Molecular Cloning and Characterization of a Transcription Regulator with Homology to GC-binding Factor*

GC-binding factor (GCF) represses transcription of certain genes and is encoded by a 3.0-kilobase mRNA (Kageyama, R., and Pastan, I. (1989) Cell 59, 815–825). The GCF cDNA hybridizes to two additional mRNA species, 4.2 and 1.2 kilobases. We have used differential hybridization to identify a cDNA clone (termed GCF2) for the 4.2-kilobase mRNA and find that it is highly expressed in HUT-102 cells. The open reading frame consists of 2256 nucleotides and encodes a protein of 752 amino acids with a calculated molecular mass of 83 kilodaltons. GCF2 expressedin vitro using reticulocyte lysates and Escherichia coli migrates as a 160-kilodalton protein in SDS-polyacrylamide gel electrophoresis but has a molecular mass of 83 kilodaltons as determined by mass spectrum analysis. GCF2 binds to epidermal growth factor receptor promoter fragments, and the major binding site is located between nucleotides −249 and −233. Cotransfection assays show that GCF2 acts to repress transcription from the epidermal growth factor receptor promoter in constructs containing the major GCF2 binding site and not when the site had been mutated. Thus, GCF2 is a newly identified transcriptional repressor with aberrant electrophoretic mobility.

Regulation of transcription is achieved by gene-specific transcription factors that bind regulatory elements in gene promoters and enhancers and alter the rate of transcription initiation (1). Most studies on the regulation of transcription focus on mechanisms of transcription activation. However, transcriptional repression is also an important factor in the regulation of many genes (2,3). For example, the transcription factor Sp1 represses basal and retinoic acid-induced expression of the osteocalcin gene by competing with the retinoic acid receptor for overlapping DNA-binding sites (4). For a given gene, the combination of cis elements and the trans-acting factors are major determinants of transcriptional activity. A large number of eukaryotic transcriptional activators and a few repressors have been identified and characterized.
The epidermal growth factor receptor (EGFR) 1 plays an important role in cell growth and development (5)(6)(7). Overexpression of the EGFR can lead to epidermal growth factor-dependent transformation (8,9). Overproduction of EGFR has been detected in several types of cancers due to gene amplification (10). Overexpression of EGFR transcripts in a variety of other tumors such as ovarian, cervical, and kidney tumors results from transcriptional or posttranscriptional mechanisms (11). A variety of agents have been shown to increase EGFR gene expression (12)(13)(14). Repression of EGFR gene transcription by different agents has also been reported (15,16). Transcriptional control must play a major role in regulation of EGFR gene expression.
The promoter of the EGFR gene lacks a TATA box and CAAT box but contains multiple GC boxes and multiple transcription initiation sites. A number of regions in the promoter have been identified that bind nuclear factors (17)(18)(19). Furthermore, Sp1, wild type p53, EGFR transcription factor, and AP2 have been shown to activate EGFR gene transcription (20 -23). Three repressor proteins, EGFR transcriptional repressor, GC-binding factor (GCF), and the Wilms' tumor supressor, also bind to sites within the EGFR promoter (24 -26).
A cDNA for GCF was isolated by screening an A431 expression library with GC-rich sequences from the EGFR promoter. GCF is a 91-kDa protein that binds to three upstream sites of the EGFR promoter. Two are between bp Ϫ270 and Ϫ225, and the other site is between Ϫ150 and Ϫ90 relative to the translational start site. Cotransfection experiments have shown that GCF can repress transcription of the EGFR promoter and several other growth-related gene promoters such as transforming growth factor-␣ and insulin-like growth factor II (27). The cDNA for GCF hybridizes to three mRNA species of 4.5, 3.0, and 1.2 kb (28). The GCF cDNA contains 2.8 kilobase pairs and is likely to encode the 3.0-kb mRNA. The larger 4.5-kb mRNA may result from homology to another gene or possibly from alternative splicing of the GCF gene. Since the 5Ј portion of the GCF cDNA that encodes the DNA binding region hybridizes very strongly to the 4.5-kb mRNA, it is possible that the cDNA for this mRNA also encodes a DNA binding protein. In this report, we present results on isolation of a cDNA that hybridizes to the larger mRNA and has homology to the 5Ј-end of the GCF cDNA. We also show that the protein encoded by this mRNA binds to the EGFR promoter and represses the activity of the EGFR, SV40, and RSV promoters.

