Combinatorial Interactions Regulate Cardiac Expression of the Murine Adenylosuccinate Synthetase 1 Gene*

The mammalian heart begins contracting at the linear tube stage during embryogenesis and continuously pumps, nonstop, throughout the entire lifetime of the animal. Therefore, the cardiac energy metabolizing pathways must be properly established and efficiently functioning. While the biochemistry of these pathways is well defined, limited information regarding the regulation of cardiac metabolic genes is available. Previously, we reported that 1.9 kilobase pairs of murine adenylosuccinate synthetase 1 gene (Adss1) 5′-flanking DNA directs high levels of reporter expression to the adult transgenic heart. In this report, we define the 1.9-kilobase pair fragment as a cardiac-specific enhancer that controls correct spatiotemporal expression of a reporter similar to the endogenousAdss1 gene. A 700-base pair fragment within this region activates a heterologous promoter specifically in adult transgenic hearts. Proteins present in a cardiac nuclear extract interact with potential transcription factor binding sites of this region and thesecis-acting sites play important regulatory roles in the cardiac expression of this reporter. Finally, we report that several different cardiac transcription factors trans-activate the 1.9HSCAT construct through these sites and that combinations result in enhanced reporter expression. Adss1 appears to be one of the first target genes identified for the bHLH factors Hand1 and Hand2.

The murine linear heart tube begins to contract spontaneously by 8 d.p.c. 1 (1) and pumps continually throughout the lifetime of the animal; therefore, the metabolic pathways that provide energy must be properly activated and regulated. Several genes that encode metabolic enzymes are activated at low levels during early cardiogenesis, then are up-regulated around birth to reach adult levels during postnatal development to deal with the increasing cardiac energy demands (2,3). Numerous metabolic pathways also utilize cardiac-specific isozymes that are present at high levels in the heart (4 -6). Although the biochemistry of cardiac metabolism is well de-fined, the genetic signaling pathways that control expression of the cardiac metabolic genes are not well understood. Few cardiac cis-regulatory elements that function in vivo have been identified and preliminary studies are just beginning to describe transcription factor interactions as genetic regulatory mechanisms. The subset of cardiac regulatory regions that has received the most attention has been those that control contractile gene transcription (7)(8)(9). Two of the best characterized cardiac structural gene enhancer and promoter regions regulate the ␣and ␤-myosin heavy chain genes, which have been used to direct expression of several other genes to the murine heart (10 -12). Less well characterized cardiac structural gene enhancers include those that control expression of the myosin light chain 1/3 (13) and 2v genes (14). Another class of cardiac elements, those regulating genes that encode enzymes of energy metabolism, is the least understood group of cardiac regulatory regions. Since the murine linear heart tube begins to beat early in development (1), these energy metabolizing pathways must be well established during cardiac formation and morphogenesis. Synthesis and control of the metabolic enzymes that provide the massive supply of energy to the constantly beating heart are areas of study that can provide further information about the molecular pathways controlling heart development and function. Analysis of only two cardiac metabolic gene flanking regions by the transgenic assay, specifically those regulating the transcription of the creatine kinase (15)(16)(17) and adenylosuccinate synthetase 1 genes (this report), has delineated some heart-specific control elements. Consequently, characterization of several energy metabolic regulatory regions is necessary and a comparison of the metabolic versus the structural cardiac enhancers will help to define the mechanisms of tissue-specific activation and enhancement.
We chose to study the adenylosuccinate synthetase 1 (Adss1) gene to characterize further heart-specific cis-acting control elements and to analyze trans-regulatory mechanisms. ADSS1, a striated-muscle specific energy metabolic enzyme, is a member of the purine nucleotide cycle (PNC), which regulates adenine nucleotide pool levels during periods of increased workload. During intense muscle contraction, the forward myokinase and AMP deaminase reactions generate additional ATP. When the myocyte recovers, ADSS1 and ADSL restore adenine nucleotide levels (18). The PNC may support energy metabolism differently in the heart. We propose that this cycle is activated under hypoxic conditions when oxidative phosphorylation pathways cannot function efficiently. The PNC generates additional ATP, while cycle intermediates positively regulate anaerobic glycolysis for substrate level phosphorylation. When oxygen levels return to normal, the return pathway of the PNC restores adenine nucleotide levels. Previously, we showed that 1.9 kb of Adss1 5Ј-flanking DNA can direct high levels of reporter gene expression to the adult transgenic mouse heart (19). In the present paper, we report that this Adss1 1.9-kb fragment regulates reporter expression in transgenic hearts similar to the endogenous Adss1 gene. This cardiac enhancer directs copy number-dependent, integration siteindependent expression to the hearts of transgenic mice at levels comparable with the endogenous gene. Also, a 700-bp fragment within the 1.9-kb 5Ј-flanking fragment activates a heterologous promoter specifically in transgenic hearts. This region contains a cluster of potential cardiac transcription factors that specifically bind proteins from a cardiac nuclear extract. Using mutational analyses, we show that these transcription factor binding sites play important regulatory roles in cardiac reporter gene expression. Finally, we report that several cardiac transcription factors trans-activate 1.9HSCAT expression in a noncardiac background specifically through these sites. Combinations of transcription factors lead to enhanced reporter expression. These results are the first to identify a potential target gene for the cardiac bHLH factors, Hand1 and Hand2.

