INTRODUCTION

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

THE AMILORIDE-SENSITIVE EPITHELIAL Na⁺ channel (ENaC) (reviewed in Refs. 12, 26, 53) is expressed in the apical membrane of salt-absorbing epithelia found in the kidney, colon, and lung where it constitutes the rate-limiting step in active Na⁺ and fluid absorption. It thus plays a major role in the regulation of salt homeostasis and blood pressure as indicated by the existence of genetic hypertensive and hypotensive diseases associated with ENaC gene mutations, such as Liddle's syndrome (45) and pseudohypoaldosteronism type 1 (5). In the lung, ENaC is important in controlling the amount of liquid in the lung airspace. Targeted "knockout" of the gene for the -subunit of ENaC in mice gave rise to newborns that suffered respiratory distress, were unable to clear their fetal lung liquid, and died within 2 days of birth despite having morphologically normal lungs (17).

ENaC, which consists of three homologous subunits, ,  and , is part of a novel cation channel superfamily (12). ENaC subunits have been cloned from many species including human and rat. In heterologous expression systems (Xenopus oocytes), amiloride-sensitive Na⁺ currents can be measured in oocytes injected with rat -ENaC (-rENaC) mRNA; coexpression of all three subunits is required for maximal activity (3). In the oocyte expression system, both a heterotetrameric structure containing two -, one -, and one -subunit (11, 23) and a nonameric structure containing three copies of each subunit (9, 47) have been proposed. Regardless of which model represents the correct protein subunit stoichiometry, no more than a twofold excess of the - over - and -subunits is predicted. However, our laboratory has found that -ENaC mRNA is expressed at markedly greater levels than - and -subunit mRNAs in respiratory tract epithelia; we observed mRNA ratios of 20:4:1 in human (31) and 50:1:5 in mouse (19) nasal turbinate by quantitative competitive reverse transcription-polymerase chain reaction (RT-PCR). This agrees with relative estimates by Northern analysis and in situ hybridization (2, 10, 27, 50) and suggests that -ENaC mRNA, under normal conditions, may be less efficiently translated than - and -subunit mRNAs. -ENaC has been shown to be capable of expressing functional channels in the absence of the - and -subunits when this subunit is artificially expressed alone (4, 21); however, the heteromultimer appears to be the preferred form of assembly, unaltered even by injection of a 100-fold excess quantity of message encoding a single subunit (11). Thus it is unlikely that formation of homomultimeric channels accounts for the excess -ENaC mRNA expression.

Several observations suggest a possible role for the 5'-untranslated region (UTR) in the translational regulation of -ENaC. First, Smith et al. (46) have recently demonstrated that -ENaC mRNA expression is detectable during the earliest stages of fetal lung development, at a time when the lung is a secretory and not an absorptive organ. Second, it has been shown that multiple -ENaC mRNAs differing in the 5'-UTR are expressed in human tissue (51). Both human and rat -ENaC 5'-UTRs are unusually long [up to 750 nucleotides (nt); most 5'-UTRs are 50-100 nt] and contain GC-rich regions, indicating a potential secondary structure (30). Finally, a study by Otulakowski et al. (30) revealed a decrease in reporter gene activity in transfected A549 cells when -rENaC promoter constructs were extended to include an additional 133 bp of the 5'-UTR. These observations suggest the presence of a negative regulatory element. Due to its location, such an element could be functionally active during either transcription or translation. Structural features in mRNA 5'-UTRs that may influence translational efficiency are secondary structure, occurrence of upstream AUG start codons, structural motifs recognized by specific binding proteins, and internal ribosome entry sites (6, 16, 24, 25, 32, 37, 49).

To study the translational regulation of -rENaC, we developed a quantitative competitive RT-PCR assay to assess reporter mRNA levels in transiently transfected cells and examined the effect of different regions of 5'-UTR on reporter gene mRNA and protein expression. Mutational analysis was used to investigate the potential influence of an upstream, out-of-frame initiation codon in the -rENaC 5'-UTR.

