A Positive GATA Element and a Negative Vitamin D Receptor-Like Element Control Atrial Chamber-Specific Expression of a Slow Myosin Heavy-Chain Gene during Cardiac Morphogenesis

Abstract

We have used the slow myosin heavy chain (MyHC) 3 gene to study the molecular mechanisms that control atrial chamber-specific gene expression. Initially, slow MyHC 3 is uniformly expressed throughout the tubular heart of the quail embryo. As cardiac development proceeds, an anterior-posterior gradient of slow MyHC 3 expression develops, culminating in atrial chamber-restricted expression of this gene following chamberization. Two cis elements within the slow MyHC 3 gene promoter, a GATA-binding motif and a vitamin D receptor (VDR)-like binding motif, control chamber-specific expression. The GATA element of the slow MyHC 3 is sufficient for expression of a heterologous reporter gene in both atrial and ventricular cardiomyocytes, and expression of GATA-4, but not Nkx2-5 or myocyte enhancer factor 2C, activates reporter gene expression in fibroblasts. Equivalent levels of GATA-binding activity were found in extracts of atrial and ventricular cardiomyocytes from embryonic chamberized hearts. These observations suggest that GATA factors positively regulate slow MyHC 3 gene expression throughout the tubular heart and subsequently in the atria. In contrast, an inhibitory activity, operating through the VDR-like element, increased in ventricular cardiomyocytes during the transition of the heart from a tubular to a chambered structure. Overexpression of the VDR, acting via the VDR-like element, duplicates the inhibitory activity in ventricular but not in atrial cardiomyocytes. These data suggest that atrial chamber-specific expression of the slow MyHC 3 gene is achieved through the VDR-like inhibitory element in ventricular cardiomyocytes at the time distinct atrial and ventricular chambers form.

The vertebrate heart, which initially forms as a linear, tubular structure, undergoes a complex series of morphogenetic movements to form a multichambered organ. The low-pressure atrial and high-pressure ventricular chambers which are formed differ in morphology, electrophysiology, and the repertoire of muscle contractile protein genes which are expressed (8, 30). The developmental and molecular mechanisms responsible for establishing and maintaining chamber-specific differences are largely unknown.

In adult animals, several members of contractile protein gene families show chamber-restricted or strong chamber-preferential expression. The myosin heavy-chain (MyHC) AMHC1 and slow MyHC 3 genes (56, 60), the myosin light-chain 1a (MLC-1a) and MLC-2a genes (21, 31), and the atrial natriuretic factor (ANF) gene (61) show atrial specificity. In small mammals, the α-MyHC gene exhibits an atrial chamber preference of expression before birth (31). Conversely, MLC-2v (44) and β-MyHC (31) are restricted to the ventricles. A common feature of genes expressed in a chamber-preferential or chamber-restricted manner in the adult is early global expression throughout the myocardium at the tubular heart stage, with subsequent chamber restriction as development proceeds (55). The establishment of chamber-restricted gene expression by downregulation in atrial or ventricular chambers, as opposed to activation in only atrial or ventricular chambers, has implications for the understanding of the underlying regulatory mechanisms.

In recent years, a growing number of regulatory gene families that are important in regulating cardiac gene expression have been described. Members of the GATA family (1, 17, 23), the tinman/Nkx2-5 family (4, 20, 29), the HAND family (47, 48), and the myocyte enhancer factor 2 (MEF2) family (13, 28) have all been shown to be cardiac transcriptional activators. Studies in cultured cardiomyocytes have implicated GATA-4 in the regulation of several cardiac genes, including those encoding α-MyHC, cardiac troponin c, ANF, and brain natriuretic peptide (15, 18, 38, 52), while in vivo injection of GATA-4, GATA-5, or GATA-6 RNA into Xenopus embryos activates expression of the genes encoding α-cardiac actin and α-MyHC (19). GATA-4 is expressed in P19 cells during their differentiation into cardiomyocytes, and the addition of antisense GATA-4 oligonucleotides blocks this differentiation (16). Although all cardiac genes examined are expressed in GATA-4 null mice, the upregulation of GATA-6 in the GATA-4 null mouse has been postulated to replace GATA-4 function and account for activation of cardiac genes (22, 39). Nkx2-5 has been shown to be essential for the activation of MLC-2v (32). Nkx2-5 null mice do not express MLC-2v (32) or the eHAND gene (3) and have markedly reduced levels of cardiac ankyrin repeat protein gene expression (64). One MEF2 family member, MEF2C, plays a key role in the activation of several cardiac genes, among which are the atrial chamber-restricted genes MLC-1a and ANF and the atrial chamber-preferential gene α-MyHC (28). Finally, there is evidence of synergy among members of several transcriptional regulatory families to activate cardiac gene expression (5, 10, 24, 46).

While mechanisms that regulate cardiac expression of specific genes are rapidly being clarified, mechanisms that regulate chamber-specific expression of cardiac genes are less clear. Recently the HAND genes, dHAND and eHAND, have been shown to be asymmetrically distributed along the anterior-posterior axis of the looped heart tube and to play an important role in the specification of the right and left ventricles, respectively (51). However, neither the HAND genes nor any of the other well-characterized cardiac transcriptional regulators is distributed in the heart in a manner which suggests a role in the establishment or maintenance of atrial as opposed to ventricular gene expression. Specifically, the cardiac transcriptional regulators GATA-4, Nkx2-5, and MEF2C are all expressed equally in both atrial and ventricular chambers (1, 15, 29, 35). Because these factors upregulate genes restricted to or preferentially expressed in the atria, such as ANF, MLC-1a, and α-MyHC (10, 28, 40), additional regulatory components must be acting in the ventricles to repress expression during development.

Currently, relatively little is known regarding molecular mechanisms that lead to atrial chamber-specific gene expression during development. The slow MyHC 3 gene is especially suitable for understanding the molecular mechanisms of atrial chamber-specific gene expression and of atrial or ventricular cell lineage diversification. The chicken homolog of the slow MyHC 3 gene, AMHC1, which encodes an atrial chamber-specific MyHC, is among the earliest cardiac genes to show chamber-specific restriction. AMHC1 is first expressed in the posterior region of the fusing chicken heart tube, the future atrial compartment, by stage 9 (60). A 160-bp region of the slow MyHC 3 gene promoter, designated ARD1, has been shown to act as an atrial chamber-specific enhancer both in cell culture and in the embryo (56). A vitamin D receptor like sequence motif present within the enhancer is required for the observed inhibition of reporter expression in ventricular but not in atrial cardiomyocytes, supporting the hypothesis that atrial chamber-specific expression is achieved by ventricular chamber-specific inhibitors (56).

