Elastin gene transcription is cell type specific and developmentally regulated, but the promoter often exhibits relatively weak activity in transient transfections of cells that express elastin at high levels. To search for positive-acting regulatory sequences, we isolated genomic clones spanning the mouse elastin gene and extensive 5′- and 3′-flanking regions. Restriction fragments of potential regulatory regions were ligated 5′ or 3′ relative to the active promoter to test for enhancer activity in transient transfections of fetal rat lung fibroblasts, which express elastin at high levels, and distal lung epithelial cells, which do not express detectable elastin. Fragments of intron 1 did not exhibit significant enhancer activity. Inclusion of the 84-bp exon 1 and adjacent 5′-untranslated region increased activity of the elastin promoter approximately sixfold compared with parental constructs. Transfections with constructs of varying promoter length showed that as little as 40 bp of the 5′ end of exon 1 confers enhanced activity in elastin-expressing rat lung fibroblasts, but these constructs had variable activity in lung epithelial cell lines. This region, localized between the transcription start site and extending into exon 1, binds Sp1 in nuclear extracts from elastin-expressing cells. These studies indicate a role for the 5′ end of the first exon of the elastin gene in regulating strong transcriptional activity in elastogenic cells.
elastin is an abundant protein in connective tissues, comprising as much as 35% of the dry weight of the adult human thoracic aorta. Expression of the elastin (ELN) gene coincides with the formation of tissues such as blood vessels and the lung during development and peaks during morphological maturation of these tissues. Lung ELN gene expression is limited to airway and vascular smooth muscle cells, vascular endothelial cells, pleural mesothelial cells, and alveolar myofibroblasts and increases >10-fold between early stages of lung development and the postnatal alveolarization period (19, 30). Elastin expression is undetectable in lung epithelial cells. Analyses of the elastin 5′-flanking region from −6 kb to the translation start site have defined a number of responsive elements, both positive and negative, but none that confers high, cell type-specific activity in transient transfections of cultured elastin-expressing cell types including neonatal lung myofibroblasts, fetal aortic smooth muscle cells, and fetal auricular chondrocytes. In transient transfection studies using cells that express both elastin and type I collagen at relatively similar levels, for example, ELN promoter constructs exhibit weak activity compared with type I collagen promoter constructs (7). Moreover, early studies using ELN promoter constructs showed approximately equivalent activity in elastogenic and nonelastogenic cells in transient transfections. Fazio and coworkers (7) found that a construct extending from −475 to −14 exhibited nearly as high activity in HT-1080 cells, which do not express elastin, as in rat aorta smooth muscle cells, which express elastin at high levels. Further functional analysis of the ELN promoter demonstrated higher promoter activity in NIH 3T3 and HeLa cells, which do not express elastin, than in rat aorta smooth muscle cells (12). More recent studies have not sought to define potential tissue-specific elements in the ELN promoter. These findings led us to the hypothesis that important positive-acting regulatory elements are located outside the region spanning from −6000 to −1 of the ELN gene.
Nonetheless, many important regulatory sites in the ELN promoter have been defined. Some positive-acting responsive elements of the ELN promoter act through loss of repression on stimulation. For example, the ELN promoter regulatory elements that respond to insulin-like growth factor (IGF) in aortic smooth muscle cells involve disruption of an IGF-I-sensitive repressor complex (10, 28, 31). Other positive-acting ELN promoter elements directly bind transactivating factors, such as the nuclear factor-1 binding site at approximately −400 of the ELN gene (5). Functional elements that repress elastin expression on exposure to basic fibroblast growth factor (bFGF) (2, 3, 27), interleukin (IL)-1β (14, 15), and tumor necrosis factor-α (TNF-α) (11) have been defined. Studies in transgenic mice reported tissue-specific activity of a transgene containing 5.2 kb of 5′-flanking sequence, although comparisons between expression patterns of the transgene and the endogenous gene were not made (9, 13, 17, 26).
To screen for regulatory elements of the ELN gene that would confer high transcriptional activity in elastin-expressing lung myofibroblasts, we isolated mouse genomic clones spanning the ELN gene and extensive 5′- and 3′-flanking regions. The region from −718 to −1 relative to the start ATG codon, was inserted 5′ of a luciferase expression cassette to make a parental expression construct. Reasoning that an enhancer may be found in the large first intron, as has been found in fibrillar collagen promoters, we tested restriction fragments of the ELN gene intron 1 for enhancer activity. These studies did not identify an enhancer in intron 1. However, a positive-acting element was found in exon 1 that markedly increases promoter activity in elastin-expressing lung fibroblasts and gives variable results in epithelial cells that do not express elastin. This element specifically binds Sp1 but not Sp3.
