Cloning of cardiac, kidney, and brain promoters of the feline ncx1 gene.

The Na+-Ca2+ exchanger (NCX1) plays a major role in calcium efflux and therefore in the control and regulation of intracellular calcium in the heart. The exchanger has been shown to be regulated at several levels including transcription. NCX1 mRNA levels are up-regulated in both cardiac hypertrophy and failure. In this work, the 5′-end of the ncx1 gene has been cloned to study the mechanisms that mediate hypertrophic stimulation and cardiac expression. The feline ncx1 gene has three exons that encode 5′-untranslated sequences that are under the control of three tissue-specific promoters. The cardiac promoter drives expression in cardiocytes, but not in mouse L cells. Although it contains at least one enhancer (−2000 to −1250 base pairs (bp)) and one or more negative elements (−1250 to −250 bp), a minimum promoter (−250 to +200 bp) is sufficient for cardiac expression and α-adrenergic stimulation.

The Na ϩ -Ca 2ϩ exchanger is an integral membrane protein involved in the maintenance of calcium homeostasis in a number of cell types. Under physiological conditions, the exchanger functions as an antiporter, moving sodium and calcium in opposite directions across the cell membrane. The exchanger electrogenetically catalyzes the exchange of one intracellular Ca 2ϩ ion for three extracellular Na ϩ ions (1,2). The exchanger is reversible and can facilitate Ca 2ϩ entry, which is capable of triggering calcium release from the sarcoplasmic reticulum. NCX1 is predominantly expressed in cardiocytes (3), neurons, and cells of the renal tubules (4 -7). However, it is expressed at lower levels in a variety of other cell types, including skeletal muscle, smooth muscle, lung, and spleen. In the heart, the exchanger is the primary mechanism for calcium efflux (8) and therefore plays a major role in cardiac contractility. In the kidney, the exchanger is involved in the reabsorption of calcium and thus in maintenance of total body calcium homeostasis. The function of the exchanger in other tissues is less clear; however, in neurons, the exchanger has been implicated in the extrusion of calcium after neurotransmitter release (4).
The cDNA for the NCX1 Na ϩ -Ca 2ϩ exchanger was originally cloned from canine heart (9) and more recently has been cloned from other tissues, such as brain and kidney, and other species, including the cow (10), rat (11), human (12), and cat (13). It shows high homology across species and tissues, and the predicted primary structure of NCX1 includes a 32-amino acid N-terminal signal sequence followed by 11 putative transmembrane domains. Between domains 5 and 6 lies a large intracytoplasmic loop that is involved in regulation of the exchanger. Near the C-terminal end of the loop is a variable domain of 110 amino acids that is generated by the tissue-specific alternative splicing of six ncx1 exons (14). Eight tissue-specific isoforms have been identified (14 -16). In addition, three distinct 5Јuntranslated regions have been identified, each with a unique tissue distribution (7,17). Regulation of exchanger isoforms is made even more complex by the discovery of two additional ncx genes in mammals. In contrast to the more ubiquitous expression of ncx1, the ncx2 (18) and ncx3 (19) genes are expressed only in brain and skeletal muscle. ncx1 expression appears to be regulated by three tissue-specific promoters, which would play a role in determining the resulting splice variants in the large cytoplasmic loop.
Exchange activity in cardiomyocytes is regulated by several factors. It is activated by cytosolic Ca 2ϩ and MgATP (20) and inhibited by cytosolic sodium (21) and ATP depletion (22). A high affinity Ca 2ϩ -binding domain has been identified in the large cytoplasmic loop (residues 371-508) that is believed to be responsible for calcium regulation (23). It is also inhibited by the exchanger inhibitory peptide, XIP, that corresponds to a 20-amino acid segment at the N terminus of the large cytoplasmic loop (24). A recent study has demonstrated that the exchanger is phosphorylated via a protein kinase C-dependent pathway and that NCX1 phosphorylation appears to coincide with up-regulation of exchanger activity (25).
In addition, the exchanger is regulated at the transcriptional level in cardiac hypertrophy, ischemia, and failure. We have demonstrated the rapid up-regulation of NCX1 mRNA in response to pressure overload (26,27). The exchanger is also upregulated in end-stage heart failure (28), but exchanger levels drop in ischemia (29). In this report, we describe the molecular cloning of the cardiac, brain, and kidney promoters; the exons encoding their respective 5Ј-untranslated regions; and the first open reading frame exon of the feline ncx1 gene. In addition, characterization of the ncx1 cardiac promoter reveals regions required for cardiac expression and up-regulation. We define the 5Ј-flanking region containing the regulatory sequences required for cardiac expression and up-regulation. This work provides a starting point for pursuing an understanding of the processes by which Na ϩ -Ca 2ϩ exchanger expression is regulated.

