Expression of human glucocorticoid receptor gene and interaction of nuclear proteins with the transcriptional control element.

We have identified sequences responsible for the expression of the human glucocorticoid receptor gene (GR gene) using a set of 5′ promoter deletion mutants in HeLa, human placenta, and human breast tumor (MCF-7) cells. The chimeric gene construct −892 5′-GAAGTGACACACTTC3′ −878-CAT was sufficient for high level of expression in HeLa and placenta cells in culture. Deletion of palindromic sequences decreased levels of GR expression in these cells. By oligonucleotide-affinity chromatography with the palindromic glucocorticoid receptor enhancing factor-binding element (GREFE), we have isolated from human placenta nuclear extract two novel proteins glucocorticoid receptor enhancing factors 1 and 2 (GREF1 and GREF2), with apparent molecular masses of 80 and 62 kDa, respectively. These proteins, similar to the DNA-binding autoantigen Ku are, like Ku, heterodimers of polypeptide subunits p80 and p62, immunologically related to factors binding to proximal sequence element 1 in the promoter of small nuclear RNA (PSE1) and transferrin receptor enhancing factors. Both Ku80 and Ku70 polypeptides were present in high concentrations in human placenta and HeLa cells. In MCF-7 cells, however, only a high level of p62 was detected. While cotransfection of pcDNA-Ku80 with pHGR(−892 to −878)-CAT potentiated the expression of CAT, introduction of pcDNA-Ku70 did not affect the expression of CAT in transfected MCF-7 cells. UV cross-linking analysis showed that only GREF1 contacted DNA directly. Supershift assays with monoclonal antibodies Ab 111 (Ku80) or Ab N3H10 (Ku70) showed a direct interaction of GREF1 and GREF2 heterodimers with the palindrome. Partial peptide fingerprinting of GREF1 and GREF2 using α-chymotrypsin and immunoblotting with Ab 111 and Ab N3H10 confirmed their identities as Ku80 and Ku70, respectively.

We have identified sequences responsible for the expression of the human glucocorticoid receptor gene (GR gene) using a set of 5 promoter deletion mutants in HeLa, human placenta, and human breast tumor (MCF-7) cells. The chimeric gene construct ؊892 5-GAAGTGACACACTTC3 ؊878-CAT was sufficient for high level of expression in HeLa and placenta cells in culture. Deletion of palindromic sequences decreased levels of GR expression in these cells. By oligonucleotide-affinity chromatography with the palindromic glucocorticoid receptor enhancing factor-binding element (GREFE), we have isolated from human placenta nuclear extract two novel proteins glucocorticoid receptor enhancing factors 1 and 2 (GREF1 and GREF2), with apparent molecular masses of 80 and 62 kDa, respectively. These proteins, similar to the DNA-binding autoantigen Ku are, like Ku, heterodimers of polypeptide subunits p80 and p62, immunologically related to factors binding to proximal sequence element 1 in the promoter of small nuclear RNA (PSE1) and transferrin receptor enhancing factors. Both Ku80 and Ku70 polypeptides were present in high concentrations in human placenta and HeLa cells. In MCF-7 cells, however, only a high level of p62 was detected. While cotransfection of pcDNA-Ku80 with pHGR(؊892 to ؊878)-CAT potentiated the expression of CAT, introduction of pcDNA-Ku70 did not affect the expression of CAT in transfected MCF-7 cells. UV cross-linking analysis showed that only GREF1 contacted DNA directly. Supershift assays with monoclonal antibodies Ab 111 (Ku80) or Ab N3H10 (Ku70) showed a direct interaction of GREF1 and GREF2 heterodimers with the palindrome. Partial peptide fingerprinting of GREF1 and GREF2 using ␣-chymotrypsin and immunoblotting with Ab 111 and Ab N3H10 confirmed their identities as Ku80 and Ku70, respectively.
The regulation of transcription in eukaryotes is dictated by the cooperative interaction between various sequence-specific DNA-binding proteins (1,2). A number of these trans-acting factors at the start of transcription have been identified (3,4), some of which are expressed in a number of cell types. The activation of gene transcription is an ability associated with a number of eukaryotic DNA-binding activators and variation in gene expression is achieved by modulating the activity of these DNA-binding transcription factors. The glucocorticoid receptor (GR), 1 a member of the nuclear receptor superfamily, mediates the actions of glucocorticoids by direct interactions and regulates the transcription of glucocorticoid responsive genes (5)(6)(7)(8)(9). Although the GR is widely distributed in nearly all cell types and has an essential role in cell metabolism and growth, its regulatory role in a number of biological events in tissues such as liver and placenta is not yet well understood.
The human placenta is a hormone responsive tissue in which excessive levels of glucocorticoids have deleterious effects on the fetus that lead to impaired fetal growth and teratogenesis, and in the adult, may predispose the individual to hypertension (10 -12). We were primarily interested in studying the transcriptional regulatory elements present in the promoter region and in examining the promoter activity (13)(14)(15)(16). Considering that the glucocorticoid effects are mediated by GR, we have constructed a series of 5Ј deletion mutants containing the promoter region of the hGR gene fused to chloramphenicol acetyltransferase (CAT) chimeric gene and introduced them into human placenta cells in culture.
