Transcriptional control of the human thromboxane synthase gene in vivo and in vitro.

Thromboxane A(2), a potent mediator of vasoconstriction and platelet aggregation, is synthesized from prostaglandin H(2) by thromboxane synthase (TXAS). We report here on promoter analyses of human TXAS using in vitro transcription and in vivo transfection methods. The 39-bp core promoter, containing both TATA and initiator elements, accurately initiates transcription in an orientation-dependent manner in a cell-free transcription system. Mutation of either TATA or initiator abolished transcriptional activity, but the upstream sequence had no effect on TXAS promoter activities in vitro, suggesting that the core promoter is sufficient for transcriptional activity from a naked DNA template. In contrast, mutation of both elements drastically decreased the promoter activity in vivo. Furthermore, TXAS proximal promoter containing the nucleotides -90 to -56 relative to the transcriptional start site was necessary for promoter transactivation in vivo. Transcriptional activities from 5'-deletion mutants indicated that the effects of the nucleotides -90/-56 were more pronounced in stably transfected cells (a 150-fold difference) than in the transiently transfected cells (an 8-fold difference), reflecting the effects of TXAS transcriptional activation from replicating and nonreplicating DNA templates. Partial micrococcal nuclease digestion indicated that the sequence -90/-56, where the NF-E2 site is located, is associated with alterations of nucleosomal structure at the regions of promoter and reporter gene but not the region further downstream. Mutagenesis and forced expression studies demonstrated a critical role of p45 NF-E2 in controlling TXAS expression under native chromatin conditions. Band shifting and chromatin immunoprecipitation assays indicated that p45 NF-E2 was bound to the TXAS promoter in vitro and in vivo. Indirect end labeling and ligation-mediated PCR analyses further demonstrated that the occupation of TXAS promoter NF-E2 site was associated with disruption of nucleosomal structure. Collectively, these results indicate that binding of NF-E2 is critical both for alteration of the nucleosomal structure and for activation of the TXAS promoter, whereas the TATA and initiator elements are essential for transcription.

Thromboxane A 2 , a potent mediator of vasoconstriction and platelet aggregation, is synthesized from prostaglandin H 2 by thromboxane synthase (TXAS). We report here on promoter analyses of human TXAS using in vitro transcription and in vivo transfection methods. The 39-bp core promoter, containing both TATA and initiator elements, accurately initiates transcription in an orientation-dependent manner in a cell-free transcription system. Mutation of either TATA or initiator abolished transcriptional activity, but the upstream sequence had no effect on TXAS promoter activities in vitro, suggesting that the core promoter is sufficient for transcriptional activity from a naked DNA template. In contrast, mutation of both elements drastically decreased the promoter activity in vivo. Furthermore, TXAS proximal promoter containing the nucleotides ؊90 to ؊56 relative to the transcriptional start site was necessary for promoter transactivation in vivo. Transcriptional activities from 5-deletion mutants indicated that the effects of the nucleotides ؊90/؊56 were more pronounced in stably transfected cells (a 150-fold difference) than in the transiently transfected cells (an 8-fold difference), reflecting the effects of TXAS transcriptional activation from replicating and nonreplicating DNA templates. Partial micrococcal nuclease digestion indicated that the sequence ؊90/؊56, where the NF-E2 site is located, is associated with alterations of nucleosomal structure at the regions of promoter and reporter gene but not the region further downstream. Mutagenesis and forced expression studies demonstrated a critical role of p45 NF-E2 in controlling TXAS expression under native chromatin conditions. Band shifting and chromatin immunoprecipitation assays indicated that p45 NF-E2 was bound to the TXAS promoter in vitro and in vivo. Indirect end labeling and ligation-mediated PCR analyses further demonstrated that the occupation of TXAS promoter NF-E2 site was associated with disruption of nucleosomal structure. Collectively, these results indicate that binding of NF-E2 is critical both for alteration of the nucleosomal structure and for activation of the TXAS promoter, whereas the TATA and initiator elements are essential for transcription.
The transcriptional start site of the human TXAS gene was mapped at nucleotide 137 upstream from the ATG translation site (15). The sequence of TXAS initiator (Inr) resembles the weakly conserved Inr element for RNA polymerase II. An upstream sequence between nucleotides Ϫ90 and Ϫ56 relative to the TXAS transcriptional start site was shown to be important for TXAS promoter activity in a megakaryotic cell line. Further detailed analysis identified transcription factor NF-E2 binding at the nucleotide region Ϫ83/Ϫ77 as being responsible for TXAS transactivation (17,18). The NF-E2 site also plays an important role in activation of the mouse and rat TXAS promoters (19,20). Transcription factor NF-E2 is a basic leucine zipper heterodimer consisting of p45 and p18 subunits, which belong to the CNC protein family and the small Maf protein family, respectively (21)(22)(23)(24). Expression of p45 NF-E2 is primarily restricted to hematopoietic cells (22), whereas small Maf (p18) appears to be ubiquitously expressed (24). p45 NF-E2 is a major enhancer mediating globin gene expression in developing and maturing erythroid cells (Ref. 25 and references therein). However, mice lacking p45 NF-E2 develop relatively normal erythropoiesis but suffer from abnormal megakaryocytic development and fail to produce platelets (26). Megakaryocytes from p45 NF-E2-null mice do not express TXAS mRNA, whereas other tissues, such as fetal liver and bone marrow, have normal levels of TXAS mRNA (18). It should also be noted that the transactivation mechanism of NF-E2 participates, at least in part, in the chromatin remodeling (27)(28)(29). Binding of the NF-E2 to the chromatin template removed nucleosomes from the ␤or ⑀-globin promoter in both in vivo and in vitro studies (30,31).
Previous studies describing transcriptional regulation of the TXAS promoter have primarily been carried out by transient transfection assays in which the template DNA is nonreplicat-ing and is unlikely to be assembled into organized nucleosomes (Ref. 32 and references therein). Stably transfected DNA templates, however, represent replicating chromatin and are assembled into repeated arrays of nucleosomes. Considering that transcriptional activities may depend on the template state, we examined TXAS promoter activities from naked DNA, nonreplicating chromatin, and replicating chromatin templates. Our results led us to propose a model in which NF-E2 induces nucleosomal alteration in the vicinity of the TXAS promoter, thereby allowing the basic transcriptional machinery to bind to the core promoter.
Cell Culture-The human megakaryoblastic cell line (MEG-01) and the human erythroleukemia cell line (HEL) were maintained in suspension in RPMI 1640 medium, and HeLa human cervical cancer cells were cultured in modified Eagle's medium. All of the media were supplemented with 10% heat-inactivated fetal bovine serum (Hyclone, Logan, UT), 2 mM glutamine, 100 units/ml penicillin, and 100 g/ml streptomycin. Cells were grown at 37°C in a 5% CO 2 humidified incubator.