EXPERIMENTAL PROCEDURES
5Ј-Rapid Amplification of cDNA Ends-RACE 1 was carried out as described previously (30 -32). 1 g of total RNA isolated from feline brain, kidney, and heart was annealed to an antisense ncx1 genespecific primer at positions ϩ781 to ϩ797 (5Ј-TTGTAAAACAG-CAGCCT-3Ј) and reverse-transcribed for 45 min at 42°C using 1 l of Superscript II reverse transcriptase (Life Technologies, Inc.) in a 25-l reaction mixture containing 0.05 M Tris-HCl, pH 8.5, 0.03 M KCl, 8 mM MgCl 2 , 0.25 mM dATP, 0.25 mM dCTP, 0.25 mM dGTP, 0.25 mM dTTP, 1 mM dithiothreitol, and 2.5 pM primer. The RNA was then degraded by addition of 1 l of RNase H and incubated for 10 min at 55°C. The cDNA was purified using the GlassMAX DNA purification system (Life Technologies, Inc.) and then tailed with dCTP at 37°C for 15 min using terminal dipeptidyltransferase (Promega). The tailed cDNA was amplified by polymerase chain reaction (30 cycles at 94, 50, and 72°C for 1 min each) using a second antisense ncx1 gene-specific primer located at positions ϩ207 to ϩ224 (5Ј-CTCTGGCAATTTTGTCTC-3Ј) and an anchor primer (5Ј-GGCCACGCGTCGACTAGTACGGGIIGGGIIGGGIIG-3Ј; Life Technologies, Inc.). Products were subjected to Southern analysis and hybridized to a [ 32 P]dCTP (3000 Ci/mmol; DuPont NEN)labeled PCR probe generated from region Ϫ26 to ϩ59 (sense primer, 5Ј-ACACTTGGAGGTCTAC-3Ј; and antisense primer, 5Ј-CAACG-GCTAACAGATGACATC-3Ј). Products that hybridized to the probe were ligated into the pT7Blue-T vector (Novagen) and transformed into competent NovaBlue cells (Novagen). Minipreps were prepared from the colonies and screened by Southern hybridization as described above. Positives were sequenced either by the dideoxy chain termination method using Sequenase 2.0 (U. S. Biochemical Corp.) or by cycle sequencing using the Ampli-Cycle sequencing kit (Perkin-Elmer) as specified by the manufacturer using an annealing temperature of 55°C.
Isolation of Total RNA and Northern Analysis-Feline brain, kidney, and heart tissues were dissected and immediately frozen in liquid nitrogen. Total RNA was then isolated from the tissue by the method of Chomczynski and Sacchi (33). Feline mRNA from heart, brain, and kidney was isolated using the FAST TRACK kit (Stratagene). Total RNA was isolated from rat neonatal cardiocytes using RNAzol (Cinna Biotex) as directed by the manufacturer. For Northern analysis, 15 g of total RNA or 1.2 g of mRNA was electrophoresed on each lane of a formaldehyde-agarose (1%) gel and transferred by the downward capillary method to a Duralon (Stratagene) membrane to which it was UV-cross-linked. The membrane was hybridized to [ 32 P]dCTP-labeled PCR probes generated as described under "Results." Membranes were then washed to a stringency of 1 ϫ SSC and 0.1% SDS at 42°C.
Screening of cDNA and Genomic Libraries-Both the feline cardiac cDNA and feline genomic libraries were screened using standard techniques (34). The libraries were plated on 10 150-mm plates at 30,000 plaques/plate. Duplicate lifts from each plate were hybridized to the [ 32 P]dCTP-labeled PCR probes. Filters underwent four 20-min washes at 42°C: the first two washes in 2 ϫ SSC and 0.1% SDS and the last two in 0.2 ϫ SSC and 0.1% SDS. The filters were then exposed to X-Omat AR film (Eastman Kodak Co.) overnight. Plaques that produced signals on both of the duplicate lifts were isolated and purified to homogeneity through three to four additional rounds of screening using the same probes and wash parameters.
A feline genomic library in the Lambda Fix II vector was purchased from Stratagene. This library was screened using the same [ 32 P]dCTPlabeled PCR probes used for Northern analysis. To subclone fragments containing exons and 5Ј-flanking regions, genomic clones were digested with EcoRI (Promega) and ligated into the pBluescript SK vector (Stratagene).
