Transcription of cystic fibrosis transmembrane conductance regulator requires a CCAAT-like element for both basal and cAMP-mediated regulation.

The cystic fibrosis transmembrane conductance regulator (CFTR) gene in man is controlled by a tightly regulated and weak promoter. The architecture of the CFTR promoter suggests regulatory characteristics that are consistent with the absence of a TATA-like sequence, including the ability to initiate RNA transcription at numerous positions. Detailed investigation of the most proximal region of the human CFTR gene promoter through deletion and mutational analysis reveals that expression is contingent on the conservation of the inverted CCAAT sequence. Basal expression of CFTR transcription and cAMP-mediated transcriptional regulation require the presence of an imperfect and inverted CCAAT element recognized as 5′-AATTGGAAGCAAAT-3′, located between 132 and 119 nucleotides upstream of the translational start site. RNA isolated from a transfected pancreatic cell line carrying integrated wild-type and mutant CFTR-directed transgenes was used to map the 5′ termini of the transgenic transcripts. Analysis of the transcript termini by ribonuclease protection analysis reflects the direct association of the conserved inverted CCAAT sequence in promoting transcript initiation. Because of the requirement for the inverted CCAAT sequence for promoting transcription of CFTR, the involvement of CCAAT-binding factors is suspected in the regulation of CFTR gene transcription. To test this, we used electrophoretic mobility shift assays to demonstrate that the majority of the binding to the inverted CCAAT element, between −135 and −116, was easily competed for by binding to cognate nucleotide sequences for CCAAT-enhancer binding protein (C/EBP). An antibody specific for the C/EBP-related protein, C/EBP, detected C/EBP as part of a nuclear protein complex bound to the inverted CCAAT sequence of the CFTR gene. Also, the detection of specific activating transcription factor/cyclic-AMP response element binding protein antigens by antibody supershift analysis of nuclear complexes suggest that species of this family of transcription factors could be involved in the formation of complexes with C/EBP within the CFTR gene inverted CCAAT-like element. These studies raise the possibility of interactions between individual members of the C/EBP and activating transcription factor/cyclic-AMP response element binding protein families potentially contribute to the tight transcriptional control rendered by the CFTR gene promoter.

The gene responsible for cystic fibrosis encodes the cystic fibrosis transmembrane conductance regulator (CFTR) 1 gene product (1,2). The expression of CFTR is predominantly confined to specific epithelial cell types and is ordinarily expressed in low levels. The low levels and cell type-specific expression of CFTR appears to be dictated, primarily, by genomic sequences 5Ј upstream of the translational start of CFTR, which correspond to functional promoter sequences (3,4). The requirements for active and cell type-specific CFTR transcription relies presumably within a narrow band of nucleotide sequences, proximal to the multiple transcript initiations (5). The levels of CFTR transcripts in individual cell types appear to be concomitant with the ability of certain epithelial cell types, which variably modulate expression of CFTR in response to 1) levels of cAMP (6), 2) the stimulation of protein kinase A and C activities (7)(8)(9), and 3) phorbol esters (10) likely through signaling pathways or mechanisms that converge ultimately on gene transcription (for review, see Ref. 11). The mechanisms that modulate CFTR gene expression through extracellular and intracellular signals may implicate activities analogous to that of both the C/EBP and ATF/CREB families of transcription factors (12).
The C/EBP-related family of nuclear transcription factors represent a class of proteins characterized by the ability to bind the CCAAT nucleotide consensus sequence and confer either transcriptional activation or repression on target genes (13)(14)(15)(16). All members in the C/EBP class of transcription factors contain a basic region and a leucine zipper (bZIP) domain, which correspond to both DNA binding (17,18) and protein dimerization potentials (19). The C/EBP family of nuclear proteins are also members of a larger superfamily of transcription factors also characterized by the bZIP motif that includes the ATF/CREB family of transcription factors (18). The ATF/CREB family has been characterized based on the recognition of several proteins to the cAMP-response element (CRE) nucleotide consensus (20 -24). It has been previously demonstrated that individual C/EBP proteins characterized thus far can homodimerize or heterodimerize within the C/EBP family of bZIP domain proteins to initiate a specific transcriptional effect (15,24) and could elicit transcriptional stimulation or repression in response to cAMP (25). It is now believed that the diverse transcriptional regulation of genes containing the recognition sites of either C/EBP and ATF/CREB may include the heterodimerization between different members in each of the C/EBP and ATF/CREB families (26). This mechanism may be utilized to respond to complex signals and transcriptional cues through single sequence elements including a response to cAMP, despite the absence of active CRE, AP1, and AP2 consensus nucleotide sequences (27,28) within the gene.
In this report we focused on the regulation of CFTR gene transcription mediated by the region encompassing only 135 nucleotides 5Ј upstream of the translational start site in pancreatic carcinoma, PANC1, cells. Through deletion analysis of the proximal region of the CFTR gene 5Ј-flanking sequences, we have identified a single cis-acting element. This cis-acting element is significant to the expression of CFTR gene transcription and represents the most proximal inverted and an imperfect CCAAT consensus essential for any detectable transcription. Analysis of the transcript termini provide evidence for the direct association of the conserved inverted CCAAT sequence in promoting and positioning transcript initiation. Thus, the core CCAAT element of the CFTR gene 5Ј-flanking region may reflect a weak initiator nucleotide consensus necessary for the generation of the multiple transcripts for CFTR gene expression. Studies to identify DNA-protein interactions were confirmed by both electrophoretic mobility shift and DNase protection assays to determine the extent of protein binding to the CFTR inverted CCAAT consensus region. Characterization of DNA-protein interactions between the CFTR inverted CCAAT sequence and nuclear proteins from PANC1 cells now provide evidence for the interaction with one member of the C/EBP family of transcription factors, C/EBP␦ (15,16) and also referred to as NFIL-6␤ (29). In addition, electrophoretic mobility shift assays also indicate the presence of CREB1 and ATF1 in the protein complexes bound to the CFTR inverted CCAAT element. Thus, these results suggest that the mechanism regulating CFTR gene transcription in PANC1 cells may include the trans-acting potentials of both the C/EBP and ATF/CREB families of transcription factors on a single inverted CCAAT element. Indication that the transcription factor C/EBP␦ (NFIL-6␤) is associated with a pancreatic cell nuclear complex bound to the CFTR inverted CCAAT element may also implicate regulation mediated by interleukin-6 (29). Hence, the complex nature of CFTR gene transcription in specific cells may require multiple interactions between bZIP families of transcription factors to direct the highly complex nature of transcriptional regulation demonstrated by the CFTR gene.

MATERIALS AND METHODS
Cell Culture-The cell line PANC1 (ATCC CCL 1469) cells and HeLa S3 (ATCC CCL 2.2) cells were cultured in high glucose (4.5 g/liter) Dulbecco's modified Eagle's medium supplemented with 10% fetal bovine serum (Hyclone Inc.) and gentamicin (1ϫ) (Boehringer Mannheim). Cell cultures were maintained in a CO 2 incubator at 37°C maintaining a 5% CO 2 humidified atmosphere. Cell culture medium was purchased through Life Technologies, Inc. To investigate the effect of cAMP on the CFTR promoter PANC1 cells transfected with plasmids containing the CFTR promoter, constructs (described below) were treated with 1 mM 8-Br-cAMP (Sigma) for the length of time indicated 36 h after DNA transfections were performed.