Plasmid Construction-DNA fragments containing the TXAS promoter sequences Ϫ36/ϩ3 and Ϫ248/ϩ3 (relative to the transcriptional start site) were obtained by PCR amplification and were inserted into the Ecl136II site of a G-free cassette vector, pC 2 AT, to obtain both the correct orientations (pC 2 AT-TX36(ϩ) and pC 2 AT-TX248(ϩ)) and the reverse orientations (pC 2 AT-TX36(Ϫ) and pC 2 AT-TX248(Ϫ)). Mutations in the TATA-box (TATA to TCCC; underline indicates mutation) and Inr (CACATT to CACGTG) were introduced by PCR-based sitedirected mutagenesis. Plasmid pGL3-248/ϩ3, which has the Ϫ248/ϩ3 TXAS promoter fragment linked to a luciferase reporter vector, was constructed first by cloning the Ϫ248/ϩ3 PCR fragment into the EcoRV site of pBluescript II SK (Stratagene) and then cloning the SacI/HindIII fragment of the resultant pBluescript into pGL3-Basic. The correct orientation was confirmed by PCR. A series of 5Ј-deletion mutants in the TXAS promoter linked to pGL3 (i.e. pGL3Ϫ280/ϩ137, Ϫ190/ϩ137, Ϫ140/ϩ137, Ϫ90/ϩ137, Ϫ56/ϩ137, and Ϫ38/ϩ137) and 3Ј-deletion mutations of pGL3Ϫ280/ϩ90, Ϫ280/ϩ75, Ϫ280/ϩ52, and Ϫ280/ϩ3 were constructed in a similar fashion. The numberings of the constructs indicate the nucleotide positions relative to the transcriptional start site. The minichromosome constructs, pMCϪ280/ϩ137, Ϫ190/ϩ137, Ϫ90/ϩ137, and Ϫ56/ϩ137, were created by first inserting the XbaI/ BamHI fragment of pGL3-Basic that contains the SV40 poly(A) signal into the XbaI/BamHI site of p220.2. Subsequently, the XbaI fragments of pGL3Ϫ280/ϩ137, Ϫ190/ϩ137, Ϫ90/ϩ137, and Ϫ56/ϩ137, containing the luciferase gene and the corresponding TXAS promoter sequence, were each inserted into the XbaI site of the p220.2-SV40 poly(A) construct. A point mutation at the NF-E2 binding site (TGATTCA to GGATCCA) in pMCϪ280/ϩ137, called pMCϪ280/ϩ137-mNF-E2, was generated by PCR. NF-E2 cDNA was obtained by reverse transcription-PCR from total MEG-01 cellular RNA (17) and first subcloned into TA vector and then into the pcDNA3.1 eukaryotic expression vector (Invitrogen) at the BamHI/HindIII site. All constructs generated by PCR were confirmed by DNA sequencing.
In Vitro Transcription Assay-HeLa cell nuclear extracts were obtained as described (36). Optimal assay conditions were defined in preliminary studies. Typically, reactions (25 l) contained 20 mM HEPES, pH 7.9, 10% glycerol, 45 mM KCl, 3 mM MgCl 2 , 0.1 mM EDTA, 0.5 mM dithiothreitol, 1 g of test template DNA, 0.5 g of control adenovirus major late promoter template pMLC 2 AT190, 10 units of RNase T1, 20 units of RNA guard (Amersham Biosciences), 3 l of HeLa nuclear extract, and the nucleotide mix (0.4 mM ATP, 0.4 mM TTP, 1 mM 3Ј-O-methyl-GTP (Amersham Biosciences), 0.016 mM UTP, and 10 Ci of [␣-32 P]UTP (800 Ci/mmol)). Reactions were initiated by adding the nucleotide mix, incubated for 60 min at 30°C, and terminated by adding 150 l of stop solution containing 0.3 M Tris, pH 7.4, 0.3 M sodium acetate, 0.5% SDS, 2 mM EDTA, and 0.45 g of yeast tRNA. Products were then extracted with phenol/chloroform/isoamyl alcohol and precipitated with ethanol. The pellets were washed with 80% ethanol, dried, and resuspended in 50% formamide and 5 mM EDTA. The samples were heated at 95°C for 2 min before they were separated on a 5% polyacrylamide gel containing 7 M urea and visualized by autoradiography.
Transient Transfection-The cationic lipids used for transient transfection were Lipofectin reagent and PLUS reagent for HeLa cells and LipofectAMINE 2000 for MEG-01 cells. Transfection was initially performed following the manufacturer's instructions and then optimized in later experiments. Briefly, 1 day before transfection, 1 ϫ 10 6 MEG-01 cells were seeded in six-well plates. MEG-01 cells were exposed to 10 l of LipofectAMINE 2000, 2 g of pGL3-TXAS reporter vector DNA, and 1 g of pCMV-␤-Gal in 2 ml of OptiMEM. For HeLa cells, 1 ϫ 10 5 cells were transfected by 0.5 ml of OptiMEM containing 2 g of pGL3-TXAS reporter vector DNA, 1 g of pCMV-␤-Gal, 20 l of Lipofectin reagent, and 10 l of PLUS reagent after washing with serum-free medium. The medium was replaced with complete medium 4 h after transfection for HeLa cells. After a 48-h incubation without changing the medium, the cells were harvested and lysed in reporter lysis buffer (Promega). Luciferase assays were performed with 20 l of lysate and 100 l of luciferase assay reagent using a MiniLumat Luminometer LB9506 (Berthold, Wildbad, Germany). ␤-Galactosidase activities were determined using a chemiluminescence assay kit.
Stable Transfection-Stable transfection of HEL cells (1 ϫ 10 6 cells in 2 ml) and HeLa cells (1 ϫ 10 5 cells in 2 ml) was carried out using 6 l of FuGENE6 and 2 g of the minichromosome DNA. One day after transfection, cells were subcultured into 20 wells of 96-well plate (HEL) or a 100-mm dish (HeLa) and selected with 400 g/ml of hygromycin B. Several clones of each construct were obtained. Copy numbers of each clone were determined by Southern blot analysis.
Electrophoretic Mobility Shift Assay (EMSA)-Nuclear extracts of HEL cells were prepared by the method described previously (37) and stored at Ϫ70°C until use. EMSA was performed based on a previously described method (38). Briefly, 5 g of nuclear extracts were incubated with a 32 P-end-labeled probe (TXAS promoter nucleotides Ϫ121 to ϩ5) in a binding reaction buffer containing 20 mM HEPES, pH 7.9, 0.2 M EDTA, 15% glycerol, 1 g of poly(dI-dC), 100 mM KCl, 2.5 mM MgCl 2 , and 1 mM dithiothreitol for 20 min at room temperature. For supershift assay, nuclear extracts were incubated with antibodies for 30 min before the labeled probe was added. The mixtures were electrophoresed in a 4.5% polyacrylamide gel and autoradiographed overnight at Ϫ70°C.
