The Gene Expression of the Amiloride-sensitive Epithelial Sodium Channel α-Subunit Is Regulated by Antagonistic Effects between Glucocorticoid Hormone and Ras Pathways in Salivary Epithelial Cells*

The functional expression of the amiloride-sensitive epithelial sodium channel (ENaC) in select epithelia is critical for maintaining electrolyte and fluid homeostasis. Although ENaC activity is strictly dependent upon its α-subunit expression, little is known about the molecular mechanisms by which cells modulate α-ENaC gene expression. Previously, we have shown that salivary α-ENaC expression is transcriptionally repressed by the activation of Raf/extracellular signal-regulated protein kinase pathway. Here, this work further investigates the molecular mechanism(s) by which α-ENaCexpression is regulated in salivary epithelial Pa-4 cells. A region located between −1.5 and −1.0 kilobase pairs of theα-ENaC 5′-flanking region is demonstrated to be indispensable for the maximal and Ras-repressible reporter expression. Deletional analyses using heterologous promoter constructs reveal that a DNA sequence between −1355 and −1269 base pairs functions as an enhancer conferring the high level of expression on reporter constructs, and this induction effect is inhibited by Ras pathway activation. Mutational analyses indicate that full induction and Ras-mediated repression require a glucocorticoid response element (GRE) located between −1323 and −1309 base pairs. The identifiedα-ENaC GRE encompassing sequence (−1334/−1306) is sufficient to confer glucocorticoid receptor/dexamethasone-dependent and Ras-repressible expression on both heterologous and homologous promoters. This report demon- strates for the first time that the cross-talk between glucocorticoid receptor and Ras/extracellular signal-regulated protein kinase signaling pathways results in an antagonistic effect at the transcriptional level to modulate α-ENaC expression through the identified GRE. In summary, this study presents a mechanism by which α-ENaC expression is regulated in salivary epithelial cells.

Sodium balance is important for the maintenance of body electrolyte, extracellular volume, and blood pressure. The amiloride-sensitive epithelial sodium channel (ENaC) 1 is expressed in airway epithelium, distal segments of kidney tubule, skin, bladder, colon, and sweat and salivary glands. In addition, ENaC is a member of the expanded degenerin/ENaC superfamily (1) and consists of at least three homologous subunits, ␣, ␤, and ␥ (2)(3)(4)(5).
Compelling functional and biochemical evidence suggests that all three subunits form a heteromultimeric complex and contribute to the optimal epithelial sodium reabsorption activity of the amiloride-sensitive sodium channel (3,6,7). However, when expressed individually in a Xenopus oocyte system, only the ␣-subunit can produce an amiloride-sensitive current. Both ␤and ␥-subunits are not functional on their own, but augment the channel activity of ␣-ENaC (3,8). Thus, the expression of a functionally active sodium channel is dependent upon the presence of the ␣-ENaC subunit. Indeed, different phenotypes were observed in ␣-ENaC(Ϫ/Ϫ) and ␥-ENaC(Ϫ/Ϫ) transgenic mice that were generated by homologous gene targeting. For example, 50% of ␣-ENaC(Ϫ/Ϫ) mice develop respiratory distress and die within 24 h after birth from an inability to clear the lung fluid (9), whereas none of the ␥-ENaC(Ϫ/Ϫ) mice die within the same time period (10).
Although the composition, structure, and activity of ENaC have been well studied, the molecular basis for the regulatory mechanisms underlying the gene expression of ENaC subunits remains unclear. Results from studies by us and others have suggested that the transcriptional control of ENaC gene expression occurs in a subunit-and tissue-specific manner (Ref. 11; reviewed in Refs. 12 and 13). For instance, the transcription of ␤and ␥-ENaC genes is up-regulated by the steroid hormone aldosterone in the colon, whereas the ␣-ENaC mRNA is constitutively expressed. In contrast, dexamethasone treatment is reported to up-regulate the steady-state mRNA levels of all three subunits in the colon and fetal lung. Moreover, vasopressin increases only ␤and ␥-ENaC mRNA levels without altering ␣-ENaC mRNA level in RCCD 1 rat cortical collecting duct cell line (14). We have previously demonstrated that TPA treatment represses the ␣-ENaC mRNA level (11). Taken together, these studies also outline a putative pathway(s) by which extracellular signals regulate cell-specific ENaC expression. Therefore, studies that more precisely define the interaction between the cellular transcriptional machinery and signaling pathways are needed to provide a framework for understanding the mechanisms underlying the regulation of ENaC expression and the resultant electrolyte homeostasis.
Nuclear hormone receptors modulate gene transcription, upon binding of their cognate ligands, by activation as well as repression (15). One of the best studied members of the nuclear hormone receptor superfamily is the glucocorticoid receptor (GR), which plays an important role in physiology and developmental biology. Transactivation by GR requires binding of receptor homodimers to specific partial palindromic sequences in the cis-regulatory region of target genes, namely glucocorticoid response elements (GREs). In recent years, it has become apparent that a variety of other factors may have a profound effect on the transactivation potential of GR. For example, the phosphorylation status of GR (16) and the formation of composite or complex GREs (17,18) can affect classical GR transactivation activity. Thus, although the GR is expressed in virtually all mammalian cell types, it is possible that the expression of a distinct set of GR-responsive genes is cell context-specific and may be modulated by other signaling pathway(s).
Ras proteins are important signaling intermediates that convey signals initiated at the cell surface to various effector pathways in the cytoplasm. Ras exerts effects on cell transformation and proliferation, cytoskeletal structure, differentiation, and apoptosis. These changes may be mediated by multiple effectors, including Raf-1, Ral-GDS, phosphatidylinositol (PI) 3-kinase, and other Ras-binding proteins (reviewed in Ref. 19), often at the transcriptional level. Even though both Ras effector pathways and gene regulation are characterized in molecular detail, the interface between canonical Ras signaling and transcriptional repression remains poorly described. Clearly, there are mechanisms, when appropriately cued, whereby signal transduction cascades operating via Ras activation would negatively modulate transcriptional events in the nucleus, ultimately leading to transcriptional repression.
