Atrial chamber-specific expression of the slow myosin heavy chain 3 gene in the embryonic heart.

The quail slow myosin heavy chain 3 (slow MyHC 3) gene is expressed in the developing heart and in slow muscles of the developing limb. It is first expressed in the pulsatile cardiac tube in the embryo, and as the heart chamberizes its expression becomes restricted to the atria. To identify regulatory elements responsible for atrial-specific expression, the 5′ upstream region of slow MyHC 3 gene was investigated. An atrial regulatory domain (ARD1) between −840 and −680 acts as an atrial cell-specific enhancer in primary cardiocyte cultures. ARD1 also specifies atrial-specific expression in vivo when the ARD1/heterologous promoter was introduced into developing chick embryos by a replication-competent retroviral vector. ARD1 is the first atrial cell-specific enhancer to be identified. Fine deletion and mutation analysis within ARD1 defined a 40-base pair vitamin D3 receptor-like element that controls atrial cell-specific expression of the slow MyHC 3 gene by inhibiting its expression in ventricular cardiocytes.

The quail slow myosin heavy chain 3 (slow MyHC 3) gene is expressed in the developing heart and in slow muscles of the developing limb. It is first expressed in the pulsatile cardiac tube in the embryo, and as the heart chamberizes its expression becomes restricted to the atria. To identify regulatory elements responsible for atrial-specific expression, the 5 upstream region of slow MyHC 3 gene was investigated. An atrial regulatory domain (ARD1) between ؊840 and ؊680 acts as an atrial cell-specific enhancer in primary cardiocyte cultures. ARD1 also specifies atrial-specific expression in vivo when the ARD1/heterologous promoter was introduced into developing chick embryos by a replication-competent retroviral vector. ARD1 is the first atrial cell-specific enhancer to be identified. Fine deletion and mutation analysis within ARD1 defined a 40-base pair vitamin D 3

receptor-like element that controls atrial cell-specific expression of the slow MyHC 3 gene by inhibiting its expression in ventricular cardiocytes.
One of the earliest organs to form, the vertebrate heart develops as a pulsatile tube by the ventral fusion of two cardiac mesoderm regions of the early embryo (1). The beating tubular heart rapidly folds to form chambers (the atria and ventricles) that differ in morphology, physiology, and in the muscle contractile protein genes each expresses (2)(3)(4)(5). Molecular mechanisms for cardiogenesis (6 -11) and for atrial and ventricular chamber specification are not well understood. Based on the coexpression of ␣and ␤-MyHC 1 genes throughout the tubular heart in mammals, it has been suggested that atrial and ventricular chamber specification occurs only when the tubular heart begins to chamberize (12). More recently, it has been proposed that diversification of atrial and ventricular cells may occur earlier because an atrial-specific MyHC gene, AMHC1, is first expressed in the posterior region of the fusing cardiac tube, the future atrial compartment of the heart (13).
To understand cardiac development, several genes expressed in the developing heart such as MLCs (14,15), ␣-MyHC (16 -19), ␤-MyHC (20 -22), cardiac troponin T (23,24), muscle creatine kinase (25), ANF (26,27), and cardiac ␣-actin (28,29) have been studied to delineate cis-elements that interact with transcription factors to control gene expression in the cardiac compartments. While cis-elements and transcription factors have been identified that regulate these genes in the heart (30 -39), distribution of the identified transcription factors in atrial and ventricular chambers of the heart has shed little light on the molecular mechanisms generating atrial and ventricular cell types.
The ventricular chamber-specific contractile protein gene, MLC-2v, has served as a model system to identify cis-elements and trans-acting factors that restrict the expression of genes to the ventricular chamber. Both positive (HF-1a and HF-1b) and negative (HF-3) cis-elements have been shown to regulate restricted expression of the MLC-2v gene in adult transgenic mouse (36). However, cis-elements have not been reported for atrial-specific genes either during embryonic development or in the adult.
We have reported the identification and characterization of a new slow myosin heavy chain gene, slow MyHC 3, that is expressed in the developing quail heart and embryonic slow skeletal muscles (40) and that is closely related (if not homologous) to the chicken AMHC1 (13). Initially the slow MyHC 3 gene is expressed throughout the tubular heart, but as the heart chamberizes, expression of the slow MyHC 3 gene in the ventricles is down-regulated, whereas expression in the atria is maintained. 2 By delineating cis-elements within the regulatory region of the gene, we can identify a transcription factor or factors that are important for specification of the atrial chamber. We have identified four regions within the slow MyHC 3 promoter that contain either positive or negative cis-elements. One of these regions, 160 bp of the 5Ј flanking sequence between Ϫ840 and Ϫ680 designated as atrial regulatory domain 1 (ARD1), functions as an atrial-specific enhancer in primary cardiocyte cultures as well as in the embryo. The function of the ARD1 in vivo was investigated by using an avian retroviral vector, RCAN/PCAT/F (41), to deliver an ARD1/SV40 promoter/ CAT construct into embryos. In developing embryos ARD1 increased expression of the CAT reporter 12.3-fold in the atria compared with other non-cardiac tissues including skeletal and smooth muscle. Deletion and mutation of the ARD1, coupled with transient transfection of atrial and ventricular cardiocyte cultures, identified a 40-base pair vitamin D 3 receptor (VDR)like element essential for atrial-specific expression. The VDRlike element residing within the ARD1 was found to regulate atrial-specific expression by inhibiting reporter expression in ventricular cardiocytes.

