Characterization of the cAMP response element of the cystic fibrosis transmembrane conductance regulator gene promoter.

A dominant negative inhibitor of the cAMP-dependent protein kinase has been shown to inhibit the basal expression of the cystic fibrosis transmembrane conductance regulator (CFTR) gene in the human colon carcinoma cell line, T84. A functional cAMP response element (CRE) was localized at −48 in the CFTR promoter, and we have analyzed the interactions of this regulatory region with transcription factors. An adjacent inverted CCAAT element (Y box) at position −60 was also investigated. Mutation of the CRE or the Y box decreases the activity of the promoter in transient transfections of T84 or JEG-3 cells. Electrophoretic mobility shift assays demonstrate that CRE-binding protein (CREB) binds to the CFTR CRE with high affinity and independently of the adjacent Y box and that the CFTR CRE binds CREB and activating transcription factor-1 in nuclear extracts of T84 and CaLu-3 cells. In transient transfections of JEG-3 cells, activation of the CFTR promoter is blocked by a dominant negative CREB mutant. The CFTR CRE will also drive cAMP-mediated expression when placed upstream of a heterologous basal promoter. These results demonstrate that CFTR is a bona fide CRE-dependent gene, and we suggest that CFTR expression levels in vivo may be responsive to hormones or drugs that activate the cAMP-dependent protein kinase system.

Many genes are induced by cAMP, including those coding for hormones, metabolic enzymes, structural proteins, and transcription factors. Although most of these genes contain a cAMP response element (CRE) 1 sequence in their promoter, other regulatory sequences, such as the AP-1 (1), AP-2 (2), Sp1 (3), Pit-1 (4), Y box (inverted CCAAT element) (5), and estrogen response element (6), may also mediate a cAMP-dependent gene activation. The CRE is bound by the cAMP response element-binding protein (CREB)/activating transcription factor (ATF) family of proteins, which can be activated by phosphorylation by the cAMP-dependent protein kinase (PKA) to stimulate gene transcription (7). A group of indirectly activated cAMP-responsive genes consists of those that respond to changes in the levels of the transcription factors induced by cAMP, and this group probably accounts for the broad changes in cellular function characteristic of physiological responses to prolonged cAMP activation.
The consensus CRE, TGACGTCA, provides the strongest and most frequently observed cAMP mediation of various promoters. Nonconsensus CREs that mediate PKA responsiveness frequently exhibit changes in the middle two bases. Examples of the variant CRE, TGACaTCA or TGAtGTCA, include the CREs found in the nonprimate glycoprotein hormone ␣ subunit (8), neurotensin/neuromedin M (9), tissue-type plasminogen activator (t-PA) (10), and retinoic acid receptor ␤2 (11) genes. Other variations of the consensus CRE site have also been shown to be functional. The variant CREs are often weaker than the consensus palindromic sequence at conferring cAMP responsiveness and may require cooperation with other sites to mediate this regulation. Occasionally, the variant CREs mediate responsiveness from other signaling stimuli, such as the phorbol ester responsiveness of the t-PA CRE (10) and the retinoic acid-mediated responsiveness of the retinoic acid receptor ␤2 site (11).
A variant CRE sequence (TGACaTCA) is present in the promoter for the cystic fibrosis transmembrane conductance regulator (CFTR) gene, and this element has recently been implicated in basal and PKA-mediated responsiveness of the promoter. We have determined that the basal activity of the CFTR gene is regulated by cAMP in the human colon caricinoma cell line, T84, and that inhibition of PKA eliminates CFTR-mediated halide efflux (12) and gene expression (13) in these cells. This regulation is mimicked by a reporter construct, CFTR(wt)-luc, T84 cells, and JEG-3 cells (13), a human choriocarcinoma cell line frequently utilized to study PKA-mediated promoter regulation. Mutation of the variant CRE sequence in the CFTR promoter decreases both basal and PKA-stimulated CFTR-luc activity in transient transfections of T84 cells (13), suggesting that this element may function as a CRE to confer cAMP regulation to CFTR gene expression.
The CFTR promoter was sequenced and initially characterized by two separate groups (14,15). These investigators identified several potential regulatory elements by sequence homology, including the variant CRE. One of these studies demonstrated that the CFTR promoter was down-regulated by phorbol ester (15), which is consistent with the effects of phorbol ester on CFTR mRNA (16). This finding was supported by a later study of the CFTR promoter, which also determined that the chromatin context and methylation of the promoter, as well as unidentified regulatory elements close to the transcription start sites, contribute to cell type-specific expression (17).
Although none has yet been identified, mutations of the CFTR promoter, including mutations in the variant CRE, might lead to the lack of CFTR expression that causes cystic fibrosis. This lack of CFTR at the plasma membrane prevents the activation of chloride efflux and inhibition of sodium influx that is necessary for proper exocrine fluid homeostasis, resulting in inspissated secretions that lead to organ failure and opportunistic infection. The potential for cAMP-mediated regulation of CFTR gene expression may offer therapeutic benefit for patients with milder forms of cystic fibrosis in which the function of CFTR may be boosted by increased expression.