MATERIALS AND METHODS
Human ELN promoter-chloramphenicol acetyltransferase constructs and transient transfections.
A series of deletion constructs containing regions of the human ELN gene promoter linked to the chloramphenicol acetyltransferase (CAT) gene were a gift from Dr. J. Uitto (Thomas Jefferson University, Philadelphia, PA). These constructs included pEP52CAT (−2260 to −14), pEP35CAT (−1553 to −14), pEP10CAT (−986 to −14), pEP27CAT (−676 to −14), and pCAT, containing the SV40 promoter and enhancer (12). These constructs were cotransfected into RFL-6 cells with a LacZ construct under control of the rous sarcoma virus promoter by calcium phosphate precipitation, followed by preparation of a cell lysate after 24 h and assay for CAT activity as previously described (23).
Cloning and characterizing mouse ELN gene.
Nylon filters containing a bacterial artificial chromosome (BAC) library representing the mouse genome were purchased from Genome Systems (St. Louis, MO) and screened with a 32P-radiolabeled rat tropoelastin cDNA clone encompassing exons 8–18 (22). Several positive clones were identified, purchased, and subsequently analyzed with exon-specific oligonucleotide probes in combination with restriction analysis, Southern blotting, subcloning, and dye-termination automated sequence analysis.
Construction of reporter vectors.
A series of pGL3 luciferase expression vectors from Promega (Madison, WI) were modified for use as reporters of ELN gene promoter activity. pGL3, utilizing firefly luciferase as a reporter cassette in combination with an SV40 splice and polyadenylation cassette, was digested with the restriction endonucleases KpnI and XhoI to allow subcloning of 5′-flanking sequences derived from the ELN gene. An 18-kb genomic subclone, 18K9B13, in pBSKSII (Stratagene), spanning from −2.2 kb through exon 4 served as template for generating inserts by PCR amplification with high-fidelity Pfu polymerase (Stratagene). The parental construct −718,−13 was constructed by PCR amplification using a 5′ primer containing the native KpnI site at −718 and a 3′ or reverse primer with an engineered XhoI restriction site at the 3′ end, subsequent restriction digestion, and ligation into pGL3, which contains the luciferase coding sequence, an intron, and a polyadenylation sequence. Because multiple transcription start sites are utilized at relatively equivalent frequency in the ELN gene, the A nucleotide of the ATG translation start codon was designated as +1 in descriptions of all constructs. Additional constructs were made with alternative 5′ primers and 3′ primers for PCR amplification. All constructs were sequenced to verify fidelity of amplification.
Cell culture and transient transfections with luciferase reporter constructs.
The rat fetal lung fibroblast cell line RFL-6 (American Type Culture Collection, Gaithersburg, MD) was used for transient transfections. RFL-6 cells were cultured in Dulbecco's modified Eagle's medium (DMEM) supplemented with nonessential amino acids, l-glutamine, penicillin-streptomycin, and 10% calf serum. Under these conditions, RFL-6 cells express high levels of tropoelastin mRNA. For transfection, RFL-6 cells were plated in six-well cluster dishes at 400,000/well 24 h before transfection and washed twice with serum-free, antibiotic-free medium. Cultures were incubated with a liposome-DNA complex containing 10 μl of Lipofectamine (Invitrogen, Carlsbad, CA) and 1 μg of supercoiled plasmid (0.5 μg test plasmid and 0.5 μg Renilla cotransfection plasmid). After 4 h an equal volume of 20% calf serum-containing medium was added, and cells were incubated overnight. The following day cells were washed and then fed DMEM supplemented with 10% calf serum.
MLE-15 cells, a mouse lung epithelial cell line, were a generous gift of Dr. Jeffrey Whitsett (University of Cincinnati, Cincinnati, OH) and were cultured in hydrocortisone, insulin, transferrin, estradiol, selenium (HITES) medium as described previously (20). For transfection, 6 × 104 cells/well were plated in 24-well plates 24 h in advance and then were incubated with a total of 2 μg of DNA and 12 μl of Superfect reagent (Qiagen) in serum-free medium for 5 h before being returned to standard HITES medium. Cells were harvested for lysate preparation and luciferase activity assay after 24 h. For all test constructs multiple plasmid preps were tested, and all transfection experiments were performed in triplicate.