MATERIALS AND METHODS
Cell Culture-Cells were maintained in Dulbecco's modified Eagle's medium supplemented with 10% fetal bovine serum (Life Technologies, Inc.) and antibiotics. Medium was removed, and cells were washed with phosphate-buffered saline without Ca 2ϩ and Mg 2ϩ prior to RNA isolation. KB cells were treated with 100 nM phorbol 12-myristate 13-acetate (Sigma) in Me 2 SO (Aldrich) for up to 7 h in the above medium.
RNA Isolation and Blotting-Total RNA was isolated by the guanidinium-thiocynate-phenol-chloroform extraction method of Chomczynski and Sacchi (29). Poly(A) ϩ RNA was selected from the total RNA population by oligo(dT)-cellulose chromatography (30). RNA concentrations were determined, and RNAs were fractionated on a 1% formaldehyde-agarose gel and transferred to nitrocellulose (31). RNA was UVcross-linked to the nitrocellulose filter, prehybridized, hybridized, and washed as described previously (32). Labeled cDNA probes were prepared by random primer extension of polymerase chain reaction (PCR)generated fragments (33).
Isolation and Sequence Analysis of GCF2 cDNA Clones-The 282-bp GCF cDNA fragment (1-282) was labeled with [␣-32 P]dCTP and used as a hybridization probe to screen an ovarian carcinoma (OVCAR-3) cell cDNA library constructed in Uni-Zap XR (Stratagene). Positive clones were purified and phagemids were excised using R408 helper phage (Stratagene). The clones were sequenced with an Applied Biosystems model 373A automated DNA sequencer. Sequence comparisons were performed with BLAST and PROSITE using the default parameter to search the National Cancer for Biotechnology Information nonredundant protein and DNA data bases (34,35).
5Ј-Rapid Amplification of cDNA Ends (RACE)-5Ј-RACE-Ready cDNA (liver) was purchased from CLONTECH (Palo Alto, CA). GCF2specific primers were selected using Oligo 4.0 (National Biosciences). Nested primers were used to enhance specificity. The 5Ј-RACE product, detected after primary and secondary amplifications, was purified by agarose gel electrophoresis, subcloned into pCRII (Invitrogen), and sequenced. The RACE products contained homology to the GCF2 cDNA clones and extended to the 5Ј-end. The full-length GCF2 cDNA was constructed by ligation of restriction fragments.
In Vitro Translation-The open reading frame of GCF2 was amplified by PCR and subcloned into pCITE2A (Invitrogen). Protein was synthesized in vitro in the presence of [ 35 S]methionine with the coupled transcription/translation system (TNT) from Promega (Madison, WI). Translated products were analyzed on SDS-polyacrylamide gels (36).
Bacterial Expression and Purification-The GCF2 open reading frame was cloned into pQE60 (Qiagen) at the BamHI site after the addition of BamHI linkers to the open reading frame by PCR. The new plasmid, pGCF2-His, was sequenced to check for mutations and used to transform JM109. JM109 cells containing pGCF2-His were induced with 1 mM isopropyl-1-thio-␤-D-galactopyranoside at A 600 ϭ 0.7 for 4.5 h. Cells were harvested and resuspended in sonication buffer (50 mM sodium phosphate, pH 8.0, 300 mM NaCl). Cells were subjected to two cycles of freezing and thawing followed by treatment with lysozyme (1 mg/ml) for 30 min on ice. The sample was then sonicated (1-min bursts/ 1-min cooling/200 -300 watts) on ice and treated with 10 g/ml RNase A for 15 min. After centrifugation at 10,000 ϫ g for 20 min, the supernatant was mixed with nickel-nitrilotriacetic acid resin for 60 min at 4°C. The mixture was loaded into a column and washed with sonication buffer followed by sonication buffer plus 0.8 mM imidazole and sonication plus 40 mM imidazole. The GCF2-His protein was eluted in sonication buffer plus 0.5 M imidazole and examined by SDS-polyacrylamide gel electrophoresis. Fractions containing GCF2-His were dialyzed versus a buffer containing 20 mM HEPES, pH 7.9, 20 mM KCl, 1 mM MgCl 2 , 2 mM dithiothreitol, and 17% glycerol. Dialyzed samples were stored in aliquots at Ϫ80°C.