Gene Constructs-Plasmid DNA constructions were initiated by amplifying DNA fragments using the polymerase chain reaction (PCR) from a genomic clone of the human CFTR gene (4) with Taq DNA polymerase according to the manufacturer's recommendations (Perkin-Elmer, Inc.). All oligonucleotide primers for the polymerase were synthesized using a DNA synthesizer (Applied Biosystems, Inc.) Primers 5Ј-ccaagcttggtaattacgcaaagcattatc-3Ј and 5Ј-gctctagatctgtatcctttcctcaaaattgg-3Ј were used to generate a genomic fragment of the CFTR gene from Ϫ912 to ϩ88 relative to the translational start site. The PCR fragment was cloned as an intermediate subclone using the TA cloning system (Invitrogen, Corp.) according to the manufacturer's directions. The resulting clone was restricted with HindIII/XbaI double digestion and was fractionated on an agarose gel. The 1-kilobase pair fragment corresponding to the CFTR genomic region was excised from the gel and purified. The purified fragment was then ligated to the pØGH vector (30) backbone, which had been previously restricted with HindIII and XbaI and designated as pCF-912. Five 5Ј end point deletions of the CFTR gene promoter were carried by PCR using the following 5Ј end primers 5Ј-ccaagcttcacctgcgggatgggcccggt-3Ј, 5Ј-ccaagcttggtcgggaaagggaggtgcg-3Ј, 5Ј-ccaagcttcgtagtgggtggagaaagccgct-3Ј, 5Ј-ccaagcttggcagtgaaggcgggggaaagagc-3Ј, 5Ј-ccaagcttgggaattggaagcaaatga-3Ј, and 5Ј-ccaagcttcatcacagcaggtcagagaaaaaggg-3Ј with the 3Ј end primer 5Ј-gctctagatctgtatcctttcctcaaaattgg-3Ј generating fragments pCF-429, pCF-262, pCF-197, pCF-135, and pCF-116, respectively, relative to the translational start site of CFTR. Each of the PCR fragment were cloned into pCRII with the TA cloning system (Invitrogen, Corp.) according to the manufacturer's directions, and intermediate subclones were obtained. The DNA sequence of the resulting subclones were verified by DNA sequencing using the Sequenase II DNA sequencing kit (U. S. Biochemical Corp.) according to the manufacturer's instructions. PCR fragments were excised from the parental vector pCR II (Invitrogen, Corp.) using HindIII and XbaI and were gel purified. The purified fragments were then ligated overnight at 12°C into the pØGH vector at the HindIII and XbaI sites. The ligated DNA was then introduced into Escherichia coli, plated on culture plates containing ampicillin, and incubated at 37°C overnight. Bacterial clones, carrying correct inserts were isolated, verified by restriction analysis, and then sequenced. The resulting clones were designated as pCF-429/hGH, pCF-262/hGH, pCF-197wt/hGH, pCF-135/hGH, and pCF-116/hGH, respectively, relative to the translational start site of CFTR. The plasmid in which the inverted CCAAT sequence of ATTGG had been mutated was performed using the intermediate pCRII construct pCF-197 to direct a mutation into the ATTGG sequence to AT-GCG in pCRII. The mutant CFTR genomic fragment of pCF-197 was then cloned into the HindIII/XbaI site of pØGH. Mutagenesis of the CCAAT sequence was carried out as described by the manufacturer using the in vitro mutagenesis system, oligonucleotide-directed version 2.2 (Amersham Corp.). An oligonucleotide (5Ј-tgatgtcatttgcttcttccccccacccaccccta-3Ј) was used to direct a triple nucleotide deletion from the M13 phage using the positive strand of the CFTR gene promoter corresponding to the construct pCF-197⌬/hGH. Preparations of plasmids were performed followed by purification of plasmid DNAs by column chromatography using a Mono Q HR column by fast protein liquid chromatography (Pharmacia Biotech Inc.). The mutations were confirmed by sequencing with the Sequenase II kit (U. S. Biochemical Corp.).
Plasmid construct ptk(pL 124 )GH 2 was used as a heterologous promoter to insert fragments of the CFTR 5Ј genomic region containing the wild-type and mutant inverted CCAAT elements. Cloned fragments, corresponding to the CFTR wild-type and mutant gene promoters pCF-197wt and pCF-197mut served as DNA templates for the generation of PCR fragments. Synthetic oligonucleotides, 5Ј-ccaagcttgggcagtgaaggcgggggaaagagcaa-3Ј and 5Ј-ccaagcttgggctcaaccctttttctctgacctg-3Ј, were used to generate a 114-bp fragment spanning from Ϫ198 to Ϫ24 of CFTR promoter sequences, containing HindIII restriction adaptors, using Vent polymerase (New England Biolabs, Inc.) according to the manufacturer's directions. These fragments were, subsequently, cloned into a HindIII site of the heterologous ptk(pL 124 )GH thymidine kinase promoter vector, and appropriate plasmid clones were verified by restriction analysis and sequenced by the Sequenase kit (U. S. Biochemical Corp.) The plasmids p[CFϩccaat]tk/hGH, p[CF-ccaat]tk/hGH, and p[CFc⌬t]tk/hGH contain, respectively, the wild-type inverted CCAAT box element of the CFTR promoter in opposing orientations independently and the mutant inverted CCAAT element of the CFTR gene promoter.
Cell Transfections-For transient transfections, monolayer of cells were plated at a density of 5 ϫ 10 5 cells/35-mm plate (approximately 75% confluent) 18 h prior to transfections. Transfections were carried out using the cationic liposome-mediated DNA transfer method (31) using the reagent N- [1-(2,3-dioleoyloxy)propyl]-N,N,N,-trimethylammonium methylsulfate (DOTAP) (Boehringer Mannheim). Briefly, 5 g of plasmid reporter and 100 ng of internal control pRSV/CAT were diluted to a final volume of 50 l in 20 mM Hepes buffer, pH 7.6. The DNA mixture was transferred to a tube containing the 30 l of DOTAP 2 G. Shue, unpublished data. diluted in 100 l of 20 mM Hepes buffer, pH 7.6. This mixture was incubated for 15 min at room temperature. This DOTAP/DNA mixture was then transferred to 1.5 ml of medium. This medium was used to replace the existing medium in the cell cultures and gently rocked, briefly, for even distribution of the DOTAP/DNA mixture. Transfected cells were incubated for 6 h at 37°C in a 5% CO 2 atmosphere. At this time, medium was replaced with fresh culture medium and cultured for an additional 48 h before assaying the cultures and culture medium for chloramphenicol acetyltransferase activity (32) and human growth hormone (hGH). The levels of human growth hormone secreted into the medium from transfected cells were assayed by hGH enzyme-linked immunosorbent assay (Boehringer Mannheim) according to the manufacturer's instructions. The values representing hGH expression were normalized to the level of chloramphenicol acetyltransferase expression directed by the Rous sarcoma virus long terminal repeat sequences in each experiment. Positive and negative controls were included in each experiment conducted here using plasmid constructs containing the cytomegalovirus promoter fused to the human growth hormone gene and the plasmid pØGH, respectively. Levels of hGH were evaluated relative to the expression of pCMV/hGH.
