Many studies have investigated the regulation of the Na+/Ca2+exchanger, NCX1, but limited data exist on transcriptional regulation of the NCX1 gene. We have identified the transcription start sites of three tissue-specific alternative promoters of NCX1 transcripts from rat heart, kidney, and brain. We have characterized the cardiac NCX1 promoter, from which the most abundant quantities of NCX1 transcripts are expressed. Transfection of primary cardiac myocytes, CHO cells, and COS-7 cells with overlapping genomic DNA fragments spanning the NCX1 cardiac transcription start site has uncovered a cardiac cell-specific minimum promoter from −137 to +85. The cardiac NCX1 promoter is TATA-less but has putative binding sites for cardiac-specific GATA factors, an E box, and an Inr as well as multiple active enhancers. The kidney NCX1 promoter has a typical TATA box and binding sites for several tissue-specific factors. The brain NCX1 promoter is very GC-rich and possesses several Sp-1 binding sites consistent with its ubiquitous expression.
- transcription start sites
- cardiac myocytes
- genetic regulation
- transcription factors
theNa+/Ca2+exchanger (NCX1) is a ubiquitous transporter that serves an important role in regulating and maintaining cellular Ca2+ balance in various tissues by exchanging three Na+ for one Ca2+ across the plasma membrane (2, 3, 33, 37). This membrane protein is most abundant in the heart, where it is involved in excitation-contraction coupling as the dominant myocardial Ca2+ efflux system (6).
In addition to NCX1, two other exchanger genes, NCX2 and NCX3, have been identified (25, 31). Tissue-specific gene expression of Na+/Ca2+exchangers is regulated at multiple levels. At the posttranscriptional level, at least 12 NCX1 and 3 NCX3 protein isoforms are generated through alternative splicing of the primary nuclear transcript. These splicing variants differ in the COOH terminus of the large intracellular loop located in the center of the protein and are expressed in a tissue-specific manner (20, 24, 35). The function and significance of these protein-splicing isoforms are under investigation.
There is also evidence indicating that regulation of NCX1 occurs at the level of transcription. In the feline model of acute right ventricular hypertrophy, NCX1 mRNA levels are upregulated following pressure overload (28). An increase in NCX1 mRNA expression was also observed in cultured cardiac myocytes following α-adrenergic stimulation by phenylephrine or exposure to veratridine (28). NCX1 mRNA and protein levels decline following birth and during normal cardiac development in rabbit and rat heart (4). In humans, end-stage heart disease has also been shown to result in an increase in NCX1 mRNA and protein expression (12, 40). This change in NCX1 expression may be a compensatory adaptation to systolic and diastolic dysfunction in the failing heart. Nevertheless, it reflects a definite change in expression of the gene in response to an environmental signal resulting from a specific disease process with poor human outcome.
The proposed mechanism by which a genetic or environmental signal is transmitted to the gene to eventuate a change in the level of expressed protein involves transcription factors that interact with cognate,cis-acting elements of the promoter to regulate transcription. Despite extensive research characterizing the functional regulation of the Na+/Ca2+exchanger at the protein level, no information regarding the transcriptional regulation of the NCX1 gene by tissue-specific factors is available.
Lee et al. (24) and Lytton et al. (26) performed rapid amplification of rat NCX1 cDNA 5′-ends (5′-RACE) and identified three distinct transcripts (designated Ht, Kc, or Br) with different transcription initiation sites presumably due to selective use of alternative NCX1 promoters (21). Similar experiments have recently been performed in cat with comparable findings (1). In rat, transcript containing Ht, Kc, or Br ends were found to be selectively expressed in rat heart, kidney cortex, or ubiquitously elsewhere (but enriched in brain), respectively. This selective use of alternative promoters could potentially orchestrate the tissue-specific expression of NCX1 protein in response to external stimuli (24). We now extend that work to identify the transcription start sites of these three tissue-specific alternative promoters. The NCX1 promoter containing the cardiac transcription start site is also further characterized to define the cell specificity and minimum promoter sequence as well as potential mechanisms of regulation. The significance of transcription factor binding sites within the sequences of the three alternative promoters is discussed.
MATERIALS AND METHODS
Preparation of Genomic DNA and Southern Blot Analysis
Southern blot analysis was performed according to established procedures (36) using rat genomic DNA isolated from liver. Briefly, DNA (5 μg) was digested with EcoR I, BamH I, or Hind III overnight at 37°C. After precipitation with ethanol, the samples were separated on a 0.8% agarose gel, transferred to a nylon membrane by diffusion, and probed with digoxigenin-labeled cDNA probes corresponding to the three different 5′-untranslated region (UTR) exons (Ht, Kc, and Br) as previously described (24).
Genomic Cloning and Sequencing
A rat genomic library in the vector λ-DASH II was obtained from Stratagene. Probes corresponding to Ht, Kc, and Br, as well as a probe from the 3′-end of the NCX1 coding region, were used to screen duplicate filter lifts of the library. Positive clones were plaque purified, and phage DNA was isolated according to established procedures (36). Maps of the different clones were obtained by restriction endonuclease digestion and hybridization with various NCX1 probes. An ∼3.6-kilobase (kb) EcoR I fragment that hybridized to the Ht probe and an ∼3.2-kb EcoR I/BamH I fragment that hybridized to both Kc and Br probes were subcloned and sequenced using a combination of subclones and primer walks with either Sequenase version 2.0 (USB) or AmpliTaq FS (Perkin-Elmer). Dye-terminator fluorescent cycle sequencing was performed on an ABI model 373 automated sequenator at the University of Calgary DNA Core Facility.
