Genomic structures of the human and murine corin genes and functional GATA elements in their promoters.

Corin is a multiple-domain type II transmembrane serine protease highly expressed in the heart. It converts pro-atrial natriuretic peptide to atrial natriuretic peptide, a cardiac hormone that regulates blood volume and pressure. Here we describe the genomic structures of the human and murine corin genes and functional analysis of their promoters. Both corin genes contain 22 exons and span >200 kb. Their intron/exon boundaries are well conserved, with most exons encoding distinct structural domains, supporting the idea that corin evolved as a result of exon duplication and rearrangement. Comparison of the 5'-flanking regions of the human and murine corin genes revealed several conserved sequences, including binding sites for TBX5, GATA, NKX2.5, and Krüppel-like transcription factors. Transfection experiments with reporter gene constructs driven by the human or murine corin 5'-flanking region indicated that the sequences from -405 to -15 in human and from -646 to -77 in mouse are sufficient to promote high levels of gene expression in murine cardiomyocytes. In contrast, these sequences produced only minimal levels of expression in HeLa cells. Within these sequences, we identified a conserved GATA element that bound to GATA-4. Mutation of the core sequence impaired both GATA-4 binding and gene expression. These data indicate that the GATA element and its binding to GATA-4 are essential for cardiac expression of the human and murine corin genes.

Corin is a member of the type II transmembrane serine protease class of the trypsin superfamily composed of multiple structurally distinct domains (1). It contains a cytoplasmic tail at its N terminus, followed by a transmembrane domain, a stem region composed of two frizzled-like cysteine-rich domains, eight low density lipoprotein receptor repeats, a macrophage scavenger receptor-like domain, and a serine protease domain at its C terminus. The overall topology of corin is similar to that of other type II transmembrane serine proteases, including hepsin (2), enterokinase (3), MT-SP1/ matriptase (4,5), human airway trypsin-like protease (6), TMPRSS2 (7), TMPRSS3/TADG-12 (8), TMPRSS4 (9), MSPL (10), and Stubble-stubbloid (11). The similar topologies with distinct modular structures suggest that these proteins compose a gene family that evolved by duplication and rearrangement of ancestral exons.
The expression of corin is abundant in tissues where atrial (ANP) 1 or B-type (BNP) natriuretic peptides are produced, predominantly in the atrium and ventricle of the heart (1). In functional studies (12,13), corin converts pro-ANP into biologically active ANP in a highly sequence-specific manner, indicating that corin is the pro-ANP convertase. In addition, corin processes pro-BNP to BNP (12). ANP and BNP regulate blood volume and pressure by promoting salt excretion, increasing urinary output, and reducing vasomotor tone (14 -20). In response to volume overload or a hypertrophic signal, the heart increases its release of these hormones, which in turn reduce blood volume and lower blood pressure. The increased release of ANP and BNP has been attributed to the increased synthesis of pro-ANP and pro-BNP (21). In principle, the level of corin expression would also affect the circulating levels of ANP and BNP because overexpression of corin increases the conversion of pro-ANP to ANP (13). It is possible that the increased synthesis of pro-ANP and pro-BNP is coordinated with the upregulation of corin expression.
The molecular mechanisms for cardiac-specific expression of ANP and BNP and their up-regulation in response to volume overload or a hypertrophic signal have been characterized (22,23). Strikingly, both genes share several common regulatory elements (such as the GATA element) in their promoters. It has been shown that the GATA element is the key regulatory element for cardiac-specific expression and possibly for their up-regulation (24,25). It is unknown, however, whether the corin gene shares similar regulatory elements with the ANP and BNP genes.
To understand the structural features of corin and its cardiac expression, we isolated the human and murine corin genes, determined their genomic structures, and analyzed the function of their 5Ј-flanking regions. The conserved feature of their genomic structures supports the concept that corin is assembled from exons encoding structurally distinct domains. The characterization of their promoter regions provides insights into the mechanism that regulates cardiac expression of corin.