EXPERIMENTAL PROCEDURES
Heart Cell Cultures-Quail eggs were purchased from Strickland Quail Farm, Pooler, GA. Embryonic atrial and ventricular cells were prepared as described by Barry et al. (42). The atria and ventricles from ED6 quail hearts were dissociated at 37°C with constant agitation during six successive 8-min incubations in 0.025% trypsin (Difco) in the Ca 2ϩ ,Mg 2ϩ -free Hanks' solution. For transfection, cells were plated at densities of 2.5 ϫ 10 6 on 60-mm collagen-coated dish (Collagen, bovine tendon, Worthington). For immunostaining, 2-5 ϫ 10 5 cells were plated on 35-mm collagen coated dishes. The atrial and ventricular cardiocytes were initially plated in heart serum-containing medium consisting of 40% M-199, 53% basic salt solution, 6% fetal calf serum (Hy-Clone), 1% glutamine, and 0.1% penicillin/streptomycin (42). On day 2 of incubation the medium on all dishes was changed to heart serum-free medium composed of Dulbecco's modified Eagle's medium/F12 containing 10 Ϫ6 M insulin, 5 g/ml transferrin, 10 Ϫ9 M selenium, 1% glutamine, 0.1% penicillin/streptomycin according to Libby (43). The cultures were fed daily after day 2 with heart serum-free medium.
Immunocytochemistry-Immunostaining for MyHC in culture was described previously (44,45). mAb F59 reacts with fast isoforms of MyHC in striated muscles, and mAb S58 reacts with slow isoforms including slow MyHC 3 (45,46). mAb F59 binding was visualized with Texas Red-labeled rabbit anti-mouse IgG (Vector) or with mAb F59 rhodamine-conjugated primary antibody when double-stained with a CAT antibody, and mAb S58 with fluorescein isothiocyanate-labeled rabbit anti-mouse IgA (Zymed). A polyclonal antibody to CAT (5 Prime to 3 Prime, Inc.) was used at 1:150 dilution and binding was detected by a fluorescein isothiocyanate-labeled goat anti-rabbit IgG.
RNA Analysis of Slow MyHC 3 Expression-Total cellular RNA was isolated from cultured cardiocytes or tissues by RNA STAT-60 according to the protocol provided by the manufacturer (TEL-TEST "B," Inc.). A total of 5 g of RNA was denatured and loaded into each slot in a slot-blot apparatus. The membrane was prehybridized in 20 ml of solution containing 6 ϫ SSC, 5 ϫ Denhardt's, 100 g/ml denatured salmon sperm DNA, and 0.5% SDS for 4 h at 42°C. The blots were probed with a slow MyHC 3Ј-specific oligonucleotide, the 3Ј-UTR (5Ј-AAG GGA ATT CAT CAG AGG TTG GGG CT-3Ј). Hybridization was conducted overnight at 42°C. Membranes were washed at 42°C for 2 h with four changes of 0.2 ϫ SSC. By using serial dilutions of atrial RNA, exposure times were chosen such that comparisons were made in the range in which the response of the atrial RNA was not saturated. The slot blots were analyzed by densitometric reading of bands on an Alpha Innotech IS-1000 image analyzer (Alpha Innotech, San Leandro, CA).
Western Blotting-Myosins were extracted from atrial and ventricular cardiocyte cultures or heart tissue according to the method of Crow and Stockdale (45). Protein concentrations were determined by Bradford assay (47). Denatured myosins were electrophoresed on a 5% SDS-PAGE gel and transferred to a nitrocellulose membrane. The blot was immunostained with mAb S58, followed by horseradish peroxidasegoat anti-mouse IgA (Zymed). The tissue extracts from fetuses that were prepared for CAT assay (see below) were also used for Western blot analysis in 12% SDS-PAGE gels to detect viral capsid protein p27 expression in specific tissues of the virus-infected fetuses. Samples were prepared for loading by mixing 1 volume of tissue extract to 1 volume of a 2 ϫ loading buffer. Separated proteins were transferred to a nitrocellulose membrane. The viral p27 was detected with a 1:1000 dilution of rabbit anti-p27 antibody followed by horseradish peroxidase-goat antirabbit IgG (Sigma). Horseradish peroxidase was visualized with diaminobenzidine tetrohydrochloride in the presence of 0.03% NiCl. Serial dilutions of concentrated RCAN/ARD1-PCAT/F retrovirus stock prepared as described by Cepko (53) were also loaded and used as standards to monitor the p27 expression level in different tissues.
Slow MyHC:CAT Chimeric Constructs-Two genomic clones, QSM4 and QSM6 (40), which overlap and contain about 17 kilobase pairs of 5Ј flanking sequence of the slow MyHC 3 gene were used for making slow MyHC-reporter constructs. The promoterless pCAT-basic vector and the pCAT-promoter vector containing a SV40 promoter without enhancer were from Promega Biotech, Madison, WI. Standard cloning procedures (49) were used to insert slow MyHC 3 genomic sequences upstream of the bacterial CAT reporter in the pCAT-basic vector. Making use of convenient restriction sites, approximately 8500 bp of 5Ј flanking sequence of the slow MyHC 3 gene was cloned into the pCATbasic. A series of 5Ј deletions removed portions of slow MyHC 3 promoter sequence. All CAT constructs in this series contained the first exon, the first intron, and part of the second exon, in addition to varying amounts of 5Ј flanking sequence. Each construct has been designated by the amount of 5Ј flanking sequence it contains.
Two heterologous promoter constructs were made as follows. A 160-bp fragment (ADR1) between Ϫ840 and Ϫ680 of the 5Ј flanking sequence of slow MyHC 3 gene was PCR-amplified (primer forward, 5Ј-ATA GAT CTC TTG TTC TGG GTG TGT AT-3Ј; primer reverse, 5Ј-TAA GAT CTT CTA TGG GTC TTT TGG GT-3Ј) and cloned into a BglII site of a pCAT-promoter vector with either forward orientation, ADR1/SV/F, or reverse orientation, ADR1/SV/R, with respect to the SV40 promoter in the vector.