Ribonuclease (RNase) protection assays were performed essentially according to Sambrook et al. (36) as modified by Bordonaro et al. (5). Briefly, riboprobes were synthesized from linearized templates, as described below, in the presence of [α-32P]UTP and were purified by ethanol precipitation. Total RNA (20 μg) isolated from various rat tissues by standard guanidinium thiocyanate extraction and cesium chloride centrifugation (36) was precipitated, dissolved together with 500,000 counts/min (cpm) of riboprobe in 15 μl of hybridization buffer [80% formamide, 400 mM NaCl, 1 mM EDTA, 40 mM piperazine-N,N′-bis(2-ethanesulfonic acid), pH 6.4], denatured at 90°C, and then hybridized overnight at 42°C. The annealed complexes were digested by adding 175 μl of 300 mM NaCl, 5 mM EDTA, 10 mM tris(hydroxymethyl)aminomethane (Tris) ⋅ HCL (pH 7.5) containing RNase A and RNase T1 (40 μg/ml and 250 U/ml, respectively) and incubating at 30°C for 1 h. After treatment with proteinase K to inactivate the RNases, the RNA was precipitated and separated on a 50% urea, 5% acrylamide sequencing gel. The dried gel was exposed to X-ray film to visualize the protected bands. The size of the protected fragments was determined by comparison with a combination of a DNA sequencing ladder and synthetic riboprobe standards of known length run side by side with the samples on the same gel. Because RNA fragments have a different mobility than DNA fragments, it was important to calibrate the DNA ladder with the RNA markers. This method introduces an error of ± 1 nucleotide in the determination of sizes. In all cases, sizes were confirmed with a minimum of two different experiments and gels.
Preparation of Riboprobes
Two riboprobes of different lengths were constructed for each 5′-end exon: Ht, Kc, and Br (see Fig. 5 for RNase protection data and schematic diagrams of the constructs used).
The genomic Hind III/Pst I fragment spanning exon 1-Ht (see Fig. 3, nucleotides 2469–2853) and a Pst I fragment from one of the heart 5′-RACE clones (24) (cDNA coordinates −52 to +97) were combined together in pBluescript II SK(−) (Stratagene) to create a hybrid species comprising the 5′-flanking sequence, exon 1-Ht, and 128 nucleotides of exon 2. The construct was linearized by digestion with Cla I and transcribed with T3 RNA polymerase to create a 564-nucleotide riboprobe (Ht-short). The Ht-short construct was combined with another cDNA fragment, thereby extending the 3′-end to an Acc I site at cDNA coordinate 250. After linearization with EcoR I, a 732-nucleotide riboprobe (Ht-long) was transcribed with T7 RNA polymerase.
The genomic Sac I fragment spanning most of exon 1-Kc (see Fig. 4, nucleotides 1479–1849) was subcloned into pBluescript II SK(−). After the construct was linearized with EcoR I, transcription with T3 RNA polymerase generated a 443-nucleotide riboprobe (Kc-short). AnSpeI/Apa I genomic fragment (see Fig. 4, nucleotides 1502–1657) was combined in pBluescript II SK(−) together with an ApaI/Hinc II fragment from rat kidney cDNA clone F1 (24) (cDNA coordinates −302 to +84). This construct was linearized with Xba I and transcribed by T7 RNA polymerase to yield a 573-nucleotide riboprobe (Kc-long).
The genomic RsaI/BamH I fragment spanning most of exon 1-Br (see Fig. 4, nucleotides 2713–3202) was subcloned into pBluescript II SK(−), linearized withTspR I, and transcribed with T3 RNA polymerase to produce a 361-nucleotide riboprobe (Br-short). The Br-short RsaI/BamH I fragment and aBamH I/Hinc II cDNA fragment from a brain 5′-RACE clone (24) (cDNA coordinates −57 to +84) were combined into pBluescript II SK(−). TheTspR I linearized template was transcribed with T3 polymerase to yield a 458-nucleotide probe (Br-long).
Confirmation of the transcriptional start site for each 5′-end was obtained by primer extension analysis essentially according to Sambrook et al. (36). The following antisense primers were used: heart primers H1 and H2, cDNA coordinates −28 to +4 and −7 to +23, respectively (24); kidney primers K1 and K2, cDNA coordinates −312 to −298 and −309 to −283, respectively (24); and brain primers B1 and B2, cDNA coordinates −151 to −123 and −125 to −97, respectively (24) (see Fig. 4). For experiments using end-labeled primer, 13 pmol of primer were labeled with 0.05 mCi of [γ-32P]ATP in a 10-μl reaction using 10 U of T4 polynucleotide kinase. After extraction with phenol and chloroform, the primer was separated from unincorporated label by passage over a Bio-Rad Bio-Spin 6 column. The specific radioactivity ranged from 0.6–1.0 × 106 cpm/pmol of primer. Total RNA (20 μg) from heart left ventricle, kidney cortex, or brain cerebrum was combined with 0.5 pmol of primer and precipitated with ethanol. The mixture was dissolved in 10 μl of hybridization buffer (seeRibonuclease Protection), heated to 95°C for 5 min, and then cooled slowly to 37°C and maintained for 5 min to 2 h. The annealed complex was then ethanol precipitated and reverse transcribed using “Expand” Maloney murine leukemia virus reverse transcriptase, essentially according to the instructions of the manufacturer (Boehringer Mannheim). For experiments involving label incorporation rather than end-labeled primer, a two-step reaction was used. Labeling was accomplished by reverse transcription in the presence of 0.5 μM [α-32P]dCTP and 1.5 μM of the other dNTPs for 20 min. The reaction was then completed in the presence of 0.5 mM dNTPs for 60 min. After reverse transcription, the template RNA was destroyed by RNase A treatment and the sample was extracted with phenol and chloroform, ethanol precipitated, and resolved on a 50% urea, 5% acrylamide sequencing gel with a DNA sequencing ladder (M13 template primed with universal primer).