Primer Extension-10 g of mRNA was reverse-transcribed at 42°C for 1 h utilizing an antisense oligonucleotide corresponding to 51-71 nucleotides 5Ј of the translational start site. The reaction consisted of 1 ϫ reverse transcriptase buffer, 0.2 mM dTTP, 0.2 mM dATP, 0.2 mM dGTP, 0.02 mM dCTP, 4 mM dithiothreitol, 200 units of Moloney murine leukemia virus reverse transcriptase (Boehringer Mannheim), and 6.5 l of [ 32 P]dCTP. The reaction was heat-inactivated at 90°C. After addition of 10 l of stop solution (U. S. Biochemical Corp.), each sample was heated at 95°C for 3 min and run on a 7 M urea, 8% polyacrylamide sequencing gel. To determine the size of primer extension products, a sequencing reaction was loaded on the gel and run simultaneously. Sequencing was carried out as described for 5Ј-RACE experiments. The gel was then dried and exposed to X-Omat AR film for 3-6 h. S1 Nuclease Protection-S1 nuclease protection was carried out according to previously published methods (35). A 57-bp end-labeled oligonucleotide was coprecipitated with 30 g of cardiac total RNA or yeast tRNA. The pelleted nucleic acids were dissolved in 80% formamide, 0.4 M Pipes, pH 6.4, 0.4 M NaCl, and 1 mM EDTA; denatured for 10 min at 90°C; and annealed for 3 h at 55-60°C. The DNA-RNA hybrids were subjected to S1 nuclease digestion in 300 l of 0.05 M ammonium acetate, pH 4.5, 0.25 M NaCl, and 0.01 M zinc chloride containing 6.25 g of denatured salmon sperm DNA. 200 -300 units of S1 nuclease was added, and samples were incubated for 30 min at 37°C. Reactions were stopped by addition of EDTA to a concentration of 5 mM, extracted with phenol/chloroform, and precipitated with ethanol. S1 products were resuspended in 10 l of distilled H 2 O and analyzed by polyacrylamide gel electrophoresis as with primer extension reactions.
Promoter-Reporter Plasmid Constructs-Digestion of the genomic subclone containing exon H1 (pBH1) with SacI (New England Biolabs Inc.) liberated a 2200-bp fragment encompassing the H1 exon and 2000 bp of 5Ј-flanking region. This fragment was cloned into the SacI site of the pGL2-Basic vector (Promega) in both forward and reverse orientations. Deletion mutants were generated by a PCR mutagenesis strategy (36). Sequencing was carried out as described for 5Ј-RACE experiments.
Cell Culture-Primary cardiocytes were obtained from day 2-3 neonatal rats by previously published methods (37). Briefly, ventricular myocardium was isolated from 40 -50 neonatal rats. The tissue was finely minced, rinsed, and placed in a suspension culture flask. A mixture of 2.4 units/ml partially purified trypsin, 2.7 units/ml chymotrypsin, and 0.94 units/ml elastase in calcium-and magnesium-free Hanks' salt solution buffered with 30 mM Hepes, pH 7.4, was repeatedly added to the suspension culture flask for six 20-min incubations at 37°C to dissociate the cardiocytes. After each incubation, dissociated cells were centrifuged at 500 ϫ g and resuspended in modified Eagle's medium supplemented with 10% newborn calf serum. To enrich for cardiocytes, the cells were pooled and subjected to differential plating for 90 min. Cardiocytes were plated at a density of 1 ϫ 10 5 cardiocytes/ cm 2 for future mRNA isolation or 1 ϫ 10 6 cardiocytes/60-mm dish for transfection. The cardiocytes were maintained overnight in modified Eagle's medium supplemented with 10% newborn calf serum. The next day, the cells were switched to serum-free maintenance medium. Cells treated with phenylephrine prior to RNA extraction were treated for 2 h with 20 g/ml after 24 h of serum-free medium. Mouse L cells were grown and passaged at low density in Dulbecco's modified Eagle's medium supplemented with 10% fetal calf serum. They were plated at a density of 1 ϫ 10 6 cells/60-mm dish 24 h prior to transfection.
Transfections-One day prior to transfection, all cell types were plated at a density of 1 ϫ 10 6 cells/60-mm dishes. The following day, the cardiocytes were placed in maintenance medium supplemented with 4% horse serum 1-4 h prior to transfection, while mouse L cells were placed in Dulbecco's modified Eagle's medium supplemented with 4% horse serum.
