Main

The amiloride-sensitive epithelial Na+ channel, ENaC, plays a fundamental role in normal pulmonary function. At birth, its participation in the Na+ transport activity facilitates the critical transition of the lung from a liquid-filled state to an air-conducting system. Mice deficient in the α subunit of ENaC die shortly after birth from their inability to actively transport salt and water from their fluid-filled lungs(1). In adult life, epithelial Na+ transport participates in control of the quantity and composition of the respiratory tract fluid and airspace fluid(2). The same ENaC is also located in the epithelia of the distal nephron, distal colon, and ducts of several exocrine glands (e.g. salivary and sweat glands)(3,4), which contribute to electrolyte and water homeostasis. Consistent with their physiologic functions, mutations in ENaC subunits have been discovered to cause Liddle syndrome, a monogenic disease distinguished by severe and early onset of salt-sensitive hypertension(5,6), and pseudohypoaldosteronism type I (PHA-I), an inherited disease marked by severe neonatal salt-wasting, hyperkalemia, metabolic acidosis, and unresponsiveness to hormonal treatment(7,8). In cystic fibrosis, an inherited disease caused by mutations in the gene encoding the CFTR protein, airway epithelia manifest increased activity of amiloride-sensitive Na+ transport(9,10). CFTR has been shown, when co-expressed, to reduce amiloride-sensitive Na+ transport(11). It has therefore been postulated that loss of CFTR results in increased activity of the Na+ channel, which contributes to the altered airway fluid and mucociliary clearance in this disease.

A functional expression approach has been used to clone three cDNA molecules each from the rat (α-, β-, and γ-rENaC) and human (α-, β-, and γ-hENaC) that encode homologous Na+ channel subunits(12–16). Expression of human or rat ENaC α subunit alone in Xenopus oocytes is sufficient to produce small amounts of amiloride-sensitive, Na+ -selective currents, probably by using low levels of endogenous oocyte β and γ subunits, whereas β- or γ-ENaC expressed by themselves failed to confer channel activity. Co-expression of all three subunits resulted in substantial increases of the channel activity that had characteristics very similar to those of the native channel(13,15,17).

Expression of the α-, β-, and γ-rENaC subunits are differentially regulated and also have different responses to corticosteroid treatments. The α subunit of rat ENaC is strongly expressed beginning in the late canalicular and saccular stages of fetal lung development, whereas β-, and γ-rENaC expression only start around birth(18). In both the lung and kidney, glucocorticoid hormone treatment in vivo increases the level of αrENaC mRNA whereas expression of β-, and γ-rENaC subunits is unaffected by corticosteroid administration(18–20). Very little, however, is known about regulation of expression of the human ENaC genes. From human fetal lung explant studies, Venkatesh and Katzberg reported that the three human ENaC genes are expressed at low levels in second trimester lung and are coordinately upregulated by high doses of glucocorticoid hormones(21). To delineate the promoter and other cis-acting elements of the human Na+ channel α subunit, and to understand mechanisms involved in its regulation, we isolated and dissected genomic clones that contain the αhENaC gene and identified regions of putative regulatory elements. Here we report the genomic organization of the gene as well as analyses on baseline and glucocorticoid hormone-regulated expression of the αhENaC promoter.

METHODS

Genomic library screening and characterization. An arrayed human genomic PAC (P1-derived artificial chromosome) library was screened using a random-primer labeled αhENaC probe, spanning cDNA nucleotide 241-547. Positive clones were rescreened by dot-blot hybridization and confirmed by Southern blot analysis using the same probe or by PCR with cDNA-derived primers. Large restriction fragments containing the 5′ flanking sequence up to intron 4 were identified by Southern analysis using the same probe, and subcloned into pBluescript vector (Stratagene, La Jolla, CA). Mapping of exon-intron boundaries from exon 5 to the end of cDNA was achieved by amplification of genomic sequences in PAC clone 97E22 using primers designed from cDNA sequence. Genomic fragments were generally amplified with Taq polymerase except intron 8, which was amplified with Advantage-GC cDNA polymerase (Clontech, Palo Alto, CA), and cloned directly into PCR2.1 vector (Invitrogen, San Diego, CA) for restriction mapping and sequencing. Nucleotide sequence analyses were performed by dideoxy sequencing reactions with [α-35S]dATP or by cycle sequencing using Thermo Sequenase (Amersham, Buckinghamshire, England) with fluorescent-labeled universal primers (LI-COR, Lincoln, NE).