Here we report data that show that the slow MyHC 3 gene is initially expressed throughout the tubular heart. While expression in the atria was maintained as the heart underwent chamberization, expression in the ventricles was downregulated. Two cis elements, a GATA-binding site and the VDR-like motif within the enhancer, were responsible for atrial chamber-specific expression of the slow MyHC 3 gene. GATA-4, but neither the transcription factor Nkx2-5 nor MEF2C, activated reporter expression from the slow MyHC 3 gene promoter in fibroblasts, suggesting that the GATA element alone is sufficient to activate cardiac chamber-specific expression. Inclusion of the VDR-like element in reporter constructs suppressed the reporter activity in ventricular but not atrial cardiomyocytes. In addition, inhibitory activity increased in ventricular cardiomyocytes during the transition from a tubular to a chambered heart. The timing of increased ventricular inhibition coincides with the downregulation of slow MyHC 3 gene expression in the ventricles during normal heart development. Overexpression of VDR, acting via the VDR-like element, duplicates the inhibitory activity in ventricular but not atrial cardiomyocytes. These data suggest that binding of the VDR to the VDR-like element is important in the downregulation of slow MyHC 3 gene expression in the ventricles during development and that the GATA element and the VDR-like element, acting together, control atrial chamber-specific expression of the slow MyHC 3 gene.

MATERIALS AND METHODS

Immunocytochemistry and whole-mount staining.

MyHC immunostaining of cultured cardiomyocytes with monoclonal antibodies F59 and S58 was performed as described previously (56). Monoclonal antibody F59 recognizes all known avian fast MyHC isoforms (6), while in the avian heart, S58 is specific for the slow MyHC 3 isoform (56). An ascites of monoclonal antibody NA8 (a gift from Everett Bandman) also is slow MyHC 3 specific in the heart (reference 25 and unpublished data). NA8 was used at a 1:2,000 dilution for whole-mount staining of quail embryos (56).

RNA analysis of slow MyHC 3 expression.

Details of total cellular RNA isolation, hybridization, and quantitation by standard protocols are as previously described (56). Five micrograms of total RNA from each developmental time point was assayed by RNA dot blotting. The blots were probed with a slow MyHC gene-specific oligonucleotide directed against a sequence in the 3′ untranslated region (5′-AAG GGA ATT CAT CAG AGG TTG GGG CT-3′).

Cells and media.

Primary cultures of atrial and ventricular cardiomyocytes, isolated from quail embryos on embryonic day 3 (ED3), ED4, and ED6, were cultured in heart serum-containing medium for 1 day. On the second day of culture, the cells were switched to a serum-free medium for two additional days of culture (56). Chicken embryonic fibroblasts (CEFs) were isolated from trunks of ED12 embryos and cultured as described previously (56).

Plasmids.

SM3CAT constructs SM3CAT:840D, SM3CAT:808D, and SM3CAT:768D contain 840, 808, and 768 bp, respectively, of the upstream slow MyHC 3 gene promoter sequence fused to the bacterial chloramphenicol acetyltransferase (CAT) reporter (56). SM3CAT:724D was made by cloning 724 bp of the upstream slow MyHC 3 gene promoter sequence into the HindIII-XbaI site of pCAT-promoter, a minimal simian virus 40 (SV40) promoter driving the CAT gene (Promega). Using PCR-mediated mutagenesis, sequence of the GATA element from positions −762 to −757 (AGATAA) in the SM3CAT:768D construct was replaced with GTCGAC to generate −768D-mGATA.

Three heterologous promoter constructs in which slow MyHC 3 gene promoter sequence was oriented 5′ to 3′ upstream of pCAT-promoter were constructed. The VDR:CAT, GATA:CAT, and VDR-GATA:CAT constructs were made by cloning the VDR-like element between positions −808 and −776, the GATA element between positions −775 and −741, and both elements between positions −808 and −741, respectively, into the BglII site of the pCAT-promoter vector. The sequence and orientation of each construct was verified by dideoxy sequence analysis. The GATA-4 expression plasmid PMT2-GATA-4 (18), the MEF2C expression plasmid (37), and the Nkx2-5 expression plasmid (5), the VDR expression plasmid (27), the retinoic acid receptor α (RARα) and RXRα expression plasmids (33, 54) were gifts from David B. Wilson, Eric Olson, Robert Schwartz, David Feldman, and Ronald Evans, respectively. A human ANF promoter construct (−1150 hANF-CAT) was provided by David Gardner (26).

DNA transfections and CAT assays.

Quail embryonic atrial and ventricular cardiomyocytes or CEFs were cultured in 35-mm-diameter dishes and transfected with 3 μg of CAT reporter plasmid plus 1 μg of psv-β-gal reference plasmid, using the calcium phosphate precipitate method (14). For cotransfection with the GATA-4, MEF2C, or Nkx2-5 expression plasmid, various amounts of the expression plasmid were used as indicated in the figure legends. Various amounts of an empty expression vector were added such that each dish was transfected with an equal amount of total DNA. Forty-eight hours after transfection, the cells were harvested and CAT and β-galactosidase assays were performed as described previously (43). All experiments were repeated at least three times, and the CAT activities were normalized to the β-galactosidase activities in order to standardize the transfection efficiency. All-trans retinoic acid (RA) was purchased from Sigma. Vitamin D₃ was obtained from Biomol (Plymouth Meeting, Pa.). Cells were treated with vitamin D₃ (10⁻⁸ M) or all-trans RA (10⁻⁶ M) for the final 24 h when transfection of the VDR expression vector or the RARα expression vector was indicated, respectively.

EMSAs.

Preparation of nuclear extracts and electrophoretic mobility shift assays (EMSAs) were performed according to the procedure reported by Zou and Chien (63). Confluent cultures of primary embryonic atrial and ventricular cardiomyocytes were grown in heart serum-containing medium for 1 day and in heart serum-free medium for two additional days after being plated at a density of 2 × 10⁷ cells per 100-mm-diameter dish. The cells in each dish were washed twice in cold phosphate-buffered saline, harvested in 0.5 ml of phosphate-buffered saline by scraping, and spun at 2,000 rpm for 4 min at 4°C to pellet cells. Each cell pellet was suspended in 400 μl of buffer A (10 mM HEPES [pH 7.9], 10 mM KCl, 0.1 mM EDTA, 1 mM dithiothreitol [DTT], and 0.5 mM phenylmethylsulfonyl fluoride). After a 10-min incubation on ice, 10 μl of 10% Nonidet P-40 was added and the mixture was vortexed at top speed for 1 min. Nuclei were pelleted by spinning the cell lysates at 6,000 rpm at 4°C for 4 min. The pelleted nuclei were suspended in 40 μl of buffer B (20 mM HEPES [pH 7.9], 0.4 M NaCl, 1 mM EDTA, 1 mM DTT, 1 mM phenylmethylsulfonyl fluoride, 1 μg of leupeptin per ml, 1 μg of pepstatin per ml, 0.2 μg of aprotinin per ml) and incubated on ice for 15 min with frequent vortexing to release the nuclear proteins. These nuclear extracts were clarified by centrifugation for 5 min at 10,000 rpm and were stored at −70°C until use.