Generation of Riboprobes and in Situ Hybridization
Analysis-In situ hybridization to sections was performed as described (20). Antisense and sense riboprobes were generated from the Adss1 cDNA and from the CAT cDNA. All of the probes were acid-hydrolyzed to an average size of 750 bp. Sections were viewed with an Olympus BX60 microscope, and photographs were captured using a SPOT digital camera (Diagnostic Instruments Inc.).
Tissue Extracts and CAT Assays-Protein concentration was determined by the Bradford method (21). CAT assays were performed as described (22). CAT specific activity was determined under linear conditions.
Isolation of RNA and Northern Blot Analysis-Total cellular RNA from tissues was isolated by the Trizol procedure as described (Life Technologies, Inc.). Northern blot analysis was performed as described previously (19).
Nuclear Extract Preparation and Electrophoretic Mobility Shift Assay-Neonatal rat cardiac nuclear extract was prepared from primary cardiomyocyte cultures as described (23). Double-stranded oligonucleotides corresponding to various regions of the Adss1 5Ј-flanking region were synthesized as follows: Nkx 2.5 (nt Ϫ1523 to Ϫ1504), MEF2 (nt Ϫ1713 to Ϫ1694), E box A (nt Ϫ1914 to Ϫ1894), E box B (nt Ϫ1893 to Ϫ1874), E box C (nt Ϫ1863 to Ϫ1834), and E box D (nt Ϫ1773 to Ϫ1754), GATA site A (nt Ϫ1423 to Ϫ1402), GATA site B (nt Ϫ1407 to Ϫ1384), GATA site C (nt Ϫ1483 to Ϫ1354), GATA site D (nt Ϫ1353 to Ϫ1324), and GATA site E (nt Ϫ1063 to Ϫ1043). The Nkx 2.5 site was mutated to taaTAtg, the MEF2 site was mutated to cGaCCtCtCg, GATA sites A-C were mutated to CAGTGC, GATA sites D and E were mutated to GCGTTG, and E boxes A-D were mutated to GGTAAC (mutated sites are capitalized). These oligonucleotides were used for electrophoretic mobility shift assay analysis (24).
Generation of Constructs-The 1.9HSCAT construct was described previously (19). The 0.6XhSCAT construct was generated by digesting the HSCAT/pSG5 vector with XhoI and SalI, blunt-ending with T4 polymerase, and was cloned into the EcoRV site of the Bluescript KSϩ vector.
The 700-bp Adss1 HindIII-XbaI fragment was blunt-ended with Klenow and cloned in the SmaI site in front of the SPAC (short promoter Ada CAT) (22) construct in both the forward and the reverse orientations. The SPAC construct is present in the Bluescript KSϩ vector backbone and consists of an 800-bp minimal Ada promoter, the CAT cDNA, and a SV40 intron and polyadenylation signal.