Stable co-transfections of pancreatic carcinoma cells were carried out using DOTAP, identically, as described above with the constructs, pCF-197wt/hGH or pCF-197⌬/hGH or the control plasmid pØGH and the selectable marker, pMC1neoA (Stratagene, Inc.) at a molar ratio of 15:1 (reporter gene/selectable marker gene) in PANC1 cells. These co-transfections were carried in the same manner as above with the exception that cells were plated on 60-mm plates at a lower density of approximately 5 ϫ 10 5 cells/plate. Following transfection, the culture medium containing the DOTAP/DNA mixture was replaced by the selection medium containing 400 g/ml Geneticin® (G418) (Life Technologies, Inc.). Stable cell populations were established after 14 days in selectable medium. Genomic DNAs were isolated from individual clonal cell populations, and the integration of the transgenes along with the copy numbers were then confirmed by Southern dot hybridization analysis using probes spanning the bacterial sequences of the transgenes. Monoclonal populations of stably transfected cells pCF-197wt/hGHpac, pCF-197⌬/hGHpac, or pØGHpac were then expanded further in 60-mm plates and maintained in selectable medium containing G418. 20 confluent plates of each stably transfected cell line containing either the pCF-197/hGH, pCF-197⌬/hGH, or pØGH transgene construct were used for the preparation of cytoplasmic RNA (33).
Ribonuclease Protection Assays-The CFTR/hGH antisense RNA probe was constructed by generating PCR fragments from the pCF-197wt/hGH, pCF-197⌬/hGH, and pØGH plasmid constructs and cloning the PCR fragments into pCRII using the TA Cloning system (Invitrogen Corp.). Individual clones were obtained to represent each of the transgene constructs integrated as stably transfected cell lines. Plasmids were sequenced and used to generate specific [␣-32 P]UTP-labeled Sp6 RNA polymerase transcripts, spanning between position Ϫ197 of the CFTR 5Ј genomic region to 137 nucleotides downstream of the BamHI site in the hGH gene. In the case with the pØGH transgene control, lacking CFTR genomic sequences, a PCR fragment of 425 bp was generated spanning between the pUC12 backbone of the pØGH vector and the hGH gene (30). The resulting 32 P-labeled RNA transcript incorporates 137 nucleotides of the human growth hormone gene from the vector backbone and 285 and 282 bp of 5Ј-flanking sequence of the CFTR gene, corresponding to the wild-type and mutant constructs, respectively. Both the 492-and 489-nucleotide [␣-32 P]UTP-labeled Sp6 RNA polymerase transcript was examined by polyacrylamide gel electrophoresis, and the RNA transcript was excised from the gel and purified from the polyacrylamide gel. Cytoplasmic RNA was isolated from stably transfected cell lines both CFTR-197wt/hGHpac, CFTR-197⌬/hGHpac, and pØGHpac. Cytoplasmic RNAs (50 g) were denatured and allowed to anneal to the antisense RNA probe (ϳ5 ϫ 10 6 cpm) for 16 h at 52°C. The RNA mixture was digested with ribonucleases A (40 g/ml) and T1 (2 g/ml) for 30 min at 30°C. Digestion of RNA was terminated with the addition of sodium dodecyl sulfate and proteinase K. RNA was extracted with phenol/chloroform mixture and precipitated, and the protected fragments of cytoplasmic RNAs were resolved on a 4% polyacrylamide gel.
Preparation of Nuclear Extracts, Electrophoretic Mobility Shift Assays, and DNase Protection Assay-Nuclear extracts from human PANC1 cells were prepared essentially as described previously (34) with modifications to the preparation of nuclear extracts described previously (35). The extracts were frozen under liquid N 2 and stored at Ϫ80°C until further use. The protein concentrations of the extracts were determined by UV absorbance. The double-stranded oligonucleotide corresponding to sequences of the CFTR-inverted CCAAT element 5Ј-tggggggaattggaagccaaatgacat-3Ј was synthesized and labeled at the 5Ј end by incubating with [␥-32 P]ATP and T4 polynucleotide kinase. The binding reactions were carried out at 4°C for 30 min with 0.1 ng of the 32 P-labeled oligonucleotide, 0.5 g of nonspecific competitor DNA (poly(dI⅐dC)) and either 2 or 10 g of PANC1 nuclear extract in the following buffer conditions: 20 mM Hepes, pH 7.9, 60 mM KCl, 0.1 mM EDTA, 5% (v/v) glycerol. For competition experiments, unlabeled competitor oligonucleotides corresponding to nucleotide sequences of the wild type CFTR inverted CCAAT element, the mutated inverted CCAAT element 5Ј-tggggggaagaagccaaatgacat-3Ј, C/EBP consensus 5Јtgcagattgcgcaatctgca-3Ј (Santa Cruz Biotechnology, Inc.), the rat albumin promoter (DEI site) 5Ј-tcgactatgattttgtaatggggc-3Ј (15), and the human c-fos C/EBP target 5Ј-gatccattgctaataatg-3Ј (36) were added in molar excess to the binding reactions prior to the introduction of labeled oligonucleotides. A control oligonucleotide (5Ј-gggtatttgcagtgatgtaatcagc-3Ј) of the human hemopexin gene A site (37) was used to determine the effectiveness of polyclonal antibodies specifically directed against C/EBP␦. DNA-protein complexes were resolved by electrophoresis on a 5% polyacrylamide/bisacrylamide (80:1) in Tris acetate/EDTA buffer at 10 V/cm at 4°C. Antibody supershift assays were performed similarly, that either polyclonal antibody (1 g) to C/EBP␦ (Santa Cruz Biotechnology, Inc.) or preimmune serum was added with bovine serum albumin (5 g) directly to the DNA/nuclear extract mixture and allowed to incubate at 4°C for an additional 30 min. DNA-protein and antibody complexes were fractionated on a 5% polyacrylamide/bisacrylamide (80:1) in Tris acetate/EDTA buffer at 10 V/cm at 4°C.
DNase protection assays were performed using nuclear extracts prepared from untreated and 8-Br-cAMP-treated cells on the 127-bp fragment homologous to the region of the CFTR promoter between Ϫ198 and Ϫ71 upstream of the translation start site. A DNA fragment was generated with the primers 5Ј-gggcagtgaaggcgggggaaagagc-3Ј and 5Јctgggtgcctgccgctcaaccctt-3Ј. The 5Ј oligonucleotide 5Ј-gggcagtgaaggcgggggaaaga-3Ј was labeled by using [␥-32 P]ATP and T4 polynucleotide kinase. Addition of the [␥-32 P]ATP-labeled oligonucleotide (ϳ 5 pmol) to the PCR was used to generate a 32 P-labeled DNA fragment. The DNA fragment was excised and purified from a 5% polyacrylamide gel. DNase protection experiments were carried out as described previously (38). To confirm the location of protection from DNase hydrolysis, corresponding Maxam-Gilbert (G ϩ A) reactions were conducted on the 32 P-labeled DNA fragment.