Cell cultures were harvested 48 h after transfection with Promega's Dual-Luciferase reagents and protocols. Cell lysates were assayed first for firefly luciferase activity and then Renilla luciferase activity. Activity of test plasmids was expressed as a ratio of firefly luciferase activity to Renilla luciferase activity.
Nuclear transcription assay.
Transcriptional activity of the ELN gene was assessed using the “run-on” transcription assay technique as previously described in detail (23), to determine whether transcriptional activity of the ELN gene is high in lung fibroblasts. RFL-6 cells (2.5 × 106) were cultured at confluence and then harvested for isolation of nuclei. The nuclear pellet was incubated for 30 min at 30°C in transcription buffer containing 32P-labeled GTP and unlabeled ATP, CTP, and TTP and then treated with guanidine-phenol to isolate nascent transcripts. After ethanol precipitation, 1 × 107 cpm of radiolabeled RNA was hybridized to nylon filters containing 5 μg of slot-blotted target plasmids corresponding to rat tropoelastin, bovine tropoelastin, actin, an Alu repeat element derived from the human β-globin cluster, and parental plasmid vector. Filters were hybridized overnight and then were washed stringently as previously described (23).
Electrophoretic mobility shift assay.
Electrophoretic mobility shift assay (EMSA, gel shift assay) was performed to learn whether nuclear proteins bind positive-acting elements in exon 1 of the ELN gene. Nuclear extracts were prepared from cell cultures by the approach described by Dignam et al. (6). Cells (108) harvested at confluence were subjected to Dounce homogenization in a sucrose buffer with detergent to disrupt membranes and then washed and centrifuged to pellet nuclei. Nuclear proteins were then extracted at 4°C in 350 mM KCl, the nuclei were lysed by homogenization, and a lysate containing nuclear proteins was prepared by ultracentrifugation as we described in detail previously (25). After overnight dialysis at 4°C in a buffer containing 100 mM KCl, the supernatant was divided into aliquots and frozen by immersion in liquid N2.
Binding reactions were performed at room temperature with 3–10 mg of nuclear extracts, 1 μg of nonspecific competitor DNA (dI:dC), and varying concentrations of unlabeled specific competitor or irrelevant competitor. Then 50,000 cpm of end-labeled, double-stranded oligonucleotide probe was added and incubated an additional 15 min. Mixtures were electrophoresed through nondenaturing polyacrylamide gels, which were dried and processed for autoradiography. For “supershift assays,” an antibody specific for Sp1 or Sp3 (Santa Cruz Pharmaceuticals, Santa Cruz, CA) was added either before addition of radiolabeled probe to test for inhibition of binding or after probe binding and incubated overnight at 4°C before electrophoresis.
Transcription of ELN gene in RFL-6 cells is readily assessed, but ELN promoter-reporter constructs have low activity in transient transfections.
Transcriptional activity of the ELN gene was assessed by nuclear transcription assay in RFL-6 cells, which express tropoelastin mRNA at high levels (21). Radiolabeled nascent transcripts from nuclei isolated from confluent RFL-6 cells were hybridized to slotted DNAs on nitrocellulose filters including parental pBluescript vector, an Alu I repeat element, β-actin, bovine ELN cDNA, and rat ELN cDNAs. Transcriptional activity for the rat ELN gene was readily assessed in RFL-6 cells (Fig. 1, left). Hybridization to the bovine ELN cDNA internal control was much weaker, likely because of the relatively low nucleotide identity (∼78%) between rat and bovine ELN mRNAs. To determine which ELN promoter elements confer high activity to the ELN gene in rat lung fibroblasts, RFL-6 cells were transiently transfected with a series of human ELN promoter-CAT constructs. Activity of pCAT, driven by the SV40 promoter and enhancer, was high as anticipated from a strong viral promoter, but all the human ELN promoter-CAT deletion constructs tested exhibited much weaker activity (Fig. 1, right). These data led to the hypothesis that positive-acting regulatory elements may reside outside this region of the human ELN gene and may be important for active ELN gene transcription in lung fibroblasts. We therefore isolated mouse genomic clones containing the endogenous ELN locus for further study.