In this report we characterize the variant CRE in the CFTR promoter and its ability to bind members of the CREB/ATF transcription factor family. Based on mutational analysis and electrophoretic mobility shift assays, we conclude that this site is indeed a bona fide CRE. Examination of other potential cAMP regulatory sites in the CFTR promoter suggests that there may be sites between Ϫ305 and Ϫ70 and that the Y box at Ϫ60 recently implicated by Pittman et al. (18) as a potential cAMP regulatory element in pancreatic cells also functions modestly as such in colonic cells. Furthermore, we identify the CRE-binding proteins in two CFTR-expressing cell lines as CREB and ATF-1, and we determine that CREB will stimulate CFTR expression through the CRE. We have also established that human lung-derived epithelial cells can regulate CFTR gene expression in response to PKA.

MATERIALS AND METHODS
Cell Culture-CaLu-3 cells were grown in Dulbecco's modified Eagle's medium supplemented with 10% fetal bovine serum and 1 mM sodium pyruvate. HTE-1 cells were generously provided by Christine Halbert (Fred Hutchinson Cancer Research Center, Seattle, WA) and were obtained from tracheal biopsy of a 50-year-old male. Cells were transformed with SV40 large T antigen and passaged on fibronectincoated plates in supplemented KGM medium (Clonetics). T84 and JEG-3 cells were grown as described previously (13).
Solution Hybridizations-CaLu-3 or HTE-1 cells were treated with 30 M N- [2-(p-bromosinnamylamino)ethyl]-5-isoquinolinesulfonamide (H-89) for 24 h, 30 M forskolin for 12 h, and 1 M isoproterenol for 12 h, or they were not treated. The concentration of vehicle (Me 2 SO) remained constant under all conditions. Total nucleic acid was obtained from CaLu-3 cells or HTE-1 cells as described previously for T84 cells (13). The human CFTR riboprobe and standards were prepared as before, and the hybridization conditions were identical to those described for CFTR hybridization of T84 cell total nucleic acid (13).
Plasmid Construction-Construction of the CFTR(wt)-luc and CFTR(mCRE)-luc, as well as the promoterless control ⌬P-luc, has been described previously (13). The CFTR(mYbox)-luc was made using the Altered Sites (Promega) protocol to change the putative inverted CCAAT element (Y box) from ATTGG to AagtG. The double mutant, CFTR(mYbox/mCRE)-luc, was constructed using the polymerase chain reaction with cftr(mCRE) as template, 5Ј-oligonucleotide primer from Ϫ650 to Ϫ633 of the CFTR promoter and 3Ј-primer from Ϫ41 to Ϫ59 of CFTR(mCRE). The fragment was digested with SacII and SpeI and ligated into the appropriately digested CFTR(mCRE)-luc. We made the (Ϫ650)CFTR-luc construct by removing the SacI-SacII fragment in CFTR(wt)-luc and religating, while (Ϫ305)CFTR-luc was constructed in a similar fashion, by removing the SacI-BssHII fragment in CFTR(wt)luc and religating. The (Ϫ151)CFTR-luc, (Ϫ70)CFTR-luc, and (Ϫ55)-CFTR-luc were all constructed using the polymerase chain reaction with cftr(wt)-luc as template. The 5Ј-oligonucleotide primers were directed against the appropriate sequences of the CFTR promoter, while the 3Ј-primer was directed against the 5Ј-end of the luciferase gene. The amplified fragments were then digested with SacI and HindIII and ligated into CFTR(wt)-luc vector that had been cut with SacI and HindIII to remove the promoter. The cftr(CRE)tk-luc and cftr(mCRE)tk-luc constructs were made by digesting inhibin-luc (19) with SacI and BamHI and cloning in the double-stranded oligonucleotides 5Ј-cTGG-AAGCAAATGACATCACAGCAGGTCg-3Ј for the wild type or 5Ј-cTG-GAAGCAAATACTAGTACAGCAGGTCg-3Ј for the mutant. The opposite strand contained the appropriate overhangs for SacI and BamHI ends. The thymidine kinase-luciferase (tk-luc) construct was recovered from inhibin-luc (19) by closing the SacI/BamHI-digested vector with SacI/XmnI/BglII linkers. The identity of all of these constructs was confirmed by DNA sequencing.
Transient Transfections-T84 cells were transfected using Lipofectin (Life Technologies) essentially as described previously (13), generally following the manufacturer's protocol. After washing in serum-free medium, 10 g of reporter or control plasmid was added, with 100 l of Lipofectin in 500 l of total medium, to 5 ml of medium on 10-cm dishes of ϳ50% confluent cells. After 24 h in 10% CO 2 , cells were trypsinized and plated in Dulbecco's modified Eagle's medium/F12 onto six-well dishes, one plate of cells/six wells, and placed in a 5% CO 2 incubator. Cells were then treated with 30 M H-89 or vehicle overnight, or with 30 M forskolin the next morning for 6 h. At this time, all cells were harvested, using 200 l of lysis buffer/well (200 mM potassium phosphate, pH 7.8, 6 mM MgSO 4 , 4 mM ATP, 0.1% Triton X-100, 1 mM dithiothreitol) and assaying 60 l for luciferase activity as described elsewhere for the JEG-3 cells (20). Luciferase activity was normalized for the protein amount (Bradford), and for those figures in which the mean of more than one experiment is presented, the basal activity of the wild-type CFTR-luc was set at 100%.
