Cell Physiology

5′ Heterogeneity in epithelial sodium channel α-subunit mRNA leads to distinct NH2-terminal variant proteins

Christie P. Thomas, Scott Auerbach, John B. Stokes, Kenneth A. Volk


The amiloride-sensitive epithelial sodium channel (ENaC) is composed of three subunits: α, β, and γ. The human α-ENaC subunit is expressed as at least two transcripts (N. Voilley, E. Lingueglia, G. Champigny, M. G. Mattei, R. Waldmann, M. Lazdunski, and P. Barbry. Proc. Natl. Acad. Sci. USA91: 247–251, 1994). To determine the origin of these transcripts, we characterized the 5′ end of the α-ENaC gene. Four transcripts that differ at their first exon were identified. Exon 1A splices to exon 2 to form the 5′ end of α-ENaC1, whereas exon 1B arises separately and continues into exon 2 to form α-ENaC2. Other variant mRNAs, α-ENaC3 and α-ENaC4, are formed by activating 5′ splice sites within exon 1B. Although α-ENaC3 and -4 did not change the open reading frame for α-ENaC, α-ENaC2 contains upstream ATGs that add 59 amino acids to the previous (α-ENaC1) protein. To address the significance of these isoforms, both proteins were expressed in Xenopus oocytes. The cRNA for each α-ENaC transcript when combined with β- and γ-ENaC cRNA reconstituted a low-conductance ion channel with amiloride-sensitive currents of similar characteristics. We have thus identified variant α-ENaC mRNAs that lead to functional ENaC peptides.

  • sodium transport
  • Xenopus oocyte
  • patch clamp
  • gene expression
  • alternate transcripts

the major route forNa+ transport in kidney collecting duct cells is via aldosterone-responsive, amiloride-inhibitable Na+ channels. These channels are also expressed throughout the airway epithelia and distal colon epithelia and in sweat ducts. Recently, several members of a family of amiloride-sensitive epithelial Na+channel (ENaC) proteins were cloned from rat and human tissue (3, 5, 19, 20, 23, 24, 40). The channel, containing α-, β-, and γ-subunits, when reconstituted inXenopus oocytes, is amiloride sensitive, highly selective for Na+, and has a single-channel conductance of 4–10 pS (3, 20). The mRNAs for these subunits are regulated by glucocorticoids, mineralocorticoids, and dietary Na+ intake and are expressed in epithelia where amiloride-sensitive Na+ transport has been identified (6, 10, 19, 29, 35, 39), providing strong evidence that these channels represent the highly selective low-conductance epithelial Na+ channel.

Recessive and dominant mutations of the ENaC complex cause human disease. Type 1 pseudohypoaldosteronism, with salt wasting, hypotension, and hyperkalemia, can occur as a consequence of homozygous inactivating mutations in any of the three subunits (7, 34). Liddle’s syndrome, an autosomal dominant form of salt-sensitive hypertension, is secondary to an activating mutation in β- or γ-ENaC (13, 31). These findings suggest that some patients with salt-sensitive hypertension may have subtle defects in ENaC function or regulation (30). To begin to evaluate this possibility in some detail, we have chosen to examine α-ENaC mRNA expression. In human kidney and lung, transcripts of at least two different sizes have been previously described for α-ENaC (24, 38), but their molecular structure, regulation, or significance has not been established. We now identify four separate transcripts for α-ENaC that arise by separate initiation of transcription and alternate splicing within the first exon. These 5′ variant mRNAs lead to distinct and functional α-ENaC peptides.


5′ Rapid amplification of cDNA ends.

Two sets of modified human cDNAs were obtained for 5′ rapid amplification of cDNA ends (5′-RACE). The double-stranded human kidney Marathon-Ready cDNA (Clontech, Palo Alto, CA) has an adapter sequence 5′-CTAATACGACTCACTATAGGGCTCGAGCGGCCGCCCGGGCAGGT ligated to both ends. The single-stranded 5′-RACE-Ready human lung cDNA (Clontech) has an anchor sequence 5′-CACGAATTCACTATCGATTCTGGAACCTTCAGAGG-NH2ligated to its 3′ end. Gene-specific reverse primers α3 [5′-TCCTCCGCCGTGGGCTGCTG; +133 to +114 from the original initiation codon; see Figs. 1 and 2 (24, 38)] and α4 [5′-CCCTGGAGTGGACTGTGGAGGGCTAG; +56 to +31 from the original initiation codon; see Figs. 1 and 2 (24, 38)] were used singly or sequentially with anchor- or adapter-specific primers in PCR reactions using Taq polymerase. Typically, the amplification occurred for 35 cycles each at 94°C for 30 s, with annealing at 61–69°C for 30 s and extension at 72°C for 3 min.

Genomic PCR.

To amplify intervening sequences between exon 1A, exon 1B, and exon 2, primers α7 (5′-AGGCCGCTGCACCTGTCAG), α8 (5′-CCTGGGGCAGAGACAGAATC), and α4 were used. To amplify genomic DNA corresponding to the 5′ portion of exon 1A, primers α18 (5′-GAGGGGGTGGCGAGGAATCA) and α19 (5′-CTGGGCAGGGGCTTTAGACG) were used; 100 ng of human genomic DNA were used in each amplification reaction, with denaturing at 94°C for 30 s, annealing at 63–67°C for 30 s, and extension at 72°C for 3–4 min for a total of 35–40 cycles.