The human GR gene contains the palindromic sequence motif 5Ј-GAAGTGACACACTTC-3Ј (Ϫ892 to Ϫ878) in the regulatory region. We designate this sequence as GREFE, the glucocorticoid receptor enhancing factor-binding element. The human GR promoter sequence reported (14,17,18) corresponds to the IC promoter of the mouse GR and promoters 1A, 1B, and IC were shown to have varying activities in different cell lines and tissues (19). Deletions of various sequence of the regulatory region of the hGR gene and subsequent analyses point to a differential promoter activity in various cells (13,15,(17)(18)(19)(20)(21).
The hGR promoter lacks the classical TATA box, but TA-AATA and CAAT boxes are present in the regulatory region located Ϫ54 nucleotides upstream of the first base in the cDNA (22). Previously, we identified by S1 nuclease mapping and primer extension the authentic GR transcription start site in the isolated genomic fragment (13,14,20).
To characterize the proteins binding to this palindromic recognition site, we have extracted nuclear proteins from human placenta and analyzed their interaction with the synthetic GREFE. Here, we describe an initial characterization of these proteins by DNA-protein interaction, purification by sequencespecific interaction chromatography, resolution of the UV cross-linked DNA-protein complexes by sodium dodecyl sulfate-polyacrylamide gel electrophoresis and by Western blotting with anti-Ku antibodies (23). Furthermore, these polypep-tides were identified as Ku80 and Ku70 by cleavage with sitespecific endopeptidase, Western blotting, and immuno detection with monoclonal antibodies (24,25). Finally, we demonstrate the sequence-specific interaction of Ku80 and Ku70 with the regulatory sequence by mobility-supershift assays using monoclonal antibodies against either Ku80 or Ku70.

MATERIALS AND METHODS
Cell Culture-Human Placenta (3A-SUB-E, CRL-1583), HeLa, and MCF-7 cells were grown in MEM medium (Sigma) supplemented with 10% fetal calf serum. Human prostate cancer cells LNCaP, were propagated in RPMI medium without phenol red. Preparation of the Nuclear Extract from Placenta-Two placentas weighing about 600 g were homogenized with a Waring blender in 1,200 ml (final volume) of Buffer A. The homogenate was centrifuged at 1,500 rpm for 15 min at 4°C to collect the crude nuclear fraction. The nuclear extract was either prepared immediately from the crude nuclear fraction or frozen at Ϫ20°C for 4 -6 weeks. The crude nuclear pellet was resuspended in 500 ml of Buffer B, centrifuged to collect the nuclear fraction and resuspended again, this time in 500 ml of Buffer C, filtered through 3 layers of cheese cloth and centrifuged to collect the nuclei. The nuclei were washed once by resuspension in 500 ml of Buffer E and centrifugation. The nuclear pellet was extracted in a total volume of 400 ml of Buffer E ϩ 350 mM NaCl (final concentration) and incubated for 60 min on ice. The residual nuclei were removed by centrifugation at 10,000 rpm for 30 min at 4°C.

Buffers-Buffer
The supernatant (250 ml) was mixed with an equal volume of Buffer D ϩ 140 g of sucrose, 700 l of 1 M DTT, 7 ml of 1 M MgCl 2 , and 1 ml of 500 mM EDTA. After dissolution of the sucrose, the final volume was 650 ml. This extract was centrifuged at 26,000 rpm in SW-28 rotor for 6 h at 4°C. The supernatant was decanted into an Erlenmeyer flask and 132 g of ammonium sulfate (132 g/330 ml) were added and mixed gently to dissolve. The pH was adjusted to 7.0 with 1 N NaOH (10.5 ml) and stirred overnight at 4°C. The precipitated nuclear proteins were collected by centrifugation at 26,000 rpm in SW-28 rotor at 4°C for 20 min. The supernatant was discarded and the walls of the tubes were wiped clean with Kleenex. The pellet was redissolved in Buffer H (total volume 25 ml) and dialyzed against two changes of Buffer H ϩ 200 mM ammonium sulfate for 6 h. The extract was clarified by centrifugation in a Beckman SW-40 rotor at 36,000 rpm for 2 h and dialyzed against Buffer H overnight at 4°C. The supernatant (40 ml) was clarified by centrifugation at 10,000 rpm for 10 min at 4°C. Protein concentration was adjusted to 15 mg/ml and stored in aliquots of 0.25, 0.5, or 1 ml in liquid nitrogen. Nuclear extracts from cells in culture were prepared essentially as described by Andrews and Faller (26).