The JEG-3 cells were transfected using a calcium phosphate procedure for 24-well dishes described previously (21). In these experiments, we used 100 ng/well for wild-type or mutant CFTR-luc reporter. For the experiments examining the effect of transfected PKA on CFTR-luc expression, we cotransfected the cells with 1 ng/well MT-CEV␣, an expression vector for the C-␣ isoform of PKA driven by the metallothionein promoter (22). For the experiments examining the effect of transfected CREB on CFTR-luc expression, we transfected 5 ng of RSV-CREB or 5 ng of RSV-KCREB. RSV-KCREB drives the expression of a mutant CREB that does not bind DNA but will bind wild-type CREB, preventing the association of CREB with the CRE (23). In all JEG-3 transfection experiments, cells were cotransfected with 50 ng/well RSV-lacZ with KS(ϩ) bringing the total amount of DNA to 250 ng/well. After a 24-h transfection, the medium was changed to Dulbecco's modified Eagle's medium and 10% fetal bovine serum, except for the experiments examining C-␣ expression, in which cells were treated with 80 M ZnSO 4 in 2.5% fetal bovine serum for 18 -24 h prior to harvesting to induce expression of C-␣ in the MT-CEV␣-transfected cells. After harvesting, cells were assayed for luciferase and ␤-galactosidase activity as before (20).
Recombinant Protein and Nuclear Extract Preparation-Recombinant CREB (24) was provided by Dr. Richard Goodman (Vollum Institute), while the recombinant ATF-2 fragment (25) was provided by Dr. James Hoeffler (Invitrogen). Nuclear extracts were prepared using an approach similar to that used by Kim et al. (26). Cells were trypsinized off the plates, the trypsin was neutralized with medium containing fetal bovine serum, and the cells were pelleted. Cells were resuspended in five packed cell volumes of 3 ϫ TGE (150 mM Tris, pH 7.4, 22.5 mM EDTA, 30% glycerol) with protease inhibitors (3 g/ml aprotinin, 1 mM 4-(2-aminoethyl)-benzenesulfonyl fluoryl hydrochloride, 0.2 mg/ml soybean trypsin inhibitor, and 2 g/ml leupeptin) and broken in a Dounce homogenizer. Nuclei were collected by a quick spin at 10,000 rpm for 10 s, the supernatant was removed, and the nuclear pellet was resuspended in two-thirds of the original packed cell volume of 3 ϫ TGE, 500 mM KCl, protease inhibitors. After 20 min at 4°C, the extracted nuclei were spun at 46,000 rpm for 23 min in an SW41 rotor, at which point the supernatant was saved and the protein concentration of the nuclear extract was determined. No dialysis was performed, so the final concentration of KCl in the electrophoretic mobility shift assay (EMSA) reflected the diluted KCl from the extraction step; the final concentration ranged from 50 to 75 mM KCl.
Electrophoretic Mobility Shift Assays-Oligonucleotide probes were end-labeled using T4 kinase and ␥[ 32 P]ATP. The probes and other oligonucleotides in these experiments were stored in 50 mM NaCl and prior to use were heated to 95°C and cooled slowly to 45°C. In all experiments, we used 0.1 ng of labeled probe, the specific activity of which was approximately 100,000 cpm/ng.
For the experiments using the 37-bp oligonucleotide probes and nuclear extracts, the conditions varied somewhat from those described above. The cftr(Ybox/CRE) oligonucleotide was labeled as the 27-mer above. These binding reactions consisted of the following: 0.1 ng of labeled probe; 2 g of poly(dI⅐dC; 1.5 g of calf thymus DNA; a 30 ϫ, 100 ϫ, or 300 ϫ excess of unlabeled competitor or buffer control in the appropriate conditions; 4 g of nuclear extract, including preincubating antibody when appropriate, or 30 ng of recombinant CREB; 12% glycerol, 12 mM HEPES, pH 7.9; 1 mM EDTA; 1 mM spermidine; 5 mM MgCl 2 ; and 1 mM dithiothreitol in a final volume of 20 l. The binding conditions for the CaLu-3 nuclear extracts differed only in that 10 g of bovine serum albumin was included. Conditions were similar for recombinant CREB or ATF-2 binding except that only 0.2 ng of poly(dI⅐dC) and no calf thymus DNA were added, and there was 60 mM KCl added to the buffer as well. Antibody incubations with anti-conalbumin (control), anti-CREB (Santa Cruz Biotechnology, Inc.; X-12), anti-CREM (Santa Cruz Biotechnology), anti-ATF-1 (Santa Cruz Biotechnology), anti-RI antiserum, or anti-ATF-2 antiserum (a kind gift of J. Hoeffler), were performed prior to binding for 16 h at 4°C on ice. The unlabeled competitors were in 50 mM NaCl, and the sequences of all doublestranded 37-mer oligonucleotides were as follows: cftr(Ybox/CRE), cTGGGGGGAATTGGAAGCAAATGACATCACAGCAGGTCg; cftr-(Ybox/mCRE), cTGGGGGGAATTGGAAGCAAATactagtACAGCA-GGTCg; cftr(mYbox/CRE), cTGGGGGGAAagtGAAGCAAATGACAT-CACAGCAGGTCg; cftr(mYbox/mCRE), cTGGGGGGAAagtGAA-GCAAATactagtACAGCAGGTCg. The gel was identical to that described above, except that electrophoresis was performed at room temperature.