Plasmid Constructs and DNA Purification
The plasmids pGL2 Basic, pGL3 Basic (pGL2B and pGL3B; Invitrogen), pMLC, pRSV, and pON249 were used for the NCX1 promoter studies. The vectors pGL2B and pGL3B contain a luciferase reporter gene but no promoter, whereas pMLC contains the myosin light chain-2 cardiac-specific promoter (13, 43) upstream of a luciferase reporter gene. The plasmid pRSV contains a strong SV40 enhancer element upstream of the promiscuous Rous sarcoma virus (RSV) promoter and a luciferase reporter gene. The plasmid pON249 containing the cytomegalovirus promoter upstream of the β-galactosidase gene was used to monitor the transfection efficiency. The 3.6-kb EcoR I DNA fragment and the 421-base pair (bp) Hind III DNA fragment containing the NCX1 cardiac transcription start site (see Fig. 3) were each ligated into pGL2B upstream of the luciferase reporter gene to create pGL2-2805 and pGL2-336, respectively (see Fig.7 B).
Similarly, the 3.1-kb EcoR I-Nhe I DNA fragment (see Fig. 3, nucleotides 1–3095) containing the cardiac transcription start site, the 1.3-kb SstI-Sma I DNA fragment (see Fig. 4, nucleotides 1959–3197) containing 1086 bp upstream from the brain transcription start site, and the 1.9-kb EcoR I-Ban I DNA fragment (see Fig. 4, nucleotides 1–1897) containing 1663 bp upstream from the kidney transcription start site were each ligated into pGL3B to create pGL3Ht-2805, pGL3Br-1086, and pGL3Kd-1663 (see Fig.7 C). All of these plasmid constructs were sequenced from both ends of the insert using Sequenase version 2.0 to confirm the orientation upstream of the luciferase gene.
Polymerase chain reaction (PCR) primers, containing additional restriction sites at the 5′-ends (underlined; Retrogen, San Diego, CA; see Fig. 4), were as follows: A1 oligonucleotide, 5′- AAGCTTTGTGTGTG-3′ (−336 to −323); B4 oligonucleotide, 5′- CCATTTCCCACCCA-3′ (−217 to −230); A2 oligonucleotide, 5′ GAATTGTGGGTGGG-3′ (−236 to −223); B3 oligonucleotide, 5′- CAACAGTAACGCCA-3′ (−117 to −130); A3 oligonucleotide, 5′- GACAGCTTGGCGTTA-3′ (−137 to −123); B2 oligonucleotide, 5′- CCCGGGAAGTTTG-3′ (−30 to −17); A4 oligonucleotide, 5′- CTCAGGACAAACTT-3′ (−36 to −23); B1 oligonucleotide, 5′- AAGCTTTGCACCTACCT-3′ (+85 to +69). These primers were used to truncate the 421-bpHind III NCX1 cardiac promoter into successively shorter pieces: 321 bp (A2-B1, −236 to +85), 222 bp (A3-B1, −137 to +85), and 121 bp (A4-B1, −36 to +85), respectively. The plasmid constructs created included pGL2-236, pGL2-137, and pGL2-36.
Similarly, PCR primers were used to create plasmid constructs containing potential NCX1 enhancer elements upstream of a heterologous thymidine kinase minimum promoter (pTK109) (32): pTK-336-217(A1-B4, −336 to −17), pTK-236-117 (A2-B3, −236 to −117), and pTK-137-17 (A3-B2, −137 to −17). These constructs were used in transfection experiments of primary myocytes. Each amplified DNA fragment was sequenced in both directions before the final plasmid construct was created.
Large-scale plasmid DNA purification was done by alkali lysis and subsequent polyethylene glycol precipitation as described by Sambrook et al. (36).
Preparation of Primary Cardiac Myocytes
After cervical dislocation of 24- to 48-h-old Sprague-Dawley rats, the hearts were trimmed, minced, and placed in 1× ADS buffer [116 mM NaCl, 20 mMN-2-hydroxyethylpiperazine-N′-2-ethanesulfonic acid, 1 mM NaH2PO4, 5 mM KCl, 0.8 mM MgSO4, 5.5 mM glucose, pH 7.35]. The cells were digested in the same buffer containing pancreatin (0.6 mg/ml) and type II collagenase enzyme (65 U/ml; Worthington) in a 37°C water-jacketed Celstir cell spinner. At the end of each of six 20-min digestions, the cells were collected by centrifugation at 700 g for 6 min in fetal bovine serum (FBS) and kept at 37°C. Primary myocytes were separated from nonmyocyte cells by Percoll gradient centrifugation for 30 min at 3,000 g. The cells were plated and maintained in Dulbecco’s modified Eagle’s medium (DMEM)-F12, 10% horse serum, and 5% FBS to a concentration of 1 × 106 cells/35-mm culture plate at 37°C in 5% CO2 for 18–24 h.
Primary cardiac myocytes were transfected by the calcium phosphate-DNA coprecipitation method (8) using 10 μg of plasmid DNA and 5 μg of pON249 and were incubated for 12–20 h in 3% CO2 at 37°C. Transfection was stopped by washing the cells with 1× phosphate-buffered saline (PBS) and returning them to fresh medium at the initial maintenance conditions for 48 h before lysing to obtain cell extract.