	MATERIALS AND METHODS

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

Promoter-reporter constructs. Cloned genomic DNA fragments containing portions of the putative -rENaC promoter were inserted upstream of the secreted alkaline phosphatase (SEAP) gene in the promoterless expression vector pSEAP2-basic (Clontech, Palo Alto, CA). A combination of naturally occurring restriction sites and fragments generated by exonuclease digestion was used to assemble the reporter constructs. The sequence ATG at +473-475 bp was mutated to TTG by sequential PCR steps. The resulting PCR fragment was cloned into pBluescript II KS(); the sequence was confirmed and then substituted for the wild-type sequence in the "dSEAP2" construct series to generate the "dmOSEAP2" constructs.

Cell culture. Madin-Darby canine kidney (MDCK) epithelial cells were maintained in Dulbecco's modified Eagle's medium (DMEM) with 10% fetal bovine serum (FBS). Mouse kidney cortical collecting duct epithelial cells (M1) were maintained in DMEM plus Ham's F-12 medium (1:1) with 5% FBS and 5 µM dexamethasone. Human lung epithelial cells (NCI-H441) were maintained in RPMI 1640 medium with 10% FBS. Mouse distal lung epithelial cells (MLE-15) were maintained in HITES medium with 2% FBS. HITES medium consisted of RPMI 1640 medium supplemented with 5 µg/ml of insulin, 10 µg/ml of transferrin, 30 nM sodium selenite, 10 nM hydrocortisone, 10 nM -estradiol, 10 mM HEPES, and 2 mM L-glutamine. MLE-15 cells were a gift from Dr. J. Whitsett (Children's Hospital Medical Center, Cincinnati, OH); all other cell lines were obtained from the American Type Culture Collection (Manassas, VA). All lines were maintained in the presence of 100 U/ml of penicillin G and 100 µg/ml of streptomycin sulfate.

DNA sequencing. Manual sequencing was carried out with the Pharmacia T7 sequencing kit (Amersham Pharmacia Biotech, Baie d'Urfé, PQ) and ³⁵S-dATP. Automated sequencing reactions were performed with a Thermo Sequenase fluorescent-labeled primer cycle sequencing kit with 7-deaza-dGTP (Amersham) and analyzed on a LICOR DNA sequencer model 4000L equipped with Base ImagIR software (LI-COR, Lincoln, NE).

Transfections. Cells were plated in six-well tissue culture dishes at 2.5 × 10⁵ cells/well for MDCK cells, 4 × 10⁵ cells/well for M1 cells, and 8 × 10⁵ cells/well for MLE-15 and H441 cells 18 h before transfection. Cells were cotransfected with pSEAP2 constructs and a Rous sarcoma virus (RSV)-driven -galactosidase (gal)-expressing plasmid (RSVgal) as an internal control for transfection efficiency. Transfections were carried out with LipofectAMINE reagent or LipofectAMINE Plus for H441 cells (Life Technologies) according to the manufacturer's recommendations. Medium was collected 48 h posttransfection for analysis of SEAP activity with a Phospha-Light chemiluminescence kit (PerkinElmer Applied Biosystems Division, Mississauga, ON). -Galactosidase activity was determined via a colorimetric assay with o-nitrophenyl--D-galactopyranoside as described previously (13) on cell extracts prepared by lysis of cultured cells in 1% Triton X-100-0.25 M Tris · HCl, pH 7.8.

Quantitative RT-PCR. A competitive internal control, RSVSEAP, for quantitative RT (QRT)-PCR in transiently transfected cells was created by deleting a 218-bp NarI fragment from the SEAP coding region of an RSV promoter-driven pSEAP construct. ENaC promoter-driven SEAP2 constructs were cotransfected with RSVSEAP at a ratio of 100:1 as described in Transfections. All transfections were carried out in triplicate. Forty-eight hours posttransfection, the cells were harvested by direct addition of lysis buffer (RLT buffer, QIAGEN, Santa Clarita, CA), and total RNA was purified with the RNeasy Mini Kit (QIAGEN). The resulting RNA was treated with RNase-free DNase and again purified on an RNeasy column. RNA was quantitated by ultraviolet spectrophotometry.