EXPERIMENTAL PROCEDURES
Materials-The murine cardiac myocytic cell line HL-5 was kindly provided by Dr. William C. Claycomb (Louisiana State University Medical Center, New Orleans, LA). Human HeLa epitheloid cells (ATCC CCL2) were purchased from the American Type Culture Collection and maintained at the Berlex Biosciences Core Facility. Antibodies against mouse GATA-1 (sc-1234x), GATA-3 (sc-268x), GATA-4 (sc-1237x), and * The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
The nucleotide sequence(s) reported in this paper has been submitted to the GenBank TM  GATA-6 (sc-7244x) were from Santa Cruz Biotechnology (Santa Cruz, CA). Other chemicals were from Sigma. Cell Culture-HL-5 cells were cultured in Ex-Cell 320 medium (JRH Biosciences, Lenexa, KS) containing 10% fetal bovine serum, 15 g/ml insulin, 50 g/ml endothelial cell growth supplement (Upstate Biotechnology, Inc., Lake Placid, NY), 1 M retinoic acid, 0.1 mM norepinephrine, 100 g/ml penicillin/streptomycin, 292 g/ml L-glutamine, and 0.1 mM minimal essential medium nonessential amino acids. HeLa cells were cultured in M199 medium (Invitrogen) supplemented with 10% fetal bovine serum. All cells were cultured at 37°C in humidified incubators with 5% CO 2 and 95% air.
PCR, DNA Isolation from Bacterial Artificial Chromosome (BAC) Clones, and Southern Analysis-PCRs were performed using the PCR reagent system from Invitrogen with 30 cycles of amplification (1-min denaturation at 94°C, 1-min annealing at 50°C, and 1-min extension at 72°C) and a final 7-min extension at 72°C. DNA isolation from BAC clones (Incyte Genomics, Palo Alto, CA) was carried out according to the manufacturer's instructions. Southern analysis was performed as described previously (26).
Cloning of Human and Mouse Corin Genes-To clone the human and murine corin genes and their 5Ј-flanking regions, we synthesized specific oligonucleotides corresponding to the 5Ј-and 3Ј-ends of corin cDNA sequences. These oligonucleotide primers were tested for amplification of specific products in PCR-based reactions using human or murine genomic DNA. The pairs of primers that successfully amplified specific PCR products were then used in a PCR-based screen to identify BAC clones containing the human and murine corin genes and/or their 5Ј-flanking regions. The identified positive BAC clones were further confirmed by Southern analysis using 32 P-labeled human and murine corin cDNA probes. The BAC clones (see Fig. 1) were either directly sequenced by a shotgun strategy or subcloned into pUC118 (PanVera/ Takara, Madison, WI) for sequencing. Assembly of the shotgun sequences was done using the Staden package (27).
Construction of Chimeric Luciferase Expression Vectors-The human corin promoter reporter constructs hCp1297LUC and hCp405LUC were generated in two steps: 1) PCR-based cloning of the 5Ј-flanking region of the human corin gene from Ϫ1297 or Ϫ405 to Ϫ15 (relative to the translation initiation codon ATG) using primers bearing SacI and Hin-dIII restriction sites, respectively; and 2) inserting the respective PCR products into the SacI and HindIII sites of the pGL3-basic vector (Promega, Madison, WI). Similarly, the murine corin promoter constructs mCp1183LUC, mCp809LUC, and mCp646LUC were also made by the PCR-based cloning approach described above. The mutant constructs hCp405mutGATA and mCp646mutGATA, carrying a mutation in the conserved proximal GATA element (GATA to CTTA), were constructed by an overlap PCR protocol (28). Briefly, two separate PCR products, one for each half of the hybrid product, were generated with either an antisense or sense mutated GATA oligonucleotide and one outside primer. The two products were purified and mixed. A second PCR was then performed using the two outside primers. The PCR product was digested with SacI and HindIII and ligated into SacI-and HindIII-digested pGL3-basic vector. All constructs were confirmed by restriction mapping and DNA sequencing.