DNA sequencing.

Amplified fragments from PCR reactions were cloned into pCR II 2.0 or 2.1 (Invitrogen, San Diego, CA), and individual clones were sequenced. DNA sequencing was performed using dye terminator cycle sequencing chemistry with Amplitaq DNA polymerase FS enzyme and was analyzed on a 373A stretch fluorescent automated sequencer (Applied Biosystems, Foster City, CA). For identification of transcription start sites, cDNA templates were sequenced by the dideoxy chain termination method using Sequenase 2.0 (Stratagene, La Jolla, CA) and [α-35S]dATP with primers α21 (5′-CTCGAGCTGTGTCCTGATTC) and α23 (5′-TCAGGCCCTGCAGAGAAGAGAGAAGAGGTC). Ribonuclease protected products were run alongside sequenced products on a sequencing gel. Transcription factor binding motifs within the human α-ENaC (α-hENaC) gene were identified with MacDNASIS (Hitachi software, San Bruno, CA).

Cell line and RNA preparation.

A human lung tumor cell line H441 [American Type Culture Collection (ATCC), Rockville, MD] was grown as a monolayer in RPMI 1640 medium supplemented with 8.5% bovine calf serum, 8.5% fetal bovine serum (FBS), 20 mMl-glutamine, 5 μg/ml insulin, 5 μg/ml transferrin, and 5 ng/ml selenium. A human colon carcinoma line HT-29 (ATCC) was grown as a monolayer in McCoy’s 5A medium supplemented with 10% FBS and 20 mM glutamine. A human lung cystic fibrosis cell line, IB3–1 (41), was grown in LHC-8 (Biofluids, Rockville, MD) supplemented with 5% FBS. A human cortical collecting duct (hCCD) cell line (26) was grown in 1:1 DMEM-Ham’s F-12 supplemented with 2% FBS, 5 nM triiodothyronine (T3), 100 nM dexamethasone, 5 μg/ml insulin, 5 μg/ml transferrin, and 5 ng/ml selenium. RNA prepared from H441, hCCD, HT-29, and IB3–1 cells and from human lung, colon, and kidney tissue were used for Northern analysis or ribonuclease protection assays (RPA). Total RNA was prepared by solubilization of monolayers or homogenized tissue in guanidinium thiocyanate buffer followed by extraction, first with Tris-saturated and then with water-saturated phenol, and precipitated with isopropanol (8).

Ribonuclease protection assay.

To assess the relative abundance of each transcript in various tissues, each of the four cDNAs identified by 5′-RACE was used to construct templates for RPA. 1) The first was an α-ENaC2 cDNA that included the 5′ end of exon 2 and the 3′ portion of exon 1B (nt +446 to +806, see Fig. 2) ligated into pCR II antisense to the T7 polymerase promoter. This is predicted to protect a 361-nt fragment corresponding to α-ENaC2 and a 188-nt fragment corresponding to α-ENaC1, -3, and -4. 2) The second was an α-ENaC1 cDNA that included the 5′ end of exon 2 and the 3′ portion of exon 1A (nt +614 to +860, see Fig. 2) ligated into pBluescript II SK (Stratagene) antisense to the T3 promoter. This is predicted to protect a 247-nt fragment corresponding to α-ENaC1 and a 188-nt fragment corresponding to α-ENaC2, -3, and -4.3) The third was an α-ENaC3 cDNA fragment that included the 5′ end of exon 2 and the contiguous portion of exon 1B seen in this form (nt +277 to +536, see Fig. 2) cloned into pCR II antisense to the T7 promoter. This is predicted to protect a 260-nt fragment corresponding to α-ENaC3, a 150-nt fragment corresponding to α-ENaC2, and a 110-nt fragment corresponding to α-ENaC1, -2, and -4. 4) The fourth was an α-ENaC4 cDNA fragment that included the 5′ end of exon 2 and the contiguous portion of exon 1B seen in this form (nt +18 to +253, see Fig. 2) cloned into pBluescript II SK antisense to the T7 promoter. This is predicted to protect a 236-nt fragment corresponding to α-ENaC4 and a 153-nt fragment corresponding to α-ENaC1, -2, and -3.

To map the transcription start site(s) of exon 1B, two separate DNA templates were constructed by PCR amplification and restriction digestion of genomic DNA: 1) a genomic fragment that included the 3′-terminal 20 nt of exon 1A and the 5′ portion of exon 1B (up to theAvr II site) and2) a second genomic fragment that also included the 3′-terminal 20 nt of exon 1A and the 5′ portion of exon 1B (up to the Xho I site; see Figs. 2 and 5). Both fragments were ligated into pCR II antisense to the T7 promoter. To map the 5′ end of exon 1A, a genomic fragment that included the putative 5′ flanking sequence and contiguous portions of exon 1A was amplified using primers α18 and α19 (see Figs. 2 and 6). A second fragment that extended to theBsu36 I site within exon 1A was created by restriction digestion of the α18-α19 fragment (see Figs.2 and 6). Both constructs were ligated into pCR II antisense to the T7 polymerase promoter.