Chinese hamster ovary (CHO) cells were a gift from Dr. P. Edwards, and African Green monkey kidney (COS-7) cells were a gift from Dr. D. Woo. Both continuous cell lines were maintained in DMEM-F12/10% FBS at 37°C in 5% CO2 and were grown to ∼60–80% confluence in 60-mm culture plates before transfection. Similarly, these cells were transfected with 10 μg of plasmid DNA and 5 μg of pON249 by the calcium phosphate-DNA coprecipitation method and were incubated for 3–4 h in 3% CO2 at 35°C. Transfection was stopped by washing the cells several times with 1× PBS and returning them to 37°C in 5% CO2 until the cells were lysed. All transfections were done in duplicate to quadruplicate.
To obtain cell extract, the cells were first washed with ice-cold 1× PBS. Cell lysis buffer [primary myocytes: 200 μl of 0.1 M potassium phosphate, 0.5% Triton X-100, and 1 mM dithiothreitol (DTT); continuous cells: 500 μl of 125 mM Tris phosphate (pH 7.80), 10 mM DTT, 10 mMtrans-1,2-diaminocyclohexaneN,N,N′,N′-tetraacetic acid, 50% glycerol, and 5% Triton X-100] was added to the culture plates. The plates were then placed on ice for 15 min; cell extract was collected into Eppendorf tubes and stored at −70°C. All assays were performed within 2 wk of extract collection.
The relative light units of 20 μl of cell extract and 100 μl of assay buffer (100 mM tricine, 10 mM MgSO4, and 2 mM EDTA, pH 7.80) containing 2 mM ATP and 10 mM luciferin were measured using the Monolight 1500 luminometer (Analytical Luminescence).
Cell extract (30 μl) from primary myocytes and 270 μl of assay buffer (100 mM MgCl2, 5 M β-mercaptoethanol, 150 mM chlorophenol red β-d-galactopyranoside, and 0.1 M Na2HPO4) were incubated for 30 min to 1 h until an orange color change was noted. The reaction was stopped with 500 μl of 1 M Na2CO3, and β-galactosidase activity was measured as absorbance at 574 nm.
Similarly, 100 μl of cell extract from CHO and COS-7 cells and 900 μl of assay buffer (0.1 M Na2HPO4, 10 mM KCl, 1 mM MgSO4, and 50 mM β-mercaptoethanol) were incubated at 37°C. After 5 min, 200 μl of O-nitrophenyl β-d-galactopyranoside (4 g/l in 0.1 M Na2HPO4, pH 7.5) was added and kept at 37°C until a bright yellow color was observed. The reaction was terminated with 500 μl of 1 M Na2CO3, and β-galactosidase activity was measured as absorbance at 421 nm. All assays were done in duplicate.
The transfection efficiency was normalized for each experiment by comparing luciferase relative light units to β-galactosidase absorbance readings, the ratio of which reflected promoter activity. Promoter activity of pMLC and the various NCX1 promoter and enhancer constructs in each cell type was then standardized by comparing their activity to maximum activity obtained from the tissue-nonspecific pRSV promoter.
The raw data were analyzed using standard statistical analyses and Student’s t-test.
Mapping of Region Encompassing NCX1 5′-UTR Sequences
To delineate the exons encoding the three unique 5′-UTR sequences (Ht, Kc, and Br) of the rat NCX1 gene previously described (24, 26), restriction endonuclease-digested rat genomic DNA was analyzed by Southern blotting using probes from the three unique sequences. As shown in Fig. 1, all three probes hybridized to an ∼20-kb BamH I fragment. The Kc and Br probes both hybridized to the same EcoR I (∼8 kb) andHind III (∼9 kb) fragments, whereas the Ht probe hybridized to an ∼3.6-kb EcoR I fragment and an ∼0.4-kbHind III fragment. The nature of the fainter bands recognized by the Br probe is unknown, but they may reflect cross-hybridization due to the high GC content of this probe.
A rat genomic library was screened with the Ht, Kc, and Br probes as well as a probe from the 3′-end of the NCX1 coding region. Clones of 3 classes were identified and purified: 15 clones recognized by the Ht probe; 3 clones recognized by both Kc and Br probes; and 5 clones recognized by the coding-region probe. In no case did the probes hybridize to clones from another group.
The exact location of the Ht, Kc, and Br regions was then determined by a combination of Southern hybridization, restriction endonuclease mapping, and sequencing (Fig. 2). Note that a BamH I site (Fig. 2) exists close to the 3′-end of the Br sequence. Because all three probes hybridized to a single ∼20-kb BamH I fragment, this suggests that the clones containing the Ht region lie upstream from those containing the Kc and Br regions. The DNA sequences for Ht and Kc/Br are shown in Figs.3 and 4, respectively. From these data it is evident that the full extent of the unique Ht, Kc, and Br sequences previously described in cDNA (24, 26) are present within the genomic clones as unique exons. These three exons form the 5′-end of NCX1 transcripts in different tissues and therefore represent three different exon 1 sequences. We will refer to these different exons as exon 1-Ht, exon 1-Kc, and exon 1-Br, respectively.
Identification of Heart, Kidney, and Brain Transcriptional Start Sites for NCX1 Gene
The location of the transcriptional start sites was determined using RNase protection and primer extension analyses of RNA from rat heart, kidney, and brain. Short and long riboprobes were prepared from 1-Ht and 1-Kc, and a long riboprobe was prepared from 1-Br (Fig.5 D) so that the 5′-site of protection could be identified unambiguously. As shown in Fig. 5 A, Kc-long and Kc-short probes each resulted in a single major protected species of 474 and 285 nucleotides, respectively, when annealed to kidney cortex RNA. These sizes correspond to the identical positions in exon 1-Kc and match very closely with earlier 5′-RACE data, as shown in Fig. 4. The Kc-long probe also gives rise to a cluster of shorter bands that correspond to the protection of only exon 2 by NCX1 transcripts from cerebellum and heart that contain alternative first exons. Primer extension data (Fig. 6) using two different primers for exon 1-Kc confirm exactly the start site determined by 5′-RACE and RNase protection assay (Fig. 4).