Gel Mobility Shift Assays-Mobility shift assays were performed as described previously (17). A double-stranded oligonucleotide containing the putative GCF2 binding site was prepared by annealing two complementary oligonucleotides containing nucleotides Ϫ249 to Ϫ229, 5Ј-CGGGCAGCCCCCGGCGCAGCG-3Ј and 5Ј-CGCTGCGCCGGGGGCT-GCCCG-3Ј, in a buffer containing 10 mM Tris, pH 8.0, 500 mM NaCl, and 1 mM EDTA. Equimolar amounts of the complementary oligonucleotides were mixed in a 1.5-ml Eppendorf centrifuge tube and placed in a heat block at 95°C. The heat block was allowed to cool to room temperature, and the sample was desalted on a G-25 Sephadex column. The double-stranded oligonucleotide and EGFR promoter fragments were end-labeled with 32 P using T4 polynucleotide kinase and ␥-ATP. For the gel shift analysis, end-labeled EGFR promoter fragments or double-stranded oligonucleotide were incubated with GCF2-His or nuclear extract at room temperature (23°C) for 15 min in the presence of 10 mM Tris, pH 7.5, 1 mM MgCl 2 , 0.5 mM EDTA, 0.5 mM dithiothreitol, 50 mM NaCl, 50 g/ml poly(dI-dC)⅐poly(dI-dC) and 4% glycerol. Nuclear extract from OVCAR-3 cells was prepared as described earlier (37). When competition assays were performed, unlabeled EGFR fragments or mutated GCF2 binding site oligonucleotides were incubated with protein and buffer for 5 min prior to the addition of the labeled oligonucleotide. The mutated GCF2 binding site oligonucleotides were purchased as single-stranded DNAs from Genosys Biotechnologies and annealed as described above. The AP2 binding site oligonucleotide was purchased from Promega. Samples (10 l) were loaded onto a 5% polyacrylamide gel and subjected to electrophoresis at 150 V for 2 h using 0.5ϫ TBE (1ϫ TBE: 89 mM Tris, 8 mM boric acid, and 2 mM EDTA, pH 8.3) as running buffer. After electrophoresis, gels were transferred to Whatman 3 MM paper and exposed to Kodak XAR film with intensifying screens at Ϫ70°C.
DNase I Footprinting-DNase I footprinting was performed according to Dynan et al. (38). The EGFR promoter fragment (Ϫ771 to Ϫ16) was labeled at the HinD III site, and a 553-base pair (Ϫ569 to Ϫ16) fragment was isolated after restriction digestion with TaqI. GCF2-His was prepared as described earlier. AP2 and Sp1 were obtained from Promega.
Transfections and Chloramphenicol Acetyltransferase (CAT) Assays-African green monkey kidney cells (CV-1) or OVCAR-3 were seeded at 5 ϫ 10 5 cells/100-mm dish incubated overnight at 37°C in a 5% CO 2 incubator. For each transfection, 2-10 g of pCMVGCF2 and 2 g of promoter-CAT DNA were mixed in 1.5 ml of Opti-MEM (Life Technologies), and a precipitate was formed using lipofectamine (Life Technologies) according to the manufacturer's recommendations. The cells were washed with serum-free Dulbecco's modified Eagle's medium, and complexes were applied to the cells for 5 h. Dulbecco's modified Eagle's medium containing 10% fetal bovine serum was added, and cells were incubated overnight. Medium was changed the following day, and cells were grown for an additional 24 h. Cells were harvested and extract was prepared as described previously (39). CAT activity was assayed in extracts using the CAT assay kit from Promega. Transfection efficiency was monitored by measuring ␤-galactosidase activity from an RSV-␤-galactosidase reporter plasmid construct that was also cotransfected. The EGFR promoter-CAT constructs, pERCAT6, pER-CAT9, and pERCAT10, were prepared previously by our laboratory (17). The SV40 early promoter CAT construct, pSV2CAT, and Rous sarcoma virus long terminal repeat CAT construct, RSVCAT, were obtained from Dr. Bruce Howard (National Institutes of Health) (39). The cytomegalovirus immediate early gene 3 promoter CAT construct, CMVCAT, was obtained from Dr. Sumitra Deb (University of Texas Health Science Center at San Antonio) (40). The EGFR-CAT constructs containing mutations in the GCF2 binding site were prepared by first using PCR with mutated oligonucleotide primers and a HindIII restriction site to generate the desired mutation in the promoter. The PCR fragment was digested with HindIII and ligated into the HindIII site of pSV0CAT. The orientation of the promoter was determined by restriction enzyme mapping, and fidelity of the sequence was confirmed by DNA sequencing using the Applied Biosystems model 373A automated sequencer.