Gel Mobility Shift Assay-Analysis of DNA-protein interaction was performed by incubating the total nuclear extract (2.5 g, diluted in Buffer G) in a final volume of 25 l in the presence of 1 g/25 l of poly(dI-dC) with 5,000 cpm (Cherenkov) of 32 P-labeled probe GREFE, for 20 min at 20°C. When indicated, a 100-fold molar excess of nonlabeled GREFE double-stranded as competitor DNA was included in the incubates (13,14). The complexes were resolved on 4% (39.2:0.8, cross-linked) acrylamide gels in Buffer F. The electrophoresis was conducted in Buffer F and at 250 V/20 ϫ 20-cm gel (100-min running time). The gels were dried under vaccum and autoradiographed using Fuji RXO-G films with intensifying screens.
Supershift Assays-The purified proteins from placenta were incubated either with 1 l of anti-human Ku70 monoclonal antibodies (N3H10) or anti-human Ku80 monoclonal antibodies (111) for 15 min at room temperature prior to the addition of labeled DNA as indicated (monoclonal antibodies were generously provided by Dr. W. Reeves, University of North Carolina, Chapel Hill). To further define the specificity, the antigen-antibody complexes were immunoabsorbed on protein A-Sepharose and the resulting supernatants were used in DNA-protein interaction assays.
Analysis of DNA-Protein Interaction with Intact hGR Gene Segment-To analyze the interaction of intact hGR gene fragment containing the authentic nuclear protein interaction site, a 280-base pair hGR gene fragment between SmaI (Ϫ1030) and SacI (Ϫ750) was purified and labeled at the dephosphorylated 5Ј SmaI end (13)(14)(15)20). An aliquot of 1,000 cpm was incubated with the purified proteins as indicated. The protein-DNA complexes were resolved on 5% nondenaturing polyacrylamide gels using 22.25 mM Tris, 22.25 mM boric acid, and 0.5 mM EDTA, pH 8.0 (0.25 ϫ TBE). The specific interaction of the proteins was assayed by depleting the proteins in the incubates with monoclonal antibodies and immunoabsorbtion as described above.
UV Cross-linking-The DNA-protein samples were incubated at 20°C in a 96-well plate and irradiated for 15 min using a Stratalinker UV cross-linker (Stratagene) at room temperature. The complexes were resolved on a 7.5% SDS-polyacrylamide gel by electrophoresis, the gels were dried and autoradiographed using Kodak X-Omat films with or without intensifying screens. Non-labeled standard protein markers (Bio-Rad) were resolved in a slot and stained with Coomassie Blue to determine the apparent molecular weight of the UV cross-linked proteins.
Preparation of the Affinity Matrix-The oligonucleotide 5Ј-GGGGT-GAAGTGACACACTTCACA-3Ј and the complementary 5Ј-TGTGAA-GTGTGTCACTTCA-3Ј were synthesized by Oligo Express, Bio/Can Scientific at a 0.8-mol scale and purified by reverse phase high performance liquid chromatography on a C-18 column. The extension of four guanosines at the 5Ј end was found to be suitable for coupling to the CNBr-activated Sepharose 4B. 1.25 mg each of the oligonucleotides were annealed together in Sequenase buffer (500 l) and an aliquot of 0.5 g was labeled with [ 32 P]ATP to determine the efficiency of coupling to CNBr-activated Sepharose 4B. For coupling, 1.5 g of activated Sepharose was used and procedures were as described by the manufacturer. The remaining active sites were blocked with 1% bovine serum albumin in 0.1 M Tris-HCl, pH 8.0, and the column was washed extensively and equilibrated in Buffer G. The coupling efficiency was 435 g of oligonucleotide/ml of packed gel (5 ml), with an overall yield of 87% measured both by counting the radioactivity and by determining the amount of oligonucleotide released in 50 l of the packed gel by boiling at 95°C. A column containing 2 ml of the packed affinity matrix was used in the purification step.
Affinity Chromatography-The nuclear extract was thawed and diluted to 2.5 mg/ml in protein concentration with Buffer G. The solution was clarified by centrifugation at 50,000 rpm in a Beckman Ti-70 type rotor for 30 min at 4°C and loaded on the affinity matrix at 1 ml/min. The flow-through was recirculated a second time at the same rate of application and the column was washed with 100 volumes of Buffer G, followed by 10 volumes of Buffer G containing 120 mM NaCl and the column was developed with a linear gradient (25 ml each) of 0.1-1 M NaCl in Buffer G. Fractions of 1 ml were collected and analyzed for specific DNA binding activity by gel mobility shift assay. Subsequently, the eluted proteins were analyzed by UV cross-linking, SDS-polyacrylamide gel electrophoresis, and Western blotting with human anti-Ku autoantibodies (27,28). The purity of the eluted proteins was determined by SDS-polyacrylamide gel electrophoresis and silver staining.