The templates were linearized and used to synthesize antisense [α-32P]UTP-labeled cRNAs. These probes were hybridized overnight with 10 μg of sample RNA or yeast RNA at 45°C in 80% formamide, 400 mM NaCl, and 40 mM PIPES, pH 6.4 (22). In some experiments, an 18S rRNA template (pTR1 RNA 18S; Ambion, Austin, TX) was used to generate an antisense riboprobe and samples were cohybridized with α-ENaC riboprobes. Samples were then digested with 30 U/ml RNase T1 and 1 U/ml RNase A (Ambion) and analyzed by denaturing PAGE. The sizes of protected fragments were determined from a radiolabeled 50-bp DNA ladder (Life Technologies, Gaithersburg, MD) run alongside.

Northern analysis.

H441 RNA samples were denatured, resolved on a 1.5% agarose and 6% formaldehyde gel, and then transferred to nylon membranes (Zetaprobe-GT, Bio-Rad, Hercules, CA). A cDNA probe (α-common) that extends from exon 2 to a Sma I site ∼1,100 bp downstream of the initiation codon was prepared from an α-ENaC cDNA clone (gift from M. J. Welsh, University of Iowa). An exon 1A-specific probe was prepared from the 3′ 350 nt of exon 1A. A cDNA fragment prepared by restriction digestion of a 5′-RACE α-ENaC2 clone that began at +128 and ended at aSac I site (+563) was used to identify transcripts that included the 3′ portion of exon 1B. APstI-Asp718 I fragment derived from exon 2 was used as an exon 2-specific probe. Random primer-extended [α-32P]dCTP-labeled cDNA probes were hybridized to immobilized RNA in 1 mM EDTA, 0.25 M NaH2PO4, pH 7.2, and 7% SDS at 65°C for 12 h. The blots were then washed in 1 mM EDTA, 40 mM NaH2PO4, pH 7.2, and 5% SDS at 65°C for 30–90 min. To identify transcripts that included the 5′ portion of exon 1B, anHphI-Xho I fragment was used as a template for the preparation of a single-stranded antisense DNA probe by extension from a specific primer, α21 (short double-stranded DNA probes from this region did not identify either transcript). This radiolabeled DNA probe was hybridized in 5× standard sodium citrate (SSC), 7% SDS, 20 mM NaH2PO4, pH 7.2, and 1× Denhardt’s solution at 60°C for 16 h and then washed first with 3× SSC, 5% SDS, 25 mM NaH2PO4, pH 7.5, and 10× Denhardt’s and then with 1× SSC and 1% SDS at 60°C.

In vitro translation.

A full-length α-ENaC1 in pcDNA3 (Invitrogen, Carlsbad, CA) with the last 24 bp that encode the COOH-terminal octapeptide replaced with the FLAG epitope (α-1-FLAG) was obtained from M. J. Welsh (University of Iowa). The cloned 5′ variant α-ENaC (exon 1B) was spliced to the 5′ end of α-ENaC1 using overlapping primers in a PCR reaction to generate α-1-FLAG. The 5′-untranslated region (UTR) in both constructs was under 15 bp, and the translation stop followed immediately after the FLAG epitope. Sequential transcription and translation of α-1-FLAG and α-2-FLAG were performed alongside pcDNA3 and H2O controls using the TnT reticulocyte lysate system (Promega, Madison, WI) following the manufacturer’s instructions. [35S]methionine was used to label the synthesized peptides, and the products were visualized by SDS-PAGE and autoradiography.

Expression in Xenopus oocytes.

The coding regions for the four hENaC subunits [α(-2), α(-1), β, and γ] were subcloned into the plasmid pGEM-HE. This vector contains the 5′- and 3′-UTRs of theXenopus β-globin gene flanking the cloning site and was engineered specifically to enhance expression of in vitro transcribed cRNAs in Xenopusoocytes (21). Each cDNA was transcribed and capped in vitro using the Message Machine kit (Ambion). The cRNAs were combined into two groups [α(-2)βγ and α(-1)βγ] so that 50-nl injections carried 2.5 ng of each hENaC subunit.

Oocytes were removed from mature female Xenopus laevis and defolliculated by incubation in a 1 mg/ml collagenase solution. After a 24-h recovery period, healthy oocytes were injected with cRNA and stored in frog Ringer solution containing 115 mM NaCl, 2.5 mM KCl, 1.8 mM CaCl2, 10 mM HEPES (pH 7.3), 5 mM sodium pyruvate, and 100 U/ml penicillin-streptomycin. Whole cell currents and single-channel recordings were measured 48 h later.

Whole cell currents were measured in frog Ringer solution using the two-microelectrode voltage-clamp technique. The OC-725C oocyte voltage-clamp amplifier (Warner Instruments, Hamden, CT) was controlled and data were acquired with the pCLAMP (Axon Instruments, Foster City, CA) software and acquisition system. Single-channel currents were recorded from cell-attached patches of oocyte membranes using pCLAMP and the Axopatch 200 patch-clamp amplifier (Axon Instruments). The bath for the single-channel recordings contained frog Ringer solution, whereas the pipette contained 140 mM LiCl, 3 mM MgCl2, and 10 mM HEPES (pH 7.35). All current recordings were performed at room temperature.