In vitro transcription and translation. Templates for in vitro transcription were constructed in pBluescript II KS(). Template L contains the -rENaC cDNA sequence from +53 nt through the poly(A)⁺ tail; S contains the -rENaC full-length cDNA sequences from +420 nt through the poly(A)⁺ tail [-rENaC open reading frame (ORF) initiates at +517 nt]. SM and LM contain the corresponding sequences, with mutation of the upstream ATG at +473-475 nt to TTG as described for the promoter-reporter constructs. Templates were linearized 3' to the poly(A)⁺ tail with XhoI. Capped mRNAs were generated with T7 RNA polymerase with mMESSAGE mMACHINE (Ambion) and quantitated by ultraviolet spectrophotometry. Aliquots were examined by electrophoresis on formaldehyde-containing agarose gels to verify length and confirm that mRNA was not degraded. We used 0.8 pmol of each capped mRNA to prime in vitro translation reactions containing [³⁵S]methionine (1,000 Ci/mmol; ICN, Cosa Mesa, CA) with a ReticLysate IVT kit (Ambion). The resulting proteins were analyzed on Laemmli SDS-polyacrylamide gels, subjected to fluorography with Amplify (Amersham Pharmacia Biotech), and exposed to Kodak X-OMAT film. For quantitation of synthesized protein, the dried gels were imaged with a Molecular Dynamics STORM 840 phosphorimager equipped with ImageQuant Software (Amersham Pharmacia Biotech).

Statistical analysis. All data are presented as means ± SE, and significances were calculated with one-way ANOVA. P < 0.05 was considered to be significant. Statistical analysis was carried out with GraphPad InStat version 3.01 (GraphPad Software, San Diego CA).

	RESULTS

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

-rENaC promoter activity in lung and kidney epithelial cell lines. In the present study, we created a series of promoter-reporter constructs (Fig. 1) extending up to 4,539 bp upstream and 499 bp downstream of the -rENaC +1 transcription start site. We characterized the basal and steroid hormone-stimulated activity of these constructs in representative kidney (MDCK and M1) and lung (MLE-15 and H441) epithelial cell lines (Fig. 2).

View larger version (20K): [in this window] [in a new window]   Fig. 1.   Rat epithelial Na+ channel -subunit (-rENaC) promoter-reporter constructs. Top line, structural organization of the -rENaC promoter (not to scale). Note the presence of a glucocorticoid responsive element (GRE) between XhoI and HindIII (30). Labels at right, promoter-secreted alkaline phosphatase (SEAP) 2 constructs representing genomic DNA fragments corresponding to the region of the -rENaC promoter directly above; nos. at right, 3'-end of the fragment; nos. at left, nucleotide number of the 5'-end of the fragment based on +1 being the exon 1A transcriptional start site identified in a previous publication by Otulakowski et al. (30). All constructs terminate upstream of the authentic -rENaC open reading frame. , Mutation of upstream ATG to TTG as described in MATERIALS AND METHODS.

View larger version (48K): [in this window] [in a new window]   Fig. 2.   Activity of -rENaC promoter-reporter constructs in kidney and lung cell lines. -rENaC promoter-SEAP2 constructs (see Fig. 1) were transiently transfected into Madin-Darby canine kidney (MDCK; A), M1 (B), MLE-15 (C), and H441 (D) cell lines. Standard medium for M1 cells contained 5 µM dexamethasone. Standard medium for MLE-15 cells contained 10 nM hydrocortisone and 10 nM -estradiol. SEAP activity in each well was first normalized to -galactosidase activity to correct for transfection efficiency and is expressed relative to the promoterless SEAP2-basic vector transfected in parallel wells. Data represent means of triplicate wells in a representative transient transfection experiment. All transient transfection experiments were carried out at least 3 times in each cell line. Differences between selected pairs of constructs were calculated with ANOVA followed by Bonferroni posttest. * P < 0.01 compared with SEAP2(+367) construct from same series. ¶ P < 0.05 compared with PvuSEAP2(+211) construct from same series.