For DNA binding assays, 10 μg of nuclear extract, 2 μg of poly(dI-dC) · poly(dI-dC) (Pharmacia Biotech), and 5 μg of bovine serum albumin were mixed with 4 μl of 5× binding buffer (200 mM KCl, 75 mM HEPES [pH 7.9], 5 mM EDTA, 2.5 mM DTT, 25 mM MgCl₂, 25% glycerol) in a total volume of 19 μl. These preincubation mixtures were incubated on ice for 30 min. Subsequently, approximately 20,000 cpm of a [γ-³²P]ATP end-labeled double-stranded oligonucleotide in 1 μl was added to the preincubation mixture, and the solution was placed on ice for 30 min. The purified oligonucleotides were purchased from Operon Technologies, Inc., and annealed at a high concentration to generate double-stranded oligonucleotides. Only the sense-strand sequences of the double-stranded oligonucleotides are shown here, as follows: GATA-4, 5′-AGGTGGGGCTGGGAGATAAGGAGGCCAGAAATAG-3′; mGATA-4, 5′-AGGTGGGGCTGGGGTCGACGGAGGCCAGAAATAG-3′, and VDR, 5′-CTTGCGAAGGACAAAGAGGGGACAAAGAGGCGGA-3′.

The samples were loaded on a 5% nondenaturing acrylamide gel and were run in 0.5× Tris-borate-EDTA buffer at 10 V/cm for 2 h. The gel was dried and autoradiographed. Supershift experiments were performed exactly as described above for standard EMSAs except that the antibody was incubated with the nuclear extract at room temperature for 30 min prior to the binding reactions. Rabbit preimmune serum and a polyclonal antibody that was made against GATA-4 but reacts with other GATA factors were provided by David B. Wilson (17). Monoclonal anti-VDR antibody was purchased from Biomol. Rabbit preimmune serum and polyclonal anti-RXRα and anti-RARα antibodies were provided by Elizabeth Allegretto of Ligand Pharmaceuticals, Inc. (53).

RESULTS

Developmental expression of the slow MyHC 3 gene.

Expression of the slow MyHC 3 gene could first be detected in the quail heart at about 35 h of embryonic development. Quail embryos at different developmental stages were fixed and processed for whole-mount immunostaining with NA8, an antibody that in the heart is specific for slow MyHC 3. Slow MyHC 3 was first faintly detected in the tubular heart of the embryo at late stage 10 (12 somites) (Fig. 1A). About 1.5 h later, at stage 11 (13 somites), slow MyHC 3 was clearly expressed uniformly throughout the tubular heart (Fig. 1B). By ED2 (18 somites), the heart tube was slightly S shaped and staining for slow MyHC 3 had become more prominent at the caudal end, the prospective atria, with less-intense but uniform staining in the remainder of the heart tube (Fig. 1C). By ED3, demarcation of the tubular heart into prospective atrial and ventricular compartments was morphologically apparent and a gradient of slow MyHC 3 expression was more evident, such that the prospective atria stained more intensely than the prospective ventricles (Fig. 1D). Only faint staining was observed in the outflow track at the most anterior portion of the heart tube. Slow MyHC 3 expression was further downregulated in the prospective ventricles during the transition to a chambered heart at ED4 (Fig. 1E). By ED6, the heart had four well-developed chambers and very little slow MyHC 3 was detected in the ventricles (Fig. 1F). Throughout morphogenesis of the heart, slow MyHC 3 expression in the prospective atria was sustained at high levels (Fig. 1D and E).

FIG. 1.

Expression of slow MyHC 3 becomes restricted to the atria as development proceeds. Staged normal quail embryos were assessed for slow MyHC 3 expression by whole-mount staining with NA8 (A to F) and by staining of frozen sections with S58 (G) and F59 (H). (A and B) Ventral views with the head on the left; (C to E) views with the dorsal side up, the head to the left, and the atria in the upper right, respectively. (A) Stage 10 (12 somites). (B) Stage 11 (13 somites). (C) ED2. (D) ED3. (E) ED4. (F) Heart isolated from an ED6 embryo. Initially expressed throughout the tubular heart (A and B), slow MyHC 3 is gradually downregulated in the ventricles during the transition to a chambered heart, while expression in the atria is maintained (C to F). (G and H) A section at the junction of the atrium and ventricle of the ED6 heart, double stained with S58 (G) and F59 (H). In the ED6 heart, slow MyHC 3 is abundant only in the atrial myocardium (G), while fast MyHC isoforms are equally abundant in both the atrial and ventricular myocardia (H).

To observe slow MyHC 3 expression within the walls of the atrial and ventricular chambers, ED6 quail hearts were sectioned and double immunostained with the slow MyHC 3-specific monoclonal antibody S58 and with F59, a monoclonal antibody which reacts exclusively with fast MyHC isoforms. A typical section through the junction of the atrium and ventricle is shown in Fig. 1G and H. Consistent with whole-mount embryo staining, slow MyHC 3 was present at a high level in the atrial myocardium and at a much lower level in the ventricular myocardium (Fig. 1G). Expression was uniformly low across the compact and trabecular layers of the ventricular wall. In contrast, fast isoforms of MyHC, recognized by F59, were uniformly distributed across the atrium and ventricle (Fig. 1H).

Expression of slow MyHC 3 mRNA decreased in the ventricles as the tubular heart chamberized. Using the 3′ untranslated region as a probe, the time course of slow MyHC 3 gene expression was investigated in developing quail atria and ventricles by RNA slot blot analysis (Fig. 2). Slow MyHC 3 transcripts were detected in the prospective atrium by ED3 and remained at high levels during fetal development and throughout posthatch life. In contrast, at ED3, slow MyHC 3 gene expression in the prospective ventricles was less than half that observed in the prospective atria. During the time of heart chamberization, between ED3 and ED6, the levels of slow MyHC 3 mRNA in the ventricles rapidly declined to reach a barely detectable level by ED10. Therefore, slow MyHC 3 gene expression in the ED10 heart is atrial chamber specific, which is consistent with previous Northern blot and in situ hybridization analyses (43).

FIG. 2.

Slow MyHC 3 gene transcript levels in the ventricle decline during the early stages of heart development. Total RNA, isolated from quail hearts between ED3 and ED14 (3d to 14d), at hatching (H), and at 4 months (4m) of development, was assayed by dot blotting with a slow MyHC 3 gene-specific probe. Each datum point is the average of data from two experiments. At each time point, the slow MyHC 3 gene levels in the atria (filled boxes) and in the ventricles (open ovals) are expressed relative to an atrial standard from a hatching embryo. The slow MyHC 3 gene is expressed continuously at high levels in the atria but is downregulated during ventricular development.

The GATA element regulates heart-specific expression of the slow MyHC 3 gene in primary cardiomyocyte cultures from tubular and chamberized embryonic hearts.