Transfection and Dual-luciferase Reporter Assays-Plasmids were prepared using an EndoFree plasmid maxi kit (QIAGEN Inc., Valencia, CA). Transfection of HL-5 cells was carried out using a Lipofectin (Invitrogen)-based method according to the manufacturer's instructions. Briefly, 10 g of DNA of each of the corin reporter constructs plus 0.1 g of pRL-SV40 (Promega) were mixed with 20 g of Lipofectin in 1 ml of Opti-MEM I reduced-serum medium. The mixture was incubated for 30 min at room temperature and then added to ϳ70% confluent HL-5 or HeLa cells cultured in one well of six-well plates. After incubation for 6 h, the medium was replaced with fresh Ex-Cell 320 medium; and 30 h later, the transfected cells were harvested and assayed for firefly and Renilla luciferase activities.
A dual-luciferase activity assay (Promega) was performed according to the manufacturer's instructions. Briefly, cell extracts were prepared by lysing the transfected cells with 250 l of freshly diluted passive lysis buffer (Promega). The lysates were frozen and thawed once before centrifugation at 13,000 rpm for 5 min to pellet the cell debris. The supernatants were transferred to a fresh tube, and a 20-l aliquot of the supernatants was assayed using the dual-luciferase reporter assay system. The luminescence of the samples was monitored by a Microplate Luminometer LB96 V (EG&G Berthold), which measured light production (relative light units) for a duration of 10 s. Each of the cell extracts was assayed in triplicate. Each transfection experiment for each construct was performed in triplicate. Firefly luciferase activity was normalized to Renilla luciferase activity.
Electrophoretic Mobility Shift Assay (EMSA)-Nuclear extracts were prepared from exponentially growing HL-5 cells (29). The doublestranded oligonucleotide probes containing two consensus GATA sequences or mutated GATA sequences (GATA to CTTA) were originally derived from the T cell receptor enhancer (30,31) and were purchased from Santa Cruz Biotechnology. The probes encompassing human or murine corin GATA or mutated GATA (GATA to CTTA) sequences were synthesized and purified by high performance liquid chromatography. The oligonucleotide probes were 5Ј-end-labeled with T4 polynucleotide kinase (Invitrogen) using [␥-32 P]ATP (3000 Ci/mmol; Amersham Biosciences). Gel mobility shift assays were performed as described previously (32). In some experiments, nuclear extracts were preincubated with antibodies (at a final concentration of 50 g/ml) at room temperature for 45 min before addition of labeled probes.

Isolation and Characterization of the Human and Mouse
Corin Genes-Four BAC clones, two each containing the human and murine corin genes, were obtained by PCR-based screening. Three BAC clones were sequenced by a shotgun strategy using dye terminator chemistry. During the course of our studies, partial human and mouse corin genomic sequences became available. Upon combination of the shotgun data and the publicly available trace file information, 2 we assembled a contiguous sequence of 340 kb containing the human corin gene and five contigs for the murine corin gene. The order of five contigs was confirmed by the existence of several mated reading pairs in respective neighboring contigs. The distance of those allowed us to determine the gap size to be Ͻ500 bp because the insert size of the public shotgun libraries was well defined. The structures of the human and murine corin genes were then analyzed. However, the 340-kb human genomic sequence did not contain the 5Ј-flanking region. We isolated an additional 4165-bp HindIII-EcoRI fragment from BAC 26540, which included the first 3919 bp of the 5Ј-flanking region, all of exon 1, and part of intron 1 (submitted to the GenBank TM /EBI Data Bank with the accession number AF521006; Fig. 3 depicts a 1596-bp portion of the sequence). Fig. 1 depicts the organization of the human and murine corin genes and the locations of BAC clones, contigs, and plasmid clones containing the corin genes and their promoter regions. Tables I and II indicate the sizes and locations of the exons and introns as well as the nucleotide sequences surrounding the splice donor and acceptor sites. The human and murine corin genes span ϳ238 and 204 kb, respectively. They consist of 22 exons and 21 introns. The relative sizes of all corresponding exons and introns are very similar. The nucleotide sequences at the 5Ј-donor and 3Ј-acceptor sites of all introns conform to the GT/AG rule (33).