Regulation of CFTR Gene Expression by PKA-
The ability of PKA to modulate CFTR gene activity remains controversial. Although we (13) and others (18,27) have observed such regulation in colonic and pancreatic cell lines, a lack of cAMP-dependent CFTR promoter activity has been reported in similar cell lines (14,15,17). This disparity may result from the use of different cell lines and transfection conditions and from an increase in sensitivity in the more recent reports. To extend our studies on the regulation of the CFTR gene by PKA, we examined the expression of CFTR in two human lung epithelial cell lines treated with PKA-modulating drugs. Fig. 1A depicts the results of solution hybridizations for CFTR mRNA in the human lung adenocarcinoma cell line, CaLu-3. Treatment with the PKA inhibitor H-89 decreased CFTR mRNA expression ϳ50%, while treatment with either the adenylate cyclase activator forskolin (30 M) or the ␤-adrenergic agonist isoproterenol (1 M) resulted in a ϳ2.5-3.0-fold increase in CFTR mRNA expression. Fig. 1B represents a similar experiment in the human tracheal epithelial cell line, HTE-1. As depicted in Fig.  1B, the basal level of CFTR expression in these cells was approximately 1 molecule/cell, which is close to the levels observed in the human respiratory tract (28). Treatment of the HTE-1 cells with either forskolin or isoproterenol resulted in increases in CFTR mRNA expression of ϳ8and ϳ5-fold, respectively. From the results presented in Fig. 1, A and B, we conclude that the CFTR gene is regulated by PKA in respiratory tissue-derived cell lines. A comparison with our previous results with the colon carcinoma cell line, T84, indicates that the fold induction over basal expression varies significantly between cell lines.
The CFTR Promoter-A diagram depicting the CFTR promoter is presented in Fig. 2A. We have previously determined that mutation of the variant CRE at position Ϫ48 diminishes the activity of the CFTR promoter in transient transfections of T84 cells (13). Recently, others have reported that an inverted CCAAT element (Y box) at position Ϫ60 mediates basal and PKA-stimulated activity of the CFTR promoter in a pancreatic cell line (18), with the contribution of the variant CRE being unclear in this cell line. To determine the relative importance of the variant CRE and Y box element, as well as to further characterize the contribution of upstream sequences, we have produced the constructs depicted in Fig. 2A, containing various CFTR promoter regions upstream of a luciferase reporter. We constructed a series of truncations, (Ϫ650)CFTR-luc, (Ϫ305)-CFTR-luc, (Ϫ151)CFTR-luc, (Ϫ70)CFTR-luc, and (Ϫ55)CFTRluc. The (Ϫ70)CFTR-luc contains both the Y box and the CRE, while (Ϫ55)CFTR-luc contains only the CRE. In addition, we have produced several promoter-reporter constructs in which there are one or more mutations in the CFTR promoter. The CFTR(cCRE)-luc contains a mutation that changes the variant CRE to a consensus CRE, while the CFTR(mCRE)-luc contains an unrelated sequence in place of the variant CRE. The CFTR(mYbox)-luc contains a mutation in the Y box that destroys the core sequence, mutating the same bases that were deleted by Pittman et al. (18) resulting in less activity of the promoter. The CFTR(mYbox/mCRE)-luc construct contains mutations in both the Y box and the variant CRE.
Transient Transfections of CFTR Promoter Truncations- Fig. 2B depicts the results from transient transfections of T84 cells, which demonstrates that the (Ϫ55)CFTR-luc is active, although only ϳ30% as active as (Ϫ2150)CFTR-luc, and is capable of approximately the same -fold stimulation by cAMP as (Ϫ2150)CFTR-luc (ϳ1.8 ϫ). The (Ϫ70)CFTR-luc is clearly less active (ϳ50%) than the (Ϫ2150)CFTR-luc promoter, and even the (Ϫ650)CFTR-luc is less active (ϳ70%) than (Ϫ2150)CFTR-luc. Fig. 2C depicts the results from transient transfections of JEG-3 cells, demonstrating that promoter truncations as small as Ϫ55 bp are active and responsive to cAMP-mediated stimulation. Moreover, these results suggest that the Y box element is not required for basal activity or PKA-mediated stimulation in JEG-3 cells. We have used the JEG-3 cells previously to study the CFTR promoter, because they exhibit a stronger cAMP-mediated regulation of gene expression than the T84 cells (13) and can be transfected more reliably than the CaLu-3 cells.
Progressive truncation from the Ϫ151 position to the Ϫ70 position produces an increase in promoter activity. This may reflect an increase in PKA-independent activity, since, as seen in Fig. 2B, H-89 treatment does not have much of an effect on the activity of Ϫ70 and Ϫ55 in T84 cells. The gain of basal activity in the shorter promoters is not unusual in promoter truncation studies. One interpretation of these results is that there are negative regulatory elements in the truncated region, although the difference in activity might also be related to the proximity of plasmid vector sequences. The results presented in Fig. 2 demonstrate that in T84 cells there are several elements mediating promoter activity, although the CRE and downstream sequences are sufficient for the PKA-stimulatable activity. In contrast, in JEG-3 cells the smallest truncation is as active as the (Ϫ2150)CFTR-luc. Perhaps the longer CFTR promoter constructs are actually more active in T84 cells as a result of upstream regulatory sequences that are inactive in JEG-3 cells. In both cell lines, however, we observe promoter activity in the truncations containing only the CRE and downstream sequences, suggesting that the CRE is sufficient for activity in the context of a minimal CFTR promoter.