A 160-bp enhancer (ARD1), positioned between −840 and −680 bp upstream of the slow MyHC 3 gene transcriptional initiation site, has been shown to restrict expression to the atria both in vitro and in the embryo (56). Deletion and mutational analyses of sequence motifs within ARD1 demonstrated the importance of a VDR-like element in restricting slow MyHC 3 gene expression to the atria and suggested that a GATA motif might be sufficient to confer cardiac specificity on this gene (56).

A cardiac tissue-specific slow MyHC 3 promoter construct, SM3CAT:768D, which contains the GATA portion of ARD1 but not the VDR-like element (Fig. 3A), was expressed at equivalent high levels in both atrial and ventricular cardiomyocytes isolated from ED6 quail hearts but at background levels in fibroblasts (Fig. 3B). However, a 5′ truncation of this construct which removed the GATA motif, SM3CAT:724D, significantly reduced expression in all cardiomyocytes (Fig. 3B). To confirm that the GATA element regulates cardiac tissue-specific expression, the core sequence of the GATA-binding motif (AGATAA) was mutated to GTCGAC in the context of SM3CAT:768D to generate −768DmGATA (Fig. 3). As observed with the deletion of GATA, mutation of the GATA motif reduced CAT expression in both atrial and ventricular cardiomyocytes (Fig. 3B). Thus, in this context the GATA element is necessary for expression in both atrial and ventricular cardiomyocytes.

FIG. 3.

Deletion or mutation of the slow MyHC 3 gene GATA element dramatically reduces expression in cardiomyocytes. (A) The sequence and location of the VDR and GATA elements within the slow MyHC 3 gene promoter are shown. A VDR-binding motif, AGGACAaagAGGGGA (box), and an RAR binding site, AGGACAaagagGGGACA (dots), have the same 5′ copy of the hexamer sequence. The mutated GATA sequence is underlined. (B) Expression of SM3CAT constructs in atrial (solid bars) and ventricular (striped bars) cardiomyocytes isolated from ED6 hearts. Each SM3CAT construct is designated by the number of base pairs, upstream from the transcriptional start site, included within that construct. As previously shown (56), inclusion of the VDR element (-808D) restricts slow MyHC 3 gene expression to atrial cardiomyocytes. Deletion of the VDR element (-768D) resulted in an increase in CAT expression in the ventricular cardiomyocytes to a level equal to that observed in the atrial cardiomyocytes but had no effect on its expression in the fibroblasts. Deletion (-724D) or mutation (-768D-mGATA) of the GATA element resulted in a sixfold reduction in CAT expression in both the atrial and the ventricular cardiomyocytes relative to the level of expression observed for -768D. The error bars represent the standard errors of the means.

Because slow MyHC 3 is expressed prior to chamberization (Fig. 1 and 2), we investigated the role of GATA at early developmental time points. By ED3, the intensity of slow MyHC 3 staining in the prospective ventricles was less than that in the prospective atria (Fig. 1D), and this difference was maintained when prospective atrial and ventricular cardiomyocytes were isolated and cultured in vitro (Fig. 4A). The 35-bp region of the slow MyHC 3 gene promoter containing the GATA element was fused upstream of a minimal SV40 promoter CAT cassette (pCAT-promoter) to generate the heterologous promoter construct GATA:CAT. When transfected into ED3, ED4, or ED6 atrial or ventricular cardiomyocytes, the GATA:CAT construct was expressed at a fourfold higher level than was the pCAT-promoter construct, which was set to unity (see the legend to Fig. 4B). In contrast, inclusion of the GATA element had no effect on expression from the SV40 promoter in fibroblasts (Fig. 4B). Thus, the slow MyHC 3 gene GATA element not only is necessary for but is sufficient to drive cardiomyocyte-specific gene expression in cardiomyocytes isolated from the developing heart. In contrast to the progressive reduction of slow MyHC 3 gene expression in ventricular cardiomyocytes as development proceeded, the GATA:CAT plasmid was expressed equally well in cultures of ventricular cardiomyocytes isolated from the tubular (ED3 and -4) or the chamberized (ED6) heart (Fig. 4B). We conclude that the slow MyHC 3 gene GATA element is sufficient for expression of the slow MyHC 3 gene in all cardiomyocytes of both the tubular and the chambered hearts.

FIG. 4.

The slow MyHC 3 gene GATA element is sufficient to activate a reporter in cardiomyocytes from the tubular and chambered heart stages. (A) Both atrial and ventricular cardiomyocytes isolated from ED3 and ED4 embryonic hearts stain strongly in culture for fast MyHC with F59 (red). Ventricular cardiomyocytes isolated from the tubular heart stage (ED3) express abundant S58-staining slow MyHC 3 (green); however, as looping proceeds (ED4), isolated ventricular cardiomyocytes express reduced levels of slow MyHC 3. Atrial cardiomyocytes from both stages stain equally well for slow MyHC 3. (B) The slow MyHC 3 gene GATA element was fused upstream of the minimal SV40 promoter in the pCAT-promoter vector to generate the GATA:CAT construct, and this construct was transfected into atrial and ventricular cardiomyocytes, as well as fibroblasts. At each of the stages of development examined, ED3, ED4, and ED6, GATA:CAT expression in the ventricular cardiomyocytes was equal to that in the atrial cardiomyocytes. The GATA element drives cardiomyocyte-specific expression, since addition of the GATA element to the pCAT-promoter vector (GATA:CAT) has no effect on expression in the fibroblasts. Values shown are relative to that of the pCAT-promoter construct, which was set to unity for each culture condition.

To determine if the GATA motif could be activated by the GATA-4 transcription factor, CEFs were cotransfected with increasing amounts of the GATA-4 expression vector (PMT2-GATA-4) and either the pCAT-promoter construct or the GATA:CAT plasmid. Cotransfection of GATA:CAT and PMT2-GATA-4 resulted in a dose-dependent increase in CAT expression in CEFs, whereas expression from the pCAT-promoter control was unaffected by cotransfection with the expression vector (Fig. 5).

FIG. 5.

GATA-4 activates expression through the slow MyHC 3 gene GATA element. Cotransfection of the GATA:CAT construct and various amounts of the GATA-4 expression vector into CEFs led to a dose-dependent activation of the reporter. In contrast, cotransfection of the GATA-4 expression vector with the pCAT-promoter construct (Promoter:CAT, lacking the slow MyHC 3 gene GATA motif) had no effect on CAT expression.

To investigate whether other cardiac transcriptional activators may play a role in regulating slow MyHC 3 gene expression in the heart, the slow MyHC 3 gene promoter construct, SM3CAT:840D, was tested for activation by GATA-4, MEF2C, and Nkx2-5 expression vectors in CEFs. GATA-4, but neither MEF2C nor Nkx2-5, activated CAT expression of SM3CAT:840D in the fibroblasts (Fig. 6A), suggesting that the GATA element alone is responsible for heart-specific slow MyHC 3 gene expression. To demonstrate that the MEF2C and Nkx2-5 expression vectors used in these analyses produced functional proteins, an hANF:CAT test plasmid was cotransfected into the fibroblasts along with each expression vector. In agreement with previously published reports on studies using the rat ANF promoter (11, 15), GATA-4 and Nkx2-5 upregulated expression from the human ANF promoter (Fig. 6B). MEF2C also increased expression (Fig. 6B).