Chromatin Immunoprecipitation (ChIP) Assay-ChIP and duplex PCR were carried out as previously described (39). HEL cells (2.4 ϫ 10 7 ) maintaining minichromosome pMCϪ280/ϩ137 were fixed with 1% formaldehyde for 10 min. Formaldehyde was neutralized by incubation with 125 mM glycine for 10 min and then washed twice with ice-cold PBS. Cells were lysed for 5 min on ice in 500 l of buffer containing 50 mM Tris, pH 8.0, 2 mM EDTA, 0.1% Nonidet P-40, 10% glycerol supplemented with protease inhibitors. Nuclei were centrifuged at 3000 rpm for 3 min and resuspended in 200 l of 50 mM Tris, pH 8.0, 5 mM EDTA, 1% SDS, and then sonicated five times for 5 s each time. The nuclei preparations were cleared by centrifugation and diluted 10 times in 50 mM Tris, pH 8.0, 0.5 mM EDTA, 0.5% Nonidet P-40, 0.2 M NaCl. Extracts were precleared for 2 h with 60 l of a 50% suspension of protein A-Sepharose 4B beads (Sigma), 40 g of sonicated salmon sperm DNA, and 80 g of bovine serum albumin. An aliquot of precleared chromatin was removed, heat-treated to reverse protein-DNA interaction, and precipitated with isopropyl alcohol (referred to as input) and used in the subsequent duplex PCR analysis. The remainder of chromatin sample was then separated into two tubes for different antibody treatments. Immunoprecipitations were carried out at 4°C for 6 h by adding 10 g of p45 NF-E2 antibody or irrelevant rabbit serum with 20 l of 50% protein A-Sepharose 4B beads (presaturated with salmon sperm DNA). The beads were washed five times with 1 ml of wash buffer (20 mM Tris, pH 8.0, 0.1% SDS, 2 mM EDTA, 1% Nonidet P-40, 150 mM NaCl) and three times with 10 mM Tris, pH 8.0, 1 mM EDTA. Protein-DNA complexes were eluted by three successive treatments with 150 l of 1% SDS, 100 mM NaHCO 3 . NaCl was then adjusted to 0.3 M, and 1 l of RNase A (10 mg/ml) was added. Protein-DNA cross-links were reversed by heating at 65°C overnight, and DNA was extracted by adding 20 l of 1 M Tris, pH 7.5, 10 l of 0.5 M EDTA, and 2 l of proteinase K (20 mg/ml) at 50°C for 30 min. After phenol/ chloroform extraction and isopropyl alcohol co-precipitation with 30 g of glycogen, the pellet was resuspended in 30 l of H 2 O. For the duplex PCR, 2 l of ChIP DNA was used as template in a total volume of 10 l of PCR buffer containing 0.5 M each of the primer pair for TXAS promoter (corresponding to the promoter sequences Ϫ90/Ϫ72 and ϩ118/ϩ137) and 0.5 M each of the primer pair for the TXAS 3Јuntranslated region (corresponding to the cDNA sequences 1581/1600 and 1901/1920 downstream from the translational start site), 7.5 Ci of [␣-32 P]dCTP, 0.2 mM dNTP and Taq polymerase. After amplification, PCR products were separated by a 4.5% polyacrylamide gel before autoradiography.
Chromatin-DNA Cross-linking, Preparation of Nuclei, and MNasecoupled Southern Blot Analysis-Nuclei from HEL cell clones were prepared according to a modification of published procedures (30). Briefly, 3 ϫ 10 7 cells were washed with phosphate-buffered saline, and chromatin-DNA was cross-linked by incubation of cells with 1% formaldehyde at room temperature for 20 min (40) before being washed twice with ice-cold phosphate-buffered saline and resuspended in 6 ml of suspension buffer (10 mM Tris, pH 7.4, 1 mM EDTA, 0.1 mM EGTA, 15 mM NaCl, 50 mM KCl, 0.15 mM spermine, and 0.5 mM spermidine) containing 0.2% Nonidet P-40 and 5% sucrose. After a 5-min incubation on ice, the suspension was centrifuged (15 min, 150 ϫ g, 4°C) through a 3.5-ml cushion of suspension buffer containing 10% sucrose. Nuclei were gently resuspended in 3 ml of suspension buffer containing 8.5% sucrose, and aliquots of 1 ml were digested with 10, 50, or 150 units of MNase in the presence of 1 mM CaCl 2 for 7 min at room temperature. The reactions were stopped by adding stop buffer at final concentrations of 0.5 M NaCl, 10 mM EDTA, 1% SDS, and 0.1 mg/ml proteinase K and then incubated at 37°C overnight. The cross-linking of DNA-protein was reversed by heating at 60°C for 5 h (41). The DNA was purified by multiple phenol/chloroform extractions, analyzed by gel electrophoresis, and subjected to Southern blot analysis using probes that were labeled by random primers.
For indirect end-label analysis, the procedures for sample preparation including nuclei preparation, chromatin-DNA cross-linking, MNase digestion, and DNA isolation, were the same as described above except that 0, 10, and 30 units of MNase were used. Purified DNAs were digested with XbaI and BclI, separated by gel electrophoresis after phenol/chloroform extraction, and analyzed by Southern blotting using the 5Ј-end of the luciferase cDNA (cut with HindIII and BclI from the pGL3-basic) as the probe.