Reporter mRNA levels in transiently transfected cells. Because the regulatory regions identified between +53 and +499 nt are located downstream from the transcriptional start site, it is impossible to tell from SEAP enzyme activity alone whether they are modifying transcription initiation, mRNA stability, or mRNA translational efficiency. To begin to clarify the process through which these elements act, we designed a control plasmid to allow quantitative RT-PCR analysis of reporter gene mRNA levels in transfected cells. RSVSEAP contains a small deletion in the region encoding the COOH terminus of SEAP protein; primers flanking this sequence were designed such that RT-PCR amplification products of full-length SEAP and SEAP mRNAs can be distinguished on the basis of size. RSVSEAP was cotransfected with 548uSEAP2, 548PvuSEAP2, 548SEAP2, and 548dSEAP2 such that it would serve both as a control for transfection efficiency and as an internal, competitive control for QRT-PCR.

View larger version (22K): [in this window] [in a new window]   Fig. 3.   Comparison of -rENaC-driven reporter gene mRNA and enzyme activity levels in M1 cells. A: M1 cells were transiently transfected with a series of fragments of the -rENaC promoter, extending 552 bp upstream from the transcription start site and terminating at various positions within the 5'-untranslated region (UTR). B: cells were cotransfected with ENaC promoter-driven SEAP and Rous sarcoma virus (RSV)SEAP. SEAP mRNA levels are expressed relative to SEAP mRNA and represent the average of triplicate wells in 2 independent transfections. C: cells were cotransfected with ENaC promoter-driven SEAP and RSV-galactosidase (gal). SEAP enzyme activity was normalized to -galactosidase activity to correct for transfection efficiency and is expressed relative to the promoterless SEAP2-basic vector transfected in parallel wells. Data represent means of triplicate wells. Differences between pairs of constructs were calculated with ANOVA followed by Tukey-Kramer posttest.

View larger version (18K): [in this window] [in a new window]   Fig. 4.   Comparison of -rENaC-driven reporter gene mRNA and enzyme activity levels in MLE-15 cells. A: MLE-15 cells were transiently transfected with a series of fragments of the -rENaC promoter, extending 552 bp upstream from the transcription start site and terminating at various positions within the 5'-UTR. B: cells were cotransfected with ENaC promoter-driven SEAP and RSVSEAP. SEAP mRNA levels are expressed relative to SEAP mRNA and represent the average of triplicate wells in 2 independent transfections. C: cells were cotransfected with ENaC promoter-driven SEAP and RSVgal. SEAP enzyme activity was normalized to -galactosidase activity to correct for transfection efficiency and is expressed relative to the promoterless SEAP2-basic vector transfected in parallel wells. Data represent means of triplicate wells. Differences between pairs of constructs were calculated with ANOVA followed by Tukey-Kramer posttest.

Effect of the +367- to +499-nt region in the context of uSEAP2 reporters. The decrease in reporter gene expression seen when extending the -rENaC 5'-UTR from +367 to +499 nt may simply be due to increasing the length of the 5'-UTR past a certain threshold. To investigate this, we deleted bases from +54 to +366 nt, effectively moving the +367- to +499-nt region directly adjacent to the 3'-end of the uSEAP2 constructs (Fig. 1, dSEAP2 constructs). As shown in Fig. 5, the dSEAP2 constructs exhibited decreased reporter gene activity compared with their corresponding uSEAP2 reporters, although this inhibition was approximately half as great as that seen with the full-length dSEAP2 constructs. The decrease failed to reach significance in the 1051-series constructs in the kidney cell lines MDCK and M1 and in the 5000-series constructs in the lung cell lines MLE-15 and H441. The effect did not appear to be modified under conditions of steroid induction of the reporter gene in M1 or MLE-15 cells (Fig. 5, B and C).

View larger version (40K): [in this window] [in a new window]   Fig. 5.   Effect of +367- to +499-nucleotide (nt) region directly abutted to uSEAP2 promoter constructs. -rENaC promoter-SEAP2 constructs (see Fig. 1) were transiently transfected into MDCK (A), M1 (B), MLE-15 (C), and H441 (D) cell lines. Standard medium for M1 cells contained 5 µM dexamethasone. Standard medium for MLE-15 cells contained 10 nM hydrocortisone and 10 nM -estradiol. SEAP activity in each well was first normalized to -galactosidase activity to correct for transfection efficiency and is expressed relative to the promoterless SEAP2-basic vector transfected in parallel wells. Data are means ± SE of 6 wells. Differences between constructs were calculated with ANOVA followed by Tukey-Kramer posttest. Significantly different compared with uSEAP2 in same construct series: * P < 0.05;  P < 0.01;  P < 0.001.