All transfections were carried out in triplicate using 16 g of ncxluciferase construct cotransfected with 8 g of cytomegalovirus promoter-driven ␤-galactosidase expression plasmid. Cells were transfected by a modified calcium phosphate precipitation technique (38). Cells were incubated with the DNA/calcium phosphate precipitate for 16 h and then washed and kept in either maintenance medium (cardiocytes) or Dulbecco's modified Eagle's medium supplemented with 10% horse serum (mouse L cells). Immediately after being placed in maintenance medium, transfected cardiocytes were treated with 100 M phenylephrine or 10 M verapamil or left untreated. After 48 h, all transfected cells were washed twice with phosphate-buffered saline and scraped in 400 l of 1 ϫ Reporter Lysis Buffer (Promega). Lysates were quickfrozen and stored at Ϫ70°C until they were assayed for luciferase and ␤-galactosidase activities.

RESULTS
5Ј-RACE was utilized to determine whether the 5Ј-UTR of the feline ncx1 gene demonstrated the same tissue-specific heterogeneity identified in the rat (7). After two rounds of amplification, four major products were clearly present from brain, two from heart, and two from kidney ( Fig. 1). RACE was repeated at least three times from different tissue samples, and separate RNA isolation yielded similar results. The 5Ј-RACE was designed so that positive clones should contain ncx1 sequence from Ϫ33 to ϩ250 bp, allowing for the screening of positive clones by Southern blotting. Interestingly, all four major products from the brain showed positive hybridization in the Southern blot, and sequencing revealed that each had the correct ncx1 sequence from Ϫ33 to ϩ250 bp, but unique 5Јends. Each of these unique 5Ј-UTR sequences was used to generate PCR probes, and Northern blot analysis revealed that only the 350-bp 5Ј-RACE NCX1 cDNA clone was present at a detectable level in the brain (Fig. 2). This clone, Br1, contains 139 bp 5Ј of position Ϫ33, has a very high GC content, and has a 65% homology to the rat brain 5Ј-UTR (7). Both products found in the kidney were positive by Southern blotting and have the correct ncx1 sequence from Ϫ250 to Ϫ33 bp, where they diverge into unique sequences. Again, only one, K1, was present in the kidney Northern analysis (Fig. 2). This clone contained 278 bp 5Ј of position Ϫ33. K1 contains an out-offrame initiation codon within a nucleotide environment showing reasonable conformity to the Kozak consensus sequence (39). A polypeptide initiated at this site would precede 17 amino acids before terminating at a stop codon. The presence of alternative upstream initiation sites has been demonstrated to reduce the efficiency of mRNA translation (39,40). This suggests that ncx1 isoforms containing K1 may be under some level of translational control. Surprisingly, the feline kidney 5Ј-UTR shows no sequence homology within the rat kidney clone (7).
In the heart, both 5Ј-RACE products were positive on Southern analysis. Extensive sequencing of numerous clones showed that the products consisted of two separate isotypes, H1 and H2. The H1 product contained 110 bases 5Ј of the Ϫ33 splice site. This isoform showed significant homology to the rat H1 and bovine P13 clones, 68 and 85%, respectively. The H2 product consisted of the entirety (278 bp) of K1 sequence 5Ј of position Ϫ33 and then continued 5Ј to include the entire 110 bp of H1 (Fig. 1). Northern analysis indicates that both H1 and K1 are present in the heart (Fig. 2). More important, 5Ј-RACE with primers within the K1 sequence always gave a product that contained all or a substantial portion of the H1 sequence. Therefore, in the normal heart, K1 is only expressed as the H2 5Ј-UTR and is never detected alone. The existence of two feline cardiac 5Ј-UTR isoforms differing only in the splicing of the intervening sequence (K1) is consistent with reports concerning bovine cardiac ncx1 (10), but differs from what has been found in the rat (7). Extensive screening of a feline cardiac cDNA library resulted in confirmation of H1 and H2 and uncovered no new alternative 5Ј-UTRs. The library yielded a ratio of three clones containing H1 to one containing H2 (out of 23 positive clones). While not quantitative, the relative number of H1 clones compared with H2 clones suggests that H1 may be the more abundant cardiac isoform. This is further substantiated by Northern analysis, as is shown below.
Tissue Distribution of Isoforms-Northern analysis indicates a tissue-selective expression of the isoforms (Fig. 2). The blot was sequentially probed with sequences corresponding to the ORF, K1, Br1, and finally the H1 exons. Comparing expression levels using a common probe generated from the ORF, the exchanger appears to be most abundant in the heart, less abundant in the brain, and still less in the kidney. The Br1 5Ј-UTR is specifically expressed only in the brain, while the other 5Ј-UTR exons show more promiscuity. The H1 5Ј-UTR is expressed in both the heart and brain. The K1 5Ј-UTR sequence is expressed in both the kidney and heart. More important, the ORF-probed blot was exposed to film for 1 day, the H1-probed blot for 2 days, and the Br1 and K1 blots for 5 days.