Mass Spectroscopy-GCF2-His was diluted 1:10 into 50% acetonitrile, 0.1% trifluoroacetic acid in water. BSA (Sigma) in the same solvent was mixed with the dilute GCF2 by trial and error to yield a comparable mass spectrum signal. A ratio of five parts GCF2 to one part BSA worked well. The protein/BSA sample was then mixed with an equal amount of sinapinic acid solution (Hewlett Packard). The sample was analyzed on the Hewlett Packard model 2025A matrix-assisted laser desorbtion ionization, time of flight mass spectrometer. The GCF2 mass was averaged from eight trials, with each trial combining the data of 20 -30 laser shots.

Differential Hybridization of GCF cDNA Fragments-We
have previously reported that GCF cDNA hybridizes to three mRNA species of 4.5, 3.0, and 1.2 kb in several cell lines. We also found that various fragments of the cDNA hybridized differently to the three mRNA species (28). A fragment containing nucleotides 1-282 and a fragment containing nucleotides 314 -961 were prepared by PCR, labeled with 32 P, and used in Northern blot hybridization analysis. Fragments were prepared by PCR to avoid the stretch of 21 adenosines in the GCF cDNA that would hybridize to many RNAs. The fragment containing nucleotides 1-282 hybridized to an mRNA of ap-proximately 4.5 kb with virtually no hybridization to other mRNAs (Fig. 1A). In contrast, a fragment containing nucleotides 314 -961 hybridized very strongly to mRNAs of 3.0 and 1.2 kb but only slightly to the 4.5-kb mRNA. This was true using RNA from both A431 and KB epidermoid carcinoma cells (Fig. 1B). Identical results were obtained using RNA from ovarian carcinoma cell line (OVCAR-3) and a T-cell lymphoma cell line (HUT-102) (data not shown).
GCF2 cDNA Isolation-To isolate the cDNA corresponding to the larger mRNA, a cDNA library prepared from ovarian carcinoma cell mRNA (OVCAR-3) was screened using the fragment containing nucleotides 1-282 as a probe. Fourteen positive cDNA clones were isolated and sequenced. The two largest clones, O (1.4 kilobase pairs) and Q (2.6 kilobase pairs) were determined to contain all of the DNA sequences present in the 14 clones (Fig. 2). The O clone sequence was found to contain an open reading frame that extended to the 5Ј-end of the clone. To obtain additional sequence present at the 5Ј-end of the cDNA, RACE was performed. The end of the open reading frame was obtained with an additional 126-bp 5Ј-untranslated region. The sequence of the combined cloned cDNAs consists of 3523 bp with an open reading frame of 2256 nucleotides (Fig. 3). The GCF2 cDNA has a region of sequence homology with the GCF cDNA of 309 bp (98% identity) (Fig. 4). The remainder of the sequence has no further significant homology to GCF or any other sequence found in GenBank TM . The deduced protein sequence of GCF2 is shown in Fig. 3. The amino acid sequence indicates the presence of potential phosphorylation sites for cAMP-dependent kinase, calcium-dependent kinase, and tyrosine kinase. Also, the presence of an N-linked glycosylation site and a nuclear localization sequence is predicted.
GCF2 mRNA Characterization-To determine the size of mRNA that the GCF2 cDNA hybridizes, Northern blot hybridization analysis was performed. Poly(A) ϩ RNA isolated from D551 (normal human fibroblast), A431 (epidermoid carcinoma), KB (epidermoid carcinoma), OVCAR-3 (adenocarcinoma), T98G (glioblastoma), Raji (Burkitts' lymphoma), and HUT-102 (T-cell lymphoma) cells was transferred to nitrocellulose and probed with a radiolabeled GCF2 cDNA probe. As compared with an RNA size ladder, a 4.2-kb mRNA hybridized to the GCF2 cDNA (Fig. 5). If comparison is made to ribosomal RNA migration, the size would be 4.5 kb, which was the original size estimate. The 4.2-kb GCF2 mRNA was detected in all cell lines with higher levels found in Raji, T98G, and HUT-102 cells.