Mouse ELN genomic clones.
Two genomic BAC clones positive by Southern blotting with radiolabeled mouse ELN exon-specific nucleotides were selected for further study. Clone 18 was found to be ∼85 kb in size by pulse-field electrophoresis and to contain exons 1–6. On the basis of the combined size of exon 1, intron 1, and exon 2 (∼12 kb), clone 18 extends ∼70 kb 5′ of exon 1. Clone 4 was found to be ∼115 kb in size and to contain exons 2–36, and it therefore extends into the 3′-flanking region at least 50 kb (Fig. 2A).
Comparison of mouse and human ELN gene 5′-flanking sequence.
A BamHI subclone 18 kb in size and extending from ∼3 kb 5′ of exon 1 through exon 4 was identified and characterized by restriction and sequence analysis. Database searches confirmed that the sequence is derived from mouse chromosome 5 and encodes the tropoelastin protein. A KpnI restriction fragment extending to −718 relative to the translational start codon was compared with the human ELN gene 5′-flanking region and exon 1 (Fig. 2B). The sequence identity over this region was 78.7% despite several insertions in the mouse ELN gene 5′-flanking region compared with the human gene. The GC content of this region of the gene is 62.4%, similar to the ELN gene 5′-flanking sequence and ELN coding sequence in other species. A number of regulatory sites identified in the human ELN promoter are found in the mouse ELN gene 5′-flanking region (data not shown).
Analysis of intron 1 fragments for enhancer-like activity.
Given the large size of intron 1 (>10 kb) in the ELN gene from several species, we hypothesized that it may contain regulatory elements that enhance promoter activity. Luciferase reporter constructs were made that contained BglII restriction fragments of intron 1 inserted 5′ of an ELN promoter fragment extending from −400 to −1 relative to the ATG start codon of the ELN gene linked to a luciferase reporter. These were transfected into RFL-6 cells, and luciferase activity was determined at 48 h relative to a Renilla luciferase vector cotransfected as a control. Some derived constructs exhibited inhibited activity compared with the parental construct, indicating that repressor-like elements reside in intron 1. However, no constructs containing restriction fragments of intron 1 exhibited activity that was significantly greater than the parental vector (see supplemental figure).1 This led to the conclusion that intron 1 of the mouse ELN gene does not contain strong enhancer-like elements.
Analysis of ELN exon 1 for transcriptional control elements.
Scanning the ELN genomic sequence for transcription factor-binding consensus sequences revealed potential regulatory sites within exon 1. To test whether exon 1, which encodes the signal peptide of tropoelastin, affects ELN gene promoter activity, additional constructs extending 5′ to a KpnI site at −718 were made and tested in transient transfections in elastin-expressing RFL-6 cells and in the MLE-15 mouse lung epithelial cell line, which does not express detectable elastin. The activity of the construct ELN-718,−13 was used as a baseline for comparing the activities of other promoter-luciferase constructs (Fig. 3A). Initially, three different constructs were made. The construct ELN-718,+84 ATG contains the 5′-untranslated region (UTR) of the ELN gene as well as the tropoelastin translational start site and signal peptide linked to the luciferase reporter cassette. The activity of ELN-718,+84 ATG was not different from that of ELN-718,−13. In this construct, retaining the ATG start codon of tropoelastin linked to the luciferase 5′-UTR and coding sequence in −718,+84 ATG could result in a mRNA difficult to translate, a fusion protein that is secreted, or altered codons derived from the luciferase 5′-UTR that result in loss of luciferase activity. No luciferase activity above background was detected in cell media with this or other constructs (data not shown), suggesting that no functional luciferase protein is secreted.
To circumvent potential pitfalls resulting from fusion of the 5′-UTRs and start codons derived from both the ELN and luciferase genes, the ELN gene ATG start codon was mutated to either TTG or AAG in the context of constructs extending 5′ to the KpnI site at −718 and 3′ through exon 1 to +84. Mutating the ATG codon to either TTG or AAG and including exon 1 resulted in a sixfold increase in promoter activity in transient transfections of fetal rat lung fibroblasts. This indicates that sequences 3′ of −13 and including exon 1 increase the activity of the ELN promoter or may be part of the functional ELN promoter.