MNase-coupled Ligation-mediated Polymerase Chain Reaction (LM-PCR) Analysis-MNase-treated nuclei were prepared, and total DNAs were isolated as described above. LM-PCR was carried out following a standard method (38). Briefly, the DNA sample (5 g) was phosphorylated with T4 polynucleotide kinase (New England Biolabs) in the presence of 1 mM ATP and ligated to the unidirectional linker (prepared by annealing the two oligonucleotides, linker 1 (5Ј-GCGGTGACCCGG-GAGATCTGAATTC-3Ј) and linker 2 (5Ј-GAATTCAGATC-3Ј)) with T4 DNA ligase (Invitrogen). The linker-ligated DNA was extracted with phenol/chloroform and precipitated with ethanol. DNA samples (0.3-0.9 g; based on the copy number of minichromosomes) were used as templates for 22 cycles of PCR amplification. The PCR mixture (40 l) contained 0.4 l of Taq DNA polymerase, 0.4 l of 10ϫ Expand HF PCR buffer with 15 mM MgCl 2 (Roche Molecular Biochemicals), 0.8 l of 10 mM dNTP, and 1.6 l each of 10 M linker 1 oligonucleotide and TXAS-specific primer P1. The reaction was carried out with the following conditions: first cycle, denature for 5 min at 95°C, with the remaining cycles for 1 min at 95°C, 2 min at optimal annealing temperature (54 -56°C, empirically determined for each primer set), 5 min at 72°C, and an additional 5-min extension at 72°C. The PCR product (5 l) was added to 15 l of a fresh PCR mixture containing 0.1 l of Taq DNA polymerase, 1.5 l of 10ϫ Expand HF PCR buffer with 15 mM MgCl 2 , 0.4 l of 10 mM dNTP solution, and 1.0 l of 1 M 32 P-end labeled TXAS-specific primer P2. The second amplification was carried out for 11-18 cycles (empirically determined for each primer set) of linear PCR amplification consisting of denaturation for 5 min at 95°C for the first cycle, with the remaining cycles for 1 min at 95°C, 2 min at optimal annealing temperature (56 -58°C, empirically determined for each primer set), 5 min at 72°C, and an additional 5-min extension at 72°C. After the second amplification, the reaction products were separated on 6% polyacrylamide, 7 M urea-sequencing gels. The sequence of the TXAS-specific oligonucleotides were as follows: 3Ј side promoter primers, P1 (5Ј-TCCTCTGAGACCCCAGAGTG-3Ј) and P2 (5Ј-GATCTCGC-CGCCCTTATGGGAAG-3Ј); 5Ј side promoter primers, P1 (5Ј-GTACAA-GTCCGTGGTTACAACC-3Ј) and P2 (5Ј-GTTTCTCAGCAAACATGGG-GAAG-3Ј). The locations of the primers in the TXAS promoter are shown in Fig. 1.

RESULTS
In Vitro Transcription of the Human TXAS Gene Promoter-In initial experiments, we used the core promoter (nucleotides Ϫ36 and ϩ3) of the TXAS gene linked to a 377-bp G-free cassette vector, pC 2 AT, as a template (pC 2 AT-TX36(ϩ)) to examine TXAS gene expression. Transcription of this template was carried out in HeLa nuclear extracts as described under "Experimental Procedures." The test template yielded a 380nucleotide transcript (Fig. 2A, lane 1), consistent with the predicted length of transcript accurately initiated from the TXAS transcriptional start site. The promoter sequence from the adenovirus major late gene (Ϫ400 to ϩ10) inserted into a 180-bp G-free pC 2 AT vector (pMLC 2 AT190) was used as control template and resulted in correct initiation and a 190-nucleotide transcript as well as "read-through" transcripts giving a 200-nucleotide transcript after digestion by the G-specific ribonuclease T1 (Fig. 2A, lane 2). The reverse orientation of the TXAS core promoter inserted into the pC 2 AT vector (pC 2 AT-TX36(Ϫ)), however, showed no promoter activity above that for the internal promoter template pMLC 2 AT190 (Fig. 2A, lane 3). We conclude that the 39 bp of TXAS core promoter is sufficient to activate correct initiation of transcription in a direction-dependent manner in the cell-free system.
To optimize concentrations of nuclear extract, templates, KCl, and MgCl 2 for the TXAS in vitro transcription, we carried out assays at different final concentrations of these components. Increasing the amount of nuclear extract from 1 to 4 l (the protein concentration was ϳ12 mg/ml) gave a linear increase in correctly initiated transcripts from both pC 2 AT-TX36(ϩ) and pMLC 2 AT190 templates (data not shown). We thus used 3 l of HeLa nuclear extract for the subsequent studies. Template levels of 1.0 g for pC 2 AT-TX36(ϩ) and 0.5 g for pMLC 2 AT190, a KCl concentration of 45 mM, and a MgCl 2 level of 3 mM were also found to maximize the in vitro transcription activities (data not shown). These conditions are similar to those used for analyzing other native and artificial promoters (42,43).
Effects of TXAS Gene Upstream Promoter Sequence on in Vitro Transcription-To test whether the upstream region is important in cell-free transcription, we linked TXAS promoter nucleotides between Ϫ248 and ϩ3 to the 377-bp G-free cassette vector (pC 2 AT-TX248(ϩ)) and compared its promoter strength with that of pC 2 AT-TX36(ϩ). Both pC 2 AT-TX36(ϩ) and pC 2 AT-TX248(ϩ) showed strong signal intensity at the position of the expected 380-nucleotide transcript, with no significant difference between the two (Fig. 2B, lanes 3 and 4). In fact, the signal from pC 2 AT-TX36(ϩ) was 30-fold higher than that from pC 2 AT-TX248(ϩ) when normalized to signals from pMLC 2 AT190. As negative controls, constructs with reversed core promoter (pC 2 AT-TX36(Ϫ)) and upstream promoter (pC 2 AT-TX248(Ϫ)) were also examined. Both constructs produced very weak transcriptional activities (Fig. 2B, lanes 1 and 2). Because HeLa cells do not express TXAS (44), nuclear extracts of HEL cells (which constitutively express TXAS) were added to the HeLa nuclear extract. However, supplementation with HEL nuclear extract did not enhance the transcriptional activity of pC 2 AT-TX248(ϩ) (data not shown). These results suggest that the core promoter of the TXAS gene, introducing as a naked DNA template in the cell-free system, is sufficient for transcription, and the upstream region is not critical for transcription.
Role of TATA Box and Initiator of TXAS Promoter in the in Vitro and in Vivo Transcription-We further examined the functional elements of TXAS core promoter that elicit transcriptional activity in vitro. Two putative elements, the TATA box and Inr, are present at nucleotides Ϫ29/Ϫ26 and Ϫ3/ϩ3, respectively (Fig. 1). The sequence of the pyrimidine-rich TXAS Inr, CACA ϩ1 TT, closely resembles a weak Inr consensus sequence, PyPyA ϩ1 N(T/A)PyPy (A is the transcriptional start site and Py is pyrimidine) (45), and is similar to the adenovirus major late Inr (CTCACT) and the ␤-globin Inr (CTTACA) (42). We first introduced single mutations in either the TATA box or in the Inr and a double mutation in pC 2 AT-TX36(ϩ). As shown in Fig. 3A, a single  ically reduced the TXAS transcriptional activity, indicating that these elements are functional and crucial in directing the transcriptional activity in vitro. In a separate experiment, dou-ble-stranded oligonucleotide composed of wild-type TATA and the Inr region (Ϫ36/ϩ3) was used as a competitor in the in vitro transcription assay. However, no competition was observed at FIG. 5. Nucleosomal structure of minichromosomes. Nuclei from HEL cells carrying minichromosome constructs, pMCϪ90/ϩ137 and pMCϪ56/ϩ137, were digested by increasing amounts of MNase (10, 50, and 150 units/ml; indicated by the wedges). Total DNAs were isolated, separated on a 2% agarose gel, and subjected to Southern blot analysis. A, linear map of minichromosome indicating the location of the PCR-generated probes, as shown in rectangles below the map. P indicates the TXAS promoter. B, ethidium bromide-stained gel. The amounts of total DNA loaded were as follows: pMCϪ90/ϩ137, 30 -35 g/lane; pMCϪ65/ϩ137, 75-85 g/lane (the relative copy number of pMCϪ90/ϩ137: pMCϪ65/ϩ137 was 2.5:1). C, hybridization with a 227-bp promoter probe extending from TXAS promoter nucleotides between Ϫ90 and ϩ137. D, hybridization with a 250-bp luciferase probe extending from nucleotide ϩ586 to ϩ835 (relative to the TXAS transcriptional start site) of the luciferase gene. E, hybridization with a 240-bp p220.2 probe extending from nucleotide ϩ2184 to ϩ2423 (relative to the TXAS transcriptional start site). a 500-fold molar excess of competitor over the template pC 2 AT-TX36(ϩ) (data not shown). We suspect that the oligonucleotide competitor is too short for the basic transcription machinery to bind and thereby fails to compete with the template plasmid.