Analysis of the genomic organization indicates that most intron/exon junctions are highly conserved between the human (Table I) and murine (Table II) corin genes. However, exon 1 diverges between human and mouse, which is consistent with the divergence in the cDNA sequences coding for the cytoplasmic tails of human and murine corin. Interestingly, the conjunction sequence ( ϩ62 GGTAAGATC ϩ70 ) between human exon 1 and intron 1 is identical to the mouse sequence ( Ϫ223 GGTAA-GATC Ϫ215 ), indicating potential alternative RNA splicing in the murine corin gene. Thus, the divergence between the cytoplasmic tails of human and murine corin found in the cDNA sequences could arise from alternative RNA splicing.
The corin cDNA sequence predicts a protein composed of a number of discrete domains. The boundaries between protein domains correspond mostly to the intron/exon boundaries of the genomic structure, as illustrated schematically in Fig. 2. The cytoplasmic tail at the N terminus is encoded by exon 1 and half of exon 2, followed by the transmembrane domain that is encoded by the other half of exon 2. The region between the transmembrane domain and the first frizzled domain is encoded by exon 3. Each of the frizzled domains is encoded by two exons, each of the eight low density lipoprotein receptors by a single exon, and the scavenger receptor cysteine-rich domain by three exons. The protease domain at the C terminus is encoded by exons 19 -22, with exon 19 coding for the sequence that includes the proteolytic activation site and the catalytic histidine residue. Exons 20 and 22 code for the sequences that include the other two catalytic residues aspartic acid and serine, respectively. The intron/exon splice junctions are split between all of the three codon phases (phases 0, 1, and 2) (Tables  I and II). This feature is conserved between human and mouse.
Comparison of the Structural Features of the 5Ј-Flanking Regions of the Human and Murine Corin Genes- Fig. 3 shows the alignment of the 5Ј-flanking sequences and the first exons of the human and murine corin genes. There is 62% sequence identity in the first 1 kb of the 5Ј-flanking regions. This degree of similarity is typical for the murine and human orthologs of a gene (34). Both genes share several putative regulatory regions, including two TBX5-binding sites (35), two GATA elements (24), two "GT" boxes for the Krü ppel-like factors (36), an NKX2.5-binding site (NKE) (37), and a TATA box (38). These elements are conserved not only in sequence, but also in relative spacing. We also found nonconserved putative binding sites for NF-AT (nuclear factor of activated T cells) and TBX5 transcription factors in the human and murine corin genes.
Functional Analysis of the 5Ј-Flanking Regions-To test whether the 5Ј-flanking regions of the corin genes have promoter activity, we prepared reporter constructs in which serially truncated fragments of the 5Ј-flanking sequence of the

ATAAAAATGGTA
human or murine corin gene were linked to a promoterless luciferase gene (Fig. 4A). These constructs were transiently transfected into murine cardiomyocytic HL-5 cells, which express corin mRNA and protein (13). Luciferase activities of the transfected cells were then measured. As shown in Fig. 4B, human constructs hCp1297LUC and hCp405LUC promoted luciferase activity that was significantly higher than the background in pGL3-basic-transfected cells. Similarly, murine reporter constructs mCp1183LUC, mCp809LUC, and mCp646LUC promoted significant luciferase activity, comparable to that promoted by the human constructs. These data suggest that the cis-sequences responsible for most of the promoter activity are located between Ϫ405 and Ϫ15 and between Ϫ646 and Ϫ77 in the human and murine corin genes, respectively.
To determine whether the constructs mediate cardiac-specific expression, we transfected them into HeLa cells, which do not express corin mRNA. In contrast to their high activities in HL-5 cells, constructs hCp405LUC and mCp646LUC had only minimal promoter activity in HeLa cells (Fig. 5). As a control, simultaneously transfected pRL-SV40 promoted higher levels of Renilla luciferase activity in HeLa cells than in HL-5 cells, indicating that HeLa cells were as readily transfected as HL-5 cells in these experiments. These results indicate that the 5Ј-flanking sequences from Ϫ405 to Ϫ15 in the human corin gene and from Ϫ646 to Ϫ77 in the murine corin gene contain elements that are sufficient for specific expression in cultured cardiomyocytes.