FIG. 6.

The slow MyHC 3 gene promoter is activated by transcription factor GATA-4 but not by transcription factor MEF2C or Nkx2-5. (A) Cotransfection of SM3CAT:840D with the GATA-4 expression vector, but not the MEF2C or Nkx2-5 expression vector, activated CAT expression in CEFs. (B) The human ANF promoter is highly activated by GATA-4 and MEF2C expression vectors and is slightly enhanced by Nkx2-5 expression in CEFs.

The slow MyHC 3 gene GATA motif is a GATA factor binding site.

Transfection experiments showed that the GATA element in the slow MyHC 3 gene enhancer can promote transcription in both atrial and ventricular cardiomyocytes. To determine if these cardiomyocytes contain a factor(s) that can bind to the GATA element in the slow MyHC 3 gene enhancer, nuclear extracts from ED6 cultured atrial and ventricular cardiomyocytes were used in EMSAs. A double-stranded oligonucleotide of slow MyHC 3 gene sequence spanning positions −775 to −741 and including a canonical GATA binding motif, AGATAA, was synthesized. This DNA fragment was ³²P radiolabeled and incubated with either atrial or ventricular nuclear extracts from ED6 heart cultures (Fig. 7).

FIG. 7.

GATA factors present in both atrial and ventricular nuclear extracts bind the GATA element of the slow MyHC 3 gene enhancer. A ³²P-radiolabeled double-stranded DNA probe of the slow MyHC 3 gene enhancer between positions -775 and -741, including a canonical GATA binding site, was used in EMSAs. (A) Nuclear extracts of both atrial and ventricular ED6 heart cultures form a major band of reduced electrophoretic mobility (arrow). No differences in the positions of the retarded bands for atrial and ventricular extracts were distinguishable (lanes 2 and 5). Binding is specific to the GATA motif. Addition of a 100-fold molar excess of cold oligonucleotide effectively eliminated binding (lanes 3 and 6), while no change in binding was observed when a 100-fold molar excess of an oligonucleotide with a mutated GATA motif was added (lanes 4 and 7). (B) The binding activity includes a GATA factor(s). Preincubation of nuclear extracts of both atrial and ventricular cell cultures with antiserum made against GATA (17) produced supershifts of probe DNA (SS, lanes 2 and 5). No supershifts were seen when rabbit preimmune serum was first incubated with the nuclear extracts (lanes 3 and 6).

Atrial nuclear extracts contained a major binding activity (at the position of the arrow in Fig. 7A, lane 2). Binding was specific because it was competed by unlabelled GATA oligonucleotide (Fig. 7A, lane 3) but a mutated GATA oligonucleotide containing six nucleotide substitutions within the GATA motif did not (Fig. 7A, lane 4). The addition of an antibody that recognizes a cardiac member of the GATA family to atrial extracts (17), but not preimmune serum, produced supershifts of the oligonucleotide (Fig. 7B, lanes 2 and 3). Thus, a cardiac member of the GATA family of transcriptional regulators is present in atrial cardiomyocytes and can bind to the GATA motif present in the slow MyHC 3 gene enhancer.

Ventricular nuclear extracts also produced a shifted band with the radiolabeled GATA probe. This band migrated to the same position as that seen in assays using atrial nuclear extracts (Fig. 7A, lane 5). The binding was specific because the shifted band was competed by the cold GATA oligonucleotide (Fig. 7A, lane 6) but the mutated GATA oligonucleotide did not (Fig. 7A, lane 7). The shifted band produced by the ventricular extracts was also supershifted by the GATA antibody (Fig. 7B, lane 5) but not by a preimmune serum (Fig. 7B, lane 6). Furthermore, the affinities of binding of atrial and ventricular nuclear extracts to the GATA probe were very similar (data not shown). Together these results suggest that binding of a GATA factor to the GATA motif present in the enhancer positively regulates slow MyHC 3 gene expression in both atrial and ventricular cardiomyocytes.

The VDR-like element acts as an inhibitory element in ventricular but not atrial cardiomyocytes, and its inhibitory activity increases as the heart chamberizes.

We have reported that the VDR-like motif serves as a negative control element in the slow MyHC 3 gene promoter in ventricular cardiomyocytes but not in atrial cardiomyocytes isolated from ED6 hearts (56). To determine if there is developmental regulation of the inhibition of slow MyHC 3 gene expression via the VDR-like element that coincides with the downregulation of slow MyHC 3 gene expression as heart development proceeds, transfections were performed in cardiomyocytes from the tubular through the chambered heart stages. Because all data suggested that the GATA element is sufficient to promote slow MyHC 3 gene expression while the VDR-like element alone can inhibit ventricular expression, we generated two test plasmids: VDR-GATA:CAT, in which the VDR-like element is positioned 5′ to the GATA element upstream of the heterologous minimal SV40 promoter (pCAT-promoter), and VDR:CAT, in which the VDR-like element alone is inserted upstream of the minimal SV40 promoter. These two constructs, as well as the parental plasmid, pCAT-promoter, were transfected into cultured atrial or ventricular cardiomyocytes from ED3 tubular hearts. VDR:CAT expression was equal to that of pCAT-promoter, which was set to unity in both atrial and ventricular cardiomyocytes (see the legend to Fig. 8). Fusion of the VDR-like element upstream of GATA had no effect on the high level of expression induced in atrial cardiomyocytes (compare Fig. 8 with Fig. 4B), but it did inhibit expression within ED3 ventricular cardiomyocytes by 36% relative to atrial expression (Fig. 8). These results demonstrate that the VDR-like element inhibits expression specifically in ventricular cardiomyocytes and that this inhibitory activity is present in the ED3 heart.

FIG. 8.

VDR-specific inhibition in ventricular cardiomyocytes increases during the transition from a tubular to a chambered heart. Two slow MyHC 3 gene promoter-containing constructs were tested in cardiomyocytes isolated during embryonic stages ED3 to ED6 of development. The VDR-like motif alone (squares) or the VDR-like plus the GATA motifs (circles) were fused to the pCAT-promoter construct to generate VDR:CAT and VDR-GATA:CAT, respectively. The VDR-like element alone (VDR:CAT) neither inhibited nor enhanced the CAT activity of the heterologous promoter in either the atrial (closed squares) or the ventricular (open squares) cardiomyocytes. In the context of VDR-GATA:CAT, however, the VDR-like element increasingly suppressed GATA activity in ventricular cardiomyocytes as development proceeded (open circles), relative to activity in atrial cardiomyocytes at the same stage. In contrast, in atrial cardiomyocytes (closed circles), VDR-GATA:CAT activity remained at a constant high level. The values shown are relative to that of the pCAT-promoter construct, which was set to unity for each culture condition.