The human ␣-ENaC gene has recently been cloned and characterized (20). A rat genomic DNA fragment harboring the ␣-ENaC promoter, first exon, partial first intron, and 5Ј-flanking region has been isolated and used very recently to investigate the ␣-ENaC gene regulation by us (11). Since more than one transcription initiation site has been identified in both human and rat ␣-ENaC genes accounting for the polymorphism of ␣-ENaC transcript size (21), 2 we have adopted a numbering convention that assigns the number ϩ1 to the first base of the translation start codon. Previously, we have shown that the regulatory elements in a 1.4-kb (Ϫ1573 to Ϫ154 bp relative to translation initiator ATG) ␣-ENaC DNA fragment are sufficient to mediate the basal and the extracellular signalregulated protein kinase (ERK) pathway-modulated expression of ␣-ENaC/reporter construct in transient transfection assays (11). In this study, we identified a 508-bp DNA fragment in the 5Ј-flanking region of the rat ␣-ENaC gene that is required for transcriptional activation of the ␣-ENaC promoter and which also confers Ras-mediated repression. By using heterologous promoter constructs and deletion analyses, we demonstrate that a functional GRE and GR are required for both transcriptional activation and Ras-mediated repression. This is the first report of an ␣-ENaC enhancer with a dual effect, suggesting that its involvement in the transcriptional regulation of ␣-ENaC gene expression in salivary epithelial cells is via the cross-talk between Ras-and GR-mediated signaling pathways. This mechanism may also be responsible for the expression of the ␣-ENaC gene in different tissues.

MATERIALS AND METHODS
Cell Culture and DNA Constructs-The rat parotid epithelial cell line Pa-4, also known as parotid C5 cell line (22), was maintained as described previously (23). Dexamethasone was obtained from Sigma, resuspended to 1 mM in ethanol, and stored in Ϫ80°C. The anti-glucocorticoid, ZK98.299 (24,25), a generous gift from Dr. M. Stallcup (University of Southern California, Los Angeles, CA), was stored at Ϫ80°C as a 1 mM stock solution in ethanol.
A series of unidirectional deletion mutants harboring the luciferase reporter gene (Fig. 3) were constructed as follows. The 0.5-kb HindIII ␣-ENaC DNA fragment (Ϫ1573 to Ϫ1066 bp) was PCR-amplified using primer pairs designed according to the ␣-ENaC DNA sequence and cloned into pCR2.1 vector with the Original TA Cloning ® kit (Invitrogen) in either sense or antisense orientation. Subsequently, each insert was excised and unidirectionally cloned into pGL2-P reporter plasmid (Promega) to generate p(Ϫ1573/Ϫ1066)GL2-P and p(Ϫ1066/Ϫ1573) GL2-P, respectively. These two constructs were then double-digested on the adjacent KpnI (3Ј-overhang) and SpeI (5Ј-overhang) sites to create a single exonuclease III-sensitive end. Unidirectional deletions were carried out by digesting DNA fragments for various periods of time with exonuclease III according to the instruction of Erase-a-Base ® system (Promega). The extent of deletion in each individual clone was determined by DNA sequence analyses. Site-specific mutations within or adjacent to the GRE in p(Ϫ1414/Ϫ1066)GL2-P were introduced by Transformer site-directed mutagenesis kit (CLONTECH) to generate Mt A and Mt B (Fig. 6A). A ␣-ENaC/CATs plasmid with point mutations on GRE site (GRE Mt; Fig. 7) was constructed by replacing a SacI/ HindIII fragment containing GRE site with the same fragment from Mt B described above that carries mutated GRE.
The DNA sequences and mutations introduced in all constructs were verified by sequence analyses utilizing T7 Sequenase 7-deaza-GTP Sequencing Kit (Amersham Pharmacia Biotech) as instructed by the manufacturer. Expression plasmids harboring Ras mutants of Ras V12, Ras S35, Ras G37, and Ras C40 were generous gifts from Dr. D. Johnson (University of Southern California, Los Angeles, CA), while glucocorticoid receptor (GR) expression construct and a tyrosine aminotransferase GRE containing reporter construct, TAT(GRE) 3 TK/Luc, were kindly provided by Dr. M. Stallcup (University of Southern California).
Transient Transfection-Plasmid DNAs were transiently transfected into Pa-4 cells using a LipofectAMINE-mediated method (Life Technologies, Inc.) as described previously (23). In all experiments, 0.1 g of Renilla luciferase plasmid, pRL-TK (Promega, WI), was included as an indicator plasmid to normalize for transfection efficiency. Total amount of DNA in each transfection of cells grown in a 35-mm dish was kept constant at 2 g by supplementing with pCMV vector, as needed. To maximize the response of cells to dexamethasone treatment, the cells 2 M. D. Zentner and D. K. Ann, unpublished observation.
were serum-starved 24 h after the start of transfection by replacing growth medium with 0.05% serum-containing medium for 8 h, followed by adding 10 Ϫ7 M dexamethasone in the same medium. All plates were harvested 16 h thereafter. The CAT reporter gene analysis was carried out as described previously (23), whereas firefly and Renilla luciferase activities were measured using Dual-Luciferase reporter assay kit (Promega) according to the manufacturer's instruction. All transient transfection assays were carried out at least three times independently.
Northern Blot Analysis-Total RNA (RNA) was isolated from Pa-4 cells using Trizol ® reagent (Life Technologies, Inc.) as instructed by the manufacturer. The quality and quantity of RNA were determined by both spectrophotometric analysis and fractionating RNA on an agarose/ formaldehyde gel (1.3% (w/v)/6% (v/v)), followed by staining with ethidium bromide to compare the ratio of 28 S to 18 S ribosomal RNAs. For Northern analyses, equal amounts of RNA (18 g/sample) were loaded onto an agarose/formaldehyde gel, fractionated by size, transferred to a 0.2-m nylon membrane (ICN Biomedicals, Inc.), and UVcross-linked. All blots were prehybridized for 1 h in QuikHyb ® (Stratagene). The hybridization was carried out according to manufacturer's instructions with 32 P-labeled ␣-ENaC probes that were prepared from an isolated rat ␣-ENaC cDNA fragment (bases 1-905) using a Random Primed DNA labeling kit (Roche Molecular Biochemicals). All blots were also reprobed with rat ␤-actin to ensure that the quality and quantity of mRNA between lanes were comparable. After hybridization, these blots were washed in a 0.1ϫ SSC (0.15 M NaCl and 0.015 M sodium citrate) solution containing 0.5% SDS at 60°C for 2 h with three changes of washing solution. The radioactive signal was visualized by autoradiography, where blots were exposed to films at Ϫ80°C overnight, using BioMax TranScreen-LE (Eastman Kodak Corp.) to improve sensitivity. Blots were also quantitated through the use of electronic autoradiography with an Instantimager 228 (Packard Instrument Co). The amount of ␣-ENaC message was quantitatively compared between lanes after normalizing against that of ␤-actin.