The RCAN/PCAT/F viral vector and Cla12 adapter plasmid, gifts from Dr. Stephen Hughes (50), were used to construct a recombinant retrovirus plasmid. The 410-bp EcoRI-HindIII fragment from ADR1/ SV/F containing the ADR1 and SV40 promoter was subcloned into a Cla12 plasmid, designated as Cla0.4. The ClaI fragment from Cla0.4 was inserted into the ClaI site of the RCAN/PCAT/F retroviral vector in a 5Ј to 3Ј orientation with respect to the CAT reporter gene embedded in the vector and was designated as RCAN/ADR1-PCAT/F. Transient Transfection and CAT Assay-Atrial and ventricular cardiocyte cultures were fed with fresh heart serum-containing medium and transfected after overnight culture using the calcium-phosphate precipitate method as described previously (40). In all transfection experiments 5 g of CsCl-purified SM3CAT plasmid DNA and 2 g of reference plasmid psv-␤-gal DNA containing the Escherichia coli ␤-galactosidase gene driven by the simian virus 40 (SV40) early promoter were cotransfected to 60-mm dishes of cardiocyte monolayers for 7 h. The cardiocytes were given a glycerol shock for 1 min after transfection and were then placed in heart serum-free medium for an additional 48 h before harvesting. The transfection of chicken embryonic fibroblasts (CEFs) was performed using the procedure as described for cardiocytes except that the CEFs were transfected for 4 h. All the transfection experiments were repeated at least three times. A separate culture dish was used for transfection with a SV40:CAT control plasmid (Promega) in every experiment.
Extracts of transfected cells from 60-mm culture dishes were prepared by 3 cycles of freeze-thaw in 30 l of 0.25 M Tris-HCl, pH 7.5. Tissue extracts from fetuses used for CAT assay and p27 expression were prepared by homogenization in 0.25 M Tris, pH 7.5, and incubated at 65°C for 10 min. Supernatants were collected after centrifugation at 10,000 rpm for 5 min at 4°C.
CAT activity was determined as described previously (40) using [ 14 C]chloramphenicol. All CAT assay were performed in the linear range of the assay. CAT assays were quantitated by excision of spots from thin layer chromatographs and liquid scintillation counting. ␤-Galactosidase activity in the extract was measured according to the method of Miller (48). ␤-Gal units were determined by comparison to a standard curve made with E. coli ␤-Gal standards. Within each experiment, the effect of differences in transfection efficiency between plates was minimized by expressing the activity of each slow MyHC 3:CAT construct relative to ␤-Gal activity. Because transfection efficiencies of CEFs are greater than atrial or ventricular cardiocytes, a separate culture dish was used for transfection with a SV40:CAT control plasmid (Promega) in every experiment. To allow comparisons among the three cell types, expression from each of the slow MyHC 3:CAT constructs is reported relative to the SV40:CAT control plasmid.
Gene Delivery in Vivo via an RCAN/PCAT/F Retrovirus-Pathogenfree fertilized chicken eggs were purchased from SPAFAS Inc. Chicken embryo fibroblasts (CEFs) were prepared from trunks of two ED12 embryos according to Hunter (51). Cells were grown in Dulbecco's modified Eagle's medium containing 10% fetal bovine serum (Life Technologies, Inc.), 2% chicken serum (Life Technologies, Inc.), and 1% penicillin/streptomycin. Cells were passaged 1:3 at confluence and were used for viral DNA transfection after three passages. The viral DNA was introduced into CEFs by calcium-phosphate transfection. One week after calcium-phosphate transfection all CEFs in the dishes were infected by retrovirus particles as judged by staining with rabbit antiviral capsid protein p27 (SPAFAS, Inc.). These virus-producing CEFs were used to infect embryos.
To infect developing embryos 100 l of culture medium containing approximately 10 6 virus-producing CEFs was injected into fertilized but unincubated chicken eggs at Hamburger and Hamilton stage 1 on the day the eggs were set (41,52). A small hole that would allow a hypodermic needle to pass was made by drilling with a small scissors near the center of the large end of eggs. Injections were targeted to the surface of egg yolk close to the blastodisc but not into the blastodisc itself. Injection sites were localized in pilot experiments by using a dye solution. The injection holes were sealed with paraffin. Infected eggs were incubated until the embryos completed fetal development at ED18. Blood was collected from these fetuses prior to sacrifice and tested for the presence of the viral capsid antigen p27 by enzyme-linked immunosorbent assay. Tissues were collected and stored at Ϫ80°C, and CAT assays were performed with extracts of these tissue samples.
Whole Mount Staining-Whole mount immunostaining was performed according to Page et al. (46) to identify embryos infected with the retrovirus. The RCAN retroviral infected embryos were incubated overnight at 4°C with a 1:2500 dilution of rabbit antiviral capsid protein p27 antibody (anti-p27) and visualized with horseradish peroxidase-goat anti-rabbit IgG (Sigma).
Mutations were made in two sites within the SM3CAT:840D by PCR mutagenesis (49). The AGGACA half-site in the VDR-like element from position Ϫ801 to Ϫ796 was replaced by a SalI restriction site, GTCGAC. Sequences from Ϫ840 to Ϫ796 and Ϫ801 to ϩ18 were PCR-amplified, respectively. The PCR products were digested by HindIII ϩ XbaI and directionally cloned into the pCAT-basic vector. The mutant construct was designated mut-VDR. Mutations were also introduced between Ϫ780 and Ϫ775 in the SM3CAT:840D in which sequence 5Ј-GGCGGA-3Ј was replaced by SalI site, 5Ј-GTCGAC-3Ј. This mutant construct served as a control for mut-VDR and was designated mut-Ctrl. All constructs were sequenced using the dsDNA Cycle Sequencing System (Life Technologies, Inc.).