Primer extension analysis. An antisense oligonucleotide primer of sequence 5′-GCTTGTTCCCCTGACAGGTGCAGCGGCC spanning bases -45 to -72 from the initiation codon was 5′ end-labeled with [γ-32P]ATP. Labeled primer was hybridized with total RNA from A549 cells at 70°C for 5 min and 52°C for 30 min in 125 mM Tris-Cl, pH 8.3, and 17.5 mM KCl. The primer was then extended with AMV Reverse transcriptase (Seikagaku) at 42°C for 1 h, in a reaction mixture that included 125 mM Tris-Cl, pH 8.3, 17.5 mM KCl, 10 mM MgCl2, 5 mM DTT, 25 mM actinomycin D, 4 mM Napyrophosphate, 0.5 mM spermidine, and 250 µM dNTPs. The sample was then precipitated with ethanol and the extended products analyzed by denaturing PAGE, along with a set of dideoxy sequencing reaction, using a subclone that contains the αhENaC promoter region and the same labeled primer.

Cell lines and transfection. A human lung epithelial type II-like cell line, A549, and an SV40 transformed African Green monkey kidney cell line, COS-1, were used to analyze αhENaC promoter activities. Both cell lines were maintained in DMEM, supplemented with 10% fetal bovine serum. The cells were transfected at 50-80% confluency with 1 µg DNA premixed with 12 µg of lipofectamine (GibcoBRL, Gaithersburg, MD) per 35 mm well in serum-free media for 6-7 h. Serum was then added back to the media containing the transfection mix, which was removed and replaced with fresh DMEM plus serum at 24 h posttransfection.

Reporter gene assay. Conditioned media were collected at 48 h posttransfection. Secreted alkaline phosphatase activities in the culture media were detected using the Phospha-Light chemiluminescent assay system (Tropix, Bedford, MA) as recommended, and measured on a luminometer (BioOrbit, Turku, Finland, for Figs. 3 and 4A; EG&G Berthold, Bad Wildbad, Germany, for Fig. 4B). In the dexamethasone induction experiments, transfected cells were placed under DMEM supplemented with active charcoal/dextran stripped fetal bovine serum, with or without 100 nM dexamethasone, from 24 to 48 h posttransfection.

Figure 3
figure 3

Analysis of αhENaC promoter activities. A549 cells were transfected with SEAPII reporter constructs containing various lengths of αhENaC 5′ UTR, with or without the first intron. The promoter lengths are depicted as fragments between restriction sites, BamHI (B), FseI (F), HindIII (H), PstI (P), and XbaI (Xb), corresponding to the schematic diagram of the genomic structure of αhENaC shown on top. Secreted alkaline phosphatase activities in the culture media harvested at 48 h posttransfection were assayed, and the expression levels are expressed as relative light units (RLU) measured with the luminometer. The results are presented as mean ± SEM (n ≥ 4).

Full size image
Figure 4
figure 4

Glucocorticoid hormone induction of expression from αhENaC promoter. (A) SEAPII constructs containing various lengths of αhENaC 5′ UTR were cotransfected with pRShGR, which supplies human glucocorticoid receptor(49), at 1:1 molar ratio into COS-1 cells. Transfected cells were placed under media supplemented with or without 100 nM dexamethasone for 24 h before assay. For positive control, we used a GRESEAPII construct in which expression of the alkaline phosphatase reporter is directed by herpes simplex virus thymidine kinase promoter connected to a GRE. (B) Similar experiments were conducted on αFXSEAPII and αFHSEAPII in comparison with their derivatives (αFXmSEAPII and αFHmSEAPII) in which the GRE consensus in the αhENaC 5′ UTR had been precisely removed (indicated as an x). The reporter activities from at least three independent experiments are shown as mean ± SEM. Note that data for panel B were obtained from a different luminometer (EG&G Berthold, Bad Wildbad, Germany), which gives readings of relative light unit at a scale 10,000-fold of those obtained from the other luminometer used for panel A in this figure and Figure 3.