Analysis of Cross-reaction with Anti-Ku Antibodies by Western Blotting-Aliquots of nuclear extracts or affinity column eluted fractions were resolved on 10% SDS-polyacrylamide mini-gels and transferred onto Immobilon-P (Millipore) membranes. The membrane was treated with a solution of 2% bovine serum albumin in 20 mM Tris-HCl, pH 7.8, 1 mM EDTA, and 140 mM NaCl (TEN) to block the reactive sites and incubated with anti-Ku serum from patient OM (generously provided by Dr. J. A. Hardin, Medical College of Georgia, Augusta, GA), diluted 1:1000 in 2% bovine serum albumin ϩ TEN overnight at 4°C. The blot was washed for 10 min each with 200 ml of TEN, 200 ml of TEN ϩ Tween 20 (500 l/liter), and once again with TEN at room temperature. The blot was incubated overnight in a sealed bag containing a 10-l aliquot of 125 I-labeled protein A (Amersham) diluted in 10 ml of 2% bovine serum albumin ϩ TEN, washed as described above, and autoradiographed using intensifying screens to visualize the labeled bands.
Transactivation Involving the HGR Palindrome-We have introduced a BamHI site at ϩ59 by oligonucleotide-directed mutagenesis and generated a set of 5Ј deletion mutants of the hGR gene regulatory region with intact hGR promoter and cap site (13-15, 17, 20). A mixture of deletions containing intact BamHI site at the 3Ј end and blunt ended 5Ј end was ligated at the SmaI/BamHI sites of a Bluescript SK(ϩ) vector. The size of the plasmid inserts was determined by digestion with HindIII/BamHI, the fragments purified by digestion with HindIII/ BamHI and ligated at the HindIII/BglII sites of pBL-CAT. The se-quence of the inserted DNA and junctions was determined with appropriate primers by sequencing the double-stranded DNA using a dideoxy sequencing kit (Pharmacia Biotech Inc.). To characterize the specificity of the palindromic sequence in the regulation, we cotransfected pHGR⌬908-CAT, pHGR⌬830-CAT, and pHGR⌬747-CAT with the ␤-galactosidase expression vector pCH110 (Pharmacia, Uppsala, Sweden) into semiconfluent MCF-7, HeLa, and human placenta cells in culture by the calcium phosphate procedure. The intact hGR gene chimera pHGR2.7-CAT used in this study is identical to the one described previously (13,14,16). Expression vectors of Ku70 and Ku80 were generated by cloning the appropriate cDNAs (provided by Dr. J. A. Hardins laboratory) at the BamHI site of the eukaryotic expression vector pcDNA 1 (Invitrogen). The specificity of the palindromic sequence was analyzed by using a reporter plasmid construct containing a single copy of the synthetic palindrome GREFE, inserted at the BamHI site of the heterologous CAT vector pBLtk-CAT. The CAT activity was determined in an aliquot of the extract containing 10 units of ␤-galactosidase activity.

RESULTS
The expression of hGR in various target tissues and cells is subject to multiple controls. We have examined the importance of the 5Ј-regulatory sequences in the promoter region of the hGR gene in an effort to identify and to characterize factors involved in the regulatory pathway (Fig. 1).
Identification of DNA Sequences Directing the Tissue-specific Expression of GR Gene-To identify sequences involved in the regulation of hGR gene, MCF-7, HeLa, and human placenta cells were cotransfected with hGR gene deletion mutants fused to the CAT reporter gene. In extracts prepared from MCF-7 cells transfected with pBL-CAT (Fig. 2, lane 1), pHGR⌬908-CAT (lane 2), pHGR⌬830-CAT (lane 3), and pHGR⌬747-CAT (lane 4), reporter gene transcription was only basal. In placenta and HeLa cells, CAT expression also remained basal with control pBL-tkCAT (Fig. 2, lanes 5 and 9) and the deletion mutant pHGR⌬747-CAT (lanes 8 and 12). However, there was a 5-fold increase in CAT activity expressed by the chimera pHGR⌬830-CAT (lanes 7 and 11) in placenta and HeLa cells compared to the levels observed with the control pBL-CAT construct. Similarly, in placenta and HeLa cells cotransfected with pHGR⌬908-CAT ( lanes 6 and 12), there was a 12-fold increase  (13)(14)(15). The deletion mutants HGR⌬908-CAT, HGR⌬830-CAT, and HGR⌬747-CAT were generated by ExoIII deletion. A double-stranded synthetic DNA containing the HGR palindrome Ϫ892 and Ϫ878 flanking BamHI sites was inserted at the BamHI site of the heterologous herpes simplex virus thymidine kinase (HSV-tk, Ϫ105 to ϩ51) promoter containing vector pBL-tkCAT. The insertion of a single unit of the palindrome was determined by DNA sequencing.
in CAT expression. These results indicate that sequences between Ϫ830 and Ϫ908 5Ј to the GR promoter are involved in the cell specific and enhanced expression of the GR gene.