Characterization of Atrial and Ventricular Cardiocytes in
Primary Culture-Embryonic atrial and ventricular cardiocytes isolated from ED6 quail hearts express myosin heavy chains when cultured on collagen-coated dishes. These cultures initially contained 90% atrial or ventricular cardiocytes and 10% non-cardiocytes as distinguished by immunostaining for fast MyHC isoforms using the mAb F59. Although both cardiocytes and non-cardiocytes increased in number in the culture period, the percentage of cardiocytes in the culture decreased to about 75% by culture day 3 and remained at this level for at least 14 days of culture. Changing to heart serum-free medium on the 2nd day of culture promoted the differentiation of atrial and ventricular cardiocytes and avoided overgrowth of noncardiocyte cells.
As determined by cell morphology, beating rates, and staining for MyHC expression, atrial and ventricular cardiocyte cultures were healthy for up to 2 weeks of incubation. All cardiocytes, whether ventricular or atrial, were identified by the expression of fast MyHC isoforms staining with mAb F59 (Fig. 1, B and E). At day 3 in culture, both atrial and ventricular cardiocytes were beating synchronously. Atrial and ventricular cardiocytes could be distinguished by differences in morphology and slow MyHC expression. Atrial cardiocytes had a less flattened morphology on the collagen substrate than ventricular cardiocytes (compare Fig. 1, A and D), and atrial, but not ventricular, cardiocytes stained with mAb S58 which recognizes slow MyHC 3 ( Fig. 1, C and F).
The amount of slow MyHC 3 transcripts in the chambers of embryonic hearts and in atrial and ventricular cultures were quantitated by RNA slot blots. Total cellular RNA were extracted from ED6 atria, ventricles, and from atrial and ventricular cultures maintained for 2, 5, and 7 days. RNA samples were probed with a 3ЈUTR oligonucleotide specific for slow MyHC 3. At ED6, slow MyHC 3 was expressed in vivo in the atrium (Fig. 2, A and B, lane A) at a 6.7-fold higher level than the ventricle (Fig. 2, A and B, lane F). Expression of the slow MyHC 3 RNA in atrial cardiocyte cultures was sustained for the culture period at a constant level (Fig. 2, A and B, lanes  B-D). In contrast, by 2 days the level of slow MyHC 3 transcripts in the ventricular cardiocyte cultures (Fig. 2, A and B, lanes G-I) was similar to control levels seen in RNAs from liver (Fig. 2, A and B, lane J) or from ED18 pectoralis major skeletal muscle (Fig. 2, A and B, lane E), a skeletal muscle that in the adult expresses fast, but no slow, MyHCs. Extending the time in culture up to 7 days did not lead to expression of slow MyHC 3 RNA in ventricular cardiocyte cultures (Fig. 2, A and B, lane I). These atrial cardiocyte cultures expressed 9-fold more slow MyHC 3 than ventricular cardiocyte cultures.
Expression of slow MyHC 3 protein was examined by Western blotting using extracts from atrial and ventricular cultures at days 2, 5, and 7 of incubation (Fig. 2C). In the cultures of atrial cardiocytes, slow MyHC 3 protein expression remained constant at high levels (Fig. 2C, lanes C, E, and G), similar to that present in ED6 atria in vivo (Fig. 2C, lane A). In contrast, slow MyHC 3 was undetectable in extracts of cultured ventricular cardiocytes at days 2, 5, or 7 (Fig. 2C, lanes D, F, and H). We conclude that atrial cardiocytes express slow MyHC 3 protein at constant high levels for at least a week in culture, whereas ventricular cardiocytes do not express slow MyHC 3 at any point during the culture period. Therefore, cultures of atrial and ventricular cardiocytes comprise a model system to investigate the mechanism of atrial-specific expression of the slow MyHC 3 gene.
Localization of an Atrial-specific cis-Element(s) in the Slow MyHC 3 Gene-To define the atrial-specific cis-element(s) of the slow MyHC 3 gene, a series of reporter constructs were made from two genomic clones, QSM4 and QSM6 (40). The slow MyHC 3 regulatory regions ranging from 290 to 8500 bp upstream from transcription start site and including intragenic sequence of exon I, intron I, and a portion of exon II were fused to the bacterial chloramphenicol acetyltransferase (CAT) gene. Convenient restriction sites, shown in Fig. 3, A and B, were employed to generate progressive deletions from the 5Ј end of the slow MyHC 3 promoter (Fig. 3C). Each of these constructs were cotransfected with psv-␤-gal into atrial cardiocytes, ventricular cardiocytes, or CEF cultures. In the atrial cardiocyte cultures, the SM3CAT:8500 construct was expressed at a high level (Fig. 3D). Activity was enhanced a small amount when 2500 bp of 5Ј flanking sequence were deleted (i.e. SM3CAT: 6000). Expression was sustained at a very high level in the SM3CAT:4500 construct but decreased 3-fold as the deletion was extended to position Ϫ3200. This level of expression was maintained in atrial cardiocytes until the regulatory region was further truncated to SM3CAT:680. Deletion to Ϫ680 reduced the CAT activity to a level comparable to the promoterless CAT construct, pCAT-basic. Interestingly, activity was dramatically increased to the highest level observed as the deletion proceeded to Ϫ290 (SM3CAT:290).