Ras Inhibits ␣-ENaC/Reporter Activity via Multiple Effector
Pathways-The Ras pathway has been recently reported to modulate epithelial sodium channel activity and expression in Xenopus oocytes (26). Moreover, we have shown that a downstream effector pathway of Ras, Raf/ERK, can down-regulate the expression of both the endogenous ␣-ENaC gene and transiently transfected reporter constructs harboring 1.4 kb (Ϫ1573/Ϫ154 bp relative to ATG) of the rat ␣-ENaC 5Ј-flanking region (11). In order to more precisely evaluate the ability of Ras and its different effector pathways to modulate ␣-ENaC expression in mammalian cells, salivary epithelial Pa-4 cells were cotransfected with an ␣-ENaC/CATs reporter construct and an expression plasmid encoding a constitutively activated form of Ras (Ras V12) or a "single-effector" mutant, i.e. Ras S35, Ras G37, and Ras C40. The ability of each Ras mutant to modulate ␣-ENaC/CATs reporter activity was assessed.
Among the three Ras "single-effector" mutants, S35 is able to activate Raf-1 but not Ral-GDS or PI 3-kinase, while G37 activates Ral-GDS but not Raf-1 or PI 3-kinase, and C40 activates PI 3-kinase but not Raf-1 or Ral-GDS (27). As shown in Fig. 1, Ras V12 markedly repressed ␣-ENaC/CATs activities when cotransfected into Pa-4 cells (lanes 3 and 4 versus lanes 1 and 2). While the C40 mutant had no effect on ␣-ENaC/CATs activities, mutants S35 and G37 appeared less effective than Ras V12 in our system (Fig. 1). Although it is possible that the known differences in the activity and expression level of these mutant proteins (28) may have contributed to these results, it is more likely that both Ral-GDS and Raf/ERK signaling pathways are involved in down-regulating ␣-ENaC expression.
Previously, we have shown that MEK inhibitor, PD 98059, was able to block the inhibition of ␣-ENaC/CATs activity by Raf/ERK activation (11). However, the MEK inhibitor had a limited effect on Ras V12-mediated inhibition, 3 suggesting that an effector pathway(s) other than Raf/ERK may also have been involved in modulating ␣-ENaC expression. Since Ras V12 elicited the maximum repression on ␣-ENaC/CATs reporter activity, we used Ras V12 to elucidate the mechanism(s) that down-regulates ␣-ENaC expression.
Identification of a Ras-repressible Enhancer Required for ␣-ENaC Expression-The results shown above and our previously published data (11) suggest that elements confined within the ␣-ENaC/CATs construct are essential for recapitulating the basal and Ras-mediated regulation of ␣-ENaC expression. To elucidate the cis-element(s) conferring Ras-mediated repression of the ␣-ENaC/CATs, we constructed a series of reporter plasmids in which truncated fragments of the ␣-ENaC promoter/enhancer were placed upstream of the promoterless CAT reporter gene. As shown in Fig. 2, removal of the first 508 bp from the parental construct, ␣-ENaC/CATs, reduced the level of CAT activity approximately 10-fold in Pa-4 cells transiently transfected with Ϫ1.0␣-ENaC/CAT. Since basal activity was so low, the ability of Ras V12 to repress reporter activity cannot be accurately assessed.
Subsequent deletions of 310 and 586 bp from Ϫ1.0␣-ENaC/ CAT, designated as Ϫ0.7␣-ENaC/CAT and Ϫ0.5␣-ENaC/CAT, respectively, resulted in minor variations of the basal CAT activity when compared with those from Ϫ1.0␣-ENaC/CATtransfected cells (Fig. 2). The reason for these variations is not clear, since they were not always reproducible. This dramatic decrease in basal activity indicated that a Ras-repressible enhancer(s) was located within the first 508 bp (Ϫ1573/Ϫ1066) of the 1.4-kb ␣-ENaC DNA fragment. To confirm this, we inserted the (Ϫ1573/Ϫ1066) fragment 5Ј to the Ϫ0.7␣-ENaC/CAT in both sense and antisense orientations and in the antisense orientation upstream of the Ϫ0.5␣-ENaC/CAT construct ( Fig.  2). High levels of basal activity were observed in cells transiently transfected with any of the three constructs (Fig. 2). Moreover, the Ras-mediated repression of reporter gene activity was quantitatively and qualitatively similar to that observed in ␣-ENaC/CATs-transfected cells. Thus, the DNA fragment Ϫ1573 to Ϫ1066 bp of ␣-ENaC enhanced the ␣-ENaC basal promoter activity in a way that could be repressed by Ras activation. The DNA sequences between Ϫ1066 and Ϫ480 bp, on the other hand, were dispensable for modulating ␣-ENaC expression.
Delineation of the Activating Sequence(s) within the Ϫ1573/ Ϫ1066 ␣-ENaC 5Ј-Flanking Region-To determine whether encoding Ras V12, and GTPase deficient (V12) forms of Ras S35, Ras G37, and Ras C40, respectively. pRL-TK plasmid (0.1 g) was cotransfected as an indicator to normalize for transfection efficiency. The total amount of plasmid transfected was kept constant at 2.0 g by supplementing with pCMV vector. CAT assay was performed after normalizing against Renilla luciferase activity. The percent conversion shown was calculated as described previously (23). Transfection with each Ras mutant expression plasmid was performed in duplicate. Similar results were obtained from three independent experiments, where one representative CAT assay is shown. the first 508-bp fragment of the parental ␣-ENaC/CATs construct alone could modulate ␣-ENaC expression, the (Ϫ1573/ Ϫ1066) fragment was inserted in both sense (Fig. 3A, construct A) and antisense (Fig. 3A, construct E) orientations upstream of a heterologous minimal SV40 promoter in pGL2-P vector. As shown in Fig. 3B, the (Ϫ1573/Ϫ1066) fragment (construct A) conferred an approximately 4-fold higher basal activity than that of vector alone (construct I). Moreover, the activity was stimulated in an orientation-independent manner, a hallmark of an enhancer element. Although it is unclear, the 508-bp fragment when in the antisense orientation (construct E) elicited a higher activity than the sense counterpart (construct A; see Fig. 3B). Moreover, coexpression of Ras V12 had little effect on attenuating reporter activity of vector alone, but markedly down-regulated the enhancer activity conferred by the Ϫ1573/ Ϫ1066 fragment.