Effect of upstream ORF on mRNA translation. The sequence of the -rENaC 5'-UTR is indicated in Fig. 6. Adjacent AUG codons initiating the -rENaC ORF are underlined at +517 nt. Note the existence of an AUG codon upstream and out-of-frame initiation codon with the -rENaC protein-coding region, initiating an upstream ORF (uORF) at +473 nt. Such upstream initiation codons have frequently been reported to exert negative effects on translation initiation at the correct, downstream initiation codon (14). None of the AUG codons exist in an optimal Kozak consensus sequence (RCCAUGG, with R at 3 nt and G at +4 nt being most influential) (25). Thus we hypothesized that the initiation of protein synthesis at the upstream AUG was inhibiting translation of the SEAP reporter (with which it is also out of frame in our constructs) in "-dSEAP2" constructs. We therefore mutated the upstream ATG sequence in our SEAP reporter constructs to TTG, thus eliminating the initiation codon of the uORF. The results of this mutation are shown in Fig. 7 (dmOSEAP2 constructs). Surprisingly, removal of the uORF had little effect, restoring activity convincingly only in kidney-derived cell lines (Fig. 7, A and B) transfected with constructs containing minimal amounts of the upstream promoter (297dmOSEAP2 and 548dmOSEAP2). In lung-derived epithelial cell lines (Fig. 7, C and D), no significant increase in reporter gene activity resulted from the uORF mutation except in MLE-15 cells transfected with 297dmOSEAP2, where a partial restoration of activity was seen.

View larger version (35K): [in this window] [in a new window]   Fig. 6.   Nucleotide sequence of the 5'-end of -rENaC mRNA. Vertical lines, end points of promoter-reporter constructs with nucleotide position above the sequence; underlined letters, methionine codons (AUG). Note the upstream open reading frame (uORF) that initiates at +473 nt and encodes a 23-amino acid peptide, overlapping the authentic -rENaC open reading frame that initiates at +517 nt.

View larger version (47K): [in this window] [in a new window]   Fig. 7.   Effect on reporter gene enzyme activity with elimination of the -rENaC uORF by mutagenesis of ATG to TTG. -rENaC promoter-SEAP2 constructs (see Fig. 1) were transiently transfected into MDCK (A), M1 (B), MLE-15 (C), and H441 (D) cell lines. Standard medium for M1 cells contained 5 µM dexamethasone. Standard medium for MLE-15 cells contained 10 nM hydrocortisone and 10 nM -estradiol. SEAP activity in each well was first normalized to -galactosidase activity to correct for transfection efficiency and is expressed relative to the promoterless SEAP2-basic vector transfected in parallel wells. Data are means ± SE of triplicate wells. Significantly different compared with dSEAP2 in same construct series: * P < 0.01;  P < 0.01;  P < 0.05.

Effects of -rENaC 5'-UTR on translational efficiency in vitro. To confirm the effects of the -rENaC 5'-UTR on translation efficiency in conjunction with the -rENaC protein-coding region, as opposed to heterologous reporter constructs, we created four templates suitable for in vitro transcription and translation of -rENaC protein (Fig. 8A). Equimolar amounts of capped -rENaC transcripts containing either 97 (S and SM) or 464 (L and LM) nt of the -rENaC 5'-UTR were used to direct protein synthesis in a reticulocyte lysate system. Radiolabeled protein products were analyzed by SDS-PAGE as demonstrated in Fig. 8B. All four templates gave rise to a specific protein product at 72 kDa, consistent with the expected size of in vitro translated -rENaC. Considerably more protein was produced from RNA transcripts containing the truncated 5'-UTR (compare protein produced from S with L). To confirm these observations, -rENaC protein was quantitated from three independent transcription or translation experiments (Fig. 8C). As in transient transfections, mutation of the upstream AUG resulted in only a small increase in translational efficiency, which was significant only in the context of the truncated UTR.