The cell specificity of the slow MyHC 3 promoter was assessed by comparing the level of CAT expression in cultured atrial cardiocytes with that in cultured ventricular cardiocytes and with that in CEFs (Fig. 4). The shortest construct, SM3CAT:290, containing 290 upstream nucleotides showed an equivalent level of expression in ventricular and atrial cardiocytes and was also expressed at a diminished but significant level in the CEFs. Hence SM3CAT:290 expression is not cardiac chamber-specific. The SM3CAT:680 construct is not expressed at significant levels in either cardiocytes or fibroblasts, whereas constructs containing 840 bp or more of upstream sequence all showed atrial cardiocyte-specific expression.
Expression of the SM3CAT:840 in atrial cardiocytes was 7-fold higher than that of ventricular cardiocytes and 14-fold higher than that of CEFs (Table I). Inclusion of additional 5Ј flanking sequence to Ϫ3200 had a modest increase in the atrial/CEF ratio and no effect on the relative expression between atrial and ventricular cultures. With the addition of slow MyHC 3 sequence between Ϫ3200 and Ϫ4500, the level of expression in atrial cultures increased to 15-fold higher than in ventricular cultures and 2 orders of magnitude higher than in CEFs, largely due to increased expression in atrial cultures.
Transfection studies have mapped four regulatory regions associated with differences in levels of reporter expression (Fig.  4). First, the 290 bp upstream from the transcription site directs high levels of non-chamber-specific expression. Second, the upstream region between positions Ϫ680 and Ϫ290 contains a negative cis-element(s) which strongly inhibits the reporter activity in all three cell types. Third, the region between Ϫ840 and Ϫ680 contains an atrial cardiocyte-specific cis-element(s), permitting expression in atrial cardiocytes but not in ventricular cardiocytes or CEFs. Fourth, the region between Ϫ4500 and Ϫ3200 further increased the level of CAT expression by 3-fold in atrial cardiocytes but had no effect on ventric- ular cardiocytes or CEFs. Therefore, the expression of slow MyHC 3 gene in cardiocytes is regulated by both positive and negative cis-elements.
Primary atrial or ventricular cell cultures contain a heterogeneous population of cells. As determined by immunostaining (Fig. 1), about 75% of the cells from either chamber are cardiocytes and about 25% are non-cardiocytes from day 3 in culture onward. To determine which cells in the cardiac cultures express slow MyHC 3 CAT constructs, CAT protein was localized immunologically. Cultures were transfected with the SM3CAT:4500 construct, which directs the highest level of atrial-specific expression, and were double-immunostained with mAb F59 to identify cardiocytes, and with a polyclonal antibody against the CAT protein (Fig. 5). In these doublelabeled cultures, CAT expression was consistently found only in atrial cardiocytes (Fig. 5C). Ventricular cardiocytes never stained for CAT (Fig. 5F) nor did the fibroblasts in either atrial or ventricular cultures. A pCAT-control plasmid containing the SV40 promoter/enhancer driving CAT, when transfected into both atrial and ventricular cell cultures, expressed detectable CAT protein in atrial and ventricular cardiocytes as well as in non-cardiocytes (data not shown). These results provide additional evidence that the slow MyHC 3 regulatory sequences within the SM3CAT:4500 construct are sufficient to restrict expression in cardiac cell cultures to atrial cardiocytes.
The ARD1 Acts as an Atrial-specific Enhancer in Cardiocyte Cultures-Inclusion of the slow MyHC 3 sequence located be-tween Ϫ840 and Ϫ680, designated ARD1, in reporter constructs increased expression 7-fold in atrial cardiocyte cultures relative to ventricular cardiocyte cultures (Table I). The 160 bp of ARD1 were PCR-amplified and fused in both the forward and the reverse orientation to a heterologous promoter, the pCATpromoter in which CAT is driven by the SV40 promoter without an enhancer. The two heterologous promoter constructs named ARD1/SV/F (forward) and ARD1/SV/R (reverse), as well as a control, pCAT-promoter, were transfected into atrial cardiocyte, ventricular cardiocyte, or CEF cultures (Fig. 6). Both constructs, with a single copy of either the forward or the reverse orientation of ARD1, increased CAT expression by 3-fold in atrial cardiocytes whereas their expression was unchanged in ventricular cardiocytes and CEFs compared with the control pCAT-promoter plasmid. Thus, sequence in the ARD1 region has attributes of an atrial cardiocyte-specific enhancer.
The ARD1 of the Slow MyHC 3 Gene Regulates Atrial-specific Expression in the Embryo-To characterize the cis-element that conferred atrial chamber-specific expression on the slow MyHC 3 gene in vivo, we employed a replication-competent retroviral vector, RCAN, for gene transfer to early developing chick embryos. The RCAN vector (41) was specifically designed to express DNA inserts from an internal promoter and to separate the expression of the insert from viral gene expression. A heterologous promoter, which contained ARD1 fused in the forward orientation to the SV40 promoter without an enhancer, was subcloned in a 5Ј to 3Ј orientation upstream of the CAT reporter (Fig. 7). The viral recombinant vector DNA was introduced into CEFs by the standard calcium-phosphate transfection. Because the retrovirus produced is replication-competent, it can infect dividing cells. All the CEFs were infected within 7 days of transfection of the culture, and the cultured cells were producing viral particles. Because pathogen-free quail eggs are not available, pathogen-free chicken embryos were used for retroviral mediated gene transfer studies. Unincubated fertilized chicken eggs were injected with 10 6 virus-producing CEFs on the first day of incubation (ED 0) at the time the embryo consisted of a blastodisc. Whole mounts were prepared of embryos harvested at ED5 and were stained with an antibody against the viral capsid protein p27 (Fig. 7A). Of the 15 em-  bryos that formed after infection, 8 (53%) were infected and 7 (47%) were not. All tissues of the eight positive embryos were ubiquitously stained (Fig. 7A), demonstrating a global infection by the viral construct. No tissues were stained in the seven p27 negative embryos, suggesting that there were no mosaic infections. A control ED5 embryo that was not injected with the virus was stained with anti-capsid protein p27 (Fig. 7B). There was no viral protein present in the control embryo.