FIG. 1. 5-RACE amplification of ncx transcripts in feline brain, heart, and kidney. The antisense 3Ј-primer (ASP) used for reverse transcription corresponded to positions ϩ781 to ϩ797 of the feline NCX1 cDNA sequence (13). The cDNA was amplified using a second antisense NCX1 cDNA-specific primer beginning at position 250 and an anchor primer. A, ethidium bromide-stained agarose gel containing products from brain, heart, and kidney reactions; B, diagram of the four unique 5Ј-UTRs spliced to the common sequence of the feline NCX mRNA in brain, heart, and kidney. ncx1 Gene Structure and Regulation Therefore, the relative band intensity indicates that, in the heart, the K1 sequence (H2) is expressed at very low levels compared with H1, again suggesting that H1 is the predominant cardiac isoform. The ORF probe also recognizes a transcript at 1.8 kb. Other probes corresponding to sequences in the 5Ј-end of the ncx1 ORF label a band at 1.8 kb (data not shown). Although these data are intriguing, the relationship of the 1.8-kb transcript to ncx1 is not clear.
Genomic Clones-To determine the structure of the 5Ј-end of the exchanger gene and to isolate its cardiac promoter region, a feline Lambda Fix II genomic library was screened using PCR probes generated from the open reading frame and each of the unique 5Ј-UTRs. Screening with a probe generated from region ϩ152 to ϩ484 yielded three clones (P1, P2, and P3) ϳ17 kb in size, each containing a 1.8-kb exon encoding nucleotides Ϫ33 to ϩ1826 (Fig. 3). Screening with probes generated from the K1 5Ј-UTR yielded one 18-kb clone (M1) that encoded both K1 and Br1 sequences. Screening with probes from the H1 5Ј-UTR yielded three identical clones ϳ15 kb in length. To further characterize the clones, each was mapped through restriction analysis. EcoRI fragments of the genomic clones were subcloned into pBluescript. These subclones were then restricted and subjected to Southern analysis, which in combination with PCR between exons and sequencing, was used to determine the arrangement of exons, distances between exons, and the ends of the clones as well as exon-intron boundaries. The data were combined to prepare the map shown in Fig. 3. The 1.8-kb exon 2 lies alone near the center of the P clones, which contain at least 6 kb of intron upstream and at least 6 kb of intron downstream. The M clone does not overlap with any of the P clones, but encodes both K1 and Br1 exons. K1 lies 1 kb upstream from Br1, with 6 kb of intron upstream and 6 kb downstream of the pair. The R clones contain only the H1 5Ј-UTR encoded as one exon located 3 kb from the 3Ј-end. The structure of the ncx1 gene is rather unique, with three upstream UTR exons, each expressed in a tissue-specific manner, followed by a 1.8-kb exon encoding over 60% of the total ORF of the transcript. There is no overlap between the R and M clones. At least 10 kb of intron lie between the 1.8-kb exon 2 and Br1 and Ͼ10 kb between K1 and H1. Hence, the 5Ј-UTR exons are spread over at least 25 kb of genomic sequence. Each exon conforms in sequence with the cDNA clones, and all were bound by conventional splice consensus sequences (Fig. 4). The recently isolated human NCX1 gene has 25 kb between what corresponds to the feline Br1 exon (exon 1) and the 1.8-kb exon (41). No other 5Ј-UTR exons were identified in the human genomic clones.
The start sites of the heart isoforms have been mapped by primer extension. Three transcriptional start sites were identified in the heart corresponding to 131, 139, and 143 bp 5Ј of the AUG codon (Fig. 5) using an antisense oligonucleotide corresponding to 51-71 nucleotides 5Ј of the AUG codon. Primer extension with a second oligonucleotide yielded products corresponding to the same initiation sites (data not shown). These were verified through S1 nuclease protection using an antisense end-labeled oligonucleotide spanning the H1 sequence from nucleotides Ϫ111 to Ϫ179 5Ј of the translational start site. S1 nuclease digestion produced protected fragments 33, 27, and 22 nucleotides in length (Fig. 5), confirming sites identified through primer extension. Control primer extension and S1 nuclease protection reaction using yeast tRNA as a template showed no products (data not shown).