Mutation of the Variant CRE and Y Box in the CFTR Promoter-To further examine the PKA-mediated regulation of the CFTR promoter, we performed transient transfections using (Ϫ2150)CFTR-luc constructs containing mutations in the variant CRE or in the Y box, changing these elements to un-related sequences. We also examined the effect of mutating the variant CRE to a consensus CRE, to determine whether this would enhance cAMP-mediated stimulation of the CFTR promoter. In Fig. 3, we demonstrate that in T84 cells, the CFTR-(cCRE)-luc construct has essentially the same activity as the wild-type construct. When the CRE is mutated to an unrelated sequence (CFTR(mCRE)-luc), both basal and forskolin-induced activity are reduced by 75%. Mutation of the Y box in the CFTR promoter also reduces activity by approximately 75%, as depicted in Fig. 3. The double mutant CFTR(mYbox/mCRE)-luc has less activity than either of the single mutants in T84 cells, suggesting that regulation through these two elements occurs separately. The double mutant retains a measurable amount of PKA-dependent basal activity, however, as do the single mutants.
The data shown in Fig. 3 demonstrate that mutation of the variant CRE to an unrelated sequence reduces the activity of the CFTR promoter, while mutation of the variant CRE to the palindromic consensus sequence does not improve the cAMP responsiveness. Furthermore, CFTR promoters truncated to contain only the CRE and downstream sequence continue to confer cAMP responsiveness. The neighboring Y box sequence appears to cooperate with the CRE in mediating promoter activity, and mutations in the Y box reduce both basal and cAMP-induced activity by ϳ75%.
Electrophoretic Mobility Shift Assays Using Nuclear Extracts-To characterize the transcription factors that interact with the CRE/Y box region, we performed EMSAs, using a 37-bp stretch of the CFTR promoter from Ϫ68 to Ϫ32 as probe (cftr(Ybox/CRE)). As depicted in lanes 2-6 of Fig. 4A, CREB binds exclusively to the CRE, and the Y box appears to exert no effect on CREB binding under these in vitro conditions. Using nuclear extracts from T84 cells, we observe two shifted bands: one band (a) comigrating with recombinant CREB, and the other (b) migrating more slowly. Interestingly, both of these bands depend on the CRE, since the shifted bands persist in the presence of excess unlabeled oligonucleotide containing the mutant CRE but are competed away by the wild-type CRE. Again, the Y box seems to have no effect on in vitro binding. As demonstrated by the last three lanes of Fig. 4A, the two bands from the T84 nuclear extracts are competed away by a shorter wild-type probe that does not contain the Y box, but not if the CRE is mutated to an unrelated sequence.
To determine the components of the two shifted bands from the T84 nuclear extracts, we included a panel of antibodies in the EMSA. As depicted in Fig. 4B (lane 5), preincubation with anti-CREB antibody causes a decrease in the intensity of the lower band and a slight increase in the intensity of the upper band resulting from a supershift; this effect is consistent with the effect of this antibody on the recombinant protein (lanes 2 and 3). Preincubation with anti-ATF-1 antibody causes a decrease in the intensity of the upper band, as well as the appearance of a faint supershifted band above the upper band.
Preincubation with the polyclonal anti-ATF-2 antiserum totally eliminates the upper band (lane 10), which is consistent with the effect of this antiserum on the recombinant protein fragment (lane 14). Preincubation with the combination of anti-CREB and anti-ATF-1 or anti-ATF-2 most clearly demonstrates that the lower band is CREB, while the upper band is either ATF-1 or -2, although closely related proteins are also possibilities.