The Nkx 2.5, MEF2, and GATA site-directed mutants were generated by a PCR megaprimer procedure. The primers that were used are as follows (mutated nucleotides are capitalized, and transcription factor sites are underlined): universal site sequencing primer, mutant MEF oligonucleotide reverse, 5Ј-tcaaagcctgagacGaGaGGtCgccaagggcccacataga-3Ј; mutant Nkx 2.5 forward, 5Ј-caagggcaagcattaaTAtgacctgaccaa-3Ј; mutant GATA A-D forward spanned nucleotides Ϫ1423 to Ϫ1324 in which sites A-C were changed to CAGTGC, and site D was changed to GCGTTG and primer B, 5Ј-tctctacctcactcagtgcca-3Ј. The first PCR reaction generated the fragment that contained the mutant site. The PCR fragment was gel-purified by the QiaexII kit (Qiagen, Inc.). That fragment was then used as the megaprimer for the second PCR reaction. PCR cycling conditions were as follows: hot start at 94°C for 10 min, then 94°C for 1 min; 55°C for 1 min, and 72°C for 2 min for 30 cycles; a final extension at 72°C for 10 min; and hold the reactions at 4°C. The large fragment was gel-purified by the QiaexII kit (Qiagen, Inc.) and was then digested with XbaI and cloned into the XbaI-digested, calf intestinal phosphatase-treated HSCAT backbone. All constructs were isolated by cesium gradient purification (25). The orientation of the cloned fragment was determined by restriction enzyme digestion, and the presence of the mutant site was confirmed by sequencing (U. S. Biochemical Corp.).
The 5Ј deletion HSCAT construct was generated by a partial ApaI digestion of the 1.9HSCAT construct. The deletion fragment was gelpurified by the QiaexII kit (Qiagen, Inc.), blunt-ended with T4 polymerase, and religated.
Transient Transfections into Rat Primary Cardiomyocytes-Transient transfection into neonatal rat primary cardiomyocytes were performed as described (29). CAT assays and ␤-gal assays were performed as described (22,30). Experimental data are presented as the mean of three independent transfection assays done in duplicate and normalized to ␤-gal activity.
Production and Analysis of Transgenic Mice-Transgene constructs were isolated by agarose gel electrophoresis and purified by either a QiaexII kit or a Qiaquick kit (Qiagen, Inc.). Transgenic mice were prepared by injecting the constructs in fertilized FVB/N oocytes (31). Founder (F0) transgenic animals were identified as described previously (19).
Cell Culture and Transfections-Cotransfections into CV1 fibroblasts were performed as described previously (27). Cells were transfected with a total of 3 g of plasmid DNA, which included 500 ng of 1.9HSCAT reporter plasmid, 1 g of CMV ␤-gal, and various concentrations of expression vector plasmids (250 ng of CMV Nkx 2.5, 250 ng of CMV GATA 4, 150 ng of CMV MEF2C, 150 ng of CMV E12, 250 ng of Hand1, and 250 ng of CMV Hand2). All transfections were balanced with a CMV empty vector. The cells were washed with phosphatebuffered saline and scraped from dishes in 400 l of a CAT assay buffer. The cells were lysed by sonication and were centrifuged at 14,000 rpm for 15 min at 4°C. The supernatant was transferred to a clean tube, and protein concentration was determined by the Bradford method (21). CAT assays and ␤-gal assays were performed (22,30). Values were the average of at least three experiments, each with duplicate plates and were normalized to ␤-gal activity.

kb of Adss1 5Ј Flank Controls Proper Spatiotemporal
Expression of a Reporter in Transgenic Hearts-Previously, we reported that 1.9 kb of Adss1 5Ј-flanking DNA possessed the ability to direct high levels of reporter gene expression to the adult murine transgenic heart (19). To determine whether this 1.9-kb Adss1 fragment possesses the capability to control correct spatiotemporal expression of a reporter, 1.9HSCAT transgenic embryos were analyzed at different developmental times. Both the endogenous Adss1 gene and the 1.9HSCAT transgene were expressed in both the atrium and ventricle of the heart at 9.0 d.p.c. (Fig. 1, A and B). At 11.5 d.p.c., expression of both the endogenous gene and the transgene were detected in both chambers of the heart and in the developing facial and jaw muscles (Fig. 1C). Interestingly, the level of CAT mRNA was lower than the level of Adss1 mRNA in the atrium (Fig. 1D). Faint expression of both the Adss1 gene and the 1.9HSCAT transgene was also detected in the fetal liver ( Fig. 1, C and D). No signal above background was detected with sense riboprobes (data not shown). To determine whether the 1.9-kb 5Ј-flanking region has the capability to maintain reporter gene expression during late embryogenesis, a founder 1.9HSCAT transgenic litter was dissected at 15 d.p.c., and cardiac and liver tissues were analyzed for CAT enzyme activity. In two transgenic embryos (11 and 12), CAT enzyme was strongly expressed in the heart, but negligible levels were detected in the liver (Fig. 1E). This experiment shows that transgene expression in the liver ceased, while transgene expression in the heart remained high. No CAT activity was detected in these tissues in nontransgenic embryos (9 and 10). Thus, the 1.9-kb Adss1 5Ј fragment possesses the capabilities to activate gene expression in cardiac progenitors and to sustain expression throughout embryonic cardiac development.