Next we sought to determine whether the positive-acting element in exon 1 is cell type specific and to delimit this element. The parental −718,−13 construct exhibited similar low activity in transient transfections of RFL-6 cells and MLE-15 cells, a murine lung epithelial cell line (Fig. 3B). The construct −718,+84 TTG exhibited a threefold increase in MLE-15 cells that did not achieve statistical significance. Together these data support the hypothesis that a positive-acting regulatory site resides in exon 1, which may have greater activity in lung cells expressing elastin. Because the element has some positive activity in MLE-15 cells, it is not strictly cell type specific. Importantly, both the AAG and TTG constructs exhibit similar, high activities in elastin-expressing cells. Therefore, it is unlikely that mutating the ATG start codon resulted in creation of a new and spurious positive-acting sequence. The deletion construct −718,+24 TTG, which contains only the 5′ end of exon 1, retained high promoter activity in RFL-6 cells. Moving exon 1 to the 5′ end of the parental construct caused a loss of activity compared with retaining exon 1 in its native position in transient transfections of RFL-6 cells. These data support the premise that the region surrounding the start codon of the ELN gene, in context, contains regulatory sequences that are important for ELN promoter activity, but this element is not a classic enhancer. This region was named transcription element in exon 1 (T-Ex1).
Binding of nuclear proteins to T-Ex1.
Nuclear extracts derived from RFL-6 cells were incubated with a series of overlapping double-stranded probes spanning −48 to +27 of the mouse ELN gene relative to the translation start site and tested in EMSA for specific binding. Probes upstream of the T-Ex1 element (−48 to −25, −37 to −13) derived from the parental ELN-luciferase construct (−718,−13) exhibited complex patterns of binding of nuclear proteins consistent with a transcriptional start site region (data not shown). To focus on the T-Ex1 region, a probe spanning −12,+12 centered on the translational start site was chosen for further study, with overlapping and nonoverlapping probes used for competition experiments (Fig. 4). Two band shift complexes were formed with nuclear extracts from the RFL-6 cells. Complex 1, of lower mobility, was diffuse, and complex 2 was comprised of several bands of higher mobility. Addition of 50× unlabeled identical probe (−12,+12) as a competitor for binding resulted in nearly complete ablation of binding. Competition with an upstream, nonoverlapping probe (−48 to −25) had little effect on binding, whereas competition with an unlabeled adjacent upstream probe (−37 to −13) slightly diminished both complexes. This may indicate that an upstream region binds the same nuclear factors as the exon 1 element. Competition with unlabeled excess −25,−1, which overlaps the 5′ half of −12,+12, caused loss of complex 1 and a band within complex 2. Competition with 50× unlabeled −13,+14, which completely encompasses the labeled −12,+12 probe, abolished much of the binding evident in both complexes. In repeated experiments, nuclear extracts isolated at different times from different cell cultures varied somewhat in the specific patterns of band shifts in mobility shift assays but always resulted in two broad band shift complexes. Complex 1 was more prominent in some cases than in others. That this region of the ELN gene binds nuclear proteins from different elastin-expressing cells specifically supports the hypothesis that it is involved in transcriptional regulation of the ELN gene.
Mutating ATG start codon does not affect binding activity.
As shown above (Fig. 3), including the region from −13 to +24 resulted in a five- to sixfold increase in ELN promoter activity when the ATG start codon was altered to TTG or AAG, to prevent complications from the presence of two start codons in the ELN promoter-luciferase constructs, one start codon derived from ELN and the other start codon from Luc (Fig. 3). To test for the possibility that these mutations altered binding of nuclear proteins to the T-Ex1 site, we tested a probe with the ATG mutated to TTG, as was used in the transient transfection studies (Fig. 4C). In this case, nuclear extracts from primary neonatal rat lung fibroblasts bound strongly to both the T-Ex1 wild “ATG” and the mutant “TTG” probe, with two prominent band shift complexes of comparable intensity with both probes.
Delineation of core binding element.
To further delimit the nuclear protein binding sites within T-Ex1, a series of double-stranded oligonucleotide probes with 5′ or 3′ deletions were generated and used in EMSA studies with RFL-6 nuclear extracts (Fig. 5A). The parental probe exhibited strong specific binding activity, resulting in two band shifts (Fig. 5B). In these experiments, the upper band shift was effectively competed by excess unlabeled competitor and the lower, fast-migrating band shift was partially competed. Deleting 4 bp from the 5′ end (probe −8,+12) resulted in a similar pattern of binding, but removing 6 bp (probe −6,+12) resulted in diminution of the upper band shift and more effective competition of binding for the lower band with excess unlabeled competitor. This suggests that a core sequence containing nuclear factor binding activity is contained within the region of −8,+12 relative to the translational start site of the ELN gene.