Since the expression pattern of the parental, ␣-ENaC/CATs, construct was functionally reproduced in heterologous promoter constructs containing the (Ϫ1573/Ϫ1066) fragment, we focused our attention on characterizing this 508-bp distal enhancer. Serial 5Ј-and 3Ј-deletion mutants of the Ϫ1573/Ϫ1066 fragment were constructed to locate the DNA motif(s) involved in Ras-repressible enhancer activity. As shown in Fig. 3B, deletions outside the Ϫ1355 to Ϫ1269 bp sequence had little effect on either enhancer activity or Ras repressibility (constructs B, C, F, and G). In contrast, constructs excluding this region (constructs D and H) exhibited a dramatic reduction in enhancer activity. In addition, Ras repression was virtually abolished (Fig. 3B, constructs D and H). These data suggest that the DNA fragment extending from Ϫ1355 to Ϫ1269 bp confers both enhancer and Ras-mediated repressor activity. Thus, the 87-bp fragment (Ϫ1355 to Ϫ1269 bp) in the 5Јflanking region appears to function as both an enhancer and a Ras-mediated repressor of ␣-ENaC expression in parotid epithelial cells.
Glucocorticoid Hormone Transactivates the Reporter Expression from the Ϫ1573/Ϫ1066 DNA Fragment-Analysis of the 87-bp fragment by Find Patterns Algorithm of the GCG software package and transcription factor sites data base (29) identified several putative transcription factor-binding sites within this region. A DNA sequence around position Ϫ1316 bp (Ϫ1323 to Ϫ1309 bp; AGAACANNNTGTCCT) with close resemblance to the consensus site for the glucocorticoid hormone receptor (GRE; AGAACANNNTGTTCT) was chosen for further investigation. To investigate whether the (Ϫ1323/Ϫ1309) DNA sequence could mediate functional induction and was necessary for the Ras-repressible enhancer activity, Pa-4 cells were cotransfected with a glucocorticoid receptor (GR) expression plasmid and a reporter construct shown in Fig. 3A. Addition of 10 Ϫ7 M dexamethasone enhanced the reporter activity of reporter constructs A and E by approximately 7-fold over those from corresponding vehicle-treated transfected cells (Fig. 3C). Moreover, despite that a high expression level was induced by exogenous GR/dexamethasone treatment, the expression of Ras V12 plasmid was still able to reduce GR/dexamethasonemediated enhancement in constructs A and E up to 5-fold (Fig.  3D). In general, shortening the ␣-ENaC DNA fragment 5Ј to Ϫ1355 bp and 3Ј to Ϫ1269 bp neither altered dexamethasone induction nor Ras-mediated repression of reporter activity (Fig.  3, C and D, constructs B, C, F, and G). However, deletion of the region containing the identified GRE virtually abolished GR/ dexamethasone-mediated enhancement as well as Ras-mediated repression of the minimal SV40 promoter (Fig. 3, C and D,  constructs D and H). These data are consistent with the results shown in Fig. 3B, suggesting that deletions outside the GRE have no effect on reporter activity, whereas deletions of the Transfections and CAT assays were performed as described in Fig. 1. The region between Ϫ1573 and Ϫ1066 bp in either sense (construct E) or antisense (constructs F and G) orientation apparently acts as an enhancer, of which effect is repressed by Ras pathway activation. A representative CAT assay from three independent experiments is shown (right panel).
GRE abolish both enhancer and repressor properties of the 508-bp fragment.
Based on these results, it appears that in Pa-4 cells p(Ϫ1066/ Ϫ1573)GL2-P reporter construct expression can be modulated by two signaling pathways; whereas GR activation stimulates reporter expression, Ras activation antagonizes GR-mediated enhancement. To examine this hypothesis, Pa-4 cells were transiently cotransfected with the reporter construct p(Ϫ1066/ Ϫ1573)GL2-P and a GR expression plasmid or an empty vector. The luciferase activity was increased approximately 1.5-fold by treatment with the GR agonist, dexamethasone (Fig. 4, lane 5  versus lane 4), suggesting that endogenous GR activation was able to increase ␣-ENaC enhancer activity modestly via a GRE located between Ϫ1573 and Ϫ1066 bp. No transcriptional stimulation by dexamethasone was observed in pGL2-P (Fig. 4, lane  3), which lacks the cloned ␣-ENaC enhancer fragment. Co-  Fig. 1. Both firefly and Renilla luciferase activities were measured simultaneously using Dual-Luciferase assay system (Promega). The relative luciferase activity from firefly luciferase reporter gene shown in each group of transfected cells was determined and normalized with the indicator Renilla luciferase activity. Each value shown is the mean Ϯ S.E. based on three independent transfection experiments. C, GR/dexamethasone (Dex) transactivates the ␣-ENaC distal enhancer in cells cultured under serum-free conditions. Pa-4 cells were cotransfected with 0.9 g of each construct shown in diagram (A) and/or 0.5 g of GR expression plasmid as described in B. Sixteen hours after the start of transfection, the cells were serum-starved for 8 h, followed by the addition of 10 Ϫ7 M dexamethasone to the culture medium, and incubated overnight. The reporter and indicator luciferase activities were determined as described in B. The level of induction, expressed as -fold induction, for each construct is calculated by dividing the normalized reporter luciferase activity in extracts from GR-transfected/dexamethasone-treated cells over that of corresponding GR-transfected/vehicle-treated cells. Error bars were calculated as in Fig. 3B. D, Ras pathway activation represses GR transactivation activity. Pa-4 cells were cotransfected with 0.9 g of each construct shown in diagram (A) and 0.5 g of GR expression plasmid in combination with or without 0.5 g of Ras V12 expression plasmid as indicated. The transfection, treatment protocol, and dual luciferase assay were carried out as described in C. The relative luciferase activity for each reporter construct in extracts, after normalization, from GR-transfected/dexamethasone-treated cells and Ras-plus GR-cotransfected/dexamethasone-treated cells, respectively, is shown. Error bars represent the standard error of the mean for each construct based on three independent transfection experiments. transfection with a GR expression plasmid in the absence of exogenous dexamethasone failed to elicit a robust enhancement on the heterologous promoter activity of p(Ϫ1066/ Ϫ1573)GL2-P (Fig. 4, lane 6 versus lane 4). However, dexamethasone substantially induced reporter activity in the same cotransfection (Fig. 4, lane 7 versus lane 4). In addition, Ras activation markedly antagonized exogenous GR/dexamethasone-stimulated expression of p(Ϫ1066/Ϫ1573)GL2-P (Fig. 4, lane 9 versus lane 7) but had little effect on p(Ϫ1066/ Ϫ1573)GL2-P alone (lane 8 versus lane 4). A similar result was also observed with sense-oriented counterpart p(Ϫ1573/ Ϫ1066)GL2-P (data not shown). These results not only support an antagonistic regulatory hypothesis, but also suggest that a direct association exists between Ras-mediated repression and GR-associated enhancement of ␣-ENaC expression. Hence, it appears that the maximal stimulating activity via the 508-bp enhancer requires the presence of a GRE in the reporter construct and the presence of glucocorticoid hormone to activate its expression. In addition, its Ras-mediated repression on ␣-ENaC promoter activity is likely to be associated with its ability to inhibit GR transactivation.