Several cardiac and muscle energy metabolic genes experience a transcriptional up-regulation during perinatal development to support the increased energy demands of the growing, mobile organism (41,42). To determine whether Adss1 cardiac expression also increases during neonatal development, total RNA was isolated from 1.9HSCAT transgenic hearts at different developmental timepoints ( Fig. 2A), and Northern analysis was performed using both Adss1 and CAT probes. Adss1 transcripts were expressed at low levels on embryonic day 17, and expression continued to steadily increase through day 21 of postnatal development. Expression of the 1.9HSCAT construct was not detected at embryonic day 17 by Northern analysis, but transcripts were visualized by day 1, and 1.9HSCAT expression continued to be up-regulated through day 21. However, the expression of the 1.9HSCAT construct appeared to be up-regulated more slowly than the expression of the endogenous Adss1 gene. The expression of both genes is depicted graphically in Fig. 2B. In conclusion, both the endogenous Adss1 gene and the 1.9HSCAT construct undergo a phase of transcriptional enhancement during perinatal cardiac development that is similar to several other muscle metabolic genes.
Important characteristics that define an enhancer element are the ability to confer integration site independence and copy number dependence (32). To determine whether the 1.9-kb Adss1 5Ј flank can confer dominant expression capability, the total CAT activity of the 1.9HSCAT construct was compared with the copy number (Fig. 2C). Copy numbers ranged from 1 to 16. The graph resulted in a linear relationship with 9/9 transgenic mouse hearts. Therefore, the 1.9 kb of Adss1 5Ј flank appears to confer both integration site independence and copy number dependence on the CAT reporter gene. Since expression of the 1.9HSCAT construct can be reasonably predicted based on copy number, we determined whether this 1.9-kb flanking region contains all the information necessary to direct the proper levels of reporter expression to the adult transgenic heart. Total RNA was isolated from various tissues of an adult 1.9HSCAT mouse containing 16 copies of the transgene (Fig. 2D). Northern analysis showed the expected pattern of Adss1 mRNA expression in skeletal muscle and heart, with the highest levels being expressed in skeletal muscle tissue (4). No expression was detected in nonmuscle tissues. A CAT probe revealed 1.9HSCAT expression specifically in adult transgenic heart tissue, but not in skeletal muscle. Therefore, the 1.9HSCAT construct directs expression specifically to the adult transgenic heart, and the levels of CAT mRNA appear to be litter (see "Materials and Methods"). Tissues examined for CAT expression were liver (liv) and heart (Ht). Embryos 9 and 10 were nontransgenic; embryos 11 and 12 were transgenic. comparable with the endogenous levels of Adss1 mRNA in cardiac tissue (see "Discussion"). In conclusion, the 1.9-kb Adss1 5Ј flank properly activates cardiac reporter expression, controls correct developmental cardiac up-regulation, and directs levels of reporter expression to the adult heart that are similar to endogenous Adss1 levels.
The 700-bp HindIII-XbaI Fragment of the Adss1 5Ј-Flanking Region Can Activate a Heterologous Promoter in Either the Forward or Reverse Orientation Specifically in Transgenic Hearts-We demonstrated that the 700 base pairs of Adss1 5Ј flank between 1.9 and 1.2 kb upstream of the transcriptional start site are essential for expression of the reporter construct in cardiac tissue (19). To determine whether this 700-bp HindIII-XbaI region can act as a cardiac enhancer, this fragment was cloned in front of the SPAC construct (see "Materials and Methods") in both the forward and reverse orientations (Fig. 3, A and B). This construct contained an 800-bp adenosine deaminase promoter (Ada), which is expressed at negligible levels in murine cardiac and skeletal muscle tissues (22). This minimal Ada promoter alone is not activated in tissues of transgenic mice, but when controlled by a placental, thymic, or forestomach enhancer, the promoter is activated specifically in those respective tissues (22,33,47). Both HX.SPAC ϩ and Ϫ constructs directed highest levels of expression of the reporter construct to cardiac tissue (Fig. 3, A and B). However, the levels of CAT enzyme activity were 500 -1000-fold lower than the 1.9HSCAT levels in the transgenic heart (data not shown). Therefore, the 700-bp Adss1 fragment possesses the ability to activate a heterologous Ada promoter in both the forward and reverse orientations in the transgenic murine heart.