View larger version (26K): [in this window] [in a new window]   Fig. 8.   In vitro transcription/translation of -rENaC. A: templates for in vitro transcription of -rENaC mRNAs were assembled in pBluescript KS() such that sense RNA could be synthesized with T7 RNA polymerase. Hatched box, open reading frame of -rENaC cDNA that initiates at +517 nt. S and SM contain 97 nt of -rENaC 5'-UTR; L and LM contain 464 nt of -rENaC 5'-UTR. ATG in S and L indicates upstream ATG. , ATG mutated to TTG in SM and LM. -rENaC cDNAs were inserted between SacII and XbaI such that all in vitro synthesized transcripts contained the same minimal (19 nt) amount of vector-derived sequence at the extreme 5'-end. B: representative SDS-PAGE analysis of [35S]methionine-labeled protein products from in vitro translation of 0.8 pmol of -rENaC cRNA. In vitro translation reaction containing 1 of the 4 -rENaC cRNAs described in A was electrophoresed on a 7.5% SDS-polyacrylamide gel, dried, and exposed to X-ray film overnight. Major band at ~72 kDa corresponds to predicted size of unglycosylated -rENaC. Benchmark-prestained protein ladder (Life Technologies) was used to estimate protein size. C: relative amounts of -rENaC protein synthesized in vitro from equimolar amounts of cRNAs differing in the 5'-UTR region. [35S]methionine incorporated into the 72-kDa -rENaC band was quantitated with a phosphorimager and is expressed relative to protein produced from L cRNA. Values are means ± SE of 3 independent in vitro transcription or translation experiments. Truncation of the UTR (S) significantly increased translational efficiency. * P < 0.01 compared with L. Mutation of the upstream AUG (SM) increased translational efficiency significantly in the context of the short UTR. ** P < 0.01 compared with S. In the context of the long UTR (LM vs. L), the increase seen was reproducible but did not reach significance.

	DISCUSSION

TOP ABSTRACT INTRODUCTION MATERIALS AND METHODS RESULTS DISCUSSION REFERENCES

We have presented evidence that suggests that the 5'-UTR of -rENaC mRNA plays a role in regulating ENaC expression. The overall inhibitory effect of the long -rENaC 5'-UTR is clear in our in vitro translation experiments. Transient transfection experiments indicated the existence of specific positive and negative translational regulatory elements within the 5'-UTR. Specifically, the region located between +53 and +211 nt increased the expression of a linked reporter gene in a variety of cell lines without increasing the amount of mRNA expressed. A second region of interest, located between +367 and +499 nt, severely inhibited translation of a linked reporter gene. In general, one would expect expression to decrease with increasing length of the 5'-UTR. The inhibitory effect of the +367- to +499-nt region was tested in the context of a shorter -rENaC 5'-UTR, i.e., in the dSEAP2 constructs (Fig. 5). Interpretation of the data from the dSEAP2 constructs is complicated because they delete the putative positive regulatory region from +54 to +211 nt. However, it is clear from Fig. 2A that the +54- to +211-nt region has no effect in MDCK cells. In Fig. 2A, expression was not affected by increasing the amount of -rENaC 5'-UTR from 53 nt (uSEAP2) to 211 nt (PvuSEAP2) or even further to 367 nt (SEAP2) in MDCK cells. Yet the dSEAP2 constructs, with only 186 nt of -rENaC 5'-UTR, generated significantly less reporter gene expression than the uSEAP2 constructs (Fig. 5A). Similar results were seen in other cell lines (Fig. 5, B-D). The inhibition was only about half as much as was seen with the intact 499-nt UTR region (dSEAP2 constructs), suggesting that either total length or possibly secondary structure interactions between the +367- to +499-nt region and the deleted sequences also play a role. Overall, the data suggest that the decrease in translation efficiency seen when extending the 5'-UTR from +367 to +499 nt is not solely due to increasing the length of the 5'-UTR.

In a previous publication, Otulakowski et al. (30) reported the existence of an additional transcription start site in this region, at position +455, detectable by conventional 5'-rapid amplification of cDNA ends and primer extension. As seen in Figs. 3 and 4, this region does not increase transcription (SEAP mRNA levels) in transiently transfected M1 or MLE15 cells, and we have subsequently been unable to demonstrate transcripts initiating at this site with a RNase protection assay (Otulakowski and O'Brodovich, unpublished data). A GC-rich sequence immediately upstream of +455 nt may have impeded RT through this region in 5'-rapid amplification of cDNA ends and primer extension.