GR Antagonist and TPA Block Dexamethasone-induced Endogenous Expression of ␣-ENaC mRNA-To investigate whether the identified enhancer DNA fragment functions in its native configuration, endogenous ␣-ENaC expression was examined in response to GR and Ras signaling pathway activation. To do this, Pa-4 cells were treated with dexamethasone in the presence and absence of GR antagonist, ZK98.299. Our preliminary results indicated that the dexamethasone EC 50 value for inducing ␣-ENaC expression was approximately 3 ϫ 10 Ϫ9 M in Pa-4 cells (data not shown). This allowed us to reduce concentration of dexamethasone used in this study to 10 Ϫ8 M in order to accommodate the usage of 100-fold excess of ZK98.299 without obvious cytotoxicity. Cells treated with 10 Ϫ8 M dexamethasone alone for 16 h exhibited an approximately 6-fold increase in ␣-ENaC mRNA level, which was completely blocked by the cotreatment with 1 M ZK98.299 (Fig. 5A). ␤-Actin mRNA expression was used to normalize ␣-ENaC mRNA level in control and treated Pa-4 cells, and to assure the viability of the treated cells (Fig. 5, A and B).
A modest decrease in the basal ␣-ENaC mRNA level by ZK98.299 treatment was also observed (Fig. 5A), demonstrating the ability of the remaining glucocorticoid hormone, after switching to serum-free conditions, to modulate ␣-ENaC expression. Since TPA has been shown to induce the Raf/ERK, Ras effector pathway, in Pa-4 cells (11), we investigated whether TPA could block dexamethasone-mediated induction of the endogenous ␣-ENaC expression. Cells treated with 10 Ϫ7 M dexamethasone for 16 h exhibited a marked increase in the steady-state ␣-ENaC mRNA level by approximately 7-fold (Fig.  5B, lane 3). Consistent with the effect of ZK98.299, TPA treatment attenuated both basal and dexamethasone-induced ␣-ENaC mRNA levels in Pa-4 cells (Fig. 5B, lanes 2 and 4). These data are consistent with the results shown above on the regulatory effects on ␣-ENaC/reporter activity by GR and Ras V12 expression plasmids, and establish the physiological relevance of this study. Together, these data suggest that glucocorticoid hormone induces ␣-ENaC expression through the putative GRE and that the GR-mediated transactivation can be antagonized by GR antagonist as well as TPA-mediated Raf/ ERK activation.

Point Mutations in GRE Abolish Antagonistic Modulation by Glucocorticoid Hormone/Ras Pathways-Our data indicate
that the Ras-mediated repressibility and dexamethasone-stimulatory effect are not operated exclusively via distinct positive and negative regulatory elements. Based on sequence analysis of the Ϫ1334 to Ϫ1306 bp DNA fragment, a second putative transcription factor binding site AP-1 was identified 15 bases upstream of the GRE (Fig. 6, from the center of putative AP-1 site to the center of putative GRE). AP-1 is activated by the Ras signaling pathway (19), and is a well established antagonist of GR-mediated transcription (17,18). Therefore, we postulated that Ras activation may repress GR/GRE association by activating AP-1 and thereby stimulating a mutual competition between AP-1 and GR for their respective response element. Second, although we showed that GR-mediated transcription is necessary for endogenous ␣-ENaC gene expression (Fig. 5A), whether or not this GR transactivation is mediated via the identified GRE is unknown. To test these hypotheses directly, Pa-4 cells were transiently transfected with one of three distinct reporter constructs: wild type, Mt A, and Mt B. Mt A differs from wild type by three nucleotide substitutions at positions Ϫ1330, Ϫ1328, and Ϫ1327, abolishing the putative AP-1 element, whereas Mt B leaves the AP-1 site unaltered but replaces five nucleotides within the putative GRE (Fig. 6A,  upper panel).