As shown in Fig. 3, alignment of sequences in the transcriptional control elements of various genes reveals the presence of a unique palindrome adjacent to a putative Sp1 binding site (13,14,20). To further delineate the sequences involved in the GR expression, we fused a single copy of this palindrome, GREFE, to the heterologous CAT reporter plasmid pBL-tkCAT. In human MCF-7, LNCaP, HeLa, and placenta cells transfected with control plasmid pBL-tkCAT, CAT expression was at comparable constitutive levels (Fig. 4, lanes 1, 2, 5, and 6, respectively). In MCF-7 and LNCaP cells (Fig. 4, lanes 3 and 4), CAT expression by the palindrome-containing reporter plasmid pHGR-CAT (Ϫ892/Ϫ878) did not rise above basal levels. In HeLa (lane 7) and placenta cells (lane 8), an 8-fold increase in CAT activity was observed with pHGR-CAT (Ϫ892 to Ϫ878), confirming that the sequences between Ϫ892 and Ϫ879 function as enhancer sequences of GR gene expression.
Purification of the Factor(s) by Sequence-specific Affinity Chromatography-The enriched nuclear extract from human placenta was clarified and depleted of the unspecific adsorbed proteins. Under the affinity chromatography conditions described above, 90% of the factor(s) bound to the stationary matrix. To detect and to visualize the proteins, fractions were analyzed by SDS-polyacrylamide gel electrophoresis and silver staining (Fig. 5A). Next, we performed UV cross-linking analyses followed by SDS-polyacrylamide gel electrophoresis and autoradiography (Fig. 5B) and finally examined the interaction of the factor(s) with the synthetic palindrome (Fig. 5, C and D) and conducted Western blotting (Fig. 5E).
UV Cross-linking Studies-To define the interaction of specific factor(s) with the palindrome, we incubated samples of various fractions with the labeled oligonucleotide, GREFE. The separation of the complexes on acrylamide gels (Fig. 5C) followed the elution pattern of the factors from the affinity column. The proteins were characterized using a combination of DNA-binding and UV cross-linking analyses. The resolution of the UV cross-linked complexes on a SDS-polyacrylamide gel, showed the presence of a major labeled band at 80 kDa (Fig.  5B), which comigrated with the silver-stained 80-kDa protein band (Fig. 5A, lanes fractions 3, 5, 7, 9, 11, and 13). Two additional low molecular bands with reduced intensity at 66 and 50 kDa (Fig. 5B) were also visible. The reduction in the intensity of the retarded complexes in the presence of 100-fold molar non-radioactive competitors in parallel incubations, con-firmed the specificity of the interactions (Fig. 5D). The two proteins will be henceforth designated as GREF1 (80 kDa) and GREF2 (62 kDa).
Immunological Similarities of GREF1 and GREF2 to Human Ku-Autoantigen-We have characterized the proteins GREF1 and GREF2 from human placenta, with the information available from previous investigations on TREF1 and TREF2. We resolved nuclear proteins from human placenta (Fig. 5E, lane 1), HeLa cells (lane 2), oligonucleotide affinity eluted fractions 7, 9, 11, and 13 (lanes 3-5, and 6, respectively) and  3,5,7,9,11, and 13 were resolved on a 7.5% SDS-polyacrylamide gel by electrophoresis and stained with silver. B shows an identical SDS-polyacrylamide gel containing 2 l of the respective fractions incubated with 5,000 cpm of 32 P-labeled double-stranded palindrome, irradiated as described under "Materials and Methods" and subjected to electrophoresis, drying, and autoradiography. The molecular weight standard bands are 98,000, 68,000, 45,000, and 30,000 from top to bottom. C shows the elution pattern of the affinity eluate assayed for DNA binding with 5,000 cpm of 32 P-labeled double stranded palindrome, GREFE, under nondenaturing conditions. Similarly, D shows the specificity of interaction of the 32 P-labeled GREFE and eluted proteins in the absence (lanes 2, 4, 6, and 8) and presence (lanes 3, 5, 7, and 9) of 100-fold molar excess of nonlabeled GREFE as competitor with the eluted peak fractions 9, 10, 11, and 12 of the affinity column. E, Western blot analysis of immunocross-reaction between crude and purified factors with human Ku autoantibodies. Following resolution of the proteins on a 10% SDS-polyacrylamide mini-gel by electrophoresis, the proteins were transferred and the blot was incubated with 1:1000 diluted human serum. The immunoreactive bands were visualized by incubating with 125 I-labeled protein A and autoradiography. bodies. We used bacterially expressed Ku70 or Ku80 (cloned in pET expression vectors) as controls in these experiments. Protein samples from placenta were digested with 0, 5, 50, 500, and 5,000 ng/ml ␣-chymotrypsin (Fig. 6A, lanes 1, 3, 2, 4, and 5, respectively) and p80 expressed in bacteria was digested with 0, 5, 50, 500, and 5,000 ng/ml ␣-chymotrypsin (Fig. 6A, lanes 6,  8, 7, 9, and 10, respectively). The samples were resolved on a 20% SDS-polyacrylamide gel and after electrophoresis, blotted onto Immobilon-P. Following treatment of the blots with monoclonal 111, washing, and incubation with 125 I-labeled protein A and autoradiography, identical peptide bands were recognized by this monoclonal antibody. The structural similarity between GREF1 and Ku80 is reflected by the degree of similarity between the peptide fragments.