TABLE I Activity ratio of slow MyHC 3 CAT constructs in atrial/ventricular cells (A/V) and atrial/CEFs (A/C)
Six chicken embryos infected with RCAN/ARD1-PCAT/F were permitted to develop to fetal stage ED18. These six fetuses were shown to be infected by using an enzyme-linked immunosorbent assay to analyze blood samples for the presence of viral capsid antigen p27. Extracts of brain, kidney, intestine, liver, lung, gizzard, stomach, atrium, ventricle, medial adductor (a slow skeletal muscle), and pectoralis major (a fast skeletal muscle) from each embryo were used in an enzymatic CAT assay (Fig. 8A). High levels of CAT activity were found only in the atria of all embryos. Levels in the atria were 12.3-fold higher then those observed in the non-cardiac tissues and 4.6-fold higher in the atria than that in the ventricles. A 2.7-fold increase was observed in the amount of CAT present in the ventricles relative to non-cardiac tissues, including skeletal and smooth muscles. Viral gene expression in each tissue was also measured directly by detecting p27 protein through Western blot analysis. Fig. 8B showed p27 expression from one viral infected embryo. Results from all six embryos showed equivalent levels of viral p27 protein expression in all tissues, confirming the global nature of infection and indicating that viral gene expression is not specifically enhanced in cardiac tissues. These results demonstrate that in vivo ARD1 can restrict transcription of a reporter gene to cardiac muscle in ED18 fetuses with preferential expression in atria and indicate that regulatory elements residing within ARD1 play a pivotal role in conferring atrial-specific expression of the slow MyHC 3 gene.  (41). Gene cassettes consisting of a promoter linked to cis-sequences of interest can be inserted at the unique ClaI cloning site positioned downstream of the retroviral structural genes gag, pol, and env to drive CAT reporter expression. The recombinant retroviral vector, RCAN/ARD1-PCAT/F, in which ARD1 linked to a heterologous promoter, SV40, was inserted into the ClaI site to drive CAT expression. The open boxes represent viral long terminal repeat. Lower, fertilized chicken eggs injected (A) with 10 6 retrovirus-producing CEFs at day of set (ED 0) or uninjected (B). Embryos were collected at ED5, and whole mount staining was performed with primary antibody against p27, followed by horseradish peroxidase-goat anti-rabbit IgG. Arrows point to hearts of the embryos.

A Vitamin D 3 Receptor-like Sequence Is an Essential Element for Atrial-specific Expression in Cardiocyte
Cultures-Sequence motifs with homology to previously identified regulatory motifs were identified in ARD1, including HF-1A (54), M-CAT (55), E-box (28), vitamin D 3 receptor (VDR), or retinoic acid receptor-like element (56), and GATA (37, 39) (Fig. 9A). To further define the cis-elements that regulate atrial-specific expression, deletions were made to remove portions of ARD1 in the context of the SM3CAT:840 construct (Fig. 9B). First, deletion of intron 1 and exon 2 sequences, to form the construct SM3CAT:840D, had no effect on the level of atrial-specific expression of the reporter (Fig. 9C). Second, primers were designed which deleted slow MyHC 3 promoter sequences from the 5Ј end of SM3CAT:840D, removing HF-1A, M-CAT, and E-box motifs (SM3CAT:808D), or HF-1A, M-CAT, E-box, and VDR motifs (SM3CAT:768D). Deletion of the sequences between Ϫ840 and Ϫ808 (SM3CAT:808D) had no effect on the level of expression in atrial cardiocytes, ventricular cardiocytes, and CEFs. However, further deletion to Ϫ768 (SM3CAT: 768D) resulted in an increase of the CAT expression in ventricular cardiocytes to a level equal to that observed in atrial cardiocytes, whereas expression in atrial cardiocytes and CEFs remained unchanged compared with SM3CAT:840D (Fig. 9C). These results suggest that the HF1A, M-CAT, and E-box motifs within ARD1 are not essential for atrial-specific expression of slow MyHC 3 but that the 40 bp between Ϫ808 and Ϫ768, including the VDR-like element, act as a negative element in ventricular cardiocytes and have no role in atrial cardiocytes and CEFs.
To demonstrate that the VDR-like element specifies atrialspecific expression of slow MyHC 3 gene, the 5Ј copy of the AGGACA direct repeat was mutated to GTCGAC in the context of SM3CAT:840D (mut-VDR, Fig. 10A). As observed with deletion construct SM3CAT:768D, CAT expression in mut-VDR increased 6-fold in ventricular cardiocytes, whereas expression in atrial cardiocytes and CEFs was sustained at a level comparable with SM3CAT:840D (Fig. 10B). In contrast, no effect on chamber-specific expression was observed when a comparable 6-bp sequence between the VDR-like element and the GATA motifs was mutated (mut-Ctrl, Fig. 10, A and B). These results demonstrate the VDR-like cis-element as an essential component of slow MyHC 3 atrial-specific expression. DISCUSSION We have found that the quail slow MyHC 3 gene is initially expressed in both atrial and ventricular chambers but that it is rapidly down-regulated in the ventricle by ED6. 2 A 160-bp enhancer, ARD1, appears to be the primary element that restricts slow MyHC 3 expression to the atria both in vitro and in vivo. Deletion or mutation of a VDR-like element within ARD1 resulted in loss of the atrial-specific expression of the reporter, implicating the VDR-like motif as a key element in the atrialspecific expression of the slow MyHC 3 gene.