5Ј-Upstream Regions-The sequence surrounding the cardiac transcriptional start site contains no apparent TATAA sequence or CCAAT box. In the first 350 nucleotides of the cardiac promoter, there are two CANNTG motifs (E boxes) at positions Ϫ173 and Ϫ155 that are potential target sites for the helix-loop-helix family of transcription factors (Fig. 4). A single M-CAT consensus element is present at position Ϫ325. M-CAT elements govern MyoD-independent cardiac transcription of several cardiac genes, including cardiac troponins T (42) and C (43), ␤-myosin heavy chain (44), and skeletal ␣-actin (45). The promoter region contains consensus sequence for two GATA boxes at positions Ϫ49 and Ϫ124. Several cardiac-specific genes such as myosin light chains IA and IV and ␤-myosin heavy chain (46,47) contain conserved GATA-binding motifs. The GATA elements in atrial natriuretic peptide (48) and ␣-myosin heavy chain (49) genes have been shown to be critical for cardiac expression. The first 350 nucleotides of the cardiac promoter also contain a single MEF-2 element at position Ϫ168. An MEF-2-like motif appears to be required for cardiacspecific expression of the rat cardiac troponin T gene.
Sequencing the 5Ј-flanking region 5Ј of the K1 exon revealed a TATAA-like motif at position Ϫ30 and a CATT box at position Ϫ133 (Fig. 4B). A consensus site for C/EBP was found at position Ϫ204. Farther upstream are putative sites for P2, AP-2, GATA-1, and two E boxes. To identify consensus sequences possibly involved in transcriptional regulation of the brain isoform, we have sequenced a portion of the 5Ј-flanking region. Sequencing proved difficult presumably due to a GC content of Ͼ75% in the region. No TATAA or CAAT boxes were evident. We have identified two overlapping Sp1 sites at positions Ϫ54 and Ϫ53, which are common in housekeeping genes (Fig. 4C). An EGR-1 consensus site begins at position Ϫ54, encompassing both Sp1 sites. At position Ϫ7 there is an E box, which is the consensus sequence for the helix-loop-helix family of transcription factors. Members of the helix-loop-helix transcription factor family have been implicated in the regulation of other neuronal genes. An AP-2 site has also been identified at position ϩ13.
Transfection Experiments-Transient transfection of freshly isolated rat neonatal cardiocytes confirmed the identity of the cardiac promoter. A fragment containing the complete H1 exon and 2 kb of the 5Ј-flanking sequence was cloned in both the forward and reverse directions into the pGL2-Basic vector. As shown in Fig. 6A, the forward construct (pH2000) reproducibly drove levels of luciferase expression 24 -55 times that of the pGL2-Basic vector alone. The reverse construct had no activity over vector alone. The chimeric luciferase gene construct con- ncx1 Gene Structure and Regulation taining the cardiac promoter was also transfected into mouse L cells. Luciferase activity in the transfected L cell extract was just above that of the pGL2-Basic vector alone (Fig. 6B). In addition, 1 kb of the kidney 5Ј-flanking sequence was also ligated into the pGL2-Basic vector. The putative kidney promoter did not drive luciferase expression significantly above background levels when transfected into neonatal cardiocytes (data not shown).
We have previously shown (26) that, in the intact heart, the levels of Na ϩ -Ca 2ϩ exchanger mRNA and protein are significantly elevated in early hypertrophy produced by pressure overload. This increase in mRNA is at least 2-fold and occurs within 1 h of pressure overload, indicating regulation of the exchanger in the manner of an immediate-early gene. ␣ 1 -Adotides of intron A. Numbering is relative to the start of transcription. H1 Px Primer is the oligonucleotide primer that was used to map the start site of transcription, and the dashed line indicates the oligonucleotide used as a probe for the S1 mapping experiment (see Fig. 5). nt, nucleotide. The asterisks indicate transcriptional start sites. The H1 5Ј-UTR contains a single alternative upstream ORF (ATG). The 5Јflanking region contains two E boxes, two GATA elements, an MEF-2 recognition sequence, and one M-CAT-binding site. B, sequence of the K1 exon and 5Ј-flanking region. The K1 5Ј-UTR contains four additional methionine-initiated ORFs, each followed by at least one in-frame stop codon (underlined). The kidney 5Ј-flanking region encodes consensus CAAT and TATA boxes at Ϫ120 and Ϫ28 bp, respectively. All other features are similar to those described for the H1 sequence. C, sequence of the Br1 exon 5Ј-flanking region. All features shown are similar to those described for the H1 sequence. D, sequence of the 1.8-kb exon 2. Numbering in this case is relative to the start of translation. The genomic sequence is identical to the feline heart cDNA sequence (13), and bases 113-1719 have not been included.  Fig. 4 and begins 71 nucleotides 5Ј of the translational start site. Three distinct extension products were observed, 92, 86, and 81 nucleotides (nt) long. The multiple start sites were confirmed by S1 nuclease mapping. 5Ј-Labeled oligonucleotide probe (position shown in Fig. 4) was annealed to 30 g of total cardiac RNA and then digested with S1 nuclease. The positions of the major protected fragments are indicated. The results of S1 nuclease protection and primer extension are summarized in the schematic, with numbers referring to the transcriptional start of exon H1.