Similar experiments performed with purified placental GREF2 and bacterially expressed Ku70 are shown in Fig. 6B. The monoclonal antibody N3H10 used in this study was raised against human Ku and binds to an epitope of the p70 subunit (18). Identical peptide fragments were recognized by the monoclonal antibodies (Fig. 6B) which is a further indication of the primary structure relationship between GREF2 and Ku70 proteins. From these results, we now unequivocally identify GREF1 and GREF2 as the human Ku80 and Ku70 factors.

DNA-Protein Interactions by Gel Mobility Shift Assay-
The ability of the synthetic GREFE to interact with factors present in nuclear extracts of various cells was determined by gel mobility-shift assay. The nuclear extract from human placenta cells in culture showed a strong complex formation (Fig. 7,  lanes 2 and 3). The presence of a gel-shifted band following the addition of 100-fold molar excess of non-labeled probe, demonstrated the specificity of this interaction (lane 3). Under identical experimental conditions, the gel-shifted complexes associ-ated with nuclear extract from MCF-7 cells migrated faster (lanes 4 and 5) than the complexes observed with the placenta nuclear extract (lanes 2 and 3). Samples of HeLa cell nuclear extract formed heavier complexes (lanes 6 and 7) which migrated with identical mobility as the complexes observed with placenta extract. pET-70 produced in bacteria, formed faster migrating complexes (lanes 10 and 11) which were much less intense, indicating that these were lighter than the larger complexes with placenta and HeLa nuclear extracts. A lighter band of fast migrating complexes was observed with the nuclear extract from human prostate cancer cell LNCaP (lanes 12 and 13) and no complexes were visible with either CV-1 (lanes 8 and 9) or COS-1 nuclear extracts (lanes 14 and 15). Similarly, no complexes were observed using labeled synthetic glucocorticoid responsive palindrome (GRE) in control experiment with the purified factors from placenta (not shown).
The Cross-reaction of Purified Factors with Anti-human Ku70 and Ku80 Monoclonal Antibodies-To assess the DNAprotein interaction with intact hGR gene segment containing the protein-binding site, we isolated a 280-base pair hGR gene fragment between 5Ј SmaI (Ϫ1030) and 3Ј SacI (Ϫ750) and used it in DNA-protein interaction assays. Specific interaction of the labeled DNA (Fig. 8A, lane 2) was abolished by immunoabsorbing the factors either with anti-Ku70 (Fig. 8A, lane 3) or anti-Ku80 (lane 4) and protein A-Sepharose. The specificity of the interaction was assayed by the addition of a 10-fold (lane 5) and 100-fold (lane 6) molar excess of non-labeled DNA as competitor. Addition of a nonspecific IgG in the depletion experiment did not affect the formation of specific DNA-protein complexes (Fig. 8B, lane 7).
The ability of monoclonal antibodies to interact with specific proteins can be used to characterize the DNA-protein interaction in supershift assays. As shown in Fig. 8B, compared to the complexes observed in the absence of monoclonal antibodies FIG. 6. A and B, partial peptide mapping with monoclonal antibodies Ab 111 (Ku80) and Ab N3H10 (Ku70). A, proteins purified from placenta by affinity chromatography (lanes [1][2][3][4][5] or Ku80 expressed in bacteria (lanes 6 -10) were cleaved in SDS sample buffer at 37°C for 1 h with 0 (lanes 1 and 6), 5 (lanes 3 and 8), 50 (lanes 2 and 7), 500 (lanes 4 and 9), and 5,000 ng/ml (lanes 5 and 10) ␣-chymotrypsin as indicated. The peptides were resolved on a 20% SDS-polyacrylamide gel, transferred onto Immobilon, treated with Ab 111 as described under "Materials and Methods." The immunocross-reaction detected in the autoradiograph shows the identity of GREF1 as Ku80. B, identical experiments as described in A are shown, but the comparison is with the bacterially expressed human Ku70. The monoclonal antibody used in this Western blot was directed to an epitope in Ku70 (Ab N3H10).

FIG. 7. Interaction of nuclear proteins with labeled palin-
drome. An aliquot of 5,000 cpm of 32 P-labeled double-stranded palindrome, GREFE, was incubated with 2.5 g of extracts from various cells in the absence and presence of 100-fold molar excess of nonlabeled competitor. The complexes were resolved on a 4% polyacrylamide gel under nondenaturing conditions as described under "Materials and Methods." The DNA-protein complexes were visualized by autoradiography. (Fig. 8B, lane 2), incubation of the proteins with either anti-Ku70 (N3H10) or anti-Ku80 (111) resulted in much slower migrating complexes (Fig. 8B, lanes 3 and 4, respectively). Although the addition of a nonspecific IgG did not affect the interaction (lane 5), depletion of these factors by protein A-Sepharose absorption, eliminated the shifted complexes in the autoradiograph (Fig. 8B, lanes 6 and 7). A control depletion analysis using a nonspecific IgG (lane 8) showed the specificity of these experiments which we further analyzed by adding 10and 100-fold molar excess of non-labeled DNA as competitor (lanes 9 and 10, respectively). In a similiar experiment, supershifted complexes with MCF-7 nuclear extract a light band was observed only with monoclonal antibody N3H10, indicating the presence of Ku70-like factor in MCF-7 cells (not shown).