We then tested whether the TATA and Inr elements were also functional in vivo. As described below, the TXAS core promoter alone yielded very low transcriptional activity in vivo. We thus included the upstream promoter region and linked the nucleotides between Ϫ248 and ϩ3 into the pGL3-Basic luciferase reporter vector as the wild-type construct. Mutations were then made in this construct at either TATA and/or Inr sites. These reporter vectors were transiently co-transfected into MEG-01 cells with pCMV-␤-Gal serving as an internal control. In addition, we used pGL3-Control as a positive control and the promoterless pGL3-Basic as a negative control. Promoter activities were normalized to those with the pGL3-Control and corrected for variations in transfection efficiency using the ratio of luciferase to ␤-galactosidase activity. In contrast to the in vitro assays, mutation at the TATA box alone had no effect on promoter activity, and mutation at Inr showed only a 30% decrease in activity (Fig. 3B). However, a significant decrease in promoter activity was observed in the double mutant construct. These data suggest that both the TXAS TATA box and the Inr are essential for transcription in vivo and, as discussed below, that they may act as compensatory elements.
In Vivo Transcription of the TXAS Gene Promoter by Transient Transfection-To further dissect the cis-elements involved in TXAS expression, we made 5Ј-and 3Ј-deletion constructs of the TXAS promoter and linked them to the pGL3-Basic vector. The parent construct in this study (pGL3Ϫ280/ ϩ137) contained nucleotides between Ϫ280 and ϩ137, which includes two putative GATA sites near nucleotide Ϫ250 (Fig.  1). For the 5Ј-deletion mutation analysis, DNA fragments containing nucleotides Ϫ190/ϩ137, Ϫ140/ϩ137, Ϫ90/ϩ137, Ϫ56/ ϩ137, and Ϫ38/ϩ137 were generated by PCR and linked to pGL3-Basic. These reporter constructs were transiently cotransfected into MEG-01 cells along with the pCMV-␤-Gal plasmid. As shown in Fig. 4A, the promoter activity of pGL3Ϫ280/ ϩ137 was potent and was about 125% of that of pGL3-Control. Deletion of nucleotides between Ϫ280 and Ϫ90 did not much affect the promoter activity, since the pGL3Ϫ190/ϩ137, Ϫ140/ ϩ137, and Ϫ90/ϩ137 constructs showed similar levels of activity. However, the removal of nucleotides between Ϫ90 and Ϫ56 in pGL3Ϫ56/ϩ137 reduced the promoter activity by 90% as compared with that of pGL3Ϫ280/ϩ137, indicating the presence of important elements in this segment.
Many promoters contain functionally important sequences in the 5Ј-untranslated region, sequences that are referred to as downstream promoter elements (46,47). Any such downstream promoter elements of TXAS cannot be studied by our in vitro transcriptional assay using the G-free cassette. To further examine the role of the 5Ј-untranslated region, a series of 3Јdeletion mutants of TXAS promoter construct were made in the ϩ1/ϩ137 segment. Luciferase reporter vectors containing DNA fragments corresponding to TXAS sequences Ϫ280/ϩ137, Ϫ280/ϩ90, Ϫ280/ϩ75, Ϫ280/ϩ52, and Ϫ280/ϩ3 (referred to as pGL3Ϫ280/ϩ137, pGL3Ϫ280/ϩ90, and so forth) were generated. A mutation at the putative Sp1 site near ϩ80, the only recognizable site in the 5Ј-untranslated region, was also made (GGGCGGG to GAAAGGG) in the pGL3Ϫ280/ϩ137 construct. These vectors were transiently co-transfected into MEG-01 cells with pCMV-␤-Gal. Promoter activities were found to grad-FIG. 6. Effect of NF-E2 on the TXAS promoter in the stably transfected cells. A, promoter activities of TXAS gene from HEL cells carrying wild type and NF-E2 mutated minichromosomes were stably transfected in HEL cells. Promoter activities were determined by normalization of luciferase activity, protein concentration, and copy number of minichromosome. Data represent means Ϯ S.D. of the relative luciferase activities, as compared with that obtained from the wild type. B, stimulation of TXAS promoter activity in stably transfected HeLa cells by forced expression of p45 NF-E2. HeLa cells carrying wild type and NF-E2 mutated minichromosomes were transiently transfected with (ϩ) or empty (Ϫ) p45 NF-E2 expression vector (pCMV-NF-E2). Data represent means Ϯ S.D. of the relative luciferase activities, as compared with that obtained from the wild type transfected with p45 NF-E2 vector.
ually decrease as the 5Ј-untranslated region became shorter (data not shown). Although the difference in promoter activities between the strongest (pGL3Ϫ280/ϩ137) and the weakest (pGL3Ϫ280/ϩ52) was about 2-fold, no remarkable reduction was observed between any two successive constructs. Furthermore, mutation at the putative Sp1 site caused only about 20% reduction in promoter activity. These results demonstrate that the effect of TXAS 5Ј-untranslated sequence on the promoter strength, albeit weak, is additive and is proportional to the length of this region. It has been speculated that, in many eukaryotic mRNAs, lengthening of the 5Ј-untranslated region can lead to a proportional increase in translational efficiency by increasing the load of the 40 S ribosomal subunits on the longer 5Ј-untranslated region (48). It should be noted that many downstream promoter elements are located within 50 nucleotides downstream of the transcriptional start site (47), and removal of these sequences drastically reduces the promoter activities in vivo and in vitro. Such a phenomenon is not observed in the TXAS 5Ј-untranslated region, since the pGL3Ϫ280/ϩ52 and pGL3Ϫ280/ϩ3 exhibited no significant variation in promoter activities.