ncx1 Gene Structure and Regulation renergic stimulated neonatal cardiomyocytes have several structural, morphological, and genetic markers of hypertrophy and provide a good starting point for our investigation of candidate cis-elements involved in the hypertrophic up-regulation of ncx1. We have already demonstrated that, within 2 h, ␣ 1adrenergic stimulation induces a 2-3-fold increase in NCX1 mRNA in neonatal rat cardiomyocytes plated at low density or pretreated with verapamil to prevent spontaneous beating (15). To determine if the rise in mRNA seen under phenylephrine stimulation is transcriptionally regulated by elements within the pH2000 construct, we treated cells transiently transfected with pH2000 with 100 M phenylephrine for 48 h. All dishes were cotransfected with a constant amount of cytomegalovirus-␤-galactosidase vector as a control for transfection efficiency. Phenylephrine treatment induced a 1.5-3-fold rise in luciferase activity (Fig. 6A) over that of untreated or verapamil-treated transfected cells, consistent with data from Northern analysis (15).
More important, cell density was an important factor in the transfection experiments. All data reported above were derived from cardiocytes plated at 1.0 ϫ 10 6 cardiocytes/60-mm plate. At this density, there is very little cell contact, and spontaneous beating is rare. Several experiments included pH2000 transfection of cells plated at higher densities. When cells were plated at densities high enough to produce rhythmic spontaneous beating, the basal level of activity of pH2000-transfected cardiocytes was measured as 1.5-2-fold greater than with transfection of less densely plated cells. However, in these experiments, the level of phenylephrine induction decreased and, in several of these experiments, was undetectable. If beating of control cells was inhibited by 10 M verapamil, basal activity decreased to levels just below that of the more sparsely plated cells, and phenylephrine up-regulation was restored.
To further delineate the promoter regions directing cardiac ncx1 gene transcription and phenylephrine-induced up-regulation, luciferase reporter constructs containing shorter portions of the cardiac 5Ј-flanking region were created through a PCR-based method (see "Experimental Procedures"). Clones containing 1250 bp (pH1250), 500 bp (pH500), and 250 bp (pH250) of 5Ј-flanking region fused to the luciferase reporter gene were generated and transfected into rat neonatal cardiocytes. In all experiments, transfection efficiency was monitored by cotransfection of a cytomegalovirus-␤-galactosidase vector, and the luciferase activity was normalized to ␤-galactosidase activity. The deletion of Ϫ2000 to Ϫ1250 bp yields a dramatic decrease in basal promoter activity. The pH1250 construct had luciferase activity just above that of the control promoterless plasmid pGL2-Basic. Constructs truncated at Ϫ500 (pH500) or Ϫ250 (pH250) bp yield luciferase activity at 15 and Ͼ50%, respectively, of the pH2000 construct. More important, both the pH250 and pH500 construct promoter activities were up-regulated at least 2-fold by ␣ 1 -adrenergic treatment. Thus, sequences within region Ϫ250 to ϩ200 of the feline ncx1 cardiac promoter confer cardiac expression and ␣ 1 -adrenergic induction. There also appears to be a cardiac-specific enhancer between Ϫ2000 and Ϫ1250 bp and one or more negative regulatory elements between Ϫ1250 and Ϫ200 bp. But these elements do not appear to be required for ␣ 1 -adrenergic stimulation.

DISCUSSION
The ncx1 gene is unusual in that it contains three promoter regions and three 5Ј-UTR exons upstream of the coding region. As a result of alternative promoter usage and alternative splicing of these 5Ј-UTR exons, ncx1 transcripts in the brain, heart, and kidney possess at least four different 5Ј-UTRs. The Br1 5Ј-UTR is expressed predominantly in the brain, but is not detectable (by Northern analysis) in the heart or kidney. Unlike the rat heart, in which only a single species of 5Ј-UTR is expressed, we have discovered two forms in the feline heart. The first isoform contains the H1 exon spliced to the common acceptor site (Ϫ33) at the beginning of exon 2. This is the predominant transcript found in the heart. The second transcript is derived by an alternative splice receptor in the K1 exon, which results in the inclusion of the entire K1 exon between the H1 exon and exon 2.