The Proximal GATA Elements Bind to GATA-4 and Are Required for Optimal Function of the Corin Promoters-The 5Ј-flanking regions from Ϫ405 to Ϫ15 in human and from Ϫ646 to Ϫ77 in mouse were sufficient to promote high levels of gene expression in cultured cardiomyocytes, but not in HeLa cells. This suggests that these regions contain regulatory elements responsible for cardiomyocyte-specific expression. Inspection of these regions revealed a conserved consensus GATA sequence (designated as the proximal GATA sequence), and the GATA element has previously been implicated in cardiac-specific expression (24).
To determine whether the proximal GATA sequences indeed bind to GATA proteins, we performed an EMSA using a well characterized consensus GATA oligonucleotide probe (30,31) and probes encompassing each of the proximal GATA sequences (Fig. 6A). As expected, the labeled consensus GATA probe formed a sequence-specific DNA-protein complex when incubated with nuclear extracts of HL-5 cells (Fig. 6B). The formation of this complex was prevented by addition of a 100fold excess of the unlabeled probe, but not of an unrelated ␥-interferon activation sequence element. The complex formation was dependent on the intact GATA sequence because mutations in the GATA sequence abolished the formation of the complex. Furthermore, the complex was not detected in the presence of a 100-fold excess of the unlabeled probe containing either the human or murine proximal GATA sequence. In contrast, a 100-fold excess of the unlabeled probes encompassing the mutated proximal GATA sequences had a minimal effect on the complex formation. These data indicate that the corin proximal GATA sequences and the consensus GATA probe bind to a common GATA protein(s).  To determine which GATA protein(s) was involved in the complex, we performed EMSA with the labeled consensus GATA probe in the presence of antibodies against members of the GATA family. As shown in Fig. 6C, an antibody against GATA-4 markedly inhibited the complex formation, whereas antibodies against GATA-1, -3, and -6 had little effect. To directly demonstrate the binding of GATA-4 to the proximal GATA sequence, we used the labeled human proximal GATA probe in the absence or presence of the same antibody against GATA-4. As shown in Fig. 6D, the antibody against GATA-4 completely inhibited the formation of a DNA-protein complex with a similar mobility to that of the complex formed with the consensus GATA probe. These data indicate that GATA-4 bound to the proximal GATA sequences, suggesting that the binding of GATA-4 to the proximal GATA sequences may contribute to the gene expression of corin in cardiac myocytes.
To corroborate whether the proximal GATA elements are actually required for the promoter activity, we mutated the wild-type sequence AGATAA to ACTTAA in the human and murine constructs with the highest promoter activity (Fig. 7). The mutations in the GATA element were the same as those made in the mutant GATA probes, which eliminated the binding to GATA-4 in the EMSAs. When transfected into HL-5 cells, the human and murine mutant constructs had 10 and 42% of promoter activities compared with their respective wild-type sequences. These results show that the proximal GATA ele- FIG. 3. Alignment of the 5-flanking regions of the human and murine corin genes. The 5Ј-flanking regions, exon 1, and part of intron 1 of the human (h) and murine (m) corin genes are aligned. The numbering is relative to the translation initiation codon ATG (boldface italics). Of note, the numbers indicated are different between human and mouse because of the divergence in the first exons. The arrowhead indicates the junction between the first exon and intron of the human corin gene, and the donor splice sequence of human intron 1 is underlined. The putative regulatory sequences are indicated in boldface (the Tbx5 site for binding to Tbx5, a T box-containing transcription factor; the NF-AT site for binding to the NF-AT transcription factor; GATA, an element for GATA proteins; the GT box for binding to the Krü ppel-like factors; the TATA box for binding to basal transcription factor IID; and NKE, a binding motif for Nkx2.5). The NKE sequence, which overlaps with the proximal GATA sequence, is also underlined.
FIG. 3-continued ments are required for constitutive expression of the human and murine corin genes in cultured cardiac myocytes. DISCUSSION We cloned the human and murine corin genes, determined their genomic structures, and analyzed the structure and function of their 5Ј-flanking regions. The conserved intron/exon boundaries and their correspondence to the boundaries of protein structural domains support the idea that corin arose from exon duplication and rearrangement. The comparison and functional analysis of the 5Ј-flanking regions reveal several conserved sequences, including a functional GATA element, providing a molecular basis for understanding the mechanism that regulates tissue-specific expression of corin.
