonucleotide,5Ј-CAGTTTTCGATTCTTAACTGGTGTG-3ЈMutatednucleotides are indicated in bold. The pRSVLuc plasmid was included in all transfections as an internal control to correct for transfection efficiencies. CAT region retain cardiac-specific activity. Because MEF2 is also a regulator of skeletal muscle, this would suggest that the flanking sequences are implicated in the binding of a complex, which is cardiac specific. Gel mobility shift assay experiments show that MEF2D binds to the sequence, with some additional interaction with MEF2A and possibly MEF2C. MEF2, and particularly MEF2C have been implicated in the regulation of a number of cardiac muscle genes, such as ␣-myosin heavy chain (39,40) or myosin light chain, MLC 2V (41), and desmin (42). The regionalization of ventricular expression is difficult to equate with the MEF2 site alone particularly because the desmin transgene, which depends on MEF2, is expressed in the right, but not the left ventricle (42). This is also the case for an MLC 2V transgene (43). A potentially critical role for MEF2C in the formation of the left ventricle is indicated by the loss of this cardiac compartment in MEF2C mutant mice (44). The enhancer described here is remarkable for its activity in adult cardiac muscle. The role of MEF2 isoforms in the adult heart has been questioned because of a report that MEF2 proteins were not detectable (45) and that MEF2 activity was shown to fall as the heart matures (46). In the gel mobility shift assay experiments reported here we show that nuclear extracts from the ventricular muscle of adult mouse hearts do contain MEF2 binding activity, in this case principally MEF2D. Func-tionally it has been shown that a transgene containing multimerized MEF2 sites from the enhancer of the MCK gene is mainly active in the embryonic heart (47). However low, but detectable, levels of expression persisted into adulthood. Indeed a detectable level of MEF2C has been reported in adult hearts and is reduced in the failing hearts of diabetic patients accompanied by a down-regulation of MEF2 target genes (48). Furthermore in mice that lack MEF2A, which binds the MEF2 site in EnC, there is an adult cardiac phenotype (49). Interestingly in these mice ␣-cardiac actin transcripts increase, possibly because of a stress response, which leads to an increase in MEF2D (49). This may also be the case for adult cardiocytes. The MEF2D mutant phenotype has not been published. However there are suggestions that this isoform is important for regulation of cardiac genes as shown by studies on the regulation of the human ␣-skeletal actin gene (50). Furthermore it is MEF2D that is implicated in the complex with Emb, a member of the POU domain family of proteins, which forms on the AT-rich HSd site of the ␣-cardiac actin gene (21).
We conclude from our analysis that the activity of the cardiac-specific enhancer, that we have identified as one of at least three enhancers regulating the mouse ␣-cardiac actin gene, depends on a MEF2 site, which in adult cardiomyocytes binds the MEF2D isoform. The striking sequence conservation around this site suggests that in this case MEF2D may participate in a cardiac-specific regulatory complex.