The same experiment was conducted on cardiomyocytes isolated from developing atria and ventricles during the transition between a tubular and a chambered heart. The VDR:CAT, VDR-GATA:CAT, and pCAT-promoter constructs were transfected into cardiomyocytes isolated from hearts when they were undergoing chamberization (ED4) or when chamberization was completed (ED6). Again, fusion of the VDR-like element to the pCAT-promoter construct (VDR:CAT) had no effect, positive or negative, on CAT expression in cardiomyocyte cell cultures at ED4 or ED6 (Fig. 8), nor did inclusion of the VDR-like element upstream of GATA (VDR-GATA:CAT) affect reporter activity in atrial cardiomyocytes from either ED4 or ED6 hearts (compare Fig. 8 with Fig. 4). By contrast, the inclusion of the VDR-like element suppressed reporter activity in ventricular cardiomyocytes from ED4 hearts by 78% and in those from ED6 hearts by 94% relative to expression in the atria (Fig. 8). We conclude that the VDR-like element is an inhibitory element, acting in a chamber-specific manner to inhibit transcriptional activation in ventricular cardiomyocytes. The increasing inhibitory activity of the VDR-like element in ventricular cardiomyocytes from ED3 to ED6 of development suggests that this element is involved in downregulating slow MyHC 3 gene expression during cardiogenesis.

Overexpression of GATA-4 in ventricular cardiomyocytes does not eliminate chamber-specific expression.

The effect of GATA-4 overexpression in ventricular cardiomyocytes was analyzed. Increasing amounts of the GATA-4 expression vector were cotransfected with SM3CAT:840D into ED6 atrial and ventricular cardiomyocytes. GATA-4 reactivated reporter expression in the ventricular cardiomyocytes in a dosage-dependent fashion (Fig. 9). At each concentration of GATA-4 tested there was a marked difference between atrial and ventricular expression. This observation suggests that the inhibitory state is not fixed in the ventricular cardiomyocytes and that expression of the slow MyHC 3 gene promoter in the ventricular cardiomyocytes is determined by a dynamic balance between positive and negative regulators.

FIG. 9.

Overexpression of GATA-4 permits expression from the slow MyHC 3 gene promoter in ventricular cardiomyocytes from chamberized hearts. Cotransfection of the SM3CAT:840D construct and various amounts of the GATA-4 expression vector activated the CAT reporter in the ventricular cardiomyocytes from ED6 hearts in a dose-dependent fashion. GATA-4 also slightly activated reporter expression in the atrial cardiomyocytes from ED6 hearts.

The slow MyHC 3 gene VDR-like motif is the binding site for VDR and RAR.

Transfection experiments showed that the VDR-like element in the slow MyHC 3 gene enhancer could suppress transcription in ventricular cardiomyocytes. EMSA was used to examine factors present in nuclear extracts from ED6 cultured atrial and ventricular cardiomyocytes that bind to the VDR-like element in the slow MyHC 3 gene enhancer. A double-stranded oligonucleotide of slow MyHC 3 gene sequence spanning positions −808 to −774 was synthesized. This region includes overlapping binding motifs (underlined) for the VDR (AGGACAAAGAGGGGA) and the RAR (AGGACAAAGAGGGGACA). This DNA fragment was ³²P radiolabeled and incubated with either atrial or ventricular nuclear extracts from ED6 heart cultures (Fig. 10A).

FIG. 10.

The VDR-like element is a binding site for VDRs and RARs. A ³²P-radiolabeled double-stranded DNA probe within the slow MyHC 3 gene enhancer sequence, spanning positions -808 to -774 and including a VDR-binding motif and an RAR binding site (Fig. 3A), was used in EMSAs. (A) Nuclear extracts of both atrial and ventricular ED6 heart cultures formed three major bands of reduced electrophoretic mobility (arrows). No differences in the positions of the retarded bands for atrial and ventricular extracts were distinguishable (lanes 2 and 5). Binding was specific for the VDR-like motif. Addition of a 300-fold molar excess of cold oligonucleotide (self) effectively eliminated binding (lanes 3 and 6), while no change in binding was observed when a 300-fold molar excess of an unrelated oligonucleotide (Non) was added (lanes 4 and 7). (B) The binding activity includes VDR, RXR, and RAR. Atrial and ventricular nuclear extracts from ED6 heart cultures were incubated with the labeled VDR-like probe (lanes 1 and 6). Preincubation of nuclear extracts of both atrial and ventricular cell cultures with monoclonal anti-VDR antibody disrupted the protein-DNA complexes (lanes 2 and 7), while preincubation with an unrelated monoclonal antibody (ctrl IgG) did not (lanes 3 and 8). Antiserum to RXRα produced a supershift of probe DNA (SS, lanes 4 and 9). No supershifts were seen when rabbit preimmune serum was first incubated with the nuclear extracts (lanes 5 and 10). Antiserum to RARα produced a supershift of probe DNA (SS, lanes 11 and 13). No supershifts were seen when rabbit preimmune serum was first incubated with the nuclear extracts (lanes 12 and 14).

Both atrial and ventricular nuclear extracts contained three major binding activities (at the positions of the arrows in Fig. 10A, lanes 2 and 5). Binding specificity was demonstrated by competition with unlabeled oligonucleotide (Fig. 10A, lanes 3 and 6) but not with an unrelated oligonucleotide (Fig. 10A, lanes 4 and 7). No differences were found in the binding affinities of the atrial and ventricular nuclear extracts for the VDR-like element (data not shown). The addition of a VDR monoclonal antibody, but not an unrelated monoclonal antibody, disrupted binding activities present in both atrial and ventricular extracts (Fig. 10B, lanes 2 and 7). Thus, VDR is present in atrial and ventricular cardiomyocytes and can bind to the VDR-like motif present in the slow MyHC 3 gene enhancer. Because nuclear hormone receptors generally bind as heterodimers with RXR (34), we tested for their presence in atrial and ventricular extracts. The addition of an antiserum against RXRα produced a supershifted band with atrial and ventricular extracts (Fig. 10B, lanes 4 and 9), while preimmune serum did not (Fig. 10B, compare lanes 5 and 10). The stronger supershifted signal evident in the ventricular extracts was not a consistent result. The results of multiple experiments suggest no consistent difference between the signals resulting from the atrial and ventricular extracts. We also examined the ability of an RAR to bind the VDR-like element. Using the same oligonucleotide probe, addition of an antiserum against RARα, but not preimmune serum, to atrial and ventricular extracts produced a supershifted band (Fig. 10B, compare lanes 11 and 13). Therefore, we conclude that both VDRs and RARs, probably as heterodimers with RXRs, can bind the VDR-like element of the slow MyHC 3 gene promoter.