Both human and murine corin genes span at least 200 kb and contain 22 exons interrupted by some large introns up to 50 kb. The corin genes are the largest genes identified so far among the trypsin-like serine protease superfamily. All but one of the intron/exon junctions are highly conserved between human and mouse. The exception is the first intron/exon, whose divergence leads to the different cytoplasmic domains of the predicted human and murine corin proteins. However, sequence analysis of the human and murine genomic structures indicates potential alternative RNA splicing of the first exon, suggesting the existence of splice variants.
The intron/exon junctions of the corin genomic structures correspond mostly to the boundaries between their protein structural domains. The codon phases used to split the junctions of their structural domains are conserved between human and mouse. The same types of codon phases are also used to split the same types of structural domains within the protein.
For example, each of the two frizzled domains is encoded by two exons, and each of the eight low density lipoprotein repeats is encoded by a single exon; the codon phases used to split these domains are the same.
The protease domain of all members of the type II transmembrane serine protease class consists of ϳ240 amino acid residues. In each gene, however, the number of exons coding for the protease domain may vary. For example, the protease domain is encoded by four exons in the corin, MT-SP1, and human airway trypsin-like protease genes; five exons in the enterokinase gene; and six exons in the human and murine hepsin (39) and the human TMPRSS2, -3, -4, and -5 genes (analysis of data from the NCBI Human Genome Database). Presumably, some of these exons have been fused (or interrupted) during the course of evolution. It is well established that exon shuffling and fusion have been a major driving force for generation of the multiplicity of domains in protein (40). The available data from the human and mouse corin genes, together with data from other members of the type II transmembrane serine protease family, support the concept that this class of proteins arose from duplication and rearrangement of preexisting exons encoding structurally distinct domains. To understand the mechanism of cardiac expression of the corin gene, we isolated the 5Ј-flanking regions of the human and murine genes. Transfection of serially truncated segments of the 5Ј-flanking region linked to a luciferase reporter gene indicated that the sequences from Ϫ646 to Ϫ77 in mouse and from Ϫ405 to Ϫ15 in human supported high level expression of luciferase, similar to longer constructs, in cardiomyocytes. In contrast, these sequences directed only minimal expression in HeLa cells. These findings suggest that these short 5Ј-flanking regions contain sufficient information to promote specific expression in cardiomyocytes. It remains to be determined whether these regions can mediate tissue-specific expression in vivo.
Within these short 5Ј-flanking regions, a number of conserved regulatory sequences were identified, including binding sites for TBX5 (35,41,42), GATA (24), NKX2.5 (37), Krü ppellike transcription factors (36,43), and an AT-rich sequence. These binding sites are conserved not only in sequence, but also in relative spacing in the context of the promoters. Furthermore, they are also present in the promoters of the ANP and/or BNP genes, as summarized in Fig. 8 (22,35,44,45).
The AT-rich sequences (Ϫ211 to Ϫ205 in human and Ϫ432 to Ϫ426 in mouse) fit well a typical TATA box (38). Interestingly, the TATA box is embedded in the conserved CAAAATATGG sequence, which resembles a serum response element (CC-(A/T) 6 GG) (46). The serum response element is present in many FIG. 7. Mutational analysis of the conserved proximal GATA elements. The same mutation (GATA to CTTA) that abolished binding to the GATA-4 protein in EMSA was introduced into luciferase (LUC) reporter constructs driven by the 5Ј-flanking regions from Ϫ642 to Ϫ77 in mouse and from Ϫ405 to Ϫ15 in human. The mutant and wild-type constructs were transfected into HL-5 cells along with the control construct pRL-SV40. The luciferase activity expressed by each construct was normalized to the Renilla luciferase activity expressed by pRL-SV40 for each transfection. The promoter activity of each mutant construct is expressed as a percentage of the corresponding wild-type construct. Each transfection experiment was performed in triplicate. The data represent the means Ϯ S.D. of three independent experiments.