Two groups have reported the primary structure of human α-ENaC mRNA and its predicted protein (24, 38). The cloned α-ENaC cDNA included 103 nt of 5′-UTR, the complete 3′-UTR, and an open reading frame of 2010 nt that encodes a 669-amino acid protein. Both groups reported that the cloned cDNA hybridized to two transcripts in several human tissues, but the identify of these transcripts was not established.

To identify the nature of the alternate transcripts previously reported, we set out to map the organization of the α-hENaC gene beginning with the 5′ end. We first determined the nucleotide sequence of the 5′ end of α-ENaC mRNA in human kidney and lung by 5′-RACE using reverse primers (α3 and α4) designed to anneal to α-ENaC RNA just downstream of the previously described translation initiation codon. Several clones were sequenced from both tissues, and remarkable heterogeneity was noted, with at least four possible cDNA forms identified. One form with a 5′-UTR of 709 nt (α-ENaC1, Fig. 1) corresponded to that previously described in kidney and lung but extended several hundred bases upstream (24, 38) and was identified in 13% of clones. The other three forms diverged from α-ENaC1 at −54 to the original translation start codon and were consistent with alternately initiated transcripts. The most abundant form (α-ENaC2, Fig. 1), representing ∼74% of the clones, added an additional 535 nt 5′ to the point of divergence. Two other forms, α-ENaC3 and -4, appeared to be internally deleted versions of α-ENaC2 consistent with splice variants and were found infrequently.

Fig. 1.

Schematic of the 5′ end of α-subunit epithelial Na+ channel (α-ENaC) gene and alternate transcripts identified by 5′ rapid amplification of cDNA ends (5′-RACE). Molecular structure of the gene is shown on the 1st line, with the 5′ splice sites indicated by ▴ and the 3′ splice site by ★. In-frame ATGs that extend the original reading frame are shown, as well as upstream in-frame stop codons (TGA and TAG). Original translation start codon in exon 2 and the 5′ translation start codon in exon 1B are denoted by * and #, respectively. The 5′ extent of each cDNA clone is depicted by • above each cDNA form. (A total of 32 clones were analyzed.) Primers used for genomic PCR and the 5′ and 3′ splice sequences are also shown. In each of the spliced forms, exons are shown, as well as the length of intervening sequence that is spliced out. Full extent of exon 1A and 1B as shown in the α-ENaC gene map (1st line) was determined by ribonuclease protection assay (RPA; see Figs. 5 and 6). Hatched bar and open bar define exons 1A and 1B, respectively, whereas solid bar indicates exon 2 and sequences further downstream.

To understand the genomic organization at the 5′ end of the α-ENaC gene, specifically to determine the exon-intron organization in this region, forward primers (α7 and α8) corresponding to the 5′ portions of form α-ENaC1 and α-ENaC3 mRNA were used with downstream primers α4 or α3 to amplify intervening sequences from genomic DNA (Fig. 1). Sequence analysis confirmed that a 665-nt intron, starting and ending with consensus GT and AG, respectively, was spliced out of α-ENaC1. This intron ends at −54 to the original translation start codon and corresponds to the point of divergence between forms α-ENaC1 and α-ENaC2 (Fig. 1). The intron sequence (intron 1) thus separates exon 1A (5′ end of α-ENaC1) from exon 2. Sequence corresponding to α-ENaC2 arises within this intron sequence and therefore represents an alternate first exon (exon 1B). This alternately transcribed sequence continues unspliced into exon 2. Forms α-ENaC3 and -4 were identified as variants of exon 1B spliced to exon 2, created by activation of 5′ splice sites within exon 1B (Figs. 1 and2). Of the four transcripts, α-ENaC2 contained four in-frame translation initiation codons that would be predicted to extend the length of the open reading frame by up to 59 amino acids, creating a variant protein α-ENaC2 (Figs. 1 and 2). All other transcripts keep the original open reading frame intact (α-ENaC1).

Fig. 2.

Nucleotide sequence data of the 5′ end of α-ENaC gene beginning with exon 1A and continuing through intron 1/exon 1B into the proximal portion of exon 2. Labeled arrows indicate PCR primers used for genomic PCR and 5′-RACE. Transcription start sites for exons 1A and 1B were determined by RPA (see Figs. 5 and 6) and are indicated with separate numbering for each of the 4 transcripts. The 5′ splice sites are indicated by ▴ and 3′ splice site by ★. Consensus splice junction sequences at the principal splice site are shown with a double underline. Asp718 I,Avr II,Bsu36 I,Hph I,Pst I,Sac I, andXho I sites used for construction of cDNA and ribonucleotide probes are indicated. Original translation start codon in exon 2 and the 5′ translation start codon in exon 1B are boxed and denoted by * and #, respectively. Untranscribed sequence in intron 1 and 5′ flanking region is shown in small letters, and transcription start sites are indicated by bent arrows.

To examine these alternate transcripts in greater detail, we performed RPA in selected tissues (Fig. 3). This was intended to confirm the existence of these variant mRNAs and to assess the relative proportion of each of these in kidney, colon, and lung, principal sites of amiloride-sensitive Na+ transport. Most studies used normal human kidney and lung tissue as well as IB3–1, HT-29, hCCD, and H441 cells. Given the complexity of these transcripts, it was not possible to design a single RPA template to distinguish all transcripts simultaneously. Therefore, separate RPA templates were constructed from each of the identified cDNA forms, designed to distinguish one of the transcripts from the other three. With the use of this method, each of the four transcripts was readily identified in all tissues examined except hCCD (Fig. 3,AD).