To identify the proteins binding to the CRE in other cell types, we performed EMSAs similar to those in Fig. 4 on nuclear extracts from CaLu-3 cells. As demonstrated in Fig. 1A, CaLu-3 cells display a more robust PKA-mediated stimulation of gene expression than the T84 cells, so we examined whether the identity of the CRE-binding proteins differed in these cells. CaLu-3 nuclear extracts produce two bands (Fig. 4C) similar to those observed with T84 nuclear extracts in Fig. 4B, although the magnitude of band a appears generally diminished with respect to the analogous band in the T84 nuclear extracts (Fig.  4 and data not shown). The CaLu-3 shifted bands show the same specificity as those from the T84 extract (Fig. 4B) when incubated with the same cold competitor oligonucleotides (data not shown). Fig. 4C depicts the CaLu-3 EMSAs after preincubation with the same antibodies used in Fig. 4B. As with the T84 nuclear extracts, bands a and b from the CaLu-3 nuclear extracts are recognized by anti-ATF and anti-CREB antibodies, respectively. Interestingly, the anti-CREB supershifting in CaLu-3 cells is essentially complete, while in T84 cells we are unable to shift the entire CREB band under the same conditions. The results depicted in Fig. 4 demonstrate that the CFTR CRE binds recombinant CREB as well as CREB and ATF-1 or -2 in nuclear extracts from two CFTR-expressing cell types. This binding is totally dependent on the CRE, and the Y box does not appear to bind a nuclear extract protein under our EMSA conditions. Electrophoretic Mobility Shift Assays Using Recombinant CREB/ATF Proteins-To further characterize the ability of the CRE in the CFTR to bind members of the CREB/ATF family in vitro, we performed EMSAs with recombinant CREB/ATF family members and serial dilutions of competitor oligonucleotide. These results provide an estimate of the relative affinity of the CFTR CRE for CREB/ATF family members and are presented in Fig. 5. Fig. 5A depicts an EMSA in which we have used a 27-bp oligonucleotide consisting of the CFTR variant CRE and surrounding sequence from Ϫ58 to Ϫ32, cftr(CRE), as the labeled probe. Fig. 5 demonstrates that recombinant CREB produced a shifted band, as expected from the results of Fig. 4, and competition with the unlabeled consensus CRE from the somatostatin promoter, ss(CRE), competes away this binding. Additionally, competition with unlabeled wild-type CFTR sequence (cftr(CRE)) or with an oligonucleotide in which the CFTR variant CRE is mutated to the consensus core sequence (cftr(cCRE)), eliminates the binding of the CFTR CRE probe to CREB in a dose-dependent manner. CREB binds equally well to either the cftr(cCRE) oligonucleotide or the cftr(CRE) oligonulceotide, as determined by comparing the dose response of the competition. Competition with an unlabeled oligonucleotide containing the mutant CRE does not compete away CREB binding, similar to the results using the cftr(Ybox/mCRE) oligonucleotide in Fig. 4. Fig. 5B demonstrates that the CFTR CRE probe will also bind a fragment of recombinant ATF-2, another member of the CREB/ATF family. Although ATF-2 is not phosphorylated or activated by PKA, this protein does bind the CRE and probably mediates stimulation through the Jun kinase signaling cascade (29,30). The strength of binding of ATF-2 appears very similar to the profile observed with CREB in Fig. 5A. These results demonstrate that the CFTR CRE will specifically bind two CREB/ATF family members with an affinity approximately equal to the affinity of these proteins for the canonical CRE. Furthermore, disruption of the CRE in the CFTR oligonucleotides eliminates the binding to CREB/ATF family members, which correlates with the decrease in basal activity of the CFTR(mCRE)-luc construct observed in Fig. 3.
Regulation of the CFTR Promoter by CREB in Transient Transfections-To determine whether CREB can activate CFTR expression in vivo, we performed transient transfections of JEG-3 cells and examined the ability of CREB and PKA to activate the wild-type and mutant CFTR promoters. As shown in Fig. 6, cotransfection with an expression vector encoding the catalytic subunit of PKA causes a stimulation of wild-type CFTR promoter, and this stimulation is enhanced by cotransfection with a CREB expression vector. Cotransfection with an expression vector encoding a dominant negative CREB mutant (KCREB) inhibits the PKA-mediated expression. Mutation of the CRE abolishes at least 80% of the basal and PKA-mediated transcriptional response, as demonstrated by the middle panel FIG. 4. Characterization of binding proteins from T84 and CaLu-3 nuclear extracts. Presented are EMSAs using labeled 37-bp cftr(Ybox/CRE) as probe on recombinant CREB, T84, or CaLu-3 nuclear extracts. A, recombinant CREB (60 ng) or T84 nuclear extract (4 g) was incubated with 0.1 ng of labeled wild-type CFTR 37-bp double-stranded oligonucleotide (cftr(Ybox/CRE)) and electrophoresed in a nondenaturing acrylamide gel as described under "Materials and Methods." Some of the reactions also included unlabeled CFTR 37-or 27-bp oligonucleotides. Lane 1, no CREB or T84 nuclear extract. Lanes 2-6, CREB binding to the cftr(Ybox/CRE) oligonucleotide competed away with no competitor, 300 ϫ unlabeled cftr(Ybox/CRE), 300 ϫ unlabeled cftr(Ybox/mCRE) in which the variant CRE has been mutated, 300 ϫ unlabeled cftr37(mYbox/CRE) in which the Y box has been mutated, and 300 ϫ unlabeled cftr37(mYbox/mCRE) in which both sites are mutated. Lanes 7-22, T84 nuclear extract binding to the cftr(Ybox/CRE) probe competed away with 0 ϫ, 30 ϫ, 100 ϫ, and 300 ϫ unlabeled cftr(Ybox/CRE), cftr(Ybox/mCRE), cftr(mYbox/ CRE), or cftr(mYbox/mCRE), as indicated. Lanes 23-25, T84 nuclear extract binding to the cftr(Ybox/CRE) probe competed away with 0 ϫ competitor, 300 ϫ cftr(CRE), or 300 ϫ cftr(mCRE), in which the CRE has been mutated and the Y box is not present. B, recombinant CREB (60 ng) or T84 nuclear extract (4 g) was preincubated overnight with various antibodies and subsequently incubated with 0.1 ng of labeled cftr(Ybox/CRE) and electrophoresed in a nondenaturing acrylamide gel as described under "Materials and Methods." Lanes 2 and 3, 60 ng of recombinant CREB preincubated with anti-conalbumin (control) or anti-CREB antibody and subsequently incubated with cftr(Ybox/CRE) and electrophoresed. Lanes 4 -12, 4 g of T84 nuclear extract preincubated with the indicated antibodies followed by incubation with labeled cftr(Ybox/CRE) and electrophoresis. Lanes 13 and 14, 8 ng of recombinant ATF-2 fragment preincubated with anti-RI (control) or anti-ATF-2 antiserum and subsequently incubated with labeled cftr(Ybox/CRE) and electrophoresed. C, CaLu-3 nuclear extract (4 g) was preincubated overnight with the indicated antibodies and subsequently incubated with 0.1 ng of labeled cftr(Ybox/CRE) and electrophoresed in a nondenaturing acrylamide gel as described under "Materials and Methods." of Fig. 6. Since our binding studies (Figs. 4 and 5) indicate that the mutant CRE has no in vitro affinity for CREB, we suggest that the residual PKA response of CFTR(mCRE)-luc is mediated by other transcription factors that bind to the CFTR promoter. The mutant Y box-containing CFTR promoter is stimulated both by PKA and by CREB, although to a lesser extent than the wild-type promoter. These results suggest that PKA and CREB stimulate wild-type CFTR promoter activity predominantly through the CRE. Mutation of the adjacent Y box does, however, cause approximately a 2-fold attenuation of the responsiveness of the CFTR promoter to PKA, although the response of this construct to CREB alone remains similar to that of the wild-type construct.