The upstream, out-of-frame initiation codon in this region seems to mediate only a minor component of the translational repression as demonstrated by the effects of point mutation of the upstream start codon in both transfected cells and in vitro translation systems. Our data thus suggest that an additional level of translational repression is conferred by this region, which may be specific to lung, because point mutation of the upstream AUG restored reporter gene activity to a lesser extent in MLE-15 or H441 cells than in MDCK and M1 cells. We do not presently understand the mechanism by which mutation of the upstream AUG restored activity only in constructs containing a "minimal" promoter, but the observation suggests that translational inhibition mediated by the 5'-UTR may be influenced by sequences in the 5'-flanking region and may be subject to cell- or tissue-specific effects. As discussed above, Fig. 5 suggests the influence of the region between +367 and +499 nt may also be dependent on promoter context and cell line in transient transfections.

The importance of control of eukaryotic gene expression at the level of translation is now well recognized (see Refs. 14, 40, 41 for recent reviews). Specific sequences within mRNA UTRs can modulate translational efficiency in a tissue- and developmental stage-specific manner. Most translational effects are exerted at the level of protein chain initiation with a combination of specific sequence features in the 5'- or 3'-UTRs of mRNAs and/or physiological regulation of initiation factor activities.

Experimental studies with both in vivo and in vitro systems clearly demonstrate that mRNAs with a high potential to form a stable secondary structure in the 5'-UTR tend to be translated inefficiently (7, 24, 25, 36). Tissue-specific use of alternative promoters to produce transcripts with different 5'-UTRs having varying amounts of secondary structure has been reported, for example, for the human -folate receptor (49) and insulin-like growth factor II (29). Mechanisms permitting the regulated release of this translational repression in response to specific signals are not completely understood. One possibility is the activation of general initiation factors involved in mRNA binding (e.g., by phosphorylation) (1). Alternative use of a shorter, more efficiently translated leader sequence may also occur in response to some developmental (6, 35) or environmental (6, 54) cue. In this context, it is interesting to note that human -ENaC mRNAs differing in the 5'-UTR through use of alternative transcriptional start sites have been reported (30, 51).

The presence of upstream AUG codons in the 5'-UTR can also affect the efficiency of translation initiation of the main ORF. In >90% of mRNAs, the AUG initiating the main ORF is the first AUG encountered when scanning from the 5'-end (25). Additional upstream AUGs tend to severely decrease the efficiency of initiation at the correct AUG, but these effects can be tissue specific (18, 20, 39).

Structural motifs in mRNA molecules can provide sites for the binding of specific proteins. In the case of the 5'-UTR, such binding can impede initiation by stabilizing an element of the secondary structure not itself strong enough to be inhibitory (15, 16). Most recently, the identification of internal ribosome entry sequence (IRES) elements in the 5'-UTR in a variety of capped eukaryotic mRNAs has led to the discovery of an alternative mechanism of translation initiation that can function as an alternative to "scanning" the 5'-UTR by the 43S preinitiation complex. Usage of IRES-mediated translational initiation can avoid the inhibitory effects of the secondary structure at the 5'-end of the mRNA and may be physiologically regulated (32).

The observation that the mRNA encoding -ENaC in both humans and rats contains an unusually long 5'-UTR burdened with GC-rich regions indicates immediately that this mRNA is probably inefficiently translated and may explain the relative abundance of -ENaC mRNA compared with - and -ENaC mRNAs in native tissue. Furthermore, the observation of tissue (51) and developmental (30) regulation of the expression of alternative 5'-UTRs suggests that translational regulation may play a role in controlling ENaC synthesis. Our data now indicate that the -rENaC 5'-UTR contains sequence elements exerting both positive and negative effects on translation of the downstream ORF. The mechanism by which these elements function is not yet elucidated, and a number of possibilities exist based on the examples given above. The positive element 53- to 211-bp region may contain sequences that bind regulatory proteins, influence secondary structure, or contain an IRES, allowing recruitment of ribosomes downstream of secondary structure at the extreme 5'-end of the mRNA. The negative effects of the 367- to 499-bp sequence appear to be only minimally mediated via the upstream AUG, suggesting that secondary structure or specific regulatory proteins may be involved in this region also. The translational efficiency of mRNAs burdened with a secondary structure in the 5'-UTR is susceptible to regulation via cell signaling cascades that modify assembly of the eukaryotic translation initiation complex eIF4F, which is responsible for unwinding the secondary structure.