Fig. 3.

RPA identifies transcripts α-ENaC1, -2, -3, and -4. Each of four cDNAs was used individually in RPA reactions and was designed to separate 1 transcript from the other 3. In all cases, kidney (kid), lung, IB3–1, and H441 RNA were used and are shown alongside yeast RNA (negative control) and free (undigested) probe. In some cases, additional RNA samples from colon, HT-29 cells, and human cortical collecting duct (hCCD) cells were used. Numbering shown in schematic (right) relates to each cDNA/RNA form and is as indicated in Fig. 2. Symbols atright are as shown in Fig. 2, and molecular weight was determined from a radiolabeled 50-bp DNA ladder.A: α-ENaC1 cRNA probe identifies 247- and 188-nt fragments. B: α-ENaC2 cRNA probe identifies 361- and 188-nt fragments.C: α-ENaC3 cRNA probe identifies 260-, 150-, and 110-nt fragments. D: α-ENaC4 and a human 18S rRNA probe were cohybridized to all samples. These probes identify 236-nt and 153-nt (α-ENaC) and 81-nt (18S) fragments. Reactions involving H441 and kidney samples were performed a minimum of 3 times.

We next determined if any or all of these transcripts corresponded in size to that previously identified by Northern analysis (24, 38). To examine this issue, a series of cDNA probes were used for Northern blotting of H441 RNA. An α-common cDNA probe corresponding to exon 2 and downstream sequences identified two equally abundant transcripts of 3.9–4.0 and 3.5–3.6 kb in H441 RNA (Fig.4 E). These are probably the same as the 3.8- and 3.4-kb transcripts reported by Voilley et al. (38) and the 3.9- and 3.2-kb transcripts reported by McDonald et al. (24). By using a variety of short exon-delimited probes, we show that both transcripts contain exon 2 (Fig.4 D), whereas an exon 1A-specific probe hybridized to the 3.9- to 4.0-kb transcript alone (Fig.4 A). A 100-nt 5′-terminal exon 1B probe that corresponds to α-ENaC2, -3, and -4 hybridized to both transcripts (Fig. 4 B), whereas a 3′-terminal exon 1B probe (Fig.4 C) that corresponds to α-ENaC2 and -3 hybridized to the 3.9- to 4.0-kb transcript alone. These results clearly indicate that, although exon 2 is contained within both transcripts, α-ENaC1, -2, and -3 mRNAs correspond to the larger transcript alone. Taken together, the data in Figs. 2-4 also suggest that α-ENaC4 mRNA corresponds to the 3.5- to 3.6-kb transcript. Because the relative amount of the 3.5- to 3.6-kb transcript identified by the exon 2 probe is much greater than that identified by the 5′-terminal exon 1B probe, we cannot exclude the presence of other transcripts that contain exon 2 but do not have portions of exon 1A or 1B within them.

Fig. 4.

Northern analysis identifies at least 2 transcripts for α-ENaC in kidney and H441 RNA. A series of exon-specific fragments was obtained by restriction digestion of cDNA or genomic DNA clones and was used to prepare DNA probes and hybridized to indicated RNA. Position of each DNA probe is shown aligned with a map of the 5′ end of the gene and its alternate transcripts. A 3.9- to 4.9-kb transcript and a 3.5- to 3.6-kb transcript were identified by an α-common probe (E). An exon 1A-specific probe (A) and a 3′ exon 1B-specific probe (C) identified the 3.9- to 4.0-kb transcript alone, whereas a 5′ exon 1B-specific probe (B) and an exon 2-specific probe (D) identified both transcripts. Symbols in schematic (top) are as described in Fig. 2. Each experiment was performed 2–4 times.

To clarify whether exon 1A and 1B have distinct transcription start sites, we performed RPAs on total RNA from selected tissues using a genomic fragment that included the terminal 20 nt of exon 1A and 5′ portions of exon 1B. Two protected fragments were seen, suggesting that exon 1B is initiated separately. The more abundant protected fragment, a 385-nt fragment, predicted a transcription initiation site for exon 1B 53 nt downstream of the 5′ splice site of exon 1A (Fig.5 A). This transcription start site was seen in kidney tissue and lung tissue and H441 and IB3–1 cells. A longer, much weaker protected fragment was also seen, suggesting that some mRNA species extended the entire length of exon 1B and into exon 1A. To confirm these findings and to accurately localize the transcription start site, a second shorter riboprobe that extended from the same 5′ end to theXho I site (100 nt downstream of the putative transcription start site) was used. The more abundant protected fragment was resolved on a sequencing gel to two closely migrating bands and indicated transcription start sites 53 and 55 nt downstream of the 5′ splice site of exon 1A (Fig.5 B). A protected fragment that was 71 nt longer indicated the existence of an mRNA species that included exon 1B and begins in exon 1A.

Fig. 5.