As shown in Fig. 6A, mutations on the putative AP-1 site (Mt A) had no obvious effect on the heterologous promoter activity when Mt A was transfected alone or cotransfected with GR and/or Ras V12 expression plasmids. Thus, the putative AP-1 site was not involved in Ras-mediated GR transrepression in Pa-4 cells. The activity of Mt B in unstimulated conditions was not affected when compared with that of the wild type or Mt A (Fig. 6A). Therefore, an intact GRE site was not required for the identified enhancer to modulate ␣-ENaC expression in the absence of glucocorticoid hormone. Its presence, however, was indispensable for dexamethasone-induced ␣-ENaC expression and for the Ras pathway to block the induction, since altering the nucleotide sequences crucial to GRE motif drastically re- FIG. 4. Hormone-dependent activation of ␣-ENaC enhancer activity by GR. Salivary Pa-4 cells were transiently transfected with 0.9 g of vector pGL2-P plasmid or p(Ϫ1066/Ϫ1573)GL2-P reporter construct with or without 0.5 g of Ras V12 expression plasmid in combination with 0.5 g of GR expression plasmids, as indicated. The amount of transfected DNA was kept constant at 2 g by supplementing with varying amount of pCMV plasmid DNA. After transfection, the cells were treated as described in Fig. 3C with 0.1 M dexamethasone or vehicle as indicated. The level of induction, expressed as -fold induction, is calculated by dividing the normalized reporter luciferase activity in each extract (lanes 2-9, respectively) over that of pGL2-P-transfected/ vehicle-treated cells, which is arbitrarily designated as 1 (lane 1) and indicated by a horizontal line. This experiment has been repeated three times, and one representative plot is shown. duced the GR/dexamethasone-induced reporter activity and the magnitude of Ras-mediated, GR-dependent repression (Fig.  6A). Moreover, AP-1 site elicited no enhancing effect on the expression of Mt B, which lacks the functional GRE site, confirming that Ras activation had no effect on the identified ␣-ENaC enhancer activity in the absence of GR/dexamethasone transactivation.
The data in Fig. 3 suggest that ␣-ENaC DNA sequences between Ϫ1355 and Ϫ1269 bp confer high GR/dexamethasoneinduced activity and Ras-mediated repression of GR/dexamethasone-dependent activity. Furthermore, the data in Fig.  6A illustrate that this Ras-repressible enhancer activity is largely abolished when the half-palindrome (nucleotides Ϫ1322 to Ϫ1318) of the GRE site is mutated, but not by mutations on the adjacent AP-1 sequence. To establish that the identifiedGREsequencealonecanserveasaGR/dexamethasonedependent enhancer and confer Ras-repressibility over the GR/ dexamethasone-mediated transactivation on a heterologous promoter, ␣-ENaC DNA sequences of Ϫ1334 to Ϫ1306 bp were cloned in both sense and antisense orientations into a luciferase reporter vector, pGL2-P, containing a minimal SV40 promoter. In transiently GR-cotransfected Pa-4 cells, the p(Ϫ1334/ Ϫ1306)GL2-P reporter construct exhibited at least a 10-fold increase in luciferase activity in the presence of dexamethasone; a ␣-ENaC GRE (Ϫ1334/Ϫ1306) copy number-dependent dexamethasone enhancement was also observed (Fig. 6B). A similar result was also observed from the antisense construct, p(Ϫ1306/Ϫ1334)GL2-P (data not shown). As illustrated in Fig.  6B, the transcriptional enhancement by GR/dexamethasone on reporter constructs harboring either one copy or two copies of ␣-ENaC GRE was inhibited by the cotransfected Ras V12 expression plasmid. Similarly, GR/dexamethasone-dependent enhancement and Ras-mediated repression were also observed in Pa-4 cells by a tyrosine aminotransferase (TAT) simple GRE  (lanes 3 and 4) to maximize ␣-ENaC induction and TPA was used at 100 ng/ml (lanes 2 and 4). Control cells (lane 1) were treated with vehicle alone. The quantitative analyses (lower panel) were performed as described in A.
sequence on the minimal TK promoter (Fig. 6B). These data clearly establish that ␣-ENaC sequences of Ϫ1323 to Ϫ1309 bp are capable of functioning as a canonical GRE and its enhancer activity is inhibited by Ras pathway activation in salivary Pa-4 cells.
To further ascertain the contribution from the identified GRE in modulating ␣-ENaC expression, we investigated the consequences of GRE mutation in its native context. Point mutations were made on the GRE of ␣-ENaC/CATs and the effects of GR/dexamethasone and Ras V12 on the wild type and GRE-mutated ␣-ENaC/CATs reporter constructs were assessed. As shown in left panels of Fig. 7, GR/dexamethasonedependent induction of wild type ␣-ENaC reporter activity was unequivocally abolished by Ras V12, although not entirely, in either serum-free condition (A) or serum-containing growth medium (B). GR/dexamethasone treatment failed to confer a robust induction on reporter activity of GRE-mutated ␣-ENaC/ CATs (right panels, Fig. 7) when compared with that of the wild type in both culture conditions. Thus, point mutations on the identified GRE not only reduced the basal reporter expression but also abrogated GR/dexamethasone-mediated induction substantially. Nevertheless, the residual reporter activity from the GRE mutant was slightly enhanced by GR/dexamethasone treatment and repressed by Ras V12 (Fig. 7, right panel). Taken together, results from Fig. 7 confirm the observation that the identified GRE plays an essential role in modulating ␣-ENaC expression. However, this fails to rule out the possibility that a second, albeit much weaker, GRE exists in the 1.4 kb of ␣-ENaC 5Ј-flanking region that could also be inhibited by Ras pathway activation. DISCUSSION The results presented herein demonstrate that the transcription of the rat ␣-ENaC gene in salivary epithelial cells depends upon an enhancer located between Ϫ1.5 and Ϫ1.0 kb upstream from the translation initiation site. Elements within this region are required for Ras pathway-mediated repression of ␣-ENaC expression (Fig. 2). We have mapped a Ras-repressible enhancer activity of this region to a DNA sequence between Ϫ1334 and Ϫ1306 bp, which encompasses a GRE site. Our data provide evidence that the overall activity of the ␣-ENaC promoter/enhancer represents an integrated response to GR/dexamethasone-dependent activation and Ras-mediated repression.