Delimitation of a Functional CFTR Gene Promoter Demonstrates Both Basal Transcription and cAMP-mediated
Regulation-Characterization of specific cis-acting elements involved in directing transcription have been complicated by the low levels of CFTR gene transcription demonstrated in most epithelial cell types (5). Therefore, we decided to test the promoter function of nucleotide sequences 5Ј upstream of the CFTR gene by performing transient DNA transfections in PANC1 cells. Unlike many other cell lines tested, PANC1 had been shown to support relatively high levels of reporter gene activity when directed by the CFTR promoter in a transfection assay (5). In this study, we utilized the hGH gene as a reporter, taking advantage of its sensitivity to transcriptional regulation and stability of its transcripts (30). A series of deletions of the CFTR gene promoter were constructed by the polymerase chain reaction using end points at selected sites up to 912 nucleotides upstream of the open reading frame. The CFTR gene promoter sequences were cloned into the pØGH plasmid vector, transfected into PANC1 cells, and assayed for hGH expression. Fig. 1 shows that the major elements of transcriptional control of CFTR gene promoter function in the pancreatic cell line are confined to sequences approximately 135 nucleotides from the translational start site. In control cultures (hatched bars) not induced by 8-Br-cAMP, the promoter sequence extending 912 nucleotides from the open reading frame produced significant levels of hGH expression, approximately 5-fold greater than control cells transfected with reporter gene constructs lacking all CFTR gene sequences (pØGH). Subsequent deletion from the 5Ј end up to 135 nucleotides from the translational start of CFTR diminished basal transcription activity only by 20%; however, deletion of a further 19 bp (pCF-116/hGH) re-duced transcriptional activity to a level 50% lower than that of the negative control, pØGH, which lacked all CFTR nucleotide sequences. This reduction of promoter activity in pCF-116/hGH to a point below the clearly detectable level of hGH expression pØGH likely indicates that cryptic promoter sites exist within the pØGH plasmid backbone or the hGH gene, and that the activity of these sites is suppressed by sequences within the proximal 116 bp of the CFTR promoter. Consistent with this interpretation is the finding that successive deletion of sequences downstream of Ϫ116 restored hGH expression to the background levels expressed by pØGH (data not shown); however, we did not perform experiments to substantiate the presence of DNA element corresponding to transcriptional repression.
In addition to supporting basal transcriptional activity, the same CFTR promoter constructs also demonstrate cAMP-stimulated expression of the hGH gene (Fig. 1, black bars). Essentially the same level of activity in the presence of 8-Br-cAMP, representing 3-4-fold stimulation over untreated controls, was seen in cells transfected with promoter constructs extending 135 or more bp upstream of the translational start site. Deletion of sequences upstream Ϫ116 bp reduced activity to below that of the pØGH control construct, abolishing the effect of 8-Br-cAMP.
A Functional Inverted CCAAT Element Is Identified by Mutagenesis-Evidence of a cis-acting element residing between a 19-bp sequence between Ϫ135 and Ϫ116, was demonstrated by the site-directed mutation of this nucleotide sequence. This 19-bp sequence of interest contains a suspected inverted CCAAT element, like those previously implicated in directing constitutive and inducible RNA polymerase II-mediated transcription of several genes (39), either in tandem with or in the absence of a TATAA element (40). We altered the inverted CCAAT sequence 5Ј-ggaattggaa-3Ј present in pCF-197 wildtype construct to a 5Ј-ggaagaa-3Ј by site-directed mutagenesis to yield the construct pCF-197⌬hGH. Cells transfected with this plasmid exhibited markedly reduced levels of basal expres-sion of hGH but also abolished the induction of hGH expression rendered by the administration of 8-Br-cAMP (Fig. 1). Taken separately, this result supports the contention that conservation of the inverted CCAAT element may be an essential requirement for the basal expression of CFTR gene transcription.
In addition to mediating basal expression of CFTR gene transcription, the inverted CCAAT element was strictly required for cAMP-mediated induction of transcriptional activity. Moreover, the absence of any demonstrable cAMP-mediated activation of CFTR gene transcription via existing sequences homologous to both CRE and AP1 suggest that this inverted CCAAT element is, alone, sufficient for cAMP responsiveness within the CFTR gene (data not shown). Recent data have indicated that the induction of CFTR gene expression was mediated by cAMP likely acting through putative cis elements of the CFTR gene promoter (6). The best studied transcriptional response by cAMP was identified through the analysis of the conventional CRE nucleotide consensus (20). Despite the presence of both CRE and AP1 consensus sites upstream in the CFTR gene 5Ј-flanking region (3,4), these sequences appeared to lack substantial cis-acting potential within the context of the CFTR gene (5,41). To map the nucleotide sequences responsible for the observed induction of CFTR by cAMP, deletion constructs ( Fig. 1) were examined for their ability to support cAMP-mediated induction of CFTR gene transcription. Constructs pCF-912/hGH, pCF-492/hGH, pCF-262/hGH, pCF-197/ hGH, and pCF-135/hGH each averaged a stimulation of approximately 3-fold, upon induction with the cAMP analog, 8-Br-cAMP (Fig. 1). This activation was detected following a 12-h treatment of cells with 8-Br-cAMP. A 19-bp deletion downstream of the Ϫ135 nucleotide position abolished cAMP induction as indicated by construct pCF-116/hGH (Fig. 1). Mutation to the inverted CCAAT sequence demonstrates the strict requirement of the conserved CCAAT sequence to mediate the induction by 8-Br-cAMP on hGH expression. Construct pCF-197⌬/hGH contains a three-nucleotide deletion in the inverted CCAAT sequence to C-T. The loss in cAMP-mediated regulation directly corresponds to the deletion of three nucleotides in the inverted CCAAT element. In addition, the absence of any CRE consensus indicates the level of cAMP-inducible transcription of CFTR is mediated by the inverted CCAAT element, alone. The cAMP-mediated induction of hGH expression directed by the CFTR gene promoter from construct pCF-197/ hGH requires the conservation of the inverted CCAAT element consensus in the CFTR gene promoter. This result is consistent with other examples of cAMP induction mediated by the CCAAT element, including the recent examination of the Gprotein i-2 subunit gene promoter (42). Thus, the results indicate here that the potential role of the inverted CCAAT element in the CFTR gene is to mediate both the basal transcription of CFTR and induction by cAMP.