RPA identifies the transcription start site for exon 1B.A: a genomic fragment including the 3′ portion of exon 1A and the 5′ portion of exon 1B was used to synthesize a 510-nt antisense cRNA probe. Yeast RNA was used as a negative control, and free (undigested) probe is shown in the last lane. A prominent 385-nt and an additional fainter 456-nt protected fragment are seen in all tissues examined.B: a second genomic fragment derived by 3′ truncation at the Xho I site was used to synthesize a 227-nt cRNA probe. Two protected fragments at 100 and 171 nt were noted. The 100-nt fragment indicated a transcription start site downstream of exon 1A, whereas the 171-nt fragment indicated an additional transcription start site in exon 1A with transcripts continuing unspliced into exon 1B. The 100-nt fragment from kidney was resolved on a sequencing gel to precisely identify this transcription start site. Riboprobe template was sequenced with a primer (α21 primer) that anneals to the 3′ end of the genomic DNA fragment. Reactions were performed at least twice.

Working separately and as part of the Human Genome Project, a group of investigators had isolated and begun to characterize a PAC clone from human chromosome 12 that contained the 5′ of the α-ENaC gene. Using the nucleotide sequence information from this clone (Raju Kucherlapati and Kate Montgomery, Albert Einstein College of Medicine, Bronx, NY, personal communication), we amplified a genomic fragment that included the 5′ portion of exon 1A and the putative 5′ flanking region. When used as a riboprobe with RNA from kidney and H441, a protected fragment ∼470 nt long was seen, suggesting that the transcription start site for exon 1A mapped within this fragment (Fig.6 A). A second, shorter riboprobe was constructed using sequence from the 5′ end to accurately localize a single transcription start site for exon 1A (Fig. 6 B) 724 nt upstream of the principal transcription start site for exon 1B.

Fig. 6.

RPA identifies the transcription start site for exon 1A.A: a genomic fragment, including the putative 5′ flanking sequences and extending to primer α19, was used to synthesize a 656-nt cRNA probe. Yeast RNA was used as a negative control, and free (undigested) probe is shown in the last lane. A single 470-nt protected fragment is seen.B: a second genomic fragment derived by 3′ truncation at the Bsu36 I site was used to synthesize a 229-nt cRNA probe. A single protected fragment at 55 nt is seen. Protected fragment from kidney was resolved on a sequencing gel to identify transcription start site. Riboprobe template was sequenced with a primer (α23 primer) that anneals to the 3′ end of the genomic DNA fragment. Reactions were performed 2–3 times.

The predicted open reading frame for α-ENaC2 is 2187 bases long and encodes an additional 59 amino acids 5′ and in-frame to that encoded by α-ENaC1 (Fig.7 A). The predicted translation start codon in α-ENaC2 has a more favorable consensus sequence for translation initiation compared with α-ENaC1 (Fig. 7 B) and should yield a protein with a molecular mass of 82 kDa. We tested whether an α-ENaC2 construct could make the corresponding protein in an in vitro transcription-translation system using reticulocyte membranes. A single band of the predicted size was seen with α-ENaC2, confirming that the 5′-most ATG is used as the translation initiation codon for α-ENaC2. In contrast, in this in vitro system, two translated products are seen with α-ENaC1, one corresponding to the longest open reading frame prediction of 76 kDa and the other indicating a second downstream translation start site.

Fig. 7.

A: nucleotide and deduced amino acid sequence of the 5′ end of α-ENaC. Translation start codons for α-ENaC1 and α-ENaC2 are boxed. Consensus sequences for myristoylation are underlined, and duplicated peptide motifs are shown in bold letters. B: translation start codons are aligned with the consensus sequence for translation initiation (14). C: α-ENaC1 and -2 cDNAs were transcribed and translated in vitro in the presence of [35S]methionine. Water and empty plasmid pcDNA3 were used as negative controls. Products were analyzed by SDS-PAGE, and molecular size was calculated from high-molecular-mass (MM) protein markers (Life Technologies). Reactions were performed twice.

To establish that these α-ENaC peptide isoforms are functional, transcripts encoding these peptides were expressed inXenopus oocytes together with β- and γ-ENaC subunits and Na+transport was analyzed. Figure 8 shows representative whole cell currents and corresponding current-voltage (I-V) relationships recorded from oocytes injected with water (Fig. 8,A andB) or ENaC (Fig. 8,CH). The currents from water-injected oocytes are relatively small (<0.5 μA at −60 mV) and have a negative reversal potential (E rev), indicating predominant contributions from endogenous K+ and Cl channels. Bath application of 10 μM amiloride had no effect on these currents (data not shown). The contribution of hENaC to the total oocyte current was determined by bath application of 10 μM amiloride as depicted in Fig.8,CH. In an oocyte expressing hENaC, theE rev is generally slightly positive, indicating contribution of a channel type with a positive E rev. Addition of amiloride to the bath reduces the current amplitude and shifts the E revto negative values, resulting in a current trace andI-Vrelationship that is similar to those from water-injected oocytes. The amiloride-sensitive currents that are due to hENaC expression are derived by subtracting the postamiloride values from the preamiloride values (Fig. 8, G andH). TheE rev of the amiloride-sensitive currents is typically near +10 mV. Because ENaC is known to be highly Na+ selective, the calculatedE rev would be substantially more positive than +10 mV, if one assumes typical intracellular ionic concentrations. It has been suggested, however, that the overexpression of ENaC in oocytes results in markedly elevated intracellular Na+ concentrations (5, 23), and this is the likely explanation for the observedE rev. Expression of either α-ENaC form along with the β- and γ-subunits resulted in time-independent, amiloride-sensitive currents with average current magnitudes andE rev values that were indistinguishable (Fig. 8 H).