Full size image

RESULTS

For analysis and mapping of complex genomic structure, PAC(22) vector systems have been well acknowledged to have several advantages, including large insert capacity and high insert stability, over traditional cosmid systems. We isolated two clones from a human genomic PAC library using an αhENaC cDNA fragment (241-547) as a probe. These two genomic clones shared the same restriction patterns corresponding to the 5′ half of the αhENaC gene by Southern analysis. One PAC clone, 97E22, containing a 130- to 140-kb insert, was further dissected to yield several overlapping subclones that add up to span an approximate 38 kb genomic sequence covering the entire αhENaC gene including 13 exons, 12 introns, 5′ and 3′ flanking sequences at least 6.5 kb and up to the end of the cDNA, respectively (Fig. 1). These subclones were restriction-mapped and sequenced, except for the internal sequences of the three large introns, introns 2, 4, and 8. The exon-intron organization and GenBank accession numbers are detailed in Table 1. The 5′ UTR is interrupted by a 667-bp intron at position -53 from the translation start codon. Exon XIII contains the last 278 bp of coding sequence and 1,100 bp of 3′ UTR up to the putative polyadenylation signal at the end of the cDNA. Table 1 also describes the exon-intron junction sequences in which the consensus splicing signals are largely conserved(23).

Figure 1
figure 1

Schematic representation of the genomic structure of the αhENaC gene. Exons are depicted as open (untranslated) and closed (translated) boxes with the transcription initiation site (bent arrow) and translation start codon, ATG, indicated. Also shown are restriction enzyme sites BamHI (B), EcoRI (E), HindIII (H), KpnI (K), XbaI (Xb), and XhoI (X).

Full size image
Table 1 Features of the exons and introns of α-hENaC
Full size table

At the amino acid sequence level, the α, β, and γ human ENaC subunits share significant homology with each other (33-37% identical) and are highly homologous (83-85% identical) to their respective rat counterparts(15). In addition, the β- and γ-ENaC sequences share distant homology with the degenerin family genes, suggesting crucial functions of these molecules. At the level of genomic organization, ENaC genes are also very well conserved among different subunits and different mammalian species. Comparison of the genomic structure of αhENaC gene with that of its rat homologue, αrENaC(24), and the reported structure of γhENaC gene(25) reveals striking similarity in exon-intron organization, especially between the human and rat α subunits (Fig. 2). However, the αrENaC gene does not contain the first intron in the 5′ UTR, which is common among the two human homologues. On the other hand, γhENaC gene lacks a 3′ intron that is shared between the α subunits (intron 10 of αhENaC and intron 9 of αrENaC genes) and has rather different intron sizes from the αENaC genes.

Figure 2
figure 2

Homology among genomic structures of αhENaC, γhENaC, and αrENaC genes. Exons are depicted as open (untranslated) and closed (translated) boxes.

Full size image

Two αhENaC cDNA clones, with identical coding regions but different lengths of the 5′ and 3′ UTRs, were independently identified by two different groups(14,16). To determine the exact transcription initiation site, primer extension analysis was performed on total RNA isolated from A549 cells, a human lung type II-like epithelial cell line, using an antisense oligonucleotide spanning bases -45 to -72 from the initiation codon. On polyacrylamide gels, the primer extension products migrated as multiple species, indicating multiple initiation sites (see discussion), while the longest and most abundant product corresponding precisely to the 5′ end of the αhENaC cDNA molecules with the longer 5′UTR(16) (data not shown). There are, however, other possible explanations for the multiple bands we observed, including degradation of the RNA samples and incomplete reverse transcription.

After identifying a subclone that contained the first two exons and a large 5′ flanking sequence of the αhENaC gene, we isolated and characterized a 6.2-kb fragment, between an FseI site at the transcription initiation and an upstream HindIII site (Fig. 3), for promoter/enhancer activity using a SEAP II reporter vector and assay system. When placed upstream of the SEAP coding region, this 6.2-kb 5′ flanking sequence directs modest but significant levels of expression in A549 cells. This basal promoter activity can be significantly induced by dexamethasone in COS-1 (Fig. 4A) and A549 (data not shown) cells. Shortening the promoter fragment to the XbaI site 2.4 kb upstream from the transcription initiation site resulted in 65% reduction of the basal expression. SEAP reporter expression directed by a 250 bp minimal promoter (up to a BamHI site) amounted to levels only 25% of those observed from the 6.2-kb fragment. These results suggest that positive regulatory elements might be present in the upstream 5′ flanking regions.