Figure 5 B shows an RNase protection experiment with the heart riboprobes. RNA from heart protects a predominant doublet at 350/351 nucleotides with Ht-long and a doublet at 197/198 with Ht-short. Fainter bands ∼11 and 30 nucleotides longer than these are also present. As shown in the sequence in Fig. 3, these data are consistent with our previous 5′-RACE results (24). In those experiments, 8 clones clustered within 1 nucleotide of the start site corresponding to the major protected doublet (residues 2805/6 in Fig. 3), whereas 3 additional clones were 11 nucleotides longer. Primer extension experiments (Fig. 6) also show a strong doublet pattern corresponding to a predominant transcriptional start site at 2805/6. Thus we have defined the A residue at position 2805 in the sequence of Fig. 3 as the transcriptional start point. RNA from kidney and brain showed a predominant doublet, consistent with partial length protection of the Ht probes with the exon 2 region of NCX1 transcripts containing alternate exon 1 sequences (i.e., exon 1-Kc and 1-Br).
In our earlier 5′-RACE experiments (24), we observed clones corresponding to the expression of exon 1-Ht in brain, with transcription initiating ∼77 nucleotides upstream from that observed in heart. Consistent with this observation, a longer exposure of the Ht-long gel of Fig. 5 B (shown only for the cerebrum lane) revealed faint bands of several different protected species, the most abundant of which maps to within one nucleotide of a cluster of four 5′-RACE clones (see Fig. 3). Similar, even fainter, bands were also observed in the lanes with heart RNA but were completely absent from both kidney and cerebellum RNA lanes. This result is consistent with our earlier Northern blotting data using the Ht probe, which showed low levels of a larger NCX1 transcript present only in cerebrum and not in other brain regions or in kidney (24, 26).
An RNase protection experiment using the Br-long probe is shown in Fig.5 C. It is evident from these data that there is a predominant protected fragment of 296 nucleotides, present only in the brain RNA lanes, that matches the transcriptional start site predicted by a cluster of three 5′-RACE clones isolated in our earlier studies (see Fig. 4). The presence of several other protected bands ranging from ∼290 to 340 nucleotides suggests that transcriptional initiation upstream of exon 1-Br may be staggered. Consistent with this result, primer extension experiments (Fig. 6) reveal a similar cluster of several bands with the predominant one corresponding to the major RNase protection product and a transcriptional start site at −215. Moreover, our earlier 5′-RACE experiments produced a collection of fainter brain RNA bands that spanned a range of 50 nucleotides or so (26). Thus all the assays used demonstrate that NCX1 transcripts bearing exon 1-Br initiate at the same set of staggered sites. The existence of multiple transcriptional start sites is not uncommon for GC-rich promoters lacking TATA boxes, such as this one (1). Partially protected products were observed with the Br-long probe in heart and kidney RNA, consistent with expectations for NCX1 transcripts bearing exon 1-Ht and exon 1-Kc. The sequence of exon 1-Br, especially toward the 5′-end, and the genomic flanking sequence is extremely GC-rich (see Fig. 4). Sequences of this nature often form very stable secondary structure, which may explain the high background observed in Fig.5 C as well as our inability to obtain reliable data from the shorter Br probe even though a variety of conditions were utilized.
Although NCX1 transcripts are expressed in all tissues, including kidney and brain, they are found most abundantly in cardiac tissue, in which the Na+/Ca2+exchanger plays a predominant role in intracellular calcium balance. Therefore, the following experiments are aimed at characterizing tissue-specific control mechanisms that lead to preferential expression from the cardiac NCX1 promoter.
NCX1 Cardiac Promoter is Tissue Specific
The 3.6 kb-EcoR I and 421-bp Hind III genomic DNA fragments containing the cardiac transcription start site were used in transfection studies to determine their promoter activity as well as cell specificity. As shown in Fig.7 A, the 421-bp DNA fragment is contained within the 3.6 kb-DNA fragment and comprises 336 bp upstream and 85 bp downstream from the cardiac transcription start site.
Plasmids pGL2-2805 and pGL2-336, containing these respectiveEcoR I and Hind III NCX1 fragments upstream of a luciferase reporter gene, were transfected into primary cardiac myocytes, CHO cells, and COS-7 cells. The plasmid pMLC, containing the cardiac-specific myosin light chain-2 promoter (13), was used as a positive control, whereas the promoterless vector pGL2B was used as a negative control. The plasmid pRSV, containing the strong and promiscuous RSV promoter, was used to define the level of cell-type independent promoter activity.
As shown in Fig. 7 B, the activity of the myosin light chain cardiac-specific promoter was three times and nine times higher in primary myocytes than in CHO or COS-7 cells, respectively. In comparison, the activity of the NCX1 promoter construct pGL2-2805 was 9 times and 18 times higher in primary myocytes than in CHO or COS-7 cells, respectively. The smaller construct, pGL2-336, gave even larger activity ratios of 13 and 31 in primary myocytes compared with CHO and COS-7 cells. These results indicate that the NCX1 DNA fragments containing the cardiac transcription start site are both transcriptionally active as well as cardiac cell specific. Figure 7 B also shows that the activity in primary myocytes is almost three times higher for the −336 promoter (pGL2-336) than for the −2805 promoter (pGL2-2805). This suggests the presence of an upstream suppressor element.