FIG. 6. Binding of nuclear proteins to the regulatory sequence encompassing the proximal GATA element. A, shown are the sequences of the upper strand oligonucleotides used as probes and competitors. The GATA motifs in each sequence are in boldface, and the mutated nucleotides are in italics. The human and murine proximal GATA elements are from the indicated regions of the corin 5Ј-flanking sequences. The consensus GATA probe containing two GATA motifs was derived from the human T cell receptor-specific enhancer region (30,31). B, the labeled consensus GATA probe or its mutant probe (Mut.) was incubated with nuclear extracts from HL-5 cells in the presence or absence of a 100-fold excess of the indicated unlabeled oligonucleotides. The arrow indicates a GATA sequence-dependent DNA-protein complex. C, the labeled consensus GATA probe was incubated with nuclear extracts from HL-5 cells in the presence of antibodies against GATA proteins. The arrow indicates a DNA-protein complex whose formation was blocked by an antibody (Ab) against GATA-4, but not by antibodies against GATA-1, -3, and -6. D, the labeled human corin GATA probe was incubated with nuclear extracts from HL-5 cells in the presence or absence of an antibody against GATA-4. The arrow indicates the DNA-protein complex whose formation was completely blocked in the presence of the anti-GATA-4 antibody.
cardiac-expressed genes, including the ANP gene, and is one of the key regulatory elements for cardiac-specific expression (47,48).
The presence of the putative binding sites for TBX5, GATA, and NKX2.5 in the human and murine corin promoters suggests that the corin gene may be a downstream target gene regulated by these transcription factors. Intriguingly, mutations in the TBX5 gene cause heart and limb malformations in Holt-Oram syndrome (35), suggesting that TBX5 regulates not only cardiac-expressed genes, but also bone-expressed genes. In fact, the corin gene is also expressed in the long bones of developing limbs (1).
Unlike GATA elements in the ANP and BNP genes, the corin GATA sequence (Ϫ40 to Ϫ35 in human and Ϫ318 to Ϫ313 in mouse) is located in the first exon, proximal to the TATA box. This GATA sequence also overlaps with a putative NKX2.5binding site. The GATA sequence is critical for gene expression of corin because mutation of this GATA sequence in the human and murine reporter constructs significantly reduced the promoter activity in transfected cardiomyocytes. The same mutation in this GATA element also eliminated binding to GATA proteins in EMSAs. We further showed that this GATA element bound to GATA-4, but not to GATA-1, -3, and -6, in cultured cardiomyocytes, indicating that GATA-4 is a major transcriptional activator for corin gene expression. This is consistent with the fact that GATA-4 expression starts at 7.0 -7.5 days post-conception, preceding the earliest expression of corin at 9.5 days post-conception in mice (49). Moreover, GATA-4 expression is maintained throughout cardiac development in both the atrium and ventricle, concurrent with the expression of the corin gene.
The role of GATA-4 in transcription of the corin gene may have physiological and pathological implications in the regulation of blood pressure and volume. It has been demonstrated that GATA-4 serves as a pivotal regulator for expression of the ANP and BNP genes (22,24,45,50). In response to fluid volume overload or a hypertrophic signal, the heart produces more ANP and BNP to reduce blood volume and lower blood pressure. One of the mechanisms for such increased production has been shown to be mediated through signaling pathways converging onto GATA elements and/or NF-AT-binding sites in the ANP and BNP promoters (50 -53). It has been proposed that myocyte stretch elevates intracellular concentrations of Ca 2ϩ , leading to activation of calcineurin and hence dephosphorylation of the transcription factor NF-AT-3 in the cytoplasm. Upon dephosphorylation, NF-AT-3 translocates to the nucleus, where it binds to NF-AT sites and/or directly interacts with the GATA-4 protein, thereby promoting expression of the ANP and BNP genes. There are at least three putative NF-AT sites in the human and murine corin promoters, although these sites are not located in the conserved regions.
The presence of several common regulatory elements in the ANP, BNP, and corin genes suggests that these genes could be regulated by similar mechanisms. It is likely that expression of the corin gene is up-regulated in response to excessive blood volume or a hypertrophic signal. The cloning of the 5Ј-flanking regions of human and murine corin genes should help us understand their tissue-specific expression and regulation.