The number of functional ENaCs on the cell surface is small and tightly regulated. Previous studies (34, 48, 52) have shown that most of the channel proteins remain in a core-glycosylated form in pre-Golgi compartments and are subsequently degraded before ever reaching the plasma membrane. This has led to the suggestion that the limiting steps for the accumulation of functional channels in the plasma membrane are the processes of assembly and maturation of the subunits in the endoplasmic reticulum (52). However, these studies have been carried out in heterologous expression systems (Xenopus oocytes and transfected mammalian cells) and used -ENaC cDNAs that lack the extended 5'-UTRs concurrently reported in both rats (30) and humans (51). Consequently, it is possible that much less -ENaC protein would be produced from the endogenous mRNAs than was seen in the heterologous systems, possibly eliminating the buildup of excess unassembled protein in the endoplasmic reticulum. Little information is available on the biosynthesis and trafficking of endogenous ENaC; however, May et al. (28) have examined the rate of synthesis of -, -, and -ENaC subunit proteins in A6 cells. After aldosterone treatment, an increase in the rate of synthesis of -ENaC protein preceded the increase in -ENaC mRNA. This led the authors to suggest that synthesis of the -subunit was a limiting factor in channel assembly and that aldosterone may selectively increase translation of existing -ENaC mRNAs.

More recently, the serum- and glucocorticoid-dependent kinase sgk has been identified as an early, aldosterone-induced gene that is a likely candidate to mediate the early increase in ENaC activity in response to aldosterone. Although aldosterone induces phosphorylation of the carboxy termini of - and -ENaC subunits when the channel is expressed in MDCK cells (44), sgk appears not to phosphorylate ENaC directly (8) nor to affect its degradation (43). Although many scenarios by which a kinase signaling cascade can lead to an increase in ENaC cell surface expression are possible, one hypothesis consistent with our observations and those of May et al. (28) would be a kinase-mediated increase in the translation initiation of ENaC mRNAs. Specifically, the eukaryotic translation initiation factor eIF4E, which has been suggested to be limiting for translation in vivo, is serine phosphorylated in response to growth factors, hormones, and mitogens (reviewed in Ref. 14). Phosphorylation of eIF4E enhances its activity, resulting in increased delivery of eIF4F (an RNA helicase) to the 5'-UTR, disrupting the secondary structure. In fact, in mammalian cells, overexpression of eIF4E results in a more efficient translation of reporter mRNAs containing structured 5'-UTRs (22) and has been shown to enhance translation of a number of mRNAs such as cyclin D1 (38) and ornithine decarboxylase with structure-rich 5'-UTRs (42). Thus it is plausible that aldosterone induces a rapid increase in ENaC protein synthesis by specifically increasing the efficiency of translation initiation of -ENaC.

Understanding of mRNA cellular biology, such as regulation of splicing, subcellular trafficking, translation efficiency, and stability, has generally lagged behind our knowledge of functional mechanisms involving DNA and proteins. However, the important role that UTRs of eukaryotic mRNAs play in gene expression is becoming more widely recognized and has led to the recent creation of a specialized database of 5'- and 3'-UTR sequences and functional elements (33). -ENaC mRNAs from both rats and humans are clearly burdened with unusually long 5'-UTRs. Our study using recombinant DNA constructs indicates that the 5'-UTR contains regions capable of regulating ENaC expression at the translational level. An investigation of the translational efficiency of endogenous -rENaC mRNA via polysome gradient analysis will be required to prove the physiological significance of these effects. Regulation at the translational level, probably in combination with ENaC assembly and trafficking, would enable ENaC-expressing tissues to respond more quickly to environmental signals than de novo transcription would permit.