Inverted CCAAT Element of the CFTR Promoter Directs cAMP-mediated Transcription in a Heterologous Gene-The inverted CCAAT element from the 5Ј-flanking sequences of the CFTR gene demonstrates both transcriptional cis activation and cAMP-mediated transactivation. To test heterologous cAMP-mediated transcription, the thymidine kinase promoter of the herpes simplex virus was fused to the human CFTR gene 5Ј-flanking sequence. Both wild-type and mutant CFTR templates (pCF-197wt/hGH and pCF-197⌬/hGH) were utilized for PCR amplification of DNA to produce a 158-and 155-bp fragment of the CFTR promoter, respectively. The human CFTR gene sequences were then fused upstream of the thymidine kinase promoter in constructs p[CFϩccaat]tk/hGH, p[CFccaat]tk/hGH, and p[CFc⌬t]tk/hGH. These constructs were then tested for the ability to confer transcriptional activation and cAMP induction of human growth hormone expression in both orientations (Fig. 2). Co-transfections were performed on PANC1 and HeLa S3 cells in the presence of an internal control plasmid pRSV/CAT with p[CFϩccaat]tk/hGH, p[CF-ccaat]tk/ hGH, or p[CFc⌬t]tk/hGH. As shown in Fig. 2, the basal levels of transcription mediated by the thymidine kinase promoter were activated by introducing the CFTR 5Ј genomic inverted CCAAT element in both orientations. A mutation, identical to that directed in construct pCF-197⌬/hGH, was placed within the context of the thymidine kinase promoter to produce the plasmid p[CFc⌬t]tk/hGH. Mutation to the inverted CCAAT consensus by deleting three nucleotides demonstrates the requirement of the CCAAT consensus for trans-activation through the thymidine kinase promoter. Deletion to the core CCAAT sequence eliminates any stimulatory activity as reflected by the negative control parental plasmid backbone, ptk/hGH. The induction by 8-Br-cAMP of human growth hormone expression in PANC1 cells transfected with p[CFϩccaat]tk/hGH and p[CFccaat]tk/hGH indicates a greater than 3-and 4-fold increase, respectively. Comparatively, HeLa S3 cells reflect a much more modest change of approximately 2-fold. Plasmid p[CFc⌬t]tk/ hGH failed to generate any increase in the expression of human growth hormone in either cell line, indicating the role of the inverted CCAAT sequence in either orientation was sufficient to confer the induction by cAMP.
The Inverted CCAAT Region Mediates Transcription from Multiple Start Sites-Previous studies of transcript initiation along the CFTR gene previously revealed several start sites (1,3,5), although their precise locations remain to be established. To determine if gene transcription from multiple initiation sites is associated with the inverted CCAAT sequence of the CFTR gene, we used a ribonuclease protection assay to identify the 5Ј termini of the transcripts. PANC1 cells were stably co-transfected with individual hGH reporter transgene constructs pCF-197wt/hGH or pCF-197⌬/hGH or the control plasmid pØGH in combination with a selectable gene. Clonal populations of stably transfected cells carrying the hGH transgenes were established and confirmed by genomic DNA hybridization analysis (data not shown), and cytoplasmic RNA was isolated. Individual antisense 32 P-labeled RNA probes, were created corresponding to the wild-type and mutant CFTR transgene constructs, pCF-197wt/hGH and pCF-197⌬/hGH, respectively. These probes, spanned the sequences between the CFTR gene promoter and the human growth hormone gene. As shown in Fig. 3, a protection assay of the wild-type CFTR transgene, isolated from two independent cell clones pCF-197wt/hGHpac-A and pCF-197wt/hGHpac-E, yielded several distinct protected fragments within the CFTR 5Ј genomic region within a 153-bp sequence upsteam of the open reading frame for CFTR. In contrast, mutant transgenes from the two independent clones, pCF-197⌬/hGHpac-G and pCF-197⌬/hGHpac-K, lacking the wild-type inverted CCAAT consensus, failed to produce transcripts from any of those end points. Control assays using native PANC1 cell cytoplasmic RNA (not shown) and PANC1 cells stably transfected with the pØGH parental vector showed no protected transcripts (Fig. 3) as expected. In Fig. 3, qualitative comparisons were made between the pattern of transcript termini rendered by the wild-type and mutant inverted CCAAT element CFTR gene constructs within the region encompassing the 5Ј-flanking nucleotide sequences of the CFTR gene. The result from this experiment indicates that despite the variable sites of transcript initiation, transcripts may be directed through a weak nucleotide consensus, which is associated with the inverted CCAAT element. The results here may imply a role of the inverted CCAAT element of the CFTR gene in stabilizing a basal transcription mechanism through the cis-acting element.
DNA-binding Proteins Interacting with the Inverted CCAAT Element Region of the CFTR Gene Promoter-To determine if CFTR gene regulation, mediated by the inverted CCAAT element, is due to specific DNA-protein interactions, electrophoretic mobility shift assays (EMSA) were performed. Fig. 4A FIG. 2. Proximal element of the human CFTR gene directs cAMP-mediated transcription through the inverted CCAAT element in a heterologous gene. Plasmid constructs containing wildtype and mutant inverted CCAAT sequence fragment of the human CFTR gene 5Ј-flanking nucleotide sequences were ligated to a thymidine kinase gene promoter linked to the human growth hormone gene (see "Materials and Methods"). Transient transfections were performed on the PANC1 (top graph) and HeLa S3 (bottom graph) cell lines with the constructs containing the inverted CCAAT element of the CFTR gene in the sense orientation (p[CFϩccaat]tk/hGH), antisense orientation (p[CF-ccaat]tk/hGH), and with the mutated CCAAT consensus, p[CFc⌬t]tk/hGH, followed by treatment of cells with 8-Br-cAMP for 24 h. The levels of human growth hormone were determined in the medium as described under "Material and Methods" and were quantitated relative to the levels of hGH generated by the pCMV/hGH as described in the legend to Fig. 1. Levels of hGH expression were normalized for transfection efficiency by the comparison to a internal control plasmid pRSV/cat. shows that nuclear extracts prepared from PANC1 cells bind the oligonucleotide, corresponding to the inverted CCAAT sequence of the CFTR gene promoter. A single major band was seen, as well as two weak slower migrating complexes. Formation of the predominant complex and one of the minor complexes was specifically depleted by increasing concentrations of unlabeled oligonucleotide. To identify DNA-binding proteins, competition with unlabeled oligonucleotides corresponding to the cognate sequence of the C/EBP protein consensus was used to confirm the specificity of DNA binding to the 32 P-labeled cognate sequence of the inverted CCAAT element of the CFTR gene. In Fig. 4B, oligonucleotides corresponding to several known C/EBP targets were used in a competition EMSA. 10fold amount of unlabeled oligonucleotide were included in the reactions prior to the addition of the 32 P-labeled CFTR inverted CCAAT element oligonucleotide. Oligonucleotide targets for C/EBP binding from the albumin gene, c-fos gene and the dyad symmetric consensus sequence for C/EBP binding (43) (16) (18), respectively, competed for binding. This result occurred despite the relatively modest excess of unlabeled oligonucleotide (10-fold). We suggest that the same or related proteins that bind to the C/EBP elements of these genes could target the CFTR gene promoter. A saturating excess (100-fold) of the oligonucleotide corresponding to a deletion mutant, without the inverted CCAAT element failed to compete for binding to the wild-type inverted CCAAT element of the CFTR gene when compared with the controls shown. Interestingly, nuclear extracts prepared from PANC1 cells treated with cAMP failed to demonstrate induction or banding patterns that differed from untreated PANC1 cells (not shown).