Primary Heart Cultures Provided a Model System in Vitro to Characterize Atrial or Ventricular-specific cis-Elements of FIG. 8. The ARD1 regulates atrial chamber-specific transcription in the embryo. A, Homogenates were prepared from 11 tissues of 6 transgenic ED18 chicken fetuses that formed from eggs infected with the retrovirus RCAN/ARD1-PCAT/F prior to incubation. Extracts were adjusted to an equivalent protein concentration, and CAT assays were performed under identical conditions. The error bars represent the standard error of the mean. Levels in the atria were 12.3-fold higher than those observed in the non-cardiac tissues and 4.6-fold higher than that in the ventricles. B, extract (20 g of proteins) of each of the tissues used in the CAT assay in A above were electrophoresed and transferred to nitrocellulose membranes. The viral capsid protein p27 was detected with a polyclonal rabbit anti-p27 antibody. Results from one embryo are shown and each tissue from all six fetuses showed equivalent amounts of the viral protein expression.
FIG. 9. Effect of 5 deletion of the ARD1 on the reporter expression in cardiocytes cultures. A, nucleotide sequence of slow MyHC 3 between Ϫ840 and Ϫ680 containing the 160-bp ARD1. Sequence motifs within the ARD1 which have been identified in other muscle genes are boxed. B, the 5Ј deletion constructs were obtained by PCR amplification of the genomic sequence from Ϫ840 to ϩ18, Ϫ808 to ϩ 18, and Ϫ768 to ϩ18, respectively, and by cloning the fragments into the pCAT-basic vector. These SM3CAT constructs were designated by a number that represents base pairs upstream from the transcription start site followed by a letter D to indicate the first intron and exon II sequence from the slow MyHC 3 gene were deleted in these constructs. C, CAT expression in atrial (filled bars), ventricular cardiocytes (open bars), and CEFs (stippled bars) transfected with these deletion constructs and the SM3CAT:840. The error bars represent the standard error of the mean. Deletion of 40-bp sequence from Ϫ808 to Ϫ768 containing the VDR-like element up-regulated the reporter expression in the ventricular cardiocytes to a level similar to that seen in atrial cardiocytes.
Chamber-specific Genes-We have used an avian atrial and ventricular primary cell culture system to study cardiac chamber-specific gene regulation. The defined heart serum-free medium we employed (43) greatly limits overgrowth of cultures by fibroblasts, ensuring that cardiocytes are the predominate cell type in the cultures. Cardiocytes retain characteristics found in vivo. In this culture system the endogenous slow MyHC 3 gene is continuously expressed in atrial cardiocytes but disappears by 2 days of incubation of ventricular cardiocyte cultures (Fig.  2). Thus, we have a model culture system with which to dissect the transcriptional control cassettes of the slow MyHC 3 gene. This system may also be useful for studying the regulatory pathways of other chamber-specific markers.
Identification of Atrial-specific cis-Sequences-ARD1, a 160-bp fragment of the quail slow MyHC 3 promoter between Ϫ840 and Ϫ680, is competent to drive expression of a heterologous CAT reporter gene in an atrial-specific fashion in primary cell cultures and embryos. This region has properties of an atrial-specific enhancer. In contrast, sequence lying between Ϫ680 and Ϫ290 acts as a silencer in cardiac and fibroblast cells, effectively repressing the high level of CAT reporter expression in these cell types seen when only 290 bp of proximal promoter sequence is present (Figs. 3 and 4). Interestingly, the SM3CAT:680 construct is expressed at a high level in embryonic skeletal muscle cultures (40), identifying a difference in the regulatory elements affecting expression of this gene in the two types of striated muscle cells. Thus, cis-sequences controlling expression of the slow MyHC 3 gene in embryonic atrial cells and embryonic slow skeletal muscle cells are segregated, as shown previously by other groups who compared cardiac and skeletal muscle-specific regulatory elements within the same gene (60 -62).
We further analyzed the ARD1 region in the developing embryo via the RCAN/ARD1-PCAT/F retrovirus. While in-fected embryos showed global viral infection throughout all tissues (Figs. 7A and 8B), reporter expression was markedly enhanced by ARD1 in atria, and to a less extent in ventricle, but not at all in skeletal or smooth muscles nor nonmuscle cells by ED18 of development (Fig. 8A). To our knowledge, this report constitutes the first characterization of cis-regulatory sequences located in a relatively small region (160 bp) that confines transcription preferentially to the atrial compartment in vivo. A number of laboratories have initiated studies in transgenic mice to identify in vivo regulatory regions in genes expressed in the atria. Among genes expressed in the atria, the best characterized regulatory regions were those of ␣-cardiac MyHC, MLC3f, and ANF. A 2-kb upstream region of the ␣-cardiac MyHC gene was shown to direct tissue-specific expression of the transgene (16), and a thyroid hormone-responsive element, TRE2, was shown to be important for reporter activity in both the atrium and ventricle (19). Recently, a 2-kb promoter region and 260-bp enhancer sequence of the MLC3f gene was shown to be sufficient to direct atrial-and left ventricularspecific expression of a reporter in adult mice (15), while a 2.4-kb regulatory region of the ANF gene was shown to direct the reporter expression equally well in both atrial and ventricular cells (26). As the cis-elements regulating atrial expression of these genes become more defined, it will be possible to determine to what extent divergent and overlapping pathways regulate atrial expression of slow MyHC 3 gene and other cardiac-expressed genes.