Figure 7 C further verified the cardiac-specific property of the NCX1 cardiac promoter. Reporter activity from primary myocytes transfected with the NCX1 cardiac promoter construct (pGL3Ht-2805) was significantly higher than activity from either the NCX1 brain (pGL3Br-1086) or kidney (pGL3Kd-1663) promoters.
Determining Cardiac NCX1 “Minimum Promoter”
Oligonucleotide primer pairs A2-B1, A3-B1, and A4-B1 were used in separate PCR reactions to truncate the 421-bp NCX1 active promoter into three successively smaller fragments (Fig.8 A), thereby creating the plasmid constructs pGL2-236, pGL2-137, and pGL2-36 that were transfected into primary cardiac myocytes. The truncated constructs pGL2-236 and pGL2-137 exhibited 61 and 35% promoter activity, respectively, compared with the parent pGL2-336 construct. Promoter activity disappeared with pGL2-36 to a level comparable with that of the promoterless vector pGL2B. As shown in Fig.8 B, the same plasmid constructs maintained their cell-specific promoter activity when transfected into CHO cells.
These results show that the information necessary for minimum promoter activity as well as cardiac-specific promoter activity of the NCX1 gene in heart lies within 137 bp upstream of the cardiac transcription start site. These data also suggest that the increase in activity in the successively larger fragments may be due to the presence of upstream,cis-acting enhancer elements.
Identification of Enhancer Elements
To locate enhancer elements, the DNA sequences between the described truncated DNA fragments were placed upstream of a heterologous herpes simplex virus thymidine kinase (TK) minimum promoter in the vector pTK109 (32). In this way, the ability of each of these DNA sequences to independently upregulate the TK promoter could be measured. The most distal fragment from the cardiac transcription start site extended from −336 to −217 (pTK-336-217) and overlapped the next fragment, which extended from −236 to −117 (pTK-236-117), which, in turn, overlapped the most proximal fragment extending from −122 to −17 (pTK-122-17).
Figure 9 Ashows the results of transfection experiments using these constructs expressed as %RSV activity. The enhancerless plasmid pTK109 had 8% RSV activity compared with 63, 18, and 64% activity in the presence of the upstream DNA elements. Thus all fragments significantly upregulated, albeit with varying strengths, the activity of the TK minimum promoter, suggesting the presence of multiple enhancer elements within the 421-bp NCX1 promoter. In a separate experiment, these DNA fragments were also ligated into the pGL2B promoterless vector and were used to transfect primary myocytes. In the absence of a minimum promoter, these DNA fragments had no independent promoter activity (Fig. 9 B). The cardiac specificity of these enhancer elements is being tested.
Structural Analysis of Cardiac NCX1 Promoter
Figure10 Ashows the results of a transcription factor search of the −336 cardiac NCX1 promoter sequence using the MatInspector database (34), which is continually updated from the TRANSFAC database. Although multiple consensus sequences for known transcription factors were found, only those which are potentially functional or cardiac specific (on the basis of established roles in the literature) are underlined. The cardiac NCX1 promoter is TATA-less, an occurrence that is repeatedly observed in several other genes with multiple transcription initiation sites (15, 19). Transcription from TATA-less promoters usually begins at an initiator (Inr) site with the consensus sequence YYANWYY, which flanks the transcription start site (17, 38). (The nucleotide abbreviations used are the standard IUPAC abbreviations for nucleotide sequences.) Thus the DNA sequence TCAATCT (−2 to +5) is the NCX1 cardiac promoter Inr sequence.
Upstream of the cardiac transcription start site (+1), there are two potential activator protein 1 (AP-1) sites (−324 and −241; consensus sequence TGASTMA) that bindfos andjun heterodimers known to be important in cell-specific transcription (11). The promoter of the rat cardiac-specific atrial natriuretic factor (ANF) gene has an AP-1 site that plays a role in induction of ANF expression during ventricular hypertrophy. There is a binding site for nuclear factor Y (NF-Y) that recognizes a CCAAT box at position −275. CCAAT boxes are found in many mammalian promoters and significantly increase the efficiency of transcription (7). The two GATA boxes (−148 and −76) belong to a subfamily of transcription factors, some of which are involved in regulating cardiogenesis (GATA-4/5/6) (23). Of particular interest is the identification of an E box (−177) with the consensus sequence CANNTG, which binds a family of transcription factors with a conserved basic helix-loop-helix structural motif. At least two such transcription factors, eHAND and dHAND, are cardiac specific (9, 39), although their target genes have yet to be defined. A serum-response factor (SRF) binding site at −107 recognizes a serum-response element that contains a CCAAT-like box and specifically regulates the human cardiac α-actin gene (23). Two nuclear factor 1 (NF-1; −214 and −130) binding sites with the TGGC motif are typically involved in cell-specific transcription in liver and other organs (27). There are two putative ETS-1 binding sites at position −212 and −4 (sequence similarity >85%). The ETS transcription factor has been implicated in upregulation of expression of β-enolase gene expression in myoblasts (41) and in developmental regulation of some genes (44).