Production and Analysis of GCF2 in Reticulocyte Lysates and Escherichia coli-As described above, the open reading frame of the GCF2 consists of 2256 residues and should encode a protein of 83 kDa. The open reading frame was cloned into the pCITE2A vector, and coupled in vitro transcription/translation was performed in the presence of radiolabeled methionine. The radiolabeled translation product was analyzed on an SDS-polyacrylamide gel. GCF2, made in vitro in reticulocyte lysates, migrates as a protein of 160 kDa, approximately twice the expected size (Fig. 6). The GCF2 open reading frame was also subcloned into a bacterial expression vector containing a His tag sequence. The protein was expressed in bacteria, and the His tag protein was purified on nickel-nitrilotriacetic acid resin. The purified GCF2-His tag protein was analyzed by SDS-polyacrylamide gel electrophoresis and was found to migrate the same as the protein made in reticulocyte lysates (Fig.  7). GCF2 has a pI of 4.4 and contains 22% acidic residues.
To further analyze the molecular mass of GCF2, mass spectroscopy was performed. Matrix-assisted laser desorbtion ionization, time of flight mass spectrum analysis of GCF2-His gave an average mass of 83,960 Da. The percentage error of the internal standard BSA was 0.01%, and the error for the GCF2-His was estimated to be 0.05%, since its mass is outside the limits of the two-point linear calibration used with BSA (doubly charged mass ϭ 33215.5 and singly charged mass ϭ 66431). DNA Binding Studies-The homology of the GCF2 cDNA and GCF cDNA is confined to the DNA binding region of GCF. To test whether GCF2 could bind to specific sites in DNA and locate the site(s), DNase I footprinting experiments were performed. GCF2 was shown to bind to one site in the EGFR promoter located between Ϫ249 and Ϫ233 (Fig. 8). As controls, AP2 and Sp1 were used to footprint the promoter, and both have footprints that overlap the GCF2 footprint. To see if there were GCF2 binding sites of lower affinity and to confirm the DNase I footprinting results, gel electrophoretic mobility shift assays were used. Three EGFR promoter fragments were endlabeled and incubated with GCF2-His. Two fragments, Ϫ384 to Ϫ167 and Ϫ105 to Ϫ16, bound GCF2 and exhibited altered mobility during polyacrylamide gel electrophoresis (Fig. 9). The binding affinity of GCF2 to the Ϫ384 to Ϫ167 fragment was 50 -100-fold greater than to the Ϫ105 to Ϫ16 fragment as measured by densitometric scanning of the retarded bands. Also, this difference in affinity was seen in competition experiments using the Ϫ105 to Ϫ16 fragment (Fig. 10A) and the Ϫ384 to Ϫ167 fragment (Fig. 10B). An EGFR promoter fragment containing residues Ϫ167 to Ϫ105 did not bind GCF2 (data not shown). Oddly, there was no footprint detected between Ϫ105 and Ϫ16. These results indicate that GCF2 binds with different affinities to the different sites. To further define the GCF2 binding site, site-directed mutations were placed at different locations throughout the binding site. Doublestranded oligonucleotides containing these mutations were used to compete with a radiolabeled GCF2 binding site oligonucleotide. The effect of these mutations is shown in Fig. 11. The mutated oligonucleotides that do not compete represent essential nucleotides in the binding site (Fig. 11, lanes 3-8). Mutated oligonucleotides that still compete contain changes that do not affect GCF2 binding (Fig. 11, lanes 9 and 10). Thus, the core of the GCF2 binding site is AGCCCCCGGCG.