In Vivo Transcriptional Studies of the TXAS Promoter Using Chromosomal Reporter Constructs-The transient transfection experiments indicated that the TXAS sequence between nucleotides Ϫ90 and Ϫ56 was important for transcriptional activity (Fig. 4A), but in vitro assay implied that this region was not critical for promoter activity (Fig. 2B). We suspected that the discrepancy was a result of differences in the template state; one is a naked DNA and the other an uncharacterized, nonreplicating nucleosomal DNA. To investigate the role of natural chromatin structure in TXAS transcription, we inserted various sizes of TXAS promoter/luciferase reporter constructs into a minichromosome containing the Epstein-Barr virus origin for self-replication. Initial attempts to develop MEG-01 cell lines harboring TXAS promoter/reporter constructs were unsuccessful. We subsequently transfected HEL cells, which constitutively express TXAS, with a series of 5Ј-deletion mutations corresponding to nucleotides Ϫ280/ϩ137, Ϫ190/ϩ137, Ϫ90/ϩ137, and Ϫ56/ϩ137 of the TXAS gene (designated as pMCϪ280/ϩ137, Ϫ190/ϩ137, Ϫ90/ϩ137, and Ϫ56/ϩ137, respectively). These episomal minichromosomes were stably maintained in HEL cells after selection by hygromycin B for 3-4 weeks. Quantitation of the expression levels of the minichromosomal reporter gene controlled by these TXAS promoter fragments revealed that cis-elements between Ϫ280/ Ϫ190 and Ϫ90/Ϫ56 are important for transcription (Fig. 4B). These results are consistent with those obtained from transient transfection experiments except that the effects exerted by these elements are more pronounced in the stably transfected cells, more than a 150-fold difference between Ϫ90/ϩ137 and Ϫ56/ϩ137 in the stably transfected cells (Fig. 4B) as compared with an 8-fold difference between the corresponding constructs in the transient transfection (Fig. 4A). The results indicate that cis-element(s) between Ϫ90 and Ϫ56 is important in regulating TXAS promoter activity in the native chromosomal structure context.
MNase-coupled Southern Blot Analyses of TXAS Promoter/ Reporter Minichromosomes-To investigate how the transcriptional activities of pMC-TXAS promoter constructs were affected by changes in nucleosomal structures, we chose the two clones carrying pMCϪ90/ϩ137 or pMCϪ56/ϩ137 for MNasecoupled Southern blot analysis, because they exhibited the largest difference in transcriptional activities (Fig. 4B). Partial digestion of chromatin by MNase, which cleaves linker DNA between nucleosome cores, produces an array of DNA fragments corresponding to various multiples of the length of nucleosomal DNA monomer. Ethidium bromide-stained total nuclear DNA digested by different concentrations of MNase showed the expected laddering pattern for bulk DNA (Fig. 5B). The positions of probes used for structural analysis in the TXAS and downstream regions of the minichromosomes are indicated in Fig. 5A. When Southern blotting was performed using TXAS promoter nucleotides Ϫ90/ϩ137 as a probe, a ladder of fragments with a repeat of about 150 bp indicated that this part of the minichromosome of pMCϪ56/ϩ137 was assembled into nucleosomes (Fig. 5C). On the other hand, a less distinct ladder and poorer hybridization were observed for the same region of the pMCϪ90/ϩ137 minichromosome, particularly at the highest MNase concentration. The results indicated that the promoter region of pMCϪ90/ϩ137 was less compact and protected than that of pMCϪ56/ϩ137. We next examined the nucleosomal structure of luciferase gene by hybridizing the same membrane with a probe containing a segment of the luciferase gene. As shown in Fig. 5D, the luciferase segment in pMCϪ56/ϩ137 adopts a nucleosomal structure, and its MNase digestion pattern is very similar to that observed for the TXAS promoter region (cf. Fig. 5, C and D). However, the luciferase FIG. 7. In vitro and in vivo binding of p45 NF-E2 to TXAS promoter. A, EMSA reactions were carried out by incubation of HEL nuclear extracts and end-labeled TXAS promoter nucleotides Ϫ121/ϩ5. Antibodies against Nrf1, Nrf2, and p45 NF-E2 were incubated with nuclear extracts before the addition of the probe. The reaction mixtures were then separated by electrophoresis. The positions of NF-E2⅐DNA complex and the supershifted band are indicated. B, autoradiogram of PCR products. ChIP and duplex PCR assays were carried out as described under "Experimental Procedures." One primer pair amplifies the TXAS 3Ј-untranslated region (3Ј-UTR), and the other pair amplifies the promoter region (Promoter). PCR templates from input DNA and ChIP DNAs treated with anti-p45 NF-E2, irrelevant serum, and no chromatin are indicated. gene region in pMCϪ90/ϩ137 reveals less nucleosomal structure (Fig. 5D), similar to that seen in the promoter region (Fig.  5C), again suggesting a disrupted nucleosomal structure. Finally, the DNA region downstream from the luciferase gene was examined by rehybridizing the same membrane with a probe of p220.2 DNA fragment. Both pMCϪ90/ϩ137 and pMCϪ56/ϩ137 showed a similar nucleosomal ladder in this downstream region (Fig. 5E). It should be emphasized that mononucleosomes can be clearly seen in both pMCϪ90/ϩ137 and pMCϪ56/ϩ137, suggesting that the region downstream from the reporter gene is well protected in both minichromosomes. Taken together, these results indicate that the disruption of nucleosomes occurs at the promoter and luciferase gene region in the minichromosome and TXAS is also actively transcribed when nucleotides Ϫ90/Ϫ56 are present. In contrast, the minichromosome maintains its intact nucleosomal structure and loses its promoter activity when the Ϫ90/Ϫ56 is deleted. These results suggest that the cis-element(s) located in the Ϫ90/Ϫ56 region of the TXAS promoter is involved, at least in part, in the disruption of nucleosomes.
Role of NF-E2 Site in TXAS Gene Transcription in a Native Chromatin Context-In transient transfection experiments, the Ϫ300/Ϫ40 TXAS upstream promoter sequence, particularly at the NF-E2 site between Ϫ83 and Ϫ77, is crucial for transcriptional activity in several types of cells (15,17,18,20). To test whether NF-E2 activates the TXAS gene in a natural chromatin context, we generated a mutation at the NF-E2 site in the pMCϪ280/ϩ137 minichromosome (pMCϪ280/ϩ137-mNF-E2) and stably transfected it into HEL cells. Cells harboring the wild-type TXAS promoter exhibited 14-fold higher luciferase activity than those harboring mutant NF-E2 pro-moter (Fig. 6A), indicating that an intact NF-E2 site dramatically increases TXAS transcriptional activity under the native chromosomal conditions. Next, we asked whether p45 NF-E2 could activate TXAS gene expression in HeLa cells, which do not express TXAS. We first carried out transient transfection experiments using the same series of 5Ј-deletion reporter plasmids described in the legend to Fig. 4A. As expected, the promoter activities of all constructs transfected into HeLa cells were very low, reaching at most only ϳ2% of the pGL3-Control (data not shown). No significant change was observed when the Ϫ90/Ϫ56 region was deleted. We also compared the promoter activities of stably transfected pMCϪ280/ϩ137 and pMCϪ280/ϩ137-mNF-E2 in HeLa cells. Luciferase activities were again very low from both wild type and mutant constructs. However, transient transfection with a p45 NF-E2 expression vector strongly stimulated luciferase activity in the cells containing pMCϪ280/ϩ137 (150fold increase) as compared with those containing pMCϪ280/ ϩ137-mNF-E2 (20-fold increase) (Fig. 6B). These results clearly demonstrated an important role for the NF-E2 site in controlling the TXAS gene expression under native chromosomal conditions.