To compare DNA binding affinities between cAMP-induced and basal levels of transcription, directed by the proximal 5Ј end of the human CFTR gene, nuclear extracts prepared from PANC1 cells treated with 8-Br-cAMP and untreated cells were used for DNase I protection assays. The footprint visualized in Fig. 5 is indicative of protection of the CFTR gene promoter, is positioned between Ϫ146 and Ϫ90 upstream of the open reading frame of CFTR, and presents a large region protected from deoxyribonuclease hydrolysis (Fig. 5). Interestingly, this footprint is primarily a characteristic of the extract derived from the untreated control and not from nuclear extract isolated from cells treated with 8-Br-cAMP. The footprint demonstrated in nuclear extracts from untreated PANC1 cells generates extensive binding, overlapping the inverted CCAAT element of the human CFTR gene. Comparatively, DNA binding affinity is slightly detectable in PANC1 cells stimulated with 8-Br-cAMP (Fig. 5) in accord with the stimulation of reporter gene transcription by cAMP as shown in Figs. 1 and 2. This result is consistent with a DNA-protein complex constitutively bound to CCAAT box sequences previously shown to contribute to the weak and basal levels of human gp91-hox gene transcription through the displacement of more potent transcriptional activators (44). This model accounts for the relative absence of change in the detection of multiple DNA-protein complexes by EMSA from various cell types despite dramatic changes in the level of gene transcription mediated by the CCAAT box (44).
Identification of a C/EBP Member as Part of the Inverted   FIG. 3. The inverted CCAAT sequence of the CFTR gene directs multiple but discrete transcript initiations. Ribonuclease protection was performed on stably transfected cells carrying the pCF-197wt/hGHpac and pCF-197mut/hGHpac transgenes, containing the CFTR promoter-human growth hormone fusion constructs as described under "Materials and Methods." Radiolabeled antisense riboprobes of 422 and 419 nucleotides in length corresponding to the wild-type and mutant CFTR transgene sequences, respectively, were generated by Sp6 RNA polymerase. A 425-nucleotide riboprobe, spanning the hGH gene sequences and pUC12 sequences of the parental vector pØGH, was used as a negative control to correct for background signals generated from the stably transfected cell line carrying the pØGHpac gene. 50 g of cytoplasmic RNA from stably transfected cells or yeast tRNA were allowed to anneal to their respective antisense riboprobes overnight. Hybridized RNA strands were hydrolyzed by ribonuclease, and the protected RNA fragments were resolved on a 6% polyacrylamide sequencing gel. The size of the protected transcripts and position of the transcript termini were mapped using a DNA sequence ladder in adjacent lanes (not shown). The schematic depicted above the representative autoradiograph indicates position of transcript termini relative to the region spanning the CFTR promoter between Ϫ154 and Ϫ48 upstream relative to the translational start site of the CFTR gene. The nucleotide sequence on top of the schematic is that of the wild-type CFTR gene. The line above the sequence refers to the inverted CCAAT sequence; the line below the sequence denotes the weak CAAT consensus found in the CFTR gene (3). Closed triangles indicate the positions of the 5Ј termini of the major transcripts, relative to the CFTR promoter, apparent from the wild-type transgene pCF-197wt/hGHpac. The bent arrows indicate the presence and direction of the transcripts detected within the 5Ј-flanking region of the CFTR gene evaluated in this assay. Down arrows refer to the position of transcript 5Ј termini emanating from minor transcriptional start sites.

FIG. 4. Competition analysis of PANC1 cell nuclear proteins bound to the CFTR gene inverted CCAAT element. Panel A,
EMSA was performed using a double-stranded oligonucleotide corresponding to sequences of the CFTR-inverted CCAAT element 5Ј-tggggggaattggaagccaaatgacat-3Ј. The oligonucleotide was synthesized and labeled at the 5Ј end by incubating with [␥-32 P]ATP and T4 polynucleotide kinase. Nuclear extract preparation, binding reactions to the radiolabeled oligonucleotide, and electrophoresis were performed as described under "Materials and Methods." 2 g of nuclear extract and unlabeled oligonucleotide DNA, identical to the 32 P-labeled oligonucleotide sequence probe and used as competitor, was introduced into the binding reaction in increasing amounts prior to the addition of the radiolabeled oligonucleotide. Values indicate the molar excess (x-fold) used relative to the amount of 32 P-labeled oligonucleotide introduced into the reaction. Panel B, competing unlabeled oligonucleotides, as indicated above the figure, were introduced into the binding reactions prior to the introduction of the 32 P-labeled inverted CCAAT oligonucleotide sequence as described above. A 10 or 100-fold excess (relative to the input of 32 P-labeled oligonucleotide) of competing oligonucleotides were added to each of the reactions. The competitor oligonucleotides used at 10-fold excess were the wild-type CFTR-inverted CCAAT element, the C/EBP binding consensus of the c-fos gene promoter, the C/EBP binding consensus of the albumin CCAAT box sequence, and the symmetrical dyad consensus for optimal binding by C/EBP proteins. The mutant CFTRinverted CCAAT sequence unlabeled oligonucleotide was added at a 100-fold excess in the binding reactions (see "Materials and Methods").
CCAAT Nucleotide Binding Complex-Experiments to identify protein species as part of the DNA-protein complex(es) associated with the inverted CCAAT element of the CFTR gene promoter were performed by an antibody supershift analysis using EMSA. The experiment conducted shown as Fig. 6A used the 32 P-labeled oligonucleotide, corresponding to the wild-type sequence of the inverted CCAAT element of the CFTR gene in the presence of nuclear extracts isolated from PANC1 cells. An antibody directed against the related C/EBP protein, C/EBP␦, was used to identify C/EBP␦ within the inverted CCAAT-bound complex. The addition of C/EBP␦-specific antiserum (1:100 dilution), but not the preimmune serum (lane 6), supershifted a single detectable band shown in the lanes 5 and 7. Using a 5-fold amount of specific competitor, oligonucleotide sequences corresponding to the C/EBP consensus of 5Ј-tgcagattgcgcaatctgca-3Ј readily competed for binding to the inverted CCAAT element of the CFTR gene. As shown in Fig. 6A, lane 7, nuclear extracts prepared from PANC1 cells following the treatment with cAMP demonstrated no apparent increase or change in the amount of C/EBP␦ in the complex bound to the inverted CCAAT element. Efforts to demonstrate supershifted complexes with antibodies specific for C/EBP␣, C/EBP␤, or CRP1 do not indicate the presence of these proteins in the DNAprotein complex (not shown), although this does not preclude the involvement of other C/EBP family members with the inverted CCAAT element of the CFTR gene promoter. As a comparative control, the antibody directed against C/EBP␦ was FIG. 5. Comparative analysis of DNase I protection of the inverted CCAAT element with nuclear extracts treated with 8-Br-cAMP. Genomic fragment of the CFTR gene encompassing nucleotides between Ϫ196 and Ϫ74 nucleotides upstream of the CFTR gene open reading frame for translation was used for the DNase protection experiment. Preparation of the 32 P-labeled DNA probe and nuclear extract from PANC1 cells were performed as described under "Materials and Methods." Nuclear extracts prepared from PANC1 cells treated with 1 mM 8-Br-cAMP and untreated PANC1 cells were incubated with the 32 P-labeled DNA fragment. The binding reactions were then subjected to hydrolysis with deoxyribonuclease I and fractionated on a 8% nucleotide sequencing gel. Numbers above the lanes indicate the amount of protein from the nuclear extract added to the reaction prior to partial deoxyribonuclease I hydrolysis. Protein amounts (100 g) remained constant in each lane by the addition of bovine serum albumin. The bracket outline indicates the region of protection from DNase I hydrolysis by nuclear extract. The footprint was mapped by comparison of Maxam and Gilbert sequence reactions performed on the identical 32 P-labeled DNA fragment (not shown) and indicated as numerical nucleotide position relative to the translational start site. The position of the inverted CCAAT element of the CFTR gene promoter is indicated by the solid box and nucleotide sequence adjacent to the figure.