Cotransfection Experiments with Promoter-CAT Constructs-The binding of GCF2 to EGFR promoter fragment indicates a possible effect of GCF2 on EGFR gene activity. Cotransfection experiments were performed to examine the FIG. 6. In vitro translation production of GCF2 in rabbit reticulocyte lysates. GCF2 and luciferase were synthesized in the presence of [ 35 S]methionine as described under "Materials and Methods." Translated products were analyzed on a 6% SDS-polyacrylamide gel. After processing, the dried gel was exposed to film at Ϫ80°C for 4 h. effect of GCF2 on EGFR gene expression. GCF2 cDNA (pCM-VGCF2) or the empty vector control (pCMV) was cotransfected with receptor plasmids containing the CAT gene under control of the EGFR promoter (pERCAT6), the SV40 early promoter (pSV2CAT), the Rous sarcoma virus long terminal repeat promoter (RSVCAT), or the cytomegalovirus IE3 gene promoter (CMVCAT). These promoter constructs were selected because they are all strong promoters and because GCF repressed the EGFR promoter but not the other three (25). As shown in Fig.  12, cotransfection with the GCF2 expression plasmid resulted in significant repression of the expression of three promoters (EGFR, SV40, and RSV) but not the CMV promoter. The control expression plasmid, pCMVGCF2R, in which the GCF2 cDNA is inserted in reverse orientation, had no effect on expression from any of these plasmids (data not shown). The extent of repression by GCF2 was similar for all three reporter plasmids, 3-4-fold at a 5:1 GCF2/CAT ratio, and was also seen when OVCAR-3 cells were transfected.
The effect of GCF2 on the EGFR promoter was further examined using EGFR promoter deletion constructs. Promoter constructs containing the major GCF2 binding site, pERCAT-6, pERCAT-7, pERCAT-8, and pERCAT-9, were repressed approximately 4-fold when cotransfected with pCMVGCF2 (Table  I). pERCAT-10, pERCAT-14, and pERCAT-15, which do not FIG. 9. Gel mobility shift assay with GCF2-His and EGFR promoter fragments. EGFR promoter fragments were end-labeled and incubated with GCF2-His as described under "Materials and Methods." Samples were analyzed on a 5% nondenaturing polyacrylamide gel. After electrophoresis, the gel was transferred to Whatman 3 MM paper and exposed to film at Ϫ80°C for 8 h. A, EGFR promoter fragment from Ϫ384 to Ϫ167. B, EGFR promoter fragment from Ϫ105 to Ϫ16. Lane 1, no addition, lanes 2-4, 150 ng of GCF2-His with competitors A1A2 (Ϫ384 to Ϫ167) or AH90 (Ϫ105 to Ϫ16). A 100-fold molar excess of competitor was used except in lane 4 of A, where a 1000-fold molar excess was used.
FIG. 10. Competition of GCF2 binding by EGFR promoter fragments. The gel mobility shift assay was performed as described under "Materials and Methods." Samples were incubated with cold competitor for 5 min before the addition of labeled EGFR promoter fragment (Ϫ384 to Ϫ167). Samples were analyzed on a 5% nondenaturing polyacrylamide gel. After electrophoresis, the gel was transferred to Whatman 3 MM paper and exposed to film at Ϫ80°C for 8 h. The molar excess of competitor is shown above the lanes. A, AH90 as competitor; B, A1A2 as competitor.
FIG. 11. Competition of GCF2 binding with mutated oligonucleotides. A, gel mobility shift assays were performed as described under "Materials and Methods" using a labeled oligonucleotide containing the GCF2 binding site (GCF2B). Unlabeled cold mutated oligonucleotides were used as competitors and are indicated at the top. B, the sequence of the GCF2 binding site oligonucleotide (GCF2B) and the mutations (thymidine (t) substitutions) found in each competitor are depicted. contain the major GCF2 binding site, were only slightly repressed, ϳ1.5-fold. These results indicate that GCF2 repression of the EGFR promoter requires a binding site located between Ϫ384 and Ϫ167. When this site is mutated in a way to prevent GCF2 binding, the repression of the EGFR promoter is also lost (Fig. 13). The M2 mutation does not allow for GCF2 binding, and the EGFR promoter reporter construct containing the M2 mutation is not repressed by GCF2 in cotransfection experiments. Conversely, the M9 mutation, which does not affect GCF2 binding does not prevent repression by GCF2. Thus, there is a direct correlation of GCF2 binding and repression.