Fig. 8.

Representative whole cell currents recorded from oocytes 48 h after injection are shown in top panels with the corresponding current-voltage (I-V) plots shown in bottom panels. Currents were elicited with voltage steps from a holding potential of 0 mV to voltages ranging from 60 to −40 mV. Dotted lines in the current traces represent 0 mA. A andB: recordings from an oocyte injected with water. C andD: recordings from an oocyte injected with α-2βγ-ENaC. E andF: recordings from the same oocyte asC andD after the extracellular application of 10 μM amiloride. G: amiloride-sensitive currents generated by subtraction ofE currents fromC currents.H:I-Vplot of amiloride-sensitive currents averaged from 18 α-1βγ and 23 α-2βγ oocytes from 4 separate preparations.

A representative example of the single-channel currents recorded from an oocyte expressing α-2βγ-ENaC is shown in Fig.9 A. Single-channel currents from α-1βγ-ENaC expressing oocytes were similar. Both displayed low conductance, slow kinetics, and variable open probabilities with no obvious differences in any of these parameters. TheI-Vrelationships for these single-channel recordings were linear between −80 and −20 mV, with a slope conductance of ∼7 pS for both groups (Fig. 9 B). This value for slope conductance is consistent with previous reports of ENaC single-channel recordings at room temperature where Li+ was used as the charge carrier (12).

Fig. 9.

Single-channel recordings from cell-attached patches in an oocyte 48 h after injection with ENaC cRNA. A: currents recorded with a holding potential of −40 mV (hyperpolarized) in an oocyte injected with α-2βγ-ENaC cRNA. Downward deflections represent channel openings and the inward flow of Li+. There were at least 2 channels in this patch. Brief inward deflections are from the endogenous stretch-activated cation channels that are present in all patches. B:I-Vrelationship of amiloride-sensitive current recorded for each voltage. Slope conductances for both forms were nearly identical (α-1βγ = 6.82 pS, n = 3, and α-2βγ = 7.15 pS, n = 5).


We have characterized four α-ENaC transcripts in various human tissues that arise by utilization of alternate first exons and by alternate splicing. Exon 1A and exon 2 are spliced together after removal of a 665-nt intron 1 to form the 5′ end of α-ENaC1 (Fig. 1). Interestingly, the exon 1B sequence is contained within intron 1 and is transcribed and processed unspliced into exon 2 to form α-ENaC2 (Fig. 1). Other variant mRNAs are formed by activation of 5′ splice sites within exon 1B, adding to the heterogeneity of RNA transcripts.

The analysis of the open reading frame of α-ENaC2 transcript reveals several in-frame ATGs upstream of the previously reported translation start codon. The most proximal start site adds a further 59 amino acids at the NH2 terminus, creating a variant protein (α-ENaC2) with a predicted molecular mass of 81.8 kDa compared with α-ENaC1 of 75.7 kDa. To assess the potential significance of this transcript, its expression was examined in several tissues (Fig. 3,AD). Although a cell line established from hCCD (26) failed to express α-ENaC, every other tissue examined showed evidence for mRNAs that encode α-ENaC1 protein (transcripts α-ENaC1, -3, -4) and α-ENaC2 protein (transcript α-ENaC2). Both forms were equally expressed in lung and colon, whereas α-ENaC2 was greater than the sum of the other transcripts in kidney. The relative proportion of the transcripts also varied within the examined cell lines; for example, α-ENaC1 was the most abundant transcript in IB3–1 and H441 cell lines (Fig. 3,A andB).

The topology for rat α-ENaC has been determined by several laboratories, and the model indicates that there are two membrane-spanning domains (M1 and M2), intracellular NH2 and COOH terminals, and a single extracellular loop (4, 28, 32). Hydropathy analyses (16) of α-ENaC1 and α-ENaC2 predict a similar topology (data not shown), suggesting that the two peptides differ in the length of the cytoplasmic NH2-terminal segment. Specifically, we can detect no additional hydrophobic segment that would suggest a third transmembrane domain nor can we detect a short hydrophobic core of apolar residues characteristic of a signal sequence (2) at the NH2 terminus of α-ENaC2.

Sequence analysis of the NH2-terminal cytoplasmic domain reveals two potential sites for myristoylation but no sites for phosphorylation or other posttranslational modifications. A motif PGLM[K/E]GNK[L/R]EEQ that is absent in bovine α-ENaC (Fig. 9 and Ref. 11) is represented once in α-ENaC1 and rat α-ENaC and repeated twice in human α-ENaC2 (Fig.10 A). The role of the NH2-terminal domain in the function of the Na+channel is not known. However, some possibilities are suggested by analogy with other integral membrane proteins. For example, heteromeric membrane proteins like the Shaker K+ channels and acetylcholine receptor complex contain structural elements in the NH2-terminal domain that regulate subunit assembly (17, 37). The NH2-terminal region of α-ENaC may also contain sites interacting with cytoskeletal elements or regulated by intracellular second messengers, and these interactions may be different for each of the NH2-terminal variants we have identified. The mechanisms by which molecules such as G proteins, actin filaments, and the protein kinases C and A interact with ENaC remain to be elucidated (1). Such interactions that regulate channel assembly or function are likely to be epithelial cell specific and may not be demonstrable in the Xenopus expression system.