To characterize the specificity of transcription activation of the hGR gene, MCF-7 cells were cotransfected with the authentic hGR gene chimera pHGR2.7-CAT (13,14). CAT activity expressed by control plasmid pBL-CAT (Fig. 10, lanes 1-4) 5-8) showed a hormonal response to treatment with 1 M R5020 (promegestone, (17␤)-17-methyl-17-(1-oxopropyl)estra4,9-diene-3-one) (lane 6) but no increase in transcription was observed by cotransfection with either Ku70 or Ku80 expression vectors alone (lanes 7 and 8, respectively). The results were considerably different in assays with the intact pHGR2.7-CAT (lanes 9 -12) and pGREFE-CAT (lanes 13-16). Transcription of the CAT gene was stimulated considerably by pcDNA Ku80 (lanes 12 and 16) but not by pcDNA Ku70 (lanes 11 and 15). The control pcDNA1 cotransfected cells demonstrated the basal transcription levels ( lanes  10 and 14). HGR-CAT constructs by transient cotransfection in MCF-7, HeLa, and human placenta cells. The high levels of CAT activity observed with pHGR⌬908-CAT and the decline with pHGR⌬830-CAT in human placenta and HeLa cells, first indicated the presence of a regulatory mechanism between these deletions that involved cis-acting sequence motifs with a putative Sp1 binding site at the 3Ј end (29).
Sequence Homologies between the HGR Promoter and Promoters of Other Genes-The GR gene sequence palindrome responsible for the high levels of GR expression in human placenta and HeLa cells is distinct from sequences described previously. However, it shares common features with transcriptional control elements present in a number of other genes. The GR gene sequence 5Ј-TGACACA-3Ј (Ϫ888/Ϫ884) at the center of the palindrome within the deletion mutants (Ϫ908/Ϫ830) essential for high level expression, is very similar to the promoters of a number of genes such as the cAMPresponsive promoter of the somatostatin gene (30), the transferrin receptor control element a and the phorbol ester-(TPA) inducible gene promoter (31)(32)(33)(34)(35)(36).
The regulatory region of the transferrin receptor promoter necessary for increased expression (37)(38)(39)(40) contains an 8-base pair sequence motif (TRA nucleotides Ϫ77 to Ϫ70). Similarly, seven nucleotides in the element B of the human U1 small nuclear RNA are identical. These elements are known to be essential in the transferrin receptor and the U1 gene transcription (41,42). The GR gene sequence under study here has also considerable homology with the conserved TPA-responsive element TRE (consensus: 5Ј-T(G/T)AGTCA(G/C)-3Ј) present in a number of TPA-inducible gene promoters (36). The TRE that confers TPA inducibility on heterologous promoters is recognized by the TPA-regulated DNA binding activity known as AP-1 (43). A 43-kDa factor ATF, which was shown to interact with the cyclic AMP-responsive element of E1a inducible promoters of adenovirus type 5, was purified from HeLa cells (33). The palindromic sequence of hGR gene promoter also has a core sequence in common with the canonical 12-base pair imperfect palindromic sequence which constitutes the binding site for the 46-kDa transcription factor MLTF or USF. This sequence is highly conserved among the human adenovirus family and is known to be essential for transcription from the MLP of adenovirus type 2 (44 -46). GREF1 and GREF2 from the Human Placenta Are Immunologically Related to Ku80 and Ku70 -Immunoblotting analyses demonstrate that GREF1 and GREF2 share immunological similarities with the autoantigen Ku. Peptide mapping with monoclonal antibodies confirmed the identity of GREF1 and GREF2 as proteins Ku80 and Ku70, respectively. Although DNA interaction with placenta nuclear extract indicated a strong DNA binding capacity, we have shown by UV crosslinking analyses with labeled oligonucleotides and purified factors, that only GREF1 interacts directly with DNA. Depletion of GREF1 and GREF2 by immunoabsorption and the disappearance of gel-shifted DNA-protein complexes in mobilityshift assays confirmed the specificity of the factors to interact with the palindrome. Although, the transcription factors CREB, the Jun-Fos/Fra complex of AP-1, and MLTF/USF appear to recognize related sequence elements (24,25,(47)(48)(49)(50)(51)(52)(53)(54)(55)(56), there is a marked size difference between these factors and GREF1 (80 kDa) and GREF2 (62 kDa). In human placenta extracts, the presence of protein PSE1, a dimer of 83-and 73-kDa subunits, was demonstrated. This PSE1 had immunological and sequence relationships with p70 and p86 of the Ku complex (57).