Proteins from a Neonatal Rat Cardiac Nuclear Extract Interact with Potential Cardiac Transcription Factor Binding Sites within the 700-Base Pair Adss1 Enhancer-To begin to determine whether cardiac proteins interact with potential cardiac binding sites present in the 700-bp HindIII-XbaI fragment from the Adss1 5Ј-flanking region, electrophoretic mobility shift assays were performed (Fig. 4). Each of the possible transcription factor binding sites was end-labeled and used as a probe to determine whether proteins present in a neonatal rat primary cardiac nuclear extract bind to the sites. First, a 9HSCAT transgene. C. Total CAT activity of nine 1.9HSCAT transgenic hearts was plotted against the copy number. The graph revealed a linear relationship. D, total RNA (10 g) from various tissues of an adult 1.9HSCAT mouse was fractionated on a denaturing 1% agarose gel and transferred to a nylon membrane. The Northern blot was probed with the 1.9-kb Adss1 cDNA and with a 200-bp HindIII-EcoRI fragment from the CAT cDNA (see "Materials and Methods"). The various tissues are skeletal muscle (Sk M), heart (Ht), kidney (Ky), liver (Liv), small intestine (Sm I), stomach (Sto), and spleen (Spl). The 28 and 18 S ribosomal RNA markers were visualized by ethidium bromide staining prior to and following blot transfer to ensure equal loading and efficient transfer of RNA.
shifted complex was evident with an Adss1 Nkx 2.5 probe (Fig.  4A, lane 1). This complex was competed away by 25 ϫ and 50 ϫ (lanes 2 and 3) of cold wild type Nkx oligonucleotide, but only partially by 100 ϫ of mutant Nkx oligonucleotide (lane 4). Next, the Adss1 MEF2 probe bound proteins from the rat cardiac nuclear extract (Fig. 4B, lane 5). 50 and 100 ϫ cold wild type MEF2 oligonucleotides competed away binding (lanes 6 and 7), but 100 ϫ cold mutant MEF2 oligonucleotides did not compete (lane 8). Third, each of the four E boxes from the Adss1 5Ј flank bound proteins present in the rat cardiac nuclear extract (Fig.  4C, lanes 9, 12, 15, and 18). Binding was competed away by 100 ϫ each of the cold wild type E box oligonucleotides ( lanes  10, 13, 16, and 19), but not by 100 ϫ cold mutant E box oligonucleotides (lanes 11, 14, 17, and 20). Multiple protein complexes formed on the E box C binding site (lanes 15-17). Finally, proteins present in the cardiac nuclear extract interacted with each of the Adss1 GATA transcription factor binding sites (Fig. 4D, lanes 21, 24, 27, 30, and 33). Binding was competed away by 100 ϫ cold wild type GATA oligonucleotides (lanes 22, 25, 28, 31, and 34), but not by 100 ϫ cold mutant GATA oligonucleotides (lanes 23, 26, 29, 32, and 35). Again, several different protein complexes were apparent on GATA sites B (lanes 24 -26) and E (lanes [33][34][35]. In summary, proteins present in a neonatal rat cardiac nuclear extract can specifically bind to each of the potential Adss1 transcription factor binding sites, but do not bind to the mutant transcription factor sites. Mutational Analysis of the 1.9-kb Flanking Region Reveals the Importance of Several Potential Transcription Factor Binding Sites for Cardiac Gene Expression-Mutations were generated in the 1.9HSCAT construct to identify binding sites that regulate Adss1 expression in cardiac muscle (Fig. 5A). The Nkx, MEF2, and GATA site mutations used for electrophoretic mobility shift assay analysis were introduced into the 1.9HSCAT construct, while the distal four E boxes were deleted from the 1.9HSCAT construct. Adss1 is endogenously expressed in rat primary neonatal cardiomyocytes (data not shown), so the 1.9-kb 5Ј flank is recognized in these cells. Each of the 1.9HSCAT mutant constructs, the wild type 1.9HSCAT construct, a minimal 0.6XhSCAT construct, and the minimal SPAC construct were transiently transfected into rat neonatal primary cardiomyocytes (Fig. 5B). The 0.6XhSCAT construct was a negative control, because it did not contain the Adss1 700-bp cardiac enhancer. The SPAC construct served as a control to ensure that a noncardiac promoter would not be ectopically activated in the cardiomyocytes. The primary cardiomyocytes offer an environment that mimics early cardiac development, since a number of embryonic genes are expressed under culture conditions (34). Activation levels of each of the 1.9HSCAT mutants were decreased compared with the wild type 1.9HSCAT construct. Both the Nkx mutant and the GATA mutant constructs were activated to only about 40 -60% of the wild type level. The MEF2 mutant was activated to only around 25% of the 1.9HSCAT level. Essentially no expression was detected from the 5Ј E box deletion construct.