In Vitro and in Vivo Binding of p45 NF-E2 to the TXAS Promoter-To further demonstrate that p45 NF-E2 is involved in the binding of TXAS promoter, we carried out EMSA and ChIP assay. EMSA results indicated that the nuclear extracts from HEL cells contained factor(s) capable of binding DNA fragments corresponding to the TXAS promoter sequences Ϫ121/ϩ5 (Fig. 7A). In a separate experiment, we performed reverse transcription-PCR experiments and found out that HEL cells expressed Nrf1, Nrf2, and Nrf3 as well as p45 NF-E2 (data not shown). Supershift analyses were therefore conducted with antibodies against p45 NF-E2, Nrf1, and Nrf2. As shown in Fig. 7A, a supershift of protein-DNA complex was observed with the addition of antibodies against p45 NF-E2 but not with Nrf1 or Nrf2 antibodies. We also carried out competition experiments using a 20-bp oligonucleotide encompassing the NF-E2 site. The shift band was completely diminished by the addition of a 10-fold molar excess of wild type oligonucleotide but was not affected by even a 100-fold molar excess of oligonucleotide with mutated NF-E2 (data not shown).
ChIP experiments were performed using the stably transfected HEL cells harboring ϳ10 copies of pMCϪ280/ϩ137. The enriched chromatin was cross-linked and sonicated to generate fragments averaging less than 500 bp. Antibodies specific for p45 NF-E2 were used to immunoprecipitate protein-DNA complexes, and a duplex PCR was carried out with one primer pair specific for the TXAS promoter sequence encompassing the NF-E2 site and a second pair specific for the TXAS 3Ј-untranslated region to generate the 227-and 340-bp PCR products, respectively. The 3Ј-untranslated region is 193 kb away from the TXAS promoter in the human chromosome 7 and does not contain the NF-E2 consensus sequence (15), thus serving as a negative control for NF-E2 binding. Duplex PCR assay was carried out under conditions of linear amplification for both products and results of 28 cycles of PCR were shown in Fig. 7B. Immunoprecipitation with anti-p45 NF-E2 antibody resulted in the recovery of considerably more promoter fragment than the irrelevant serum control or the control with no chromatin added to the protein A-Sepharose beads. In contrast, the 3Јuntranslated region, which lacks the NF-E2 binding site, was not detectable with anti-p45 NF-E2 antibody or control samples in the same experimental conditions. In comparison with input DNA sample, p45 NF-E2 binding to TXAS promoter was significantly enriched relative to an irrelevant locus, i.e. 3Јuntranslated region. Taken together, these results demon- FIG. 9. MNase hypersensitivity analysis of TXAS promoter by indirect end labeling. Nuclei from HEL cells carrying stably transfected pMCϪ280/ϩ137 and pMCϪ280/ϩ137-mNF-E2 were digested by 0, 10, or 30 units of MNase as indicated by the wedges. After purification of the DNA and digestion with XbaI and BclI, the products (7-15 g of DNA) were separated on a 1.2% agarose gel and analyzed by Southern blotting using a HindIII/BclI probe. HS, X, and the horizontal arrow indicate hypersensitive site to MNase, the NF-E2 site, and the transcription start site, respectively. M, 1-kb size marker. strate that p45 NF-E2 binds TXAS promoter in vitro and in vivo.
Indirect End Label and MNase-coupled LM-PCR Analyses of TXAS Promoter/Reporter Minichromosomes-To examine the nucleosomal organization of the TXAS promoter encompassing the NF-E2 region in more detail, we carried out MNase-coupled LM-PCR using the minichromosomal constructs in stably transfected HEL cells. For the pMCϪ56/ϩ137 minichromosome, MNase digestion revealed two prominent sites at nucleotide positions 130 and 280, indicating a 150-bp nucleosome core particle (Fig. 8A). A weaker band was also seen at nucleotide position 105, suggesting a 25-bp linker. MNase digestion of the pMCϪ90/ϩ137 minichromosome, however, revealed several additional cleavage sites not observed with pMCϪ56/ ϩ137, namely at nucleotide positions 185, 180, and 147 (Fig.  8A). These results are consistent with the MNase-coupled Southern blot analysis in which intact nucleosomes were readily seen in the promoter of pMCϪ56/ϩ137 but not in pMCϪ90/ϩ137, indicating a different nucleosomal structure when nucleotides Ϫ90/Ϫ56 were present.
To ascertain whether binding at the NF-E2 site was associated with the nucleosomal remodeling, we examined the nuclei harboring the pMCϪ280/ϩ137 and pMCϪ280/ϩ137-mNF-E2 minichromosomes by indirect end label analysis. The minichromosomes were partially digested with MNase, cleaved with XbaI and BclI to obtain the 1.03-kb parent fragment, and then subjected to Southern blot analysis using the 0.58-kb 5Ј-end luciferase gene as a probe (Fig. 9). MNase-sensitive sites were observed in the wild type minichromosome. The positions of MNase cleavage sites of the wild type minichromosome were mapped approximately between the size of 900 and 750 bp, corresponding to TXAS nucleotides Ϫ180 and Ϫ30. Mutation of NF-E2 site, however, resulted in loss of the MNase-hypersensitive sites (Fig. 9). We thus conclude that an intact NF-E2 site is associated with nucleosomal disruption.