FIG. 6. C/EBP␦ is detected in the nuclear protein complex bound to the human CFTR gene promoter inverted CCAAT element. EMSA experiments shown were carried out as described under "Materials and Methods." Panel A, EMSA using the oligonucleotide containing the inverted CCAAT element of the human CFTR gene as a probe (see legend to Fig. 4). Nuclear extracts (10 g) prepared from untreated and 8-Br-cAMP-treated PANC1 cells were added along with the radiolabeled probe followed by the addition of 1 g of antiserum directed against C/EBP␦ or rabbit preimmune serum in the lanes indicated. Competitor oligonucleotide DNA representing the typical dyad consensus for C/EBP binding was introduced into the binding reactions in the lanes indicated at a 25-fold molar excess relative to the amount of 32 P-labeled probe used. Arrows (A and BЈ), depict the bandshifts of DNA-protein complexes. A indicates the position of the prominent DNA-protein complex generated from the PANC1 cell nuclear extract. BЈ indicates the position of the DNA-protein-antibody supershift complex detected in the autoradiograph. Panel B, an EMSA experiment using nuclear extracts prepared from PANC1 cells and an oligonucleotide probe carrying a prototypical C/EBP␦ consensus site from the human hemopexin gene A site (37). 1 g of C/EBP␦-specific antibody or preimmune serum was added to the binding reactions where indicated. Arrows, depict the bandshifts of DNA-protein complexes. tested with PANC1 nuclear extract bound to the cognate oligonucleotide sequence of a C/EBP␦-inducible gene (Fig. 6B). As the results indicate from Fig. 6, we can readily detect the C/EBP␦ protein from the polyclonal antibodies used in these experiments.
ATF1 and CREB1 Appear in Complexes Bound to the Inverted CCAAT Element of the CFTR Gene-In an effort to understand the regulation of CFTR transcription mediated by cAMP, investigation of the CREB and the ATF families of transcription factors would provide evidence for a role of these factors in response to cAMP-mediated transcription. An EMSA antibody supershift was performed using nuclear extracts prepared from PANC1 cells. Polyclonal antibodies against CREB1, CREB2, ATF1, ATF2, and ATF3 were used to detect the presence of these proteins in protein complexes bound by the CFTR gene-inverted CCAAT element. As shown in Fig. 7 bandshifts were detected as the major DNA-protein complex in each of the lanes containing nuclear extract, but only antibodies to CREB1 and ATF1 could supershift the complex in their corresponding lanes. This would indicate the presence of CREB1 and ATF1 in DNA-protein complexes formed within the CFTR gene inverted CCAAT element. DISCUSSION The expression of CFTR is controlled by different regulatory mechanisms (6,7,9,(45)(46)(47)(48), suggesting that an array of signals could be required to maintain the activity of the CFTR cAMP-mediated Cl Ϫ channel. Intracellular signaling by cAMP is a major component in the modulation of CFTR activity (49). With regard to the processes that modulate the function of CFTR through cyclic AMP-mediated activation of Cl Ϫ channels, a complementary model has included the stimulatory effects mediated by cAMP on CFTR gene transcription (6), therefore providing further provocation of epithelial cells to direct the expression of sufficient levels of CFTR through signaling pathways mediated by cAMP. In this report, the CFTR gene 5Ј-flanking region serves as a model promoter to examine the role of nucleotide sequences proximal to both transcriptional and translational start sites in the complex regulation of CFTR gene transcription. The expression of CFTR gene tran-scription is consistent with the characteristically weak promoter that utilizes multiple initiation sites for transcription (3,4). Through deletion and mutagenesis of the 5Ј-flanking sequences of the CFTR gene, we demonstrate that the basal transcription of CFTR in PANC1 is conserved by using sequences extending only 135 nucleotides upstream of the open reading frame for translation. Experiments were performed and indicate the role of an inverted CCAAT sequence in directing both basal and cAMP-inducible regulation of CFTR. The presence of the cis-element containing both an inverted and imperfect repeat of the canonical CCAAT sequence of ATTG-GAAGCAAAT within the deleted 19-bp element indicates a possible relationship between the inverted CCAAT element and transcriptional regulatory function. Analysis of the inverted CCAAT element, proximal to many transcript initiations directed by the CFTR gene (3)(4)(5), demonstrates that conservation of the inverted CCAAT consensus is essential for maintaining basal levels of reporter gene activity in a transient expression assay. Based on this characterization of the CFTR gene promoter, mutation of nucleotides within the inverted CCAAT sequence virtually abolished the expression of reporter gene activity. Although these studies cannot account for cell type-specific differences in the expression of CFTR gene transcription, this provides some fundamental understanding of the nucleotide requirements for directing CFTR gene transcription.
Despite the absence of nucleotide consensus for CRE, AP1, or AP2 activity within the context our promoter constructs, we have evaluated reporter gene levels of transfected PANC1 cells stimulated with cAMP. The cis-acting element, which apparently directs basal transcription, can also account for the induction of CFTR gene transcription by cAMP in these studies. These unexpected results from our transfection studies performed in this report indicate that cAMP-mediated regulation resides within the nucleotide sequences also directing basal levels of transcription. To address the logical flaw in interpreting the cAMP-mediated regulation conferred by the same nucleotide sequences directing basal expression of CFTR gene transcription, we had constructed heterologous gene promoter constructs using the CFTR gene 5Ј-flanking element fused to the herpes simplex virus thymidine kinase promoter. Our results suggest that the same sequence elements promoting basal CFTR gene transcription confer cAMP-mediated regulation through the context of a heterologous gene. Although models that demonstrate the ability to direct cAMP-mediated transcription in the absence of conventional cAMP-responsive regulatory elements are relatively few, some recent investigations implicate the activity of the CCAAT element in directing both basal and cAMP-mediated transcription (50). This was also previously demonstrated with the human G-protein i-2 gene promoter (42). It has been more recently described that the regulation of the human tryptophan hydroxylase gene transcription requires an inverted CCAAT motif for maintaining basal expression and cAMP-mediated transcription (51). The absence of a CRE, AP1, or AP2 consensus sequence within the region of the CFTR promoter that we have examined here reflects some similarities to these unconventional genetic models for cAMP-mediated regulation. The kinetics of cAMP-mediated transcription in these models (42,51) bears some similarities to the cAMP induction of CFTR gene expression (6) in relationship to the time required for stimulation to occur (Ͼ6 h). This is in contrast to stimulation mediated by cAMP through a conventional cis-acting CRE, such as that previously demonstrated with the somatostatin gene (52), requiring only minutes for transcriptional stimulation. This would suggest that transcription mediated by cAMP can occur through sev- eral different mechanisms, accounting for differences in the kinetics of gene transcription mediated by cAMP. The delayed response to intracellular increases of cAMP to the activation of gene transcription appears to fall into a category of genes possibly requiring de novo protein synthesis (51). This is in contrast to the rapid transcriptional response to intracellular levels of cAMP mediated by the CRE not requiring new protein synthesis (53). We indicate cAMP-responsive transcription within the CFTR gene promoter is likely due to the inverted CCAAT sequence. This result is indicated from co-transfection studies despite the absence of more conventional sequences responsive to cAMP such as the CRE (52), AP1 (54), and AP2 (55) sites within the nucleotide elements studied in this report. Interestingly, the inverted CCAAT element of the CFTR gene was able to stimulate both basal levels as well as cAMP-mediated transcription within the context of a heterologous gene promoter. Although the placement of such cis-acting sequence elements of the CFTR gene within the context of a heterologous gene promoter may demonstrate a misleading effect with regard to actual CFTR promoter function, here we demonstrate the ability of the inverted CCAAT element to only confer cAMPresponsive transcriptional control.