The binding site for GCF2 either overlaps or is in close proximity to binding sites for AP2 and Sp1. To examine the binding of proteins to this region, nuclear extracts from OVCAR-3 cells and a labeled GCF2 binding site oligonucleotide were prepared. A gel mobility shift assay was performed, and cold oligonucleotides were used to compete for the binding (Fig.  14). The M2 and M3 oligonucleotides competed for some but not all of the binding (Fig. 14, lanes 3 and 4). An oligonucleotide containing an AP2 binding site also competed for some but not all of the binding at a 100-fold molar excess (Fig. 14, lane 7). These results indicate that GCF2 and AP2 can bind to this region.
We have previously shown that increased EGFR gene expression due to phorbol ester treatment was mediated via AP2 (23). Since GCF2 and AP2 can bind to overlapping sites in the promoter, we examined the effect of phorbol ester treatment on GCF2 expression and compared it to EGFR expression. Total RNA was isolated from treated cells, and the expression of GCF2 and EGFR was analyzed by Northern blotting. GCF2 mRNA levels decreased as low as 10-fold during the time course (Fig. 15). In contrast, the EGFR mRNA increased approximately 5-fold during the same time frame. Also, the decrease in GCF2 mRNA preceded the increase in EGFR mRNA. Thus, there is an inverse correlation of GCF2 and EGFR expression following phorbol 12-myristate 13-acetate treatment. DISCUSSION Transcriptional repression is an important factor in the regulation of many genes. Transcription of specific genes can be modulated downward in various ways. Most of the steps required for activation of transcription can be altered by transcriptional repressors. Repressors fall into two basic categories, passive and active. Passive repressors down-regulate the activity of one or more transcriptional activators by competing for binding sites or by binding the activator. Active repressors have an intrinsic repressing activity and inhibit transcription initiation directly.
GCF2 Has Homology to GCF-In this paper, we describe a new repressor that has a DNA binding activity and represses the activity of the EGFR gene. The GCF2 cDNA was isolated and cloned based on homology to GCF. The initial 309 base pairs of the GCF cDNA are homologous to residues 1382-1690 of GCF2. There is a 98% identity, 305 of 309 base pairs, be-tween the GCF and GCF2 cDNAs in this region. The region of homology is restricted to the 5Ј-end of the GCF cDNA but is internal, nucleotides 1382-1690, to the GCF2 cDNA.
GCF2 Migrates with an Altered Mobility during SDS-Polyacrylamide Gel Electrophoresis-The GCF2 deduced protein  FIG. 14. Analysis of GCF2 binding in nuclear extracts from OVCAR-3. A gel mobility shift assay using nuclear extract from OVCAR-3 cells and the labeled GCF2B oligonucleotide was performed as described under "Materials and Methods." The protein-DNA complex is denoted as Bound, and the competitors are indicated above the lanes. M2 and M3 are GCF2 binding site mutations from Fig. 11. NS is a nonspecific oligonucleotide (21-mer). AP2 contains the binding site for activator protein 2.
FIG. 15. Correlation of GCF2 and EGFR mRNA levels. KB cells were treated with 100 nM phorbol 12-myristate 13-acetate, and total RNA was isolated at various time points. Northern blot analysis was performed, and the blot was probed with labeled cDNAs for EGFR and GCF2. After hybridization, the blot was exposed to film, and the film was scanned using a laser densitometer. RNA levels are plotted relative to control RNA levels that were treated with Me 2 SO (0 h). sequence contains a DNA binding and nuclear localization motif similar to GCF (Fig. 3). GCF2 expressed from the open reading frame migrates as a 160-kDa protein on SDS-polyacrylamide gels. However, the calculated molecular mass is 83 kilodaltons. This could be due to the acidic nature of the protein (p ϭ 4.4 and 22% acidic residues) or to an unusual ability to form very stable dimers. -70 from E. coli also migrates at a significantly higher molecular mass (90 kDa) on SDS-polyacrylamide gels as compared with the calculated molecular mass of 70 kDa (41). Production of protein from deletion mutants in reticulocyte lysates revealed that the altered migration during SDS-polyacrylamide gel electrophoresis is associated with the protein sequence between residues 490 and 530 (data not shown). This region includes the putative DNA-binding region and the nuclear localization signal. It contains a sequence stretch of residues where 11 out of 14 are lysine. Charge interactions between this region and acidic regions may result in a protein conformation that has an aberrant migration on SDS-polyacrylamide gels.