Next, we analyzed each of the 1.9HSCAT mutant constructs in transgenic mice, because the construct is integrated into the chromatin superstructure and must undergo the same developmental influences that the endogenous Adss1 gene undergoes to be activated and expressed. Cardiac expression of each of the 1.9HSCAT mutants was decreased compared with the wild type 1.9HSCAT construct (Fig. 5C). Expression of the Nkx 2.5 mutant was lowered by approximately 50%, while expression of the MEF2 mutant was lowered by about 90%. No ex- pression of either the GATA mutant (see "Discussion") or the 5Ј E box deletion mutant was seen. In conclusion, both the cardiomyocyte transfections and the transgenic mouse assay defined important roles for each of the transcription factor binding sites in the cardiac expression of the 1.9HSCAT reporter, especially for the GATA sites and the E boxes.
Combinatorial Interactions between Several Cardiac Transcription Factors Activate CAT Reporter Gene Expression in CV1 Fibroblasts-We have characterized several cis-regulatory sites within the 700-bp Adss1 fragment that contribute significantly to cardiac expression of this gene, using both biochemical and functional analyses. The next step was to assess whether candidate cardiac transcription factors can contribute to Adss1 cardiac expression. Using noncardiac CV1 fibroblasts as a "test tube", several candidate cardiac transcription factors were cotransfected with the 1.9HSCAT reporter to test their trans-activation potential. Nkx 2.5 (26), MEF2C (28), and GATA 4 (27) were chosen as candidate factors, because they consistently bind their consensus sequences that are present in the Adss1 700-bp fragment, and they are expressed throughout prenatal cardiac development (35). Information about binding specificities and activation capabilities of Hand1 and Hand2 bHLH factors is sparse; therefore, we wanted to determine whether Adss1 could be a potential target gene. Fig. 6A shows that individual transcription factors modestly trans-activate the reporter 3-7-fold over background. When each transcription factor is cotransfected with the specific mutant 1.9HSCAT construct, the expression of the reporter construct is decreased. Therefore, each factor appears to act through a specific binding site within the Adss1 flank, and each protein appears to contribute a distinct level of transcriptional regulation to cardiac gene expression. It

DISCUSSION
This report and our previous paper (19) show that the 1.9 kb of Adss1 5Ј flank provides proper developmental activation of reporter genes in cardiac muscle precursors. Atrial CAT expression appeared low in the 11.5 d.p.c. transgenic embryo, which could mean that the 1.9-kb 5Ј flank directs higher reporter levels to the embryonic ventricles than the atria. This Adss1 1.9-kb flank also confers proper up-regulation of reporter expression during perinatal cardiac development. Interestingly, the endogenous Adss1 gene reached a higher level of expression earlier in postnatal development than the 1.9HSCAT transgene did. This discrepancy could be attributed to a lack of elements that fully control perinatal cardiac upregulation. Overall, we show that the Adss1 1.9-kb fragment confers proper developmental activation and neonatal enhancement in cardiac lineages.