To further explore the mechanism by which the VDR-like element inhibits slow MyHC 3 expression in ventricular cardiomyocytes, the effect of VDR or RARα overexpression was analyzed. The reporter construct SM3CAT:840D is expressed at a low level in ventricular cardiomyocytes isolated from ED4 (56) (Fig. 11A). Cotransfection of SM3CAT:840D with a VDR expression vector, alone or in combination with an RXRα expression vector, resulted in 3.5- and 4.2-fold inhibition, respectively, in ventricular cardiomyocytes isolated from ED4 hearts (Fig. 11A). In contrast, cotransfection of SM3CAT:840D with an RARα expression vector, alone or with the RXRα expression vector, had little effect on expression of the reporter in the ED4 ventricular cardiomyocytes (Fig. 11A). Reporter expression from the SM3CAT:840D construct was unaffected by cotransfection of VDR, RARα, VDR-RXRα, or RARα-RXRα into ED4 atrial cardiomyocytes (Fig. 11B). Together these results suggest that it is the VDR, rather than the RAR, that regulates atrial chamber-specific expression of the slow MyHC 3 gene. The importance of the VDR-like element was shown by its mutation in the context of SM3CAT:840D (mVDR construct) (Fig. 11C). The VDR expression vector alone, or the VDR and RXRα expression vectors, was cotransfected into ventricular cardiomyocytes with the mVDR construct. Only slight inhibition of reporter expression was detected, implicating the VDR-like element in the inhibition of SM3CAT:840D seen with overexpression of the VDR expression vector (Fig. 11A).

FIG. 11.

Overexpression of the VDR, but not the RAR, specifically inhibited reporter expression in the ventricular cardiomyocytes from ED4 heart. (A) Cotransfection of SM3CAT:840D with expression vectors encoding VDR (or VDR plus RXRα), but not RARα (or RARα plus RXRα), inhibited reporter expression in the ventricular cardiomyocytes. (B) VDR, VDR plus RXRα, RARα, or RARα plus RXRα had no effect on reporter expression in the atrial cardiomyocytes. (C) The suppression by the VDR was via the VDR-like element, since cotransfection of mVDR, in which the VDR-like element was mutated in the context of SM3CAT:840D, with the VDR (or VDR plus RXRα) expression vector only slightly inhibited reporter expression.

DISCUSSION

Within a relatively short period of time, the developing vertebrate heart undergoes a complex series of morphogenetic movements that transform a simple, straight tube into a complex four-chambered organ. Distinct lineages of prospective atrial and ventricular cardiomyocytes emerge in the posterior and anterior regions of the heart tube, respectively. An early marker of the atrial cardiomyocyte cell lineage is expression of the AMHC1 or slow MyHC 3 contractile protein gene (43, 60). We have used the identification of cis elements and trans-acting factors that regulate atrial chamber-specific expression of this gene as a means of investigating the mechanism(s) underlying diversification of early cardiogenic cells into atrial or ventricular cell lineages and as an approach to the analysis of cardiac morphogenesis.

This study confirmed the very early onset of slow MyHC 3 gene expression, by the tubular heart stage (HH10) of avian development (60). In contrast to that in the chicken embryo (60), slow MyHC 3 expression in the quail embryo is initially nearly uniform throughout the tubular heart. Subsequently, expression of slow MyHC 3 becomes restricted to cardiomyocytes of the anterior heart tube as they give rise to the atria. Previous work identified a 160-bp enhancer in the slow MyHC 3 gene promoter responsible for the observed chamber-specific expression of this gene (56). Furthermore, deletion and mutational analyses identified a VDR-like motif as a critical element in its regulatory control (56).

Several groups have reported that the GATA element is important for the regulation of expression of several cardiac genes (15, 18, 38, 52). Here we have shown that the GATA element in the promoter of the slow MyHC 3 gene is an activator of transcription in cardiomyocytes and that GATA factors positively regulate slow MyHC 3 gene expression throughout the tubular heart and subsequently in the atria. During the transition from a tubular to a chambered heart, restriction of slow MyHC 3 gene expression to the atria is mediated by the VDR-like element, which acts as an inhibitory element in ventricular, but not atrial, cardiomyocytes. Thus, as the heart completes morphogenesis, the final pattern of atrial chamber-specific slow MyHC 3 gene expression in the heart results from the positive action of the GATA element in the atria and from inhibition via the VDR-like element in the ventricles. The progressive loss of slow MyHC 3 gene expression in the embryonic ventricle is concurrent with an increased inhibitory activity associated with the VDR-like element.

An anterior-posterior gradient of slow MyHC 3 expression develops as cardiac morphogenesis proceeds.

An anterior-posterior gradient of slow MyHC 3 expression is clearly visible by ED3, when the demarcation of the tubular heart into prospective atrial and ventricular segments becomes morphologically evident (Fig. 1). Analyses of steady-state mRNA levels (Fig. 2) suggest that the observed gradient of slow MyHC 3 expression is regulated at the transcriptional level. High steady-state levels of slow MyHC 3 mRNA are maintained in atrial cardiomyocytes from the posterior end of the cardiac tube, while prospective ventricular cardiomyocytes from the anterior end show a gradual reduction in the amount of slow MyHC 3 mRNA throughout the embryonic period of heart development (Fig. 2).

Similar to the mammalian β-cardiac/slow MyHC gene, the slow MyHC 3 gene is expressed in both slow skeletal and cardiac muscle cells, and the two genes have a high level of sequence homology in their coding regions (43). However, whereas the slow MyHC 3 gene becomes atrial chamber restricted, the β-cardiac/slow MyHC 3 gene is a ventricular chamber-specific gene (31). The slow MyHC 3 gene exhibits an equally high level of sequence homology to the mammalian α-MyHC gene (43, 60), but again the pattern of slow MyHC 3 gene expression is different from that of its mammalian counterpart. Like the slow MyHC 3 gene, α-MyHC is initially expressed throughout the tubular heart (7, 30, 31), and high levels of expression are maintained in the atria. However, following a transient downregulation in the ventricles between 10.5 and 16.5 days postcoitum, α-MyHC expression increases in the ventricles and ultimately replaces β-MyHC in all postnatal ventricular cardiomyocytes (31). Thus, neither α-MyHC nor β-MyHC shows the atrial chamber specificity that the slow MyHC 3 gene does in birds.

In mammals, the gene to demonstrate the earliest atrial chamber restriction during development is MLC-2a (21). Similarly to the slow MyHC 3 gene, MLC-2a is initially expressed throughout the tubular heart at ED8 in the mouse embryo and is downregulated in the ventricular segment during chamber formation. The downregulation of MLC-2a in the ventricular chamber is initiated by ED9 and is completed by ED12. In contrast, MLC-1a shows relatively late chamber restriction during mouse development. Downregulation of MLC-1a in the ventricles begins during fetal development, but detectable levels are observed in the ventricles even after birth (30). The mechanisms for the downregulation of MLC-2a and MLC-1a in the ventricles are not clear.