The existence of two feline cardiac 5Ј-UTR isoforms differing only in the splicing of the intervening sequence (K1 exon) is consistent with what was reported for the bovine cardiac ncx1 p17 clone (10). Bovine p13 and the 5Ј-end of the bovine p17 clone are identical and have 85% homology to the feline H1 sequence. Interestingly, the p17 clone contains a 156-bp insert that, although not homologous to the feline K1 sequence, has significant homology to the rat kidney 5Ј-UTR sequence (7). The K1 sequence was always found as an alternative exon inclusion with H1 and was never detected alone. In addition, transfection of neonatal cardiocytes with the kidney promoterluciferase fusion construct never gave luciferase expression above background levels. Therefore, K1 expression in the heart is not regulated by the kidney promoter. Only one species of transcript is present in the kidney, in which the K1 exon is spliced directly to the common receptor site (Ϫ33) at the beginning of exon 2.
Several other genes have multiple promoters and undergo complex alternative splicing events that are regulated in a tissue-specific or developmental manner. In some cases, the alternative promoters control transcription of the same protein, but diverge only in the 5Ј-UTR sequences, such as the human mineralocorticoid receptor gene (50) and lck gene (51). The parathyroid hormone receptor gene (52) is regulated by two promoters, one tissue-specific and the second controlling ubiquitous expression. The pem homeobox gene is regulated by either an androgen-dependent or -independent promoter in a tissue-specific manner (53). In some cases, the alternative splicing events alter both the 5Ј-UTR and coding regions. This is what is occurring with the ncx1 gene (Refs. 7 and 15 and this work). Schulze and co-workers (15) have identified several tissue-specific isoforms of ncx1. These variants are produced by the alternative splicing of a cluster of six exons (termed A-F) coding for the variable region in the C terminus of the large intracellular loop. To date, nine isoforms have been identified from cDNAs and reverse transcription-PCR in a wide variety of cells and tissue (7,15,19). The role of each promoter and 5Ј-UTR splice variant in determining alternative splicing in the variable region of the large cytoplasmic loop is currently being investigated. More important, the two alternatively spliced 5Ј-UTR species found in the heart yield two loop isoforms of the exchanger. 2 Therefore, unlike the rodent (7,15), the cat expresses two distinct ncx1 isoforms in cardiac tissue. The cytoplasmic loop has been demonstrated to contain domains involved with intracellular calcium regulation and exchangerinhibitor peptide interaction. Given the diverse roles that the exchanger plays in the kidney, brain, and heart, it is believed that its regulation would differ between tissues. Thus, the variable region may play a role in the tissue-specific regulation (7,15) or possibly in subcellular targeting of the exchanger.
The ncx1 brain and heart promoters have several features common to a number of housekeeping genes, including a high GC content, a high frequency of the dinucleotide CpG, and the absence of TATA and CAAT boxes. Only the kidney promoter contains canonical CAAT and TATA boxes. This is consistent with the finding that the brain and heart isoforms are expressed at very low levels in several tissues (7). Although not conclusive, 5Ј-RACE products and transfections of the kidney promoter support the premise that kidney promoter expression is tissue-specific. The kidney promoter has several consensus sequences, including two E boxes and C/EBP, P2, AP-2, and GATA-1 sites. Only 88 bp of the brain promoter sequence have been obtained presumably because of extensive secondary structure. This proximal end of the brain promoter has a very high GC content (Ͼ75%) and two Sp1 elements. As discussed above, although the heart promoter has several features of a housekeeping gene, it also has several elements that have been demonstrated to be required or important for cardiac expression of other genes. These include two GATA elements, two E boxes, and an M-CAT element.
To begin to characterize the ncx1 cardiac promoter, a DNA fragment containing 2000 bases of the 5Ј-flanking region and 122 bases of the H1 exon plus 80 bp of the first intron was fused to the luciferase reporter gene and transfected into rat neonatal cardiocytes. Deletion analysis demonstrated that there may be one or more enhancer elements between Ϫ2000 and Ϫ1250 bp and at least one negative element between Ϫ1250 and Ϫ250 bp. But more important, the first 250 bp of the 5Ј-flanking region contain all that is requisite for cardiac-directed expression and ␣ 1 -adrenergic stimulation of the ncx1 gene. This region contains the two GATA elements, an Sp1 site, a single AP-2 element, and two E boxes. More important, the M-CAT element at position Ϫ380 is not required for basal expression or ␣ 1 -adrenergic stimulation. This is clearly different from what has been reported for both the ␤-myosin heavy chain (54) and skeletal ␣-actin (55). The ␣-adrenergic induction of the the ␤-myosin heavy chain requires M-CAT and possible AP-1 elements, and the induction of the skeletal ␣-actin promoter also requires M-CAT as well as c, AT-rich G sequence (CArg) and Sp1 elements. Studies are currently in progress to identify the specific elements required for cardiac expression and the upregulation of the ncx1 gene in cardiac hypertrophy.