Elements within the 1.9-kb Adss1 5Ј-flanking fragment not only control proper developmental expression, but also direct correct adult cardiac expression levels in a dominant fashion over the surrounding chromatin environment. The characteristics of copy number dependence and integration site independence have been associated with classical enhancers, such as the globin locus control region (32). However, elements that open chromatin are being separated from the elements that direct high levels of expression (36). Regulatory elements do not always possess both abilities, like the globin locus control region and the Adss1 cardiac enhancer. The murine Ada placental enhancer directs reporter gene expression specifically to the trophoblast lineages of the placenta, but expression levels vary significantly depending on the integration site of the  9, 12, 15, 18). The extract was preincubated with either 100 ϫ unlabeled wild type Adss1 E box oligonucleotides (lanes 10, 13,16,19) or with 100 ϫ unlabeled mutant Adss1 E box oligonucleotides (lanes 11,14,17,20). The letters A, B, C, and D under the lanes refer to each of the E boxes present in the 700-bp Adss1 fragment. D, 10 g of the rat extract was incubated with each of the Adss1 GATA oligonucleotides (lanes 21, 24, 27, 30, 33). The extract was preincubated with either 100 ϫ unlabeled wild type Adss1 GATA oligonucleotides (lanes 22,25,28,31,34) or with 100 ϫ unlabeled mutant Adss1 GATA oligonucleotides (lanes 23,26,29,32,35). The letters A, B, C, D, and E under the lanes refer to each of the GATA sites within the 700-bp Adss1 fragment. transgene (33). Interestingly, the 1.9HSCAT graph did not pass through zero on the x axis; it appeared to pass closer to 2. Two or more copies of a transgene may be necessary to "insulate" each other from the surrounding chromatin (37,38). Overall, elements within the 1.9-kb Adss1 5Ј flank confer copy number dependence as well as integration site independence on the transgenic mice were generated and analyzed as described under "Materials and Methods." Four-week-old founder transgenics were dissected and analyzed for levels of CAT expression in the heart. CAT activities are the average of expression levels in eight 1.9HSCAT mice, three Nkx mutants, two MEF2 mutants, three E box deletion mutants, and three GATA mutants. CAT reporter construct in the transgenic murine heart. When comparing endogenous and transgenic expression in an adult, a contrasting pattern results. The endogenous Adss1 gene is most highly expressed in skeletal muscle, with slightly lower levels being expressed in the heart. However, the 1.9HSCAT construct expresses specifically in the adult transgenic heart, but not in skeletal muscle. Therefore, regulatory elements for adult skeletal muscle expression must be located outside of the 1.9-kb region. By correcting for a variety of factors, we propose that the 1.9-kb flank directs expression levels comparable with the endogenous levels in the adult transgenic heart. Unfortunately, we cannot make any conclusive statements about mRNA stability.
We demonstrated previously that the 700-bp fragment between 1.9 and 1.2 kb upstream of the Adss1 transcriptional start site was essential for cardiac reporter expression (19). This fragment directs the highest levels of expression of a heterologous promoter to the transgenic heart from both the forward and reverse orientations. Comparatively, though, the level of CAT activity in the HX.SPAC transgenic hearts was 500 -1000-fold lower than the levels of CAT activity in the 1.9HSCAT transgenic hearts (data not shown). These results demonstrate that the 700-bp Adss1 HindIII-XbaI fragment confers cardiac specificity on a noncardiac promoter, but information within the remaining 1.2 kb of Adss1 5Ј-flanking region is necessary to provide the high level of expression. This ability, in conjunction with the characteristics of proper developmental control, copy number dependence, and integration site independence, define the Adss1 5Ј-flanking region as a cardiac enhancer with characteristics of an locus control region.
Few cardiac regulatory elements have been identified that possess these features. Two of the most extensively characterized cardiac enhancers control the ␣and ␤-myosin heavy chain genes' expression. Specific elements control cardiac expression, such as TRE, MEF2, MCAT, and Be3 (7)(8)(9). Flanking regions of the myosin light chain 2v (13) and 1/3 (14) genes have also been analyzed in transgenic mice. Only one other cardiac metabolic gene, muscle creatine kinase, has been analyzed in vivo. Transgenic analysis described a 1.2-kb 5Ј-flanking region of MCK that directs both cardiac and skeletal muscle expression, while preliminary mutational analysis revealed the importance of only three E boxes for cardiac expression (17). Definition of the cis-acting elements that regulate cardiac gene expression can provide a starting point to construct genetic signaling pathways. Within the 700-bp Adss1 cardiac enhancer, multiple potential cardiac transcription factor binding sites were defined by homology, specifically E boxes (39), an Nkx 2.5 site (40), a MEF2 site (41), and GATA sites (42). Each of the possible Adss1 transcription factor binding sites interacted specifically with proteins from a neonatal rat primary cardiac nuclear extract. Interestingly, multiple protein complexes were noticed on the E box C oligonucleotide, which may consist of ubiquitous E proteins and/or MEF family members, along with the Hand transcription factors (39). Several complexes also formed on the GATA B and E oligonucleotides. GATA factors bind DNA as homo-and heterodimers and have been shown recently to interact with Nkx 2.5 (27) and SRF (43) transcription factors to regulate cardiac gene expression. These biochemical experiments revealed protein-DNA interactions within the 700-bp cardiac enhancer that may control Adss1 expression. Unfortunately, protein-DNA complexes identified by biochemical assays may not be the same complexes that actually regulate a target gene. Therefore, the effects of specific mutations on 1.9HSCAT reporter expression levels were analyzed in both neonatal primary cardiomyocyte cell culture and in transgenic mice. Cardiomyocytes in culture can express nu-merous embryonic markers and may offer an early developmental environment to study gene activation (34). The transgenic mouse assay provides an adult environment to analyze enhancement. The results of the two experiments generally agree; mutation of transcription factor binding sites decreases 1.9HSCAT reporter expression in the heart. Interestingly, the GATA mutant construct retained activity in cells, but did not express in transgenic mice. This result could mean that the GATA mutant can be activated, but cannot be up-regulated to adult levels. In summary, each of the cardiac transcription factor binding sites present in the Adss1 700-bp enhancer specifically interacts with proteins present in a cardiac nuclear extract, and these cis-acting sites confer varying degrees of control on the expression of the 1.9HSCAT construct in the murine transgenic heart.