Regulation of a Basal Heterologous Promoter by the CFTR Variant CRE-To determine whether the CFTR CRE can bestow cAMP regulation on a non-PKA-regulated basal promoter, we constructed reporter constructs containing the 27-mer oligonucleotide used for the EMSAs in Fig. 5 upstream of the basal tk promoter. This approach is similar to that employed by Pei et al. (19) for the CRE in the inhibin promoter, and since our constructs originally derived from the inh(CRE)tk-luc, we included this reporter in our transient transfections of JEG-3 cells. Fig. 7 depicts the -fold stimulation of each of our reporters by the cAMP-elevating agent forskolin. Both the inh(CRE)tkluc and cftr(CRE)tk-luc were stimulated ϳ10-fold by forskolin, while the cftr(mCRE)tk-luc was only stimulated ϳ2-fold and the basal tk-luc was stimulated ϳ1.5-fold. These results demonstrate that the CFTR CRE is capable of conferring a cAMP induction similar to that for other CREs and that the adjacent Y box is not required for this response under these conditions. DISCUSSION The CFTR promoter contains a variant CRE (TGACaTCA) at position Ϫ48 that we have shown to have all of the properties expected of a functional CRE. The CFTR CRE binds CREB and ATF family members with high affinity, and nuclear extracts from CFTR-expressing cell lines contain CREB/ATF family members that bind this sequence in electrophoretic mobility shift assays. Mutation of the variant CRE within the context of the Ϫ2.2-kilobase pair CFTR promoter reduces basal activity when assayed in T84 cells and prevents much of the PKA and CREB-mediated induction when assayed in JEG-3 cells. The CFTR CRE can be transferred to a heterologous promoter, where it confers robust cAMP-induced promoter activity, which is lost if the CRE is mutated.
Clearly, other elements within the CFTR promoter are also important for transcriptional activity and probably act cooperatively with the CRE in governing basal and cAMP-mediated CFTR expression in vivo. The persistent basal and cAMPinducible activity of the CFTR(mCRE) promoter suggests that some of these additional regulatory regions may in fact be bound by transcription factors that are regulated by cAMP. These regions include putative AP-1 sites at Ϫ1058, Ϫ976, Ϫ745, and Ϫ283, putative Sp1 sites at positions Ϫ335 and Ϫ256, and a putative AP-2 site at Ϫ1108 (17). Although the AP-1 site generally confers Fos/Jun-mediated activation and the Sp1 site generally modulates basal activity, both the AP-1 (1, 31, 32) and Sp1 (3,33) sites have been shown to mediate PKA responsiveness in certain promoters. AP-2 sites mediate basal and phorbol ester-stimulated activity and have also been , the wild-type CFTR CRE; cftr(cCRE), the CFTR CRE mutated to create a palindromic CRE in the context of the CFTR CRE; and cftr(mCRE), which replaces the CFTR core CRE with unrelated sequence. The various unlabeled competitors were used at concentrations of 30 ϫ, 100 ϫ, and 300 ϫ relative to the labeled probe, as indicated above the autoradiograph. B, similar to A, except using 8 ng of recombinant ATF-2 fragment in each reaction.
Our data suggest that one additional element important for the activity of the CFTR promoter is the Y box at position Ϫ60. This element was originally identified as a modulator of basal and PKA-stimulated activity of the CFTR promoter in PANC1 cells (18), and our results in JEG-3 and T84 cells support these findings. We have demonstrated that the CRE and Y box are independently regulated in T84 cells, although we have not observed binding of nuclear extract proteins to the Y box. This discrepancy may result from a low level of Y box-binding proteins in our nuclear extracts that is below the threshold for detection in the in vitro EMSA, or it may indicate that the Y box-binding proteins require additional sequences outside of the sequence of our probes for optimal binding. Our experiments have demonstrated, however, that there is no contribution by the Y box to CREB/ATF binding to the CFTR CRE and that the CREB-mediated stimulation of the promoter occurs almost exclusively through the CRE.