3′ Deletions of the T-Ex1 probe were performed next (Fig. 5C). Changes in band shift patterns were noted with nuclear extracts isolated at different times, but binding of RFL-6 nuclear extracts to the parental probe always resulted in two band shifts. Removing as little as 2 bp from the 3′ end of T-Ex1 (probe −12,+10) diminished the intensity of the faster-mobility band shift. The upper band shift persisted when 6 bp was removed from the 3′ end (probe −12,+6) and then diminished markedly with further deletions. Removal of 12 bp resulted in nonspecific binding patterns that did not compete with excess unlabeled probe. This loss of specificity may be due in part to the short length of the probe. Together with the 5′ deletions, these studies indicate that a core binding element within T-Ex1 is located between −8 and +8 relative to the ATG and results in the upper band shift detected in these studies. The lower band shift appears to be somewhat less specific and to require the sequences from −6 to +10 relative to the ATG translational start site.
To begin to determine whether specific nucleotides are critical for binding, mutant double-stranded probes were used in competition experiments in EMSA studies employing radiolabeled T-Ex1 probe for binding (Fig. 6A). Competition with 50× unlabeled wild-type probe abolished the upper band shift and diminished the lower band shift as seen before (Fig. 6B). Unlabeled competitor with mutations in bases −8 and −9 competed weakly for binding with the wild-type probe. Mutating bases −6 and −7 resulted in virtual loss of competitive activity, indicating that these C-C bases are critical for the binding of the wild-type T-Ex1 probe. These data reinforce the finding that removing 6 bases from the 5′ end diminishes binding activity. Mutating the C nucleotide alone in the unlabeled competitor at base −5 or −1 resulted in partial diminution of competitive activity. A probe derived from outside the region did not compete for binding and resulted in prominent band shifts.
The region spanning −12 to +12 relative to the translational start codon is highly conserved between the human and murine ELN genes, differing between the two species only at −12 and −10, which our deletion studies indicate are not critical for binding. A probe spanning −12 to +12 of the human ELN gene was used in EMSA studies with RFL-6 nuclear extracts and exhibited an extended band shift that was specifically competed with excess mouse or human T-Ex1 probe (Fig. 6C). This indicates that the specific binding activity of the T-Ex1 region is similar between species. Similar results were obtained with nuclear extracts from fetal bovine chondrocytes and human dermal fibroblasts (data not shown). To determine whether similar binding activity is found in cells not expressing elastin, nuclear extracts from MLE-15 cells were incubated with the T-Ex1 probe and showed binding activity (Fig. 6D). This suggested that a widely expressed transcription factor(s) may bind this region of the ELN gene.
Binding of Sp1 to T-Ex1.
Recognizing that the GC-rich T-Ex1 site may bind members of the Sp1 family of transcription factors, we performed supershift EMSA assays using antibodies to Sp1 and Sp3. The antibodies were either preincubated with the nuclear extracts before addition of the radiolabeled probe or added after binding of probe to nuclear extract (Fig. 7). As in previous studies, incubating the T-Ex1 probe with RFL-6 nuclear extracts resulted in two band shifts, with the upper band completely competed by 50× unlabeled probe and the lower diminished by competitor. Preincubating the nuclear extract with antibody to Sp1 resulted in a near-identical pattern as addition of 50× cold competitor, indicating that the antibody prevented formation of the complex represented by the upper band shift and diminished formation of the complex represented by the lower band shift. Incubation with the anti-Sp1 antibody after complex formation occurred did not affect binding. As this antibody is directed against a peptide contained within the DNA binding site of Sp1, it does not typically result in a supershift or appearance of a ternary complex containing the probe, Sp1, and the antibody. An antibody directed against Sp3 had no effect on gel shift patterns whether incubated with the extract before or after addition of the probe. This likely means that there is little Sp3 present in the nuclear extracts used, or that the amount of Sp1 present is sufficient to shift most of the probe. Alternatively, there may be different affinities for Sp1 and Sp3 binding to this sequence.