To determine whether the first intron located in the 5′ UTR harbors any cis-acting elements, we engineered reporter constructs containing the three different lengths of the αhENaC 5′ flanking sequence with the first intron linked in the genomic configuration. In all cases, the presence of the first intron resulted in decreases of expression levels ranging from 2- to 4-fold (Fig. 3), suggesting a negative regulatory element in the first intron. Alternatively, the presence of an intron upstream of the reporter coding region might introduce a splicing anomaly that could affect RNA stability or transport and hence the levels of protein product.

It has been shown, both in human and in rat, that the expression of the α subunits of the epithelium Na+ channel can be upregulated by glucocorticoid hormones in vivo and in cell culture. Of the 6.2-kb αhENaC promoter, a 3.1-kb proximal region has been sequenced, revealing a dyad GRE motif AGAACAGAATGTCCT at position -783 to -769, a 93.3% match to the GRE consensus AGAACANNNTGTTCT(26). Secreted alkaline phosphatase activities from cells transfected with constructs containing this putative GRE motif (with the 2.4- and 6.2-kb promoter fragments) increased 8- to 20-fold after 24 h of dexamethasone treatment (Fig. 4A). The magnitudes of dexamethasone induction from these αhENaC promoter fragments are larger than that (3- to 4-fold) observed from a control construct in which expression of the reporter is driven by the herpes simplex virus thymidine kinase promoter connected to a GRE(24). The minimal 250-bp αhENaC promoter, which does not contain the GRE motif, showed no response to dexamethasone induction. To verify the role of this GRE motif, we performed PCR site-directed mutagenesis to precisely remove the 15-bp GRE motif. The deletion completely abolished the response to dexamethasone induction in the reporter constructs containing the 2.4- and 6.2-kb promoter fragments (Fig. 4b). Taken together, these results suggest that the distal region of the αhENaC promoter contains cis-acting elements that can be directly regulated by corticosteroids.

DISCUSSION

We sequenced the αhENaC 5′ flanking region for up to 3.1 kb from the transcription initiation, continuous through the first exon, the first intron, the second exon, and into the second intron. Close examination of the sequence around the transcription initiation site reveals none of the conventional CCAAT or TATA-like promoter elements but, rather, three GC boxes, two at the transcription start site and one slightly upstream, which might serve as binding sites for transcription factor Sp1(27). These features resemble the γhENaC promoter as well as other promoters that are associated with constitutively expressed genes that have housekeeping and growth-related functions(28). Of these TATA-less promoters, some contain specific initiator sequences for accurate positioning of RNA polymerase II(29,30) and, therefore, accurate transcription initiation; the others use multiple sites of initiation. The αhENaC promoter appears to belong to the latter group for its lack of an initiator element and for the apparent heterogeneous transcription initiation events. With a sequence analysis program, DNA-SIS, other potential transcription factor binding sites were also identified, including consensus binding sites for homeodomain factor Nkx-2(31), proto-oncoprotein c-Rel(32), heat-shock factor HSF2(33), E1A-associated 300 kD protein p300(34), the δ-crystallin enhancer-binding protein δEF1(35), κE2 sequence binding protein E47(36), transcription factors AP-1(37,38), GATA-1,2(39), USF(40), oct-1 and oct-x(41), in the 250 bp proximal 5′ UTR (Fig. 5). Further upstream, there are putative binding sites for the proto-oncoprotein Ets-1/PEA3(42) and the transcription factor NF-κB(32), in addition to a GRE motif(26).

Figure 5
figure 5

(A) Proximal 5′ flanking genomic sequence of αhENaC gene between positions -1 and -250 from the transcription initiation site. Sequence data obtained from both strands are consistent. Consensus binding motifs for transcription factors are indicated. (B) Depiction of putative regulatory elements in the distal αhENaC 5′ flanking region between BamHI (-250) and XbaI (-2, 400) sites, including consensus motifs for GRE and for NFκB or Ets-1/PEA3 binding sites.