Heterodimers of Ku80 and Ku70-like Proteins Regulate hGR Gene Transcription in MCF-7 Cells-The cotransfection of pcDNA1-Ku70 with pGREFE-CAT did not elevate CAT activity, although we detected a high concentration of an immunologically related polypeptide in MCF-7 cells with polyclonal antibodies. Western blotting showed the presence of a single protein of 62 kDa in MCF-7 cells and DNA-protein interactions with MCF-7 nuclear extract showed a lighter, faster moving complex than with HeLa or placenta extracts. Interestingly enough, the introduction of Ku80 into MCF-7 cells resulted in an elevation in CAT activity. This would indicate that the Ku-like complex, with an acidic and a leucine zipper region commonly associated with transcription factors (58) and DNAbinding proteins (47), acts in the form of a heterodimer (24, 25, 48 -56).
Several studies have demonstrated that protein dimers of 80 -85 kDa/70 -75 kDa influence transcription. Footprinting experiments showed that the proximal sequence element in the promoter of small nuclear RNA genes that influence transcription activity, was made of two subunits of 83-and 73-kDa proteins (57). The promoter region of the human collagen type IV genes was specifically recognized by the CTC factor CTCBF, a dimer of 85 and 75 kDa subunits. This dimer has sequence homology with 80-and 70-kDa proteins of Ku and is involved in the control of divergent transcription of collagen genes (47).
The Ku protein, a heterodimer of 70-and 80-kDa subunits, was first identified as a DNA-binding protein that was recog- nized by autoantibodies from certain patients with scleroderma-poliomyositis (48). It is also implicated in DNA replication, repair, and transcriptional control (59). The 72-kDa subunit is the product of a gene family that produces different variants of the dimer (60). In the placenta, for example, the protein dimers are made of 80-and 62-kDa subunits. Ku, an abundant protein (4 ϫ 10 5 copies/HeLa cell), is mainly nuclear. In CV-1 and LNCaP cells, no Ku-like proteins were detected. Characteristics such as sequence homology with other transcription factors, recognition of promoter sequences, DNA binding capacity, dimer formation, and transcription activation capabilities suggest that it could be a general regulating factor (61,62). Several studies also point to a possible link between Ku and the cell cycle. A Ku80 was recently identified as a somatostatin receptor that regulates a PP2A phosphatase. PP2A in turn, dephosphorylates specific GR sites (63) and is also involved in the dephosphorylation of histone H1 in a cascade of phosphorylation/dephosphorylation events.
Phosphorylation of histone H1 is necessary for chromosome condensation at the M phase of the cell cycle (64). p34 cdc2 , the catalytic component of the maturation/mitosis promoting factor, in complex with a regulatory cyclin protein, is a cell cycle protein. Inappropriate synthesis of cyclins, crucial to the propagation of the cell cycle, could result in the untimely stimulation of cell division kinases and so lead to unwarranted cell proliferation. The dephosphorylation of Tyr-15 and Thr-14 of p34 cdc2 is achieved by the specific phosphatase p80 cdc25 (65). By maintaining this protein in an inactive form, PP2A inhibits its activity and maturation/mitosis promoting factor cannot phosphorylate H1. Consequently, chromosome condensation does not take place and the cell cannot proceed through mitosis. The Ku that regulates PP2A is also a dimer of 86-and 70-kDa subunits (66). Ku genes are activated during late G 1 phase and Ku80 dissociates from the chromosomes during mitosis (67). Ku, also part of a DNA protein kinase, has a regulatory component that phosphorylates RNAPII in the presence of transcription factors TFIID, TFIIB, and TFIIF (68).
In the same manner that Ku80 is implicated in the cell cycle, Ku70 in MCF-7 cells and GREF1 and GREF2 from placenta could also play a role in the regulation of cell cycle kinases and phosphatases. Many cells are unresponsive to glucocorticoids during the G 2 phase of the cell cycle and hence some activities of the GR may be subject to cell cycle control (63). Progesterone induces the expression of p39 mos , a kinase which regulates a protease which in turn cleaves the cyclin part of the maturation/mitosis promoting factor complex. In a similar pathway, glucocorticoids could be involved in a cascade of events that include kinases, phosphatases, and GR. Although a number of properties of Ku have been identified, we now attribute to Ku a role in the regulation of housekeeping genes such as the GR. The wide distribution of GR in nearly all cell types and tissues implies a major role in metabolism and growth. In this study, we report the isolation of two novel proteins, GREF1 and GREF2 immunologically identical to Ku80 and Ku70, from human placenta and we demonstrate the direct involvement of GREF1 with DNA. The elevation of CAT activity in MCF-7 cells by the introduction of pKu80 and pGREFE-CAT or pHGR 2.7-CAT but not with MMTV-CAT, suggests that the dimer of Ku-proteins is implicated in the regulation and cell specific expression of the hGR gene.