In many in vivo and in vitro models elastin expression is regulated primarily by abundance of the tropoelastin mRNA, but steady-state mRNA levels for tropoelastin are regulated at both the transcriptional and posttranscriptional levels (21, 23). The induction of elastin expression in development appears to be transcriptionally mediated, and downregulation is often exercised by increasing the turnover of the tropoelastin transcript (24, 30). Some factors such as TGF-β, which is likely involved in the onset of elastin expression in development, may affect both tropoelastin transcription and mRNA stability (13, 16, 32). The present study was designed to investigate what ELN promoter elements are used to confer high transcriptional activity to cells expressing elastin at high levels.
The overall size and structure of the ELN gene is similar among mammals, comprised of 34–36 relatively small exons encoding alternating hydrophobic and cross-linking domains over ∼40 kb. In contrast, in fibrillar collagen genes, which commonly retain 90% or greater identity in coding sequences between mammalian species, the coding sequence of the murine tropoelastin mRNA is 70% identical to the human tropoelastin mRNA. The high level of identity found between the human and mouse ELN gene 5′-flanking region (78% in the region studies) suggests that this portion of the gene is under functional constraints that limit divergence. The regulatory elements responsive to IGF, cell cycle, bFGF, TNF-α, and IL-1β found in the human ELN gene promoter are also present in the murine ELN gene, and the T-Ex1 motif described here is also present in exon 1 of other vertebrate ELN genes. Previously, a repressor element was found to reside ∼450 bp downstream of the transcription start region (18). Our survey of this region, seeking positive regulatory elements, found none.
These studies have identified a positive transcriptional element in exon 1 of the ELN gene. Although the T-Ex1 element enhances transcriptional activity of the ELN promoter, it is not a classic position- and orientation-independent enhancer such as SV40, and may be viewed as part of the ELN promoter necessary for high activity. The location of T-Ex1 within the coding region of exon 1 posed technical difficulties in engineering promoter-reporter constructs and necessitated mutagenesis of the ATG codon within the responsive element in constructs for transfection. However, −718,+85 TTG and −718,+85 AAG both exhibited similar increased activity compared with constructs lacking exon 1 sequences. This argues that these point mutations per se did not create spurious binding sites for positive-acting factors. Furthermore, gel shift assays with mutated probes identified nucleotides flanking the ATG as critical for binding Sp1.
Sp1, a member of the Kruppel-like family of transcription factors, is sometimes viewed as a “constitutive” or “universal” transcription factor, but in fact its activity varies as much as 100-fold between cell types (29) and its binding and functional activity often involve cooperation or antagonism with other transcription machinery elements. Thus Sp1 contributes to cell type- or developmental stage-specific expression of many genes. Moreover, Sp1 is implicated in the IGF-1 stimulation of elastin expression (4, 10), where it and Sp3 compete with other factors for binding to a retinoblastoma control element in an IGF-1-dependent manner. The binding of Sp1 but not Sp3 in the supershift studies may indicate that there is preferential binding of Sp1 to the exon 1 element based on the sequence present. Alternatively, it is possible that Sp3 is present in lower concentration and the Sp1 in the extracts comprises most of the binding activity. In that case, neutralizing the Sp3 in the extracts from binding by using a blocking antibody would not diminish the amount of probe shifted.
It was noted previously that the ELN gene is highly GC rich, lacks a strong TATA element, and utilizes multiple transcription start sites (1). These are characteristics of a promoter that may not tightly bind the transcription machinery at one specific site to initiate transcription. The 5′-UTR of the ELN gene is short, placing the coding sequence of exon 1 and the T-Ex1 element near the TATA consensus site. We hypothesize that Sp1 binding within the nearby T-Ex1 element stabilizes a transcription complex and thus enhances transcription of the ELN gene. This mechanism is proposed for an element within the Col1A1 gene that is also downstream of the transcription start site, binds Sp1, and enhances transcription in a cell type-specific manner (8). Following this hypothesis, the cell type-specific enhancement of transcriptional activity conferred by the T-Ex1 element may require the combination of Sp1 binding to the T-Ex1 element downstream of the transcription start site region and interacting transcription factors binding at the transcription start sites.
This work was supported by National Heart, Lung, and Blood Institute Grant R01-HL-54049.
We gratefully acknowledge the assistance of Karen Schwander in the Division of Biostatistics at Washington University School of Medicine.
↵1 The online version of this article contains supplemental data.
The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
- Copyright © 2006 the American Physiological Society