Since cardiac-specific cis-acting sites that control the Adss1 gene were delineated, we began to identify trans-acting factors that regulate expression. Based on the biochemical and functional data, we chose candidate transcription factors to analyze trans-activation potential on the 1.9HSCAT construct. Nkx 2.5, MEF2C, GATA 4, and the Hand1 and Hand2 proteins are all expressed in the precardiac mesoderm and in the developing heart (35), therefore they may play a role in genetic activation. The Adss1 Nkx 2.5, MEF2, and GATA sites are consensus sequences, and the respective factors can consistently bind them to regulate numerous cardiac target genes (26,41,42). Little is known about the DNA binding sites or the regulatory ability of the Hand factors, but it has been suggested that they may bind nonconsensus E boxes (44). To date, no target genes have been identified for the Hand factors. The results of the cotransfection experiments demonstrate that individual cardiac transcription factors moderately trans-activate, and mutation of specific binding sites abolishes trans-regulatory ability. Therefore, each factor appears to regulate Adss1 expression through a specific site, and each protein confers a distinct level of transcriptional control. Combinations of factors strongly activate the 1.9HSCAT reporter in CV1 fibroblasts. Other interactions have been identified (Nkx 2.5 and GATA 4 (27, 45); Nkx 2.5 and SRF (43); MEF2 and TR (46); and MEF2, E12, and MyoD (28)) that regulate short or synthetic promoters. The activation of the 1.9HSCAT construct may not be as strong as these reports, but to date, this is the longest endogenous DNA flanking region that responds to trans-activation by multiple cardiac transcription factors. The Adss1 gene is one of the first reported potential targets of the cardiac bHLH factors Hand1 and Hand2. It appears to respond most strongly to Hand2, either alone or in combination with other factors. The combinations of the factors may provide insight into mechanisms of transcriptional activation in the heart. As the heart must respond to changing energy demands, numerous signaling pathways are activated. The cardiomyocyte can modulate gene expression by regulating the levels and modifications of various transcription factors. The specific type of protein complex that is formed may confer different levels of gene expression. This is the largest protein-protein interaction that has been shown to trans-activate, utilizing four different families of cardiac transcription factors. How this protein-DNA complex interacts with accessory proteins and with proteins at the proximal promoter remains as the elusive question in gene regulation.
In conclusion, the 5Ј-flanking region of the Adss1 gene contains the information to regulate gene expression in cardiac lineages. Multiple cis-regulatory sites play important roles in cardiac expression of the Adss1 gene, while combinations of several cardiac transcription factors interact to strongly activate Adss1 in the heart. Future aims include definition of the factors and/or complexes that control Adss1 gene activation versus perinatal enhancement in the heart. Several cardiac transcription factors are expressed in the precardiac mesoderm and in early cardiac development, such as Hand1 and Hand2, GATA 4, GATA 5, and GATA 6, MEF2A to MEF2D, and Nkx 2.5 (35), which may be responsible for Adss1 activation. Some of these factors are shut off by late embryogenesis and are replaced by other factors that continue to be expressed throughout postnatal development (35). Different combinations of transcription factors at certain developmental times and places may specifically control gene activation versus enhancement. Definition of the mechanisms by which cardiac metabolic genes are activated and enhanced during murine development will aid in understanding how the heart responds to changing energy demands.