Sequence Analysis of Kidney and Brain NCX1 Promoters
Figure 10 B shows upstream regions from kidney and brain NCX1 transcription start sites that are comparable in length to the cardiac NCX1 promoter shown in Fig.10 A. The kidney NCX1 promoter has a typical TATA box (−28). Potentially important transcription factor sites are underlined. Upstream from +1, there are two tandem GATA boxes (−196 and −202), an NF-Y binding site (−104), and two CAAT enhancer binding proteinlike (C/EBP-like) sites at −122 and −196 (sequence similarity >90%) with consensus sequence TKNNGYAAK (16). Position −279 has tandem binding sites for adenosine 3′,5′-cyclic monophosphate-response element binding protein (CREB; consensus sequence TGACGYMR), activating transcription factor (ATF), and AP-1. Two additional octamer sites (Oct-1) are located at −67 and −12. Oct-1 is preferentially expressed during the S phase of the cell cycle (11). Although each of these tissue factors may not be specific to the kidney, their unique combinations may lead to kidney-specific transcription regulation.
Unlike the heart and kidney NCX1 transcripts, which are tissue exclusive, the brain NCX1 transcript is ubiquitous, but most abundant, in brain tissue (24). This region is highly GC-rich, containing 80% G+C bases with several specific protein 1 (Sp-1) and AP-2 binding sites. The CpG island in this region of the brain NCX1 promoter represents a site of transcription regulation in which housekeeping genes (as opposed to tissue-specific genes) remain unmethylated, and therefore turned on, during early embryogenesis (10). This promoter also has no TATA box.
Previous data have shown that transcripts from the Na+/Ca2+exchanger NCX1 are expressed in a tissue-specific manner (24), which suggested the existence of alternative promoters. This study has identified the transcription start sites of three NCX1 promoters that express distinct transcripts in heart and kidney and a ubiquitous transcript that is enriched in brain. This situation has been reported previously in other genes such as the homeotic antennapedia gene ofDrosophila melanogaster (22) and the cytoplasmic tyrosine kinase gene p56 (42). Similar to the Na+/H+exchanger, the cystic fibrosis transmembrane regulator, and the Cl−/ exchanger, the NCX1 isoform of the Na+/Ca2+exchanger is present in almost all cell types, is involved in general cellular ionic homeostasis, and may be classified as a “housekeeping gene” (31, 35). In addition, however, very high levels of NCX1 are present in heart and kidney, where the protein appears to play very specialized roles in excitation-contraction coupling and relaxation (heart) and in epithelial calcium reabsorption and systemic calcium homeostasis (kidney). The presence of alternative promoters for NCX1 may allow the gene to maintain a housekeeping role while still permitting a mechanism for cell type-specific regulation.
The different rat NCX1 transcriptional start sites result in three potential first exons spread out over a genomic region of as much as 20 kb, with the exon present in NCX1 transcripts from heart significantly further upstream than those present in kidney or brain. Each first exon is then spliced in a mutually exclusive and tissue-specific fashion to a common exon 2 sequence >10 kb further downstream. Mapping of the human NCX1 gene (21) has identified what is likely to be the equivalent of our rat exon 1-Br, separated by an intron of ∼30 kb from coding exons. We anticipate that extended mapping of the human NCX1 gene further upstream will also reveal exons equivalent to the 1-Kc and 1-Ht found in rat. The sequences of each rat NCX1 exon 1 appear to end at position −32 (cDNA coordinates), and exon 2 appears to begin at position −31. A comparison of these sequences (Figs. 3 and 4), however, reveals that all three exon 1 sequences have three nucleotides in common at their 3′-ends, whereas exons 1-Kc and 1-Br share an additional two residues. As a consequence, the fragments corresponding to protection of only exon 2 regions of NCX1 transcripts with alternate exon 1 sequences (see Fig. 5) are of different length in different tissues. In addition, however, the partial length Kc-long fragment protected by brain cerebrum RNA is two or three nucleotides longer than predicted. Moreover, in almost all cases of partial length protection, a faint doublet, or even triplet, is observed. The identity of the transcripts giving rise to these unexpected bands is not known. Possibly, NCX1 species bearing exon 1 sequences other than 1-Ht, 1-Kc, or 1-Br are expressed at low levels. It is also possible that some heterogeneity exists in the precise choice of splice junction. Because this site lies within the 5′-UTR of the NCX1 transcripts, no loss of protein reading frame ensues from such a lack of fidelity in splicing. Importantly, however, in each tissue examined, NCX1 transcripts bearing the corresponding exon (i.e., exon 1-Ht for heart, exon 1-Kc for kidney, and exon 1-Br for brain) constitute ≥90% of the total.
Both NCX1 mRNA and protein levels respond to various signals (18, 28). Although much research has focused on the mechanism of regulation of NCX1 at the posttranslational level, few studies have directly addressed the regulation of NCX1 expression at the transcriptional level. Characterization of the rat NCX1 cardiac promoter by functional studies and structural analysis has uncovered potential mechanisms to permit tissue-specific protein expression and stimulus-responsive transcription.
We show that a sequence 2805 bp upstream from the cardiac transcription start site of the NCX1 promoter can direct tissue-specific promoter activity in primary neonatal cardiac myocytes (Fig. 7,B andC). Furthermore, a significantly shorter DNA sequence 336 bp upstream from the cardiac transcription start site demonstrates cardiac-specific promoter activity that is almost three times higher than that of the longer DNA fragment. This implies that augmented activity in the proximal promoter region is suppressed by upstream elements. Further analysis has identified a minimum promoter consisting of 137 bp upstream from the cardiac transcription start site that is strongly cardiac cell specific. A recent publication by Barnes et al. (1) has described the structure of the 5′-end of the cat NCX1 gene. In this study, a sequence 2,000 bp upstream from the cardiac transcription start site was capable of tissue-specific transcription with a tissue-specific minimum promoter of 500 bp (−250 to +250).