To further map the MNase-hypersensitive sites in this region, MNase-coupled LM-PCR analysis using the nested primer set 3Ј-P1 and 3Ј-P2 (Fig. 1) was carried out. LM-PCR defined a nucleosome core particle for both wild type and mutant minichromosomes between nucleotide positions 130 to 280 and a linker at nucleotide positions 130 -105 (Fig. 8B), which correspond to the TXAS promoter nucleotides Ϫ40 to Ϫ190 and Ϫ40 to Ϫ15, respectively. This position of the nucleosome particle identified by LM-PCR is consistent with that by indirect end label analysis (Fig. 9). In this nucleosome particle, MNasehypersensitive sites at nucleotide positions 230, 188, 184, and 172 (corresponding to TXAS promoter nucleotides Ϫ140, Ϫ98, Ϫ94, and Ϫ82, respectively) were found in pMCϪ280/ϩ137 minichromosome but not in pMCϪ280/ϩ137-mNF-E2 minichromosome or naked pMCϪ280/ϩ137 (Fig. 8B). Furthermore, many MNase cleavage sites were observed near the transcriptional start site (nucleotide position 90 in Fig. 8B) of pMCϪ280/ ϩ137 and were absent in pMCϪ280/ϩ137-mNF-E2. We subsequently examined the nucleosomal structure from the 5Ј-end of TXAS promoter using the nested primer set 5Ј-P1 and 5Ј-P2 (Fig. 1). As shown in Fig. 8C, MNase cleavage sites at nucleotide position 96 and the 68 -74 region (corresponding to the TXAS promoter nucleotide Ϫ81 and the Ϫ103 to Ϫ108 region, respectively) were more pronounced in the wild type minichromosome than in the mutant. In contrast, two cleavage sites at nucleotide positions 152 and 135 were seen at similar levels of intensity for both wild type and mutant minichromosomes. These two sites corresponding to TXAS promoter nucleotides Ϫ25 and Ϫ42 define a linker region, in agreement with LM-PCR using the nested primer set 3Ј-P1 and 3Ј-P2. Moreover, these MNase cleavage sites were not found using the naked wild type minichromosome DNA (Fig. 8, B and C).
The results of LM-PCR thus place the NF-E2 site at one nucleosome particle (N1 in Fig. 8, A-C), TATA at a linker (L2), and Inr at another nucleosome particle (N2). When the NF-E2 site is occupied, both nucleosomes N1 and N2 are disrupted (Fig. 8, B and C). It should be noted that in Fig. 5C, significant nucleosomal changes at the promoter region were observed at 150 units of MNase but were less evident at 50 units of MNase. Because the MNase concentrations used in the LM-PCR and indirect end labeling assays were at the range of 10 -60 units, gross nucleosomal changes were not seen, since the linker regions were still prominent (Fig. 8, A and B). However, the MNase-hypersensitive sites were clearly detected at this concentration of MNase. Taken together, these results demonstrated that NF-E2 binding was associated with the disruption of the nucleosomal structure of TXAS promoter, particularly in the NF-E2 region. DISCUSSION The strength of promoter and accuracy of initiation have been shown, in general, to largely depend on the sequence and relative distance of TATA and Inr elements in many genes driven by RNA polymerase II (Ref. 49 and references therein). Although the consensus Inr sequence is loosely defined, a pyrimidine at Ϫ1, an A at ϩ1, and a T or A at ϩ3 are the most critical nucleotides for determining the strength of an Inr (50). These critical nucleotides are conserved in the TXAS Inr, CA ϩ1 TT. In addition, the distance of 25 bp between TATA and Inr, also the case in the TXAS core promoter, was found to give the highest level of in vitro transcription from the artificial templates (51). These factors may thus contribute to a strong TXAS core promoter activity in vitro. Lee et al. identified the transcription start site and TATA box of the TXAS gene at the positions corresponding to our nucleotides Ϫ98 and Ϫ127/ Ϫ124, respectively (52), whereas Miyata et al. (53) identified the transcription start site corresponding to our nucleotide Ϫ2. However, no functional analysis was reported with respect to the Inr and TATA elements. We showed here that mutation of either element considerably reduced the promoter strength in the cell-free transcription but did not affect the accuracy of initiation, suggesting that the TATA and Inr elements played similar roles in determining the transcription initiation site and directing the transcription machinery to the TXAS promoter. In contrast to in vitro studies, mutation of either the TATA or Inr element in the presence of upstream sequence had little effect on promoter activity, whereas mutation of both elements abolished TXAS promoter activity. These findings suggested that the TATA and Inr elements of the TXAS promoter were functionally redundant in vivo. This is consistent with the notion that a preinitiation complex can be formed through either a TATA or an Inr element, followed by a common step leading to the gene activation (54).
Our results support the model that the upstream activator (i.e. NF-E2) alters the nucleosomal structure and opens up the core promoter. Because NF-E2 lacks the chromatin-modifying activity, NF-E2 activation upon TXAS promoter is probably aided by recruiting other proteins. In this aspect, evidence for NF-E2 involved in chromatin disruption has been extensively studied in the globin genes (30,31). The mechanism by which NF-E2 alters nucleosomal structures was elucidated from several studies. In the glutathione S-transferase pull-down experiments, p45 NF-E2 was shown to interact directly with CBP/ p300, a protein that possesses histone acetyltransferase activity (55). Transient transfection experiments showed that NF-E2 transactivation activity was enhanced by expression of CBP/p300 and was inhibited by E1A, an inhibitor of CBP/p300 (56). Interestingly, both subunits of NF-E2 were acetylated by CBP, and the DNA binding and transcriptional activities were hence increased (57). It is noteworthy that, despite the evidence of direct interaction between NF-E2 and histone-modifying enzymes in the globin gene transactivation, no evidence of this direct interaction is presented for the TXAS promoter. Furthermore, although our hypothesis favors the model in which NF-E2 induces the nucleosomal changes of the TXAS promoter, it is also plausible that NF-E2 increases transcription of the TXAS gene by recruiting the transcriptional machinery before the nucleosomal structures are altered.
The importance of NF-E2 in TXAS expression can probably explain why TXAS is expressed at high levels in blood cells but much lower levels in the nonblood cells (58). It should be noted that considerable functional redundancy has been found among NF-E2 sites. The NF-E2 site is recognized not only by p45 NF-E2 but also by other family members, such as Nrf-1 (59), Nrf-2 (60), Nrf-3 (61), Bach1, and Bach2 (62). Small Maf proteins include MafK, MafG, and MafF (23). Both MafK and MafG could activate globin gene expression (63,64). A combination of these factors with others, such as zinc finger transcription factors, may thus account for the diversity of TXAS expression in different cells (65). For example, Rat macrophage TXAS is suppressed by peroxisome proliferator-activated receptor ␥ ligands through an interaction of Nrf2 and peroxisome proliferator-activated receptor binding at the NF-E2 site (20). p18 NF-E2 can form a heterodimer with Fos to act as a repressor for NF-E2 activity (66). Furthermore, the gene regulation can also be controlled by the localization of NF-E2 transcription factors. A recent work showed that p18 NF-E2 was located in the centromeric heterochromatin compartment and bound globin gene as a homodimer in the repressed stage. Upon induction, p18 NF-E2 was relocated to the euchromatin compartment and formed a heterodimer with p45 NF-E2 (67). We are currently studying the TXAS transcriptional activity in the nonblood cells. Preliminary results indicated that the NF-E2 site is also important for TXAS promoter activity in these cells. 2 It is therefore plausible that through the NF-E2 site TXAS is regulated not only for certain activators or inhibitors but also for the cell preference expression.