A number of physiological, pathological, and pharmacological stimuli are known to regulate ENaC function (reviewed in Refs. 12 and 13). In order to establish how these stimuli work in concert, it is necessary to define the signal transduction pathways and transcriptional control paradigms that are activated by various stimuli and to assess the integrated response to these stimuli. Previously, we have reported that TPA-mediated ERK activation appears to be required for regulating several biological responses in salivary epithelial cells includ-FIG. 6. Functional analyses of putative GRE as a regulatory element in GR-mediated transactivation and Ras-mediated repression. A, function of the putative GRE in the context of heterologous promoter. DNA sequences of the rat ␣-ENaC distal enhancer region extending from Ϫ1334 to Ϫ1306 bp are shown. The GRE-and AP-1-like recognition motifs are boxed. Mt A and Mt B represent mutant variants used in transfection assays. Sequences that are identical to the wild type are shown as dashes, whereas the mutated DNA sequences are shown in letters. Reporter construct (0.9 g) was cotransfected with 0.4 g of Ras V12 and/or 0.4 g of GR expression plasmids as indicated. Luciferase activities of the wild type and mutant variants were assayed and analyzed as described in Fig. 3. B, the putative GRE in ␣-ENaC confers glucocorticoid response in the context of heterologous promoter. Three different reporter constructs were tested. Two constructs carry one and two copies of ␣-ENaC GRE encompassing sequences, Ϫ1334 to Ϫ1306 bp, upstream of the minimal SV40 promoter-containing luciferase reporter gene and the third construct, TAT(GRE) 3 TK/Luc plasmid, harbors three copies of simple GRE from tyrosine aminotransferase gene upstream of a minimal TK promoter. Pa-4 cells were cotransfected with 0.9 g each of the above three con-structs in combination with Ras V12 and GR expression constructs as indicated. Transfection with the empty vector pGL2-P is included as a control. The transfection, treatment protocol, and dual luciferase assays were carried out as described in Fig. 3. The luciferase activity in each extract of p(Ϫ1334/Ϫ1306)GL2-P-, p(Ϫ1334/Ϫ1306) 2 GL2-P-, TAT (GRE) 3 TK/Luc-, and pGL2-P-transfected cells, respectively, is arbitrarily designated as 1 (lanes 1, 5, 9, and 13). The level of induction for each construct, expressed as -fold induction, is calculated by dividing the normalized reporter luciferase activity in extracts from Rascotransfected cells (lanes 2, 6, 10, and 14), GR-cotransfected/ dexamethasone (Dex)-treated cells (lanes 3, 7, 11, and 15), and Ras-plus GR-cotransfected/dexamethasone-treated cells (lanes 4, 8, 12, and 16), respectively, over that of corresponding vehicle-treated transfected cells (lanes 1, 5, 9, and 13). One representative data set out of three independent experiments is shown.
ing the modulation of ␣-ENaC mRNA level (11,30). In this context, a partial aim of the present study was to identify the regulatory elements of the rat ␣-ENaC gene responsible for Raf/ERK-mediated down-regulation of ␣-ENaC expression in salivary epithelial cells. The exact steps that follow ERK activation to attenuate ␣-ENaC mRNA steady-state level are unclear at present. The activity of a number of transcription factors, including AP-1, Ets, and SRF (31)(32)(33)(34) are known to be modulated by the Raf/ERK pathway. However, whether or not their activation is associated with ERK-mediated repression of ␣-ENaC expression has not been established.
To delineate the mode of ␣-ENaC gene regulation, we have examined the ability of different Ras effector pathways, essential cis-elements, and their cognate trans-factors to regulate ␣-ENaC expression. A 508-bp fragment located between Ϫ1573 and Ϫ1066 bp of the rat ␣-ENaC gene was found to activate transcription of ␣-ENaC promoter (Fig. 2). Moreover, the observed enhancing activity was promoter-and orientation-independent, based on its ability to increase the transcription via either the SV40 minimal promoter or the proximal ␣-ENaC promoter in both sense and antisense orientations, the hallmark of a classical enhancer. One novel aspect of the identified ␣-ENaC enhancer is that both the activation and repression elements reside within the same region, Ϫ1355 to Ϫ1269 bp, i.e. deletion of the Ϫ1355 to Ϫ1269 bp region substantially decreases both enhancement and Ras-mediated repression (Fig. 3), implying that elements located within this region enhances the homologous/heterologous promoter activity and that Ras pathway activation antagonizes its enhancing effect. Moreover, site-specific mutations of nucleotides Ϫ1322 to Ϫ1318 abrogate both enhancing and repressing effects (Fig.  6A), strongly suggesting that the enhancement and repression are mediated via the same element. In this regard, we further demonstrate that a single copy of this element alone is sufficient to recapitulate the dexamethasone-stimulated and Rasinhibitable effects on the heterologous promoter (Fig. 6B). In addition, specific mutation on the identified GRE in its native context drastically attenuated the Ras-repressible and GR/dexamethasone-inducible ␣-ENaC/CATs reporter activity (Fig. 7). We conclude that Ras pathway activation represses GR-mediated transactivation of ␣-ENaC via the identified GRE located in nucleotides Ϫ1334 to Ϫ1306 of the ␣-ENaC 5Ј-flanking region.
This novel finding extends the potential role of Ras path-way(s) in modulating various biological responses. The exact mechanism(s) by which Ras blocks GR/dexamethasone-mediated transactivation used in modulating ␣-ENaC expression is unknown. Currently, we are investigating whether Ras pathway activation down-regulates the GR protein level or attenuates its ability to interact with the ␣-ENaC GRE in Pa-4 cells.
A similar functional antagonism between the activation of JNK and ERK kinases and dexamethasone-stimulated GRE/reporter activities has recently been reported (35). These authors showed that phosphorylation of GR at Ser-246 by JNK led to the inhibition of GR-mediated transcriptional activation, whereas ERK did so indirectly. Our present work extends the conclusion by Rogatsky et al. (35) by demonstrating for the first time that the transcription of the endogenous ␣-ENaC gene as well as ␣-ENaC/reporter construct(s) is regulated via the opposing effects of GR-and Ras-mediated signaling pathways. Furthermore, the findings that endogenous ␣-ENaC expression is induced by glucocorticoid hormone and attenuated by TPA treatment (Fig. 5) support the physiological and pharmacological relevance of our data. Moreover, it rules out the possibility that the observed transrepression is a result of transcriptional squelching caused by overexpression of Ras V12 and/or GR expression plasmids. Substantial progress has been made in understanding the mechanisms through which Ras exerts its biological effects. For example, there are multiple effector pathways stimulated by Ras activation, which vary in different cells. Recent studies have shown that the activation of Ras signaling pathway regulates total RNA and protein synthesis (36 -38). A wide array of promoters have been reported to be up-regulated by Ras activation as a result of a stimulatory effect on the basal transcriptional apparatus of the cells as well as on specific transcription factors (reviewed in Ref. 39). Using the attenuated transcription of ␣-ENaC by the Raf/ERK pathway as a paradigm, we have uncovered a mammalian signal transduction mechanism underlying Ras-mediated gene repression in epithelial cells. These results emphasize a fundamental difference between activation and inhibition of gene expression mediated by the same pathway, and reinforce the importance of studying Ras action in both the global and the specific enhancement/ repression of gene expression.