Fig. 10.

A: comparison of the 5′ amino acid sequence of human, bovine, and rat α-ENaC. Identical residues are shaded. Sequence is shown up to the first transmembrane domain.B: transcription factor binding sites at the 5′ end of human α-ENaC gene are shown. Motifs on the antisense strand are shown below, whereas those on the sense strand are shown above the line depicting the gene. Transcription initiation sites are shown as bent arrows.

Although they may exist, 5′ variant transcripts have not been reported in other mammalian species. Alternate transcripts that have been described in mouse and rat appear to be splice variants that differ in their origin and in the nature of their translated products from what we have identified in humans. The rat splice variant, α-rENaCa, initially identified by RT-PCR in taste tissues, is also expressed as a minor transcript in kidney and lung. This splice variant introduces a premature stop codon before the second transmembrane domain of the translated protein and when expressed inXenopus oocytes does not generate an amiloride-sensitive Na+ current (18). The mouse α-ENaC gene is expressed as a 3.5-kb transcript in kidney, colon, and lung. In addition, a 1.2-kb transcript was identified only in the prenatal and early postnatal mouse colon and to the exclusion of the normal transcript (9). This transcript hybridizes to cDNA fragments from the 3′ portion of α-ENaC but not to 5′ fragments that encode the NH2 terminus, the first transmembrane domain, and the proximal portion of the extracellular domain. Although not directly tested, the translated product of this transcript is thus unlikely to reconstitute an Na+ channel. These data clearly show that there are species-specific variations in the nature of the α-ENaC transcripts expressed with likely widely different consequences.

The fact that human α-ENaC mRNA heterogeneity is determined in part by distinct transcription start sites raises questions regarding the nature of the promoter(s) for these two transcripts. Generally, promoters characterized by a TATA box are found within the first 25–50 nt of the transcription start site. TATA-less promoters are typically GC-rich in the 5′ flanking region and may have GC boxes with which TFIID and other TATA-binding proteins interact to initiate transcription (27). The γ-hENaC gene is TATA-less and contains two GC boxes immediately upstream of the transcription start site (36). In the α-ENaC gene, we have amplified and sequenced ∼500 nt of genomic sequence directly upstream of exon 1B (within exon 1A) and have not identified a TATA box or a GC box. Sequences directly upstream of exon 1A are GC rich and contain one motif with homology to a GC box (Fig.10 B). It is possible that sequences within exon 1A function as promoter elements for exon 1B, with sequences further upstream functioning as promoter elements for exon 1A, thus providing separate regulatory control. In this regard, several transcription factor binding motifs are identified within these regions, including AP2, Sp1, PEA3, and nuclear factor-κB motifs (Fig. 10 B). Alternatively, both transcripts may be directed by a common promoter upstream of exon 1A and transcriptional regulation provided by distincttrans-acting factors. Separately, translational control of each α-ENaC peptide may be provided by the length, composition, and secondary structure of its 5′-UTR (15). Whether the variant 5′-UTRs of α-ENaC impose constraints on translational efficiency is yet to be determined.

The biophysical characteristics of α-ENaC1 and α-ENaC2 when coexpressed with β- and γ-subunits in oocytes are very similar (Figs. 8 and 9). Both isoforms produce small-conductance channels with slow kinetics, typical of ENaC expressed in oocytes (5, 33, 38) and of naturally occurring ENaC channels in the mammalian collecting duct (25,39). Although the functional difference between these isoforms is not evident in the Xenopus expression system, we speculate that in its natural state in human epithelial cells these variants may confer distinct properties to the channel that could have an impact on tightly regulated Na+ transport.


We thank Raju Kucherlapati and Kate Montgomery for sharing prepublication nucleotide sequence information from a genomic clone that contained portions of the human α-ENaC gene. We also thank Michael Welsh for α-ENaC1 and β- and γ-ENaC cDNA, E. R. Liman for pGem-HE, Pam Zeitlin for IB3–8 cells, Prof. Pierre Ronco for hCCD cells, and Andrew Russo and Curt Sigmund for critical reading of the manuscript. We acknowledge the DNA synthesis and sequencing services provided by the University of Iowa DNA core facility.


  • Address for reprint requests: C. P. Thomas, Division of Nephrology, Dept. of Internal Medicine, Univ. of Iowa Hospitals and Clinics, 200 Hawkins Drive, Iowa City, IA 52242-1081.

  • This work was supported in part by grants from the Cystic Fibrosis Foundation Gene Therapy Center at the University of Iowa, by the March of Dimes Foundation, by an American Heart Association grant-in-aid, and by National Institute of Diabetes and Digestive and Kidney Diseases Grant DK-52617.

  • The nucleotide sequences reported in this paper have been submitted to GenBank with accession number U81961.

  • Portions of this work have been presented at the American Society of Nephrology annual meeting in 1996 and 1997.


View Abstract