GCF2 Binds DNA and Represses Transcription-GCF2 binds to EGFR promoter fragments with different affinities. The promoter fragments, Ϫ384 to Ϫ167 and Ϫ105 to Ϫ16, were labeled to similar specific activities, but GCF2 binding was stronger to the Ϫ384 to Ϫ167 fragment (Fig. 8). DNase I footprinting detected a single footprint in the Ϫ384 to Ϫ167 region. We have not been able to localize the GCF2 binding site in the Ϫ105 to Ϫ16 fragment due to a much lower affinity. The binding site detected by DNase I footprinting resembles binding sites for GCF. This is not unexpected, since there is sequence homology to the DNA binding region of GCF. The GCF2 binding site overlaps binding sites for Sp1 and AP2 (Fig. 8). Also, p53 has been reported to bind the EGFR promoter between Ϫ265 and Ϫ239 (42). Thus, this location is very important in the regulation of EGFR promoter activity by transcriptional activators. Binding of GCF2 may provide a mechanism to turn down promoter activity after activation or to prevent unwanted activation. This could be mediated through direct binding to DNA or through protein-protein interactions with Sp1, AP2, and p53. GCF2 represses activity from three different promoters in cotransfection assays. Two of these promoters, EGFR and SV40, have GC-rich regions. The EGFR promoter has been shown previously to be repressed by GCF, while the SV40 promoter was shown not to be repressed (25). It is interesting to note that although the overall effect of GCF2 on the promoters was about the same, at lower GCF2/reporter ratios the effect was more evident on the SV40 and RSV promoters. This may be due to the greater activity of these promoters, which are 5-10fold stronger than the EGFR promoter in CV-1 cells. GCF2 may be acting as either an active repressor or a passive repressor. In either case, it appears to be a general repressor, since it is active on both cellular and viral promoters. To determine whether GCF2 is an active or passive repressor requires detailed studies of DNA-protein interactions and protein-protein interactions. Recently, NAB1 was isolated and shown to be a repressor of NGFI-A-and Krox20-mediated transcription (43). NAB1 interacts directly with the activators to prevent transcriptional activation. Nucleolin was recently purified and shown to bind to the B-motif of the ␣-1 acid glycoprotein and to repress transcription of the ␣-1 acid glycoprotein promoter in transfection assays (44). Analysis of GCF2 will aid in increasing understanding of gene regulation and in extending our knowledge of the mechanisms of action of eukaryotic transcriptional repressors.
The Relationship of the GCF and GCF2 cDNAs-One unresolved issue is the high degree of homology between the 5Ј-end of the GCF cDNA and residues 1382-1690 of the GCF2 cDNA. Attempts to isolate GCF cDNA clones from cell lines other than A431 have resulted in isolation of cDNAs that do not contain the GCF 5Ј-region (data not shown). Furthermore, we have not been able to isolate an identical GCF cDNA from an A431 cDNA library but have isolated additional cDNA clones that contain internal deletions. These cDNAs all contain 5Ј-ends that begin around nucleotide 320 of the GCF cDNA (data not shown). This indicates that the GCF 5Ј-region is possibly a cloning artifact or is a mutation found in A431 cells that have a number of chromosomal abnormalities. Also, P1 clones containing GCF genomic DNA sequences do not hybridize with the 5Ј-end of the GCF cDNA (data not shown). However, the melting temperature of the GCF cDNA around residue 320 is greater than 85°C. This results in problems in amplifying the cDNA by PCR and may interfere with additional experimentation, including cDNA synthesis using reverse transcriptase. The origin of the GCF cDNA requires a more detailed analysis of gene structure that includes genomic clones from normal tissues and A431 cells.
EGFR levels are decreased by a number of factors including retinoic acid and nerve growth factor. Recently, it has been shown that the decrease of EGFR in PC12 cells is controlled at the transcriptional level (45). Also, correlating with the decrease in EGFR transcription is an increase in GCF2. In agreement with these studies is the correlation of a decrease in GCF2 mRNA and an increase in EGFR mRNA by phorbol 12-myristate 13-acetate reported in this study (Fig. 15). This suggests that GCF2 may have a physiological role in regulating EGFR transcription. Additional studies on GCF2 and EGFR expression will aid in understanding the complex nature of EGFR gene regulation.