Within ARD1, sequences homologous to five known cis-elements were identified (Fig. 9A). From 5Ј to 3Ј they are as follows: a HF-1A element, a M-CAT element, an E box, a vitamin D 3 receptor (VDR) or retinoic acid receptor-like element, and a GATA element. The HF-1A is an element that contributes to cardiac-specific expression of the MLC-2v gene in the mouse (54). The M-CAT, initially described as a crucial FIG. 10. Mutation of the VDR-like element in ARD1 led to loss of atrial cell specificity of the SM3CAT:840D constructs in cardiocytes cultures. A, the mutation constructs. The mut-VDR were generated by creating a 6-base pair substitution mutation from position Ϫ801 to Ϫ796 in the SM3CAT:840D by PCR mutagenesis. The mut-Ctrl, in which a comparable 6-bp sequence from position Ϫ780 to Ϫ775 downstream of the VDRlike element was mutated in the SM3CAT:840D, served as a control for mut-VDR. B, CAT expression of atrial (filled bars), ventricular cardiocytes (open bars), and CEFs (stippled bars) of the mutation constructs and the SM3CAT:840D. Mutation of the VDR-like element has no influence on the CAT activity in the atrial cardiocytes and CEFs (compare SM3CAT: 840D, mut-VDR, and mut-Ctrl), while mutation of the VDR-like element increased CAT activity in ventricular cells to 75% of the level in atrial cells. The mutation downstream of the VDR-like element (mut-Ctrl) has no effect on reporter expression in the three cell types. The error bars represent the standard error of the mean. element for cardiac troponin T gene expression in cardiac cells (55), has been shown to regulate a number of contractile protein genes in both cardiac and skeletal muscles (60). Some evidence suggests an E box-dependent pathway in the regulation of some cardiac genes including ␣-cardiac actin (28), muscle creatine kinase (25), and ␣-MyHC (63). The recent discovery of new bHLH proteins, eHAND and dHAND, supports this notion (8). However, deletion results (Fig. 9C) suggest that the E box, as well as HF-1A, M-CAT motif within ARD1 is not essential for atrial-specific expression of the slow MyHC 3 gene.
Both deletion (Fig. 9) and mutation (Fig. 10) of the VDR-like element within the ARD1 of the slow MyHC 3 gene led to up-regulation of the reporter expression in ventricular cardiocytes, suggesting that the VDR-like element controls atrialspecific expression of the slow MyHC 3 gene in embryonic heart by negative regulation. It would be extremely interesting to know how the VDR-like element inhibits expression of slow MyHC 3 in ventricular cardiocytes. Members of the steroid and thyroid hormone (T 3 ) receptor superfamily of transcriptional regulators have been implicated in many critical aspects of vertebrate development, including heart morphogenesis (64). Retinoic acid receptors (RXRs and RARs), vitamin D receptors, and thyroid hormone receptors bind similar cis-elements (56). T 3 has been shown to regulate both ␣and ␤-MyHC expression during cardiac development (19,62), whereas vitamin D 3 negatively regulates the expression of ANF in cardiac cells (31).
Also located within ARD1 is a GATA element (Fig. 9A). Recently a GATA element has been shown to regulate cardiac muscle-specific expression of the ␣-MyHC gene (37) and cardiac troponin C (57). Inclusion of the GATA element (SM3CAT: 768D) drives reporter expression in ventricular cells to a level equivalent to that in atrial cells. Thus, the GATA motif and/or sequence between Ϫ768 and Ϫ680 is able to relieve the inhibition observed in the SM3CAT:680 construct. The ARD1 does not contain a CArG box, which was shown to be important in expression of the ANF gene in both atrial and ventricular cells (27), or a MEF2 site, which is important in control of the expression of a number of contractile protein genes in cardiac, skeletal, and smooth muscle (9,10,65).
The Slow MyHC 3 Gene Can Serve as a Molecular Marker of Atrial Specification-Several myosin heavy and light chain genes have been used as markers to study atrial and ventricular chamber specification. In mice, two cardiac MyHC isoforms, ␣-MyHC and ␤-MyHC, are initially coexpressed in the atrium and ventricle. ␤-MyHC showed preferential expression in the ventricles by day 9.5 postcoitum, while ␣-MyHC is continuously expressed in both atrial and ventricular chambers (58). Similarly, studies of myosin light chain genes (MLC-1a, MLC-1v, MLC-2a) have shown that chamber-restricted expression occurs late in development (58,59), with initial coexpression in the atrium and ventricle in embryonic development. Investigators have also shown, using AMHC1 as an atrial and MLC-2v as a ventricular marker, respectively, that regional expression is established early in the tubular heart stage (13,14). Several features make regulation of the slow MyHC 3 gene particularly interesting in this regard. First, its expression is established early in the tubular heart stage, and thus it can be used as a marker to study early positional and/or molecular cues leading to cardiac tube formation. Second, its expression is sustained throughout development in the atria, and, therefore, it can be used as a marker to understand the mechanism(s) that maintains its expression. Third, its atrial chamber-specific expression is accompanied by down-regulation in the ventricular chambers. Because down-regulation occurs in the future ventricles when the tubular heart begins to chamberize, inves-tigation of slow MyHC 3 gene regulation may provide clues to formation of the ventricles as well. The mechanism(s) that establishes the early chamberization of the heart remains unknown. Identifying the ARD1 as an enhancer for atrial-specific expression of the slow MyHC 3 gene and the importance of the VDR-like element in this process allows us to study molecular cues that mediate the early regionalization of the heart.