Based on our data, a putative model on the regulation of ␣-ENaC expression in the salivary epithelial cells is depicted as an integrated response to both Ras-and GR-mediated signaling FIG. 7. The ␣-ENaC GRE confers glucocorticoid response in the context of its own promoter. Salivary Pa-4 cells were transiently transfected with 0.9 g of ␣-ENaC/CATs (Wt) or its GRE mutant (GRE Mt) with point mutations, as described in Fig. 6A, in the presence or absence of 0.5 g of Ras V12 and 0.5 g of GR expression plasmids, as indicated. Following transfection, the cells were either serum-starved for 8 h and treated with 10 Ϫ7 M dexamethasone (Dex) overnight (A), or cultured in the growth medium and treated with dexamethasone overnight (B). Transfection, normalization, and treatment protocols were as described in Fig. 3, whereas CAT assay was as described in Fig. 1. pathways (Fig. 8). We propose that Ras activation leads to the activation of ERK kinases as well as a less well defined Ral-GDS effector pathway. Subsequently, these two effector pathways may work in concert or individually to inhibit GR-stimulated ␣-ENaC transcription in a direct and/or indirect fashion. This model is supported by the failure of MEK inhibitor, PD 98059, to fully block Ras-mediated repression, indicating that other effector pathway(s) may participate in parallel with the elucidated Raf/ERK pathway to mediate GR-dependent transrepression of ␣-ENaC expression. The results obtained from transient transfection studies employing mutant constructs with homologous (Fig. 2) and heterologous (Fig. 3) promoters support the hypothesis that GR activation is necessary for Ras-mediated repression of ␣-ENaC promoter. In addition, part of the Ras-mediated GR transrepression is independent of the presence of the adjacent putative AP-1 site (Fig. 6A, Mt A). Moreover, the expression of Ras V12 inhibits the luciferase activity in cells transfected with reporter constructs containing either ␣-ENaCor TAT-GRE (Fig. 6B). These data suggest that Ras-mediated GR/dexamethasone-dependent repression may be a common mechanism modulating the expression of GRtarget genes in epithelial cells. It is possible, however, that there is other yet unidentified transcription factor(s) "X" interacting with DNA motif(s) within the cloned 1.4-kb ␣-ENaC 5Ј-flanking region that may likewise be modified by Ras pathway activation to inhibit ␣-ENaC expression (Fig. 8).
In this model, we also propose a novel mechanism in which growth factor or cytokine stimulation via Ras could counteract GR-transcriptional enhancement (Fig. 8). This regulatory mode could provide a physiological and pharmacological balance in many cell types where GR activation exerts a potent antiinflammatory and anti-proliferative effects through counteracting the AP-1 signaling pathway (40). For instance, hormoneactivated nuclear receptors have been shown to prevent c-Jun phosphorylation and AP-1 activation by blocking the induction of one of the Ras-effector pathways, JNK cascade (41). Thus, Ras activation not only increases transcriptional activity through the JNK pathway, i.e. AP-1 (19), but also decreases transcriptional activity of the GR (this report). Therefore, the cross-talk between Ras-and GR-mediated signaling pathways appears to be bidirectional. Whereas GR activation affects the Ras pathway(s), Ras exerts a distinct effect on GR-regulated gene expression. Although we failed to detect any effect by the PI 3-kinase pathway, via Ras C40 mutant (Fig. 1), on ␣-ENaC gene expression in salivary epithelial cells, we cannot completely exclude the role of PI 3-kinase pathway in counteracting GR-mediated events since different Ras effector pathway(s) may be utilized at distinct efficiencies in different cell types.
In conclusion, our data on the regulation of ␣-ENaC gene expression by Ras-and GR-mediated pathways in salivary epithelial cells represent the first detailed study on the role(s) of regulatory elements in the transcriptional control of any of the three subunits of ENaC expressed in epithelia. The present work indicates that the cross-talk between Ras-and GR-medi- FIG. 8. A model for the regulation of ␣-ENaC expression in salivary epithelial cells by glucocorticoid hormone and Ras signaling pathways. Transcription of ␣-ENaC gene in parotid salivary epithelial cells is activated by glucocorticoid hormone treatment and repressed by GR antagonist, ZK98.299. GR/dexamethasone-mediated induction is also antagonized by Ras activation of at least two distinct effector pathways, Ral-GDS and Raf/ERK. In addition, treatment of cells with TPA represses ␣-ENaC transcription possibly through Raf/MEK pathway. Previously, we have shown that TPA-mediated down-regulation of ␣-ENaC is attenuated by the MEK inhibitor, PD 98059 (11), which is consistent with our current results, indicating that the Ras/Raf/MEK/ERK pathway activation antagonizes ␣-ENaC expression. Whether Ral-GDS down-regulates ␣-ENaC expression via another downstream effector(s) or MEK-independent ERK is unclear. It is possible that the Ras/ERK pathway not only suppresses GR/dexamethasone-mediated induction, but also simultaneously induces and/or activates an unidentified protein "X" to repress ␣-ENaC transcription. This scheme depicts an integrated molecular model of ␣-ENaC gene regulation, illustrating the antagonistic effects between glucocorticoid hormone and Ras pathways. By the same mechanism, Ras pathway activation by cytokine or growth factor may also influence the expression of other GR-targeted genes in salivary epithelia. ated pathways dictate the overall transcriptional regulation of the ␣-ENaC gene. The challenge now is to understand how Ras-mediated signals, transient or persistent, are directed at distinct effector pathway(s) and the interface between each effector pathway and transcriptional control to convert GR/ dexamethasone-mediated transactivation to transrepression in different cell types. Studies on the signaling pathway(s) and transcriptional mechanism(s) governing ␣-ENaC gene expression provide a unique paradigm for understanding the interplay between different signaling pathways as well as the biological consequences of such events. It is conceivable that these events may allow cells to respond accordingly to changing cues in different cellular environments. Moreover, the results obtained from this study should be useful for future studies on the expression of the ␣and other subunits of ENaC in various epithelial cells. Finally, it may aid in devising strategies to manage or modulate the concerted expression of all three ENaC subunits simultaneously in epithelia.