To address the relevance of the inverted CCAAT sequence in directing the transcriptional machinery to the CFTR gene, stably transfected cells were obtained using two independent CFTR promoter transgene constructs. RNA isolated from the transgenic cells, containing wild-type and mutant inverted CCAAT sequence fused to reporter sequences, were analyzed by ribonuclease protection mapping. Results from the ribonuclease protection assay indicate the direct association of the conserved inverted CCAAT sequence with the positioning and pattern of the transcription start sites, using antisense probes specific for each of the constructs tested shown in Fig. 3, thus providing evidence for the involvement of an inverted CCAAT nucleotide consensus in directing basal transcription. Although many examples have alluded to the role of CCAAT sequences in directing RNA polymerase II-dependent transcription (40), there are no apparent studies that demonstrate a direct role for a CCAAT consensus nucleotide sequence as a component of the basal transcription apparatus. One explanation for the lack of evidence is due to the absence of example promoters and cell types that could represent a model environment demonstrating both basal and inducible transcription involving a CCAAT nucleotide consensus, exclusively. Another explanation may involve multiple and complex levels of regulation directed at specific CCAAT elements, making the evaluation between basal and inducible regulation difficult to interpret. Due to the relatively weak promoter function of the CFTR gene, this may well represent a novel feature of transcriptional regulation characteristic of the CFTR gene. In addition, specific 32 Plabeled RNA antisense probes were created to span specifically between the CFTR promoter and the human growth hormone gene. Therefore, the qualitative evaluation of the protected fragments was restricted to transcripts generated only from the CFTR gene promoter fragment from the window of nucleotides depicted in Fig. 3. A study by Koh et al. (5) indicated that the presence of an additional putative exon upstream of the CFTR promoter, referred to as exon 1a, may utilize alternative or cryptic starts in transcription. Although this may represent an additional level of regulation for CFTR gene expression, we have no indication of this yet in the cell type we have studied in this report (data not shown). Through the context of the promoter sequences examined, we indicate the possible significance of the inverted CCAAT sequence in basal regulation of CFTR gene transcription. This would indicate that the conservation of the inverted CCAAT sequence is associated with the transcriptional start site selection and represents a sequence requirement for basal transcription of the CFTR gene.
Studies to compare the affinity of DNA binding overlapping the inverted CCAAT sequence between cAMP-mediated and basal transcription indicate extensive protection of DNase hydrolysis in untreated cells when compared with cAMPstimulated cells. Contrary to our own expectations, extensive protection of the nucleotide sequences overlapping the inverted CCAAT element of the human CFTR gene was more characteristic of basal expression and not of cAMP-stimulated expression in the PANC1 cell type. Although EMSA experiments performed here do not yet account for some form of displacement of transcriptional activation through binding to the the inverted CCAAT element, one explanation may be due to the sensitivity of the DNase protection assay to detect subtle changes in DNA binding affinities. Arguably, the constitutive binding of a repressor to the CCAAT element, which may act to displace transcriptional activation as shown previously (44), could function in tandem with other nuclear protein(s) to allow some access of more potent trans-activators to the CCAAT sequence. Thus, this mechanism may allow more or less activation of CFTR gene transcription through the competition between positive and negative effectors of CFTR gene transcription. The relief of repressor activity by other factors may represent a plausible mechanism for activation of CFTR gene transcription, but this hypothesis has not yet been experimentally tested.
Gel shift and competition analysis of the CCAAT element indicate that in fact CCAAT binding protein(s) are responsible as least in part for the interaction with the CFTR gene-inverted CCAAT nucleotide consensus. Interestingly, competition with nucleotide consensus elements corresponding to both the albumin and c-fos gene CCAAT sequences (56) (36), respectively, readily competed for the CFTR gene-inverted CCAAT element, suggesting that specific CCAAT enhancer binding proteins may target the CFTR gene promoter. A further attempt to characterize such interaction by specific gene products was performed by antibody supershift analysis to immunologically characterize the C/EBP species from PANC1 cell nuclear protein bound to the inverted CCAAT element of the CFTR gene. Results of experiments shown in Fig. 6 indicate the presence of C/EBP␦ in nuclear complexes bound to the CFTR-inverted CCAAT sequence. Due to the involvement of C/EBP␦ (NF-IL6␤) in a response to cytokine-mediated inflammation by interleukin-6 (57, 37), we cautiously speculate the involvement of interleukin-6 in a signaling pathway in regulating CFTR gene transcription.
Immunological detection of CREB1 and ATF1 associated with protein complexes bound to the inverted CCAAT element of the CFTR gene promoter (Fig. 7) may indicate another level of regulation associated with cAMP-responsive transcription factors despite the absence of a CRE nucleotide consensus within the sequences examined in this report. Although the absence of a CRE certainly may not preclude CREB or ATF proteins from targeting promoters devoid of such cis-acting elements, a recent example by Vallejo et al. (58) demonstrates that regulation of the somatostatin gene requires proteinprotein interaction between both C/EBP and ATF/CREB factors to elicit a response by cAMP through the somatostatin gene CRE. Inversely, C/EBP proteins have been shown to bind, specifically, to the phosphoenolpyruvate carboxykinase gene CRE with high affinity to promote cAMP-mediated transcriptional activation (59). With regard to the CFTR gene, this paradox may be accounted for by a reversed mechanism, suggesting the inverted CCAAT element as the cis-acting target directing multiple protein interactions between members of the C/EBP and ATF/CREB families of transcription factors. In addition, studies by Roesler et al. (60) demonstrate C/EBP as an effector of cAMP-mediated transcription through the combined interactions with liver-specific transcription factors to modulate the cAMP regulation of phosphoenolpyruvate carboxykinase. The models described here may represent complex features analogous to CFTR gene regulation. Therefore, through the identification of trans-acting factors responsible for directing CFTR gene transcription, we can begin to elucidate the mechanisms regulating CFTR gene expression that may help us better understand the pathology associated with cystic fibrosis.