Full size image

Previous studies have shown that expression of αhENaC in cultured human fetal lung explants can be upregulated by dexamethasone at the transcriptional level, through the glucocorticoid receptor, without de novo protein synthesis, suggesting that the αhENaC gene is a primary responder to glucocorticoid hormones(21). Consistent with these findings, we have identified a GRE motif in the 5′ flanking region of αhENaC gene and demonstrated that only reporter constructs containing this GRE motif exhibit inducible expression in response to dexamethasone. In general, reporter constructs containing longer 5′ flanking sequences showed higher basal expression activities and greater response to glucocorticoid hormone treatment, indicating the presence of cis-acting elements in the distal region. Interestingly, the higher basal expression from the longer promoter fragments was lost when the transfected cells were placed under charcoal-stripped and hence steroid-free serum (Figs. 4A and B, -DEX). Such discrepancy can be explained by the presence or absence of low levels of glucocorticoid hormones in the regular serum versus the stripped serum. Alternatively, there might be other regulatory elements in the distal promoter region whose activation depend on certain extracellular factors that are removed from the serum by the stripping. Among all the human ENaC subunit gene promoters, ours is the first isolated and characterized to be sufficient for expression analysis. Identification of the transcription factors that specifically interact with and regulate the αhENaC promoter will lead to a better understanding of the regulation of the sodium channel gene expression.

As this manuscript was under preparation, a parallel, independent study using a genomic PCR technique was published by Ludwig et al. on the structural organization of αhENaC gene(43). Although our results and this recent report agree largely on the exon-intron organization and junction sequences, there are significant discrepancies regarding the size of several introns. In each case we observed larger introns - 10 kb, 6.4 kb, and 4.9 kb for introns 2, 4, and 8, respectively, according to our data versus 3.7 kb, 6 kb, and 622 bp as published. Besides naturally occurring polymorphism, one possible explanation of this apparent disagreement could be the presence of Alu repeats in introns 2-4 and 8. Alu repeats have been known to be involved in homologous recombination events in several genes(44–47), resulting in deletions and insertions, and could be potentially problematic for genomic PCR. In their study, Ludwig et al. noted difficulty in amplifying introns 2-4 and 8, which could only be obtained by long αhENaC gene by restriction mapping of overlapping genomic subclones, instead of polymerase chain reactions, which are intrinsically sensitive to repetitive sequences and secondary structures. We have also encountered problems while amplifying intron 8; however, after adopting the Advantage Genomic PCR kit by Clontech, we observed reproducibly a 4.9-kb product. Moreover, the PAC (P1-derived artificial chromosome) genomic library system we used has been well acknowledged to have several advantages, including large insert size and high insert stability, over traditional cosmid systems(22), making it more favorable for reliable analysis and mapping of complex genomic structures. Taken together, we believe that our results on genomic organization still provide valuable information of interest to scientists studying the sodium channels.

Two groups have separately reported cloning and characterization of the αhENaC cDNA. The cDNA cloned by Voilley et al. from human lung contains a 98-bp 5′ UTR before the open reading frame and about 1 kb 3′ UTR up to a polyadenylation signal followed by the poly(A) tail(16). From a human kidney library McDonald et al. isolated a cDNA molecule with an identical coding region but only partial 5′ and 3′ UTRs(14). Our results confirmed that the cDNA molecule isolated from human lung is a complete clone that extends 5′ to the initiation of transcription and 3′ to the poly(A) tail. However, there is a notable discrepancy between the sizes of the full-length cDNA molecule and the specific messengers observed by Northern analyses. In lung, kidney, pancreas, and colon, there appeared to be two species of αhENaC mRNA, one migrating at 3.8-3.9 kb and the other at 3.2-3.4 kb(14,16). Although the 3.1-kb cDNA of Voilley et al. with the potential addition of multiple adenine residues during RNA processing might be considered the faster migrating species, it cannot account for the longer transcript. Three possibilities could result in the longer mRNA species: utilization of an alternative downstream polyadenylation signal, utilization of a different transcription initiation site, or alternative splicing at the 5′ untranslated region, which cannot be resolved by our results. On the other hand, a recent report on 5′ heterogeneity of αhENaC mRNA(48) has addressed these hypotheses in depth.