A search for transcription factor binding sites within the minimum promoter identifies elements for potentially binding the tissue-specific factors NF-1, (27), SRF, and GATA (23), as well as an Inr. Activation and initiation of eukaryotic transcription typically require a TATA box within the first 25–30 bp upstream of the transcription start site (7). However, in TATA-less promoters, such as the cardiac NCX1 promoter, transcription activation and initiation may be executed by a complex interaction between the Inr and upstream,trans-acting proteins, including Sp-1 (38), NF-1, and USF (29). Both an Inr and NF-1 are located within the NCX1 minimum promoter, potentially providing sufficient information for basal transcription. Consistent with this concept, the DNA sequence −36 to +85 is unable to support basal promoter activity because binding sites for suchtrans-acting elements are absent.
Upstream of the minimum promoter, there are binding sites for additional tissue-specific transcription factors that could allow for diversity in the level of NCX1 expression. Established interactions between bHLH transcription factors such as myoD, myogenin, and MRF-4 and their specific skeletal muscle genes have been recognized (14). However, there are no data to define the target genes for cardiac-specific bHLH transcription factors. E boxes at position −177 and three additional upstream sites in the cardiac NCX1 promoter may furnish targets for novel cardiac-specific bHLH proteins, such as eHAND or dHAND. Furthermore, the transcription factor GATA-4 has been shown to restrict expression of the α-myosin heavy chain gene (30) to cardiac myocytes. The two GATA sites within the NCX1 cardiac promoter may function in a similar way to restrict expression of cardiac NCX1.
Three enhancer sequences are recognized within the 421-bp NCX1 cardiac promoter that span positions −336 to −217, −236 to −117, and −137 to −17. Within the enhancers there are NF-Y, SRF, and NF-1 elements that are capable of enhancer activity (7,11). AP-1 sites present within the proximal NCX1 promoter provide sites that may mediate c-fos-induced responses to hypertrophy (28) and stretch (18).
Analysis of the structural organization of the kidney NCX1 promoter reveals several features consistent with its tissue-specific pattern of expression. Unlike the cardiac NCX1 promoter, the kidney NCX1 promoter has a conventional TATA box and CAAT enhancer sites. ETS binding sites in both promoters are consistent with reports of developmental regulation of NCX1 in the heart (4) and in the kidney (35). Consensus sequences for C/EBP, CREB, ATF, AP-1, and Oct-1 provide potential sites by which tissue-specific expression and regulation of NCX1 transcripts could be achieved.
Unlike the cardiac and kidney NCX1 promoters, the brain NCX1 promoter is structured differently. The brain NCX1 promoter has features that are most consistent with a potential housekeeping role of the protein, because there are potential binding sites for ubiquitous transcription factors such as Sp-1 and nuclear factor-κB. Three binding sites for AP-2 imply that, in addition to its housekeeping role, the brain NCX1 promoter could also exhibit cell-specific properties similar to the nerve growth factor gene (37). This arrangement of alternative promoters with different structural characteristics could account for the ubiquitous expression of NCX1 in all tissues while still affording regulation of NCX1 transcription in a tissue-specific manner from additional cardiac and kidney promoters.
A comparison between the promoters of the rat and cat NCX1 gene reveals a >90% sequence identity from −250 to −50 upstream of the cardiac transcriptional start site as well as significant similarity between cat exon H1 and rat exon 1-Ht. Analysis of this region, for both rat and cat, reveals elements known for cardiac cell-specific expression including GATA factors and E-box consensus sites. The cardiac promoter from both animals is TATA-less. Enhancer elements have been identified within the rat, but not the cat, cardiac NCX1 promoter. There is also significant similarity between the cat kidney K1 exon and its upstream sequences, and our rat genomic sequence. The region of identity, however, lies downstream from the rat kidney 5′-end exon 1-Kc. This may explain the lack of tissue specificity observed for the cat K1 exon compared with the rat exon 1-Kc. Reverse transcription-coupled PCR experiments have confirmed the existence in rat kidney of NCX1 transcripts containing an exon 1 sequence essentially identical to cat K1 (data not shown). Our RNase protection experiments with long probes that span the boundary between exon 1 and exon 2 (Fig. 5) provide an upper limit on the fraction of transcripts in heart and kidney containing this species. In both cases, full-length protection corresponds to >90% of the signal, meaning that transcripts bearing a K1 equivalent or any other exon 1 sequence are <10% of total NCX1 transcripts. Also, the cat Br1 exon has significant homology with the rat exon 1-Br, and the promoter region in both animal species is GC rich, consistent with the ubiquitous pattern of tissue expression (24).
In summary, we have characterized the promoter of the NCX1 gene with emphasis on the cardiac NCX1 promoter. The study lays the groundwork for a better understanding of the mechanism of NCX1 gene regulation at the transcriptional level. This understanding could lead to novel approaches toward correcting genetic or environmental perturbations in NCX1 gene regulation.
This work was supported by grants from the Robert Wood Johnson Foundation (to S. B. Nicholas), the American Heart Association, and the Medical Research Council of Canada (to J. Lytton) and by National Heart, Lung, and Blood Institute Grant HL-48509 (to K. D. Philipson). J. Lytton is an Established Investigator of the American Heart Association and an Alberta Heritage Foundation for Medical Research Scholar.
Address for reprint requests: S. B. Nicholas, Cardiovascular Research Laboratories, MRL Bldg. 3–645, UCLA School of Medicine, 675 Circle Dr. S., Los Angeles, CA 90095-1760.
The work was started while J. Lytton and S.-L. Lee were in the Renal Division of the Departments of Medicine at the Brigham and Women’s Hospital and Harvard Medical School, respectively.
- Copyright © 1998 the American Physiological Society