The regulation of the CFTR gene by PKA has been established now in several cell lines, including the T84 (13), HT-29 (27), CaLu-3, and HTE-1 cells. The level of inducibility of the endogenous CFTR gene by cAMP is dependent on cell type. We have observed only modest induction in T84 cells, although the high basal activity of the promoter in these cells is almost completely dependent on PKA activity. In contrast, the human tracheal epithelial cell line, HTE-1, expresses CFTR basally at very low levels, while the gene is induced 5-10-fold by cAMP. The CaLu-3 cells fall between the T84 and HTE-1 cells, with a moderately high degree of basal expression and a cAMP inducibility of ϳ3-fold. This range of regulation may reflect different "set points" for basal PKA activity, or it may be due to different compositions of cAMP-regulated transcription factors in the various cell lines.
The relative ratio of CREB to ATF proteins may govern the range of inducibility of the CFTR promoter in various cell lines. In general, we have observed a greater proportion of ATF with respect to CREB in the less cAMP-regulatable T84 cells in our EMSAs. Regardless of whether the ATF protein is ATF-1 or ATF-2, these proteins have been shown previously to be less responsive to cAMP than CREB. ATF-2 is not activated by PKA, and it forms heterodimers with Fos or Jun but not with CREB or ATF-1 (36), suggesting that ATF-2 bound to the CFTR CRE may mediate signaling through pathways other than the cAMP pathway. CREB and ATF-1 form heterodimers (36), and in cells with a high ratio of ATF-1 to CREB the magnitude of cAMP-mediated gene expression is attenuated (37). Clearly, however, there are several other possible mechanisms that could lead to the range of cAMP inducibility between cell lines, including differences in the amount and types of PKA present, and the final profile of regulation probably depends on multiple factors.
The potential for cAMP to regulate both the transcriptional activity of the CFTR gene and the functional activity of the CFTR protein provides a dual mechanism of control. Under normal physiological conditions, the CFTR protein is phosphorylated directly by PKA and thus regulated by hormones that signal stress. A greater augmentation of CFTR activity might occur when the signal is sufficient for an increase in CFTR gene expression. The clinical significance of the transcriptional effect is as yet unknown, but patients are often treated with cAMP-modulating drugs for a variety of conditions. For example, long term treatment of asthmatic patients with ␤2-adrenergic agonists might cause a chronic stimulation of the PKA pathway and an increase in CFTR gene expression.
Chronic treatment of cystic fibrosis patients with cAMPmodulating drugs might be beneficial if there is an increase in the level of CFTR at the plasma membrane, especially for those patients with milder disease that have mutations that produce lower levels or moderately functional CFTR. A recent report that overexpression of the ⌬F508 CFTR mRNA leads to functional expression (38), is interesting in this regard, since ⌬F508 FIG. 6. Effect of cotransfected CREB on CFTR-luc mutant constructs. JEG-3 cells in 24-well dishes were transiently transfected with 100 ng/well CFTR(wt)-luc, CFTR(mCRE)-luc, or CFTR(mYbox)-luc in addition to 50 ng/well RSV-lacZ. Cells were also cotransfected with 1 ng/well PKA catalytic subunit expression vector MT-CEV␣, 5 ng/well CREB expression vector RSV-CREB, and/or 5 ng/well dominant negative CREB expression vector RSV-KCREB, as indicated below the x axis. After a 24-h treatment with 80 M ZnSO 4 , cells were harvested and assayed for luciferase activity and ␤-galactosidase activity. Data are presented as a ratio of luciferase to ␤-galactosidase activity; error bars represent S.D. of wells treated in triplicate, individually transfected with the same DNA precipitate solution. The experiment presented is representative of five similar experiments.
FIG. 7. Ability of CFTR variant CRE to direct PKA-mediated expression from a basal promoter. JEG-3 cells were transfected with luciferase reporter constructs containing the CRE sequence from the ␣-inhibin promoter upstream of the basal tk promoter (inh(CRE)tkluc), 27 bp from the CFTR promoter containing the wild-type CRE upstream of the basal tk promoter (cftr(CRE)tk-luc), 27 bp from the CFTR promoter containing a mutated CRE upstream of the basal tk promoter (cftr(mCRE)tk-luc), or a short linker sequence upstream of the basal tk promoter (tk-luc). Luciferase constructs (5 ng) were cotransfected with 50 ng of RSV-lacZ for 24 h using the calcium phosphate procedure. After an 18-h recovery, cells were treated with forskolin (10 M) or vehicle control for 6 h, at which point cells were harvested and assayed for luciferase and ␤-galactosidase activity. Data are presented as -fold stimulation of forskolin-treated cells relative to untreated cells. Error bars represent S.D. of wells treated in triplicate transfected with the same DNA precipitate.
is the most common mutation in cystic fibrosis patients. Clinical studies examining the effects of PKA-stimulating bronchodilators, such as the ␤2-adrenergic agonist albuterol, on cystic fibrosis patients suggest that there may indeed be some therapeutic benefit with these drugs. Hordvik et al. (39) observed that long term albuterol treatment significantly improved the respiratory function of cystic fibrosis patients in a manner that was unrelated to the patients' asthma. Interestingly, the response required 2 weeks of treatment before most patients' conditions improved, and after the drug was discontinued many patients rebounded to conditions worse than before the onset (39). These changes are consistent with changes in gene expression. Our examination of the regulation of CFTR gene expression by PKA suggests that the potential mechanism by which drugs such as the ␤-adrenergic agonists may be beneficial could involve regulation through the CRE in the CFTR promoter.