As development progresses, the ANF gene demonstrates changes in expression in the chambers of the heart. Expression of ANF is first detected at ED8 of mouse development (61). Throughout embryonic and fetal development, ANF is expressed along the anterior-posterior axis of the heart tube, in both atrial and ventricular cardiomyocytes. Soon after birth, ANF expression in ventricular cardiomyocytes declines rapidly to a low but detectable level (approximately 1% of that of the adult atria) (2). While atrial chamber-restricted expression of ANF also results from a downregulation in the ventricles, the timing of downregulation is different from that of the slow MyHC 3 gene. ANF is downregulated after complete chamberization and in cells already expressing a ventricular cardiomyocyte phenotype, while the slow MyHC 3 gene is downregulated prior to the formation of distinct cardiac chambers and concurrent with commitment of early cardiogenic cells to the ventricular cardiomyocyte cell lineage. For this reason, the slow MyHC 3 gene is especially well suited for investigations into early events leading to the diversification of cardiomyocytes to an atrial or ventricular lineage. Furthermore, because this diversification occurs concomitant with morphogenesis of the heart, it will be of interest to determine if, or how, these two processes are interrelated.

An anterior-posterior gradient has also been observed in the ventricles in the expression of a MLC-2v transgene with a lacZ reporter (45). Initially the lacZ reporter is expressed in a bilaterally symmetrical manner in the prospective ventricles at the headfold stage. Subsequently there is a higher level of lacZ expression in the right ventricle than in the left ventricle, although the endogenous MLC-2v gene is uniformly expressed throughout the ventricles (45).

Control of atrial chamber-specific expression.

The establishment of an anterior-posterior gradient and, subsequently, atrial chamber-specific expression of the slow MyHC 3 gene could be achieved by downregulation of cardiac transcriptional activators in the ventricles, upregulation of ventricular chamber-specific inhibitors, or both mechanisms acting simultaneously. To our knowledge, there is no evidence to suggest that the known cardiac transcriptional activators GATA-4/5/6, Nkx2-5, and MEF2 are developmentally downregulated in the ventricles (13, 17, 19, 20, 23, 29, 36, 41, 42). In the case of the slow MyHC 3 gene, our data suggest that GATA alone is sufficient to direct heart-specific expression and that Nkx2-5 and MEF2C play no direct role (Fig. 6). Because EMSA identified GATA binding activity in both atrial and ventricular cell extracts (Fig. 7) and transfection studies found no temporal or spatial differences in GATA activity (Fig. 4), quantitative differences in factors that bind to the GATA element alone cannot account for the gradient of slow MyHC 3 gene expression that develops during cardiac morphogenesis.

The alternative hypothesis, that the anterior-posterior gradient in the early heart is driven not by the distribution of positive factors but by the imposition of restraints on expression, appears to be the mechanism involved. We found that inclusion of the VDR-like element upstream of the GATA-binding site inhibited reporter activity driven by the heterologous SV40 promoter by 36% in ventricular cardiomyocytes from ED3 tubular heart (Fig. 8) and that the magnitude of inhibition gradually increased during the transition from a tubular to a chambered heart. This suggests that specific inhibition acting through the VDR-like element in the anterior portion of heart is responsible for the gradient of slow MyHC 3 gene expression. Our data suggest that atrial chamber-specific expression of the slow MyHC 3 gene is achieved by stimulatory activity through the GATA element in the atria and by specific inhibition through the VDR-like element in the ventricles.

An inhibitory element is also found in the rat ANF promoter (11). In contrast to the slow MyHC 3 gene, an Nkx2-5 response element, termed the NKE, is required for expression of ANF promoter constructs in atrial cardiomyocytes (11). Interestingly, a deletion removing a small region of the ANF promoter, including the NKE, leads to upregulation of a reporter in cultured ventricular but not cultured atrial cardiomyocytes. This suggests that the NKE, or an adjacent site, binds an inhibitor restricted to ventricular cardiomyocytes (11). The sequence of the inhibitory element in the ANF promoter has not been defined, nor is the mechanism by which the inhibitory element suppresses ANF expression in the ventricle clear.

The VDR-like motif in the slow MyHC 3 gene enhancer has homology to binding sites for a family of nuclear hormone receptors, including RAR and RXR, VDR, and thyroid hormone receptor (34). Slight differences in primary sequence appear to dictate how well individual members of this family can bind, and there is evidence that they bind primarily as heterodimers, with RXR acting as one of the pair (34). Many studies have suggested a role for RA in establishing an anterior-posterior axis in the developing heart (12, 49, 50, 60). Bader and colleagues (59, 60) found that addition of RA to explants of undifferentiated mesoderm isolated from the anterior (prospective ventricles) cardiac region of the gastrulating embryo evoked the development of an atrial phenotype, as evidenced by the activation of the AMHC1 gene. In the zebra fish, application of RA causes a preferential deletion, first of the ventricle and then of the atrium (49), providing further evidence that the nuclear hormone receptor family may play an important role in chamberization of the heart. Most recently, RA was shown to block differentiation of the myocardium after heart specification in Xenopus laevis (9), suggesting that RA may suppress the function of cardiac transcription factors which activate differentiation. RA also suppress both phenylephrine- and endothelin-stimulated ANF upregulation and hypertrophy of neonatal rat cardiomyocytes (62). Both vitamin D₃ and RA antagonize endothelin-induced ANF upregulation and hypertrophy of neonatal cardiomyocytes (57). Although the heart is not thought to represent a classical target for vitamin D₃, this vitamin has been shown to inhibit ANF expression in atrial cardiomyocytes (26, 58). We found that both VDRs and RARs present in atrial and ventricular nuclear extracts can bind (probably as heterodimers with RXRs) to the VDR-like motif in the slow MyHC promoter (Fig. 10). However, transfection of VDR, but not of RAR, was able to inhibit reporter expression in the ventricular cardiomyocytes (Fig. 11A). Mutation of the VDR-like motif demonstrated that the observed inhibition in ventricular cardiomyocytes by overexpression of VDR functions through the VDR-like element (Fig. 11C). Overexpression of either VDRs or RARs had no effect on reporter activity in the atrial cardiomyocytes (Fig. 11B). These data suggest a role for the VDR as an inhibitor of slow MyHC 3 gene expression in the ventricles during chamber formation.

EMSA did not detect consistent differences between atrial and ventricular nuclear extracts with regard to binding activity to the slow MyHC 3 VDR-like element (Fig. 10). At least two mechanisms may be involved in the VDR-dependent inhibition observed in the ventricles. First, the VDR-RXR heterodimer could interact with a ventricle-specific transcriptional repressor. In this regard, a transcriptional repressor, SMRT, has been shown to inhibit transcription via binding to the unliganded RAR (34). Alternatively, posttranslational modifications of the VDR in atrial or ventricular cardiomyocytes could affect transcriptional activation. Additional studies of the VDR in vivo will provide further insights into the patterning of gene expression in the tubular heart and into early cardiomyocyte lineage diversification.