Ligand-dependent transcriptional enhancement by DNA curvature between two half motifs of the estrogen response element in the human estrogen receptor α gene

Gene 294 (2002) 279–290 www.elsevier.com/locate/gene

Ligand-dependent transcriptional enhancement by DNA curvature between two half motifs of the estrogen response element in the human estrogen receptor a gene Xiao-Man Li a, Yoshiaki Onishi a, Kentaro Kuwabara b, Jeung-yon Rho b, Yuko Wada-Kiyama b, Yasuo Sakuma b, Ryoiti Kiyama a,* a

Research Center for Glycoscience, National Institute of Advanced Industrial Science and Technology, AIST Central 6, 1-1-1 Higashi, Tsukuba, Ibaraki 305-8566, Japan b Department of Physiology, Nippon Medical School, Sendagi, Bunkyo-ku, Tokyo 113-8602, Japan Received 23 March 2002; received in revised form 3 June 2002; accepted 19 June 2002 Received by J.L. Slightom

Abstract We previously reported five DNA bend sites (ERB-4 to -1, and ERB 1 1) in the promoter region of the human estrogen receptor a (ERa) gene [FEBS Lett. 444 (1999) 117]. One of these sites, ERB-2, was accompanied by two half motifs of the estrogen response element (ERE) and several short poly(dA) .poly(dT) tracts including an A4 tract located next to a half ERE motif. This A4 tract and the 20 bp immediate flanking sequence containing a half ERE motif (T3B) exhibited DNA curvature. Transcription assays using luciferase as a reporter gene indicated that T3B sequence conferred positive estrogen responsiveness. Mutations introduced in this sequence indicated that both bendability and estrogen responsiveness were synergistically associated with the A4 tract located next to the half ERE motif. This motif and a mutant sequence, GGTTA, had affinity for ERa protein, which seems to account for ERa protein binding to the region without an ERE motif. These findings suggest that some DNA curvature acts as a transcriptional modulator by modifying the state of ligand effects. Published by Elsevier Science B.V. Keywords: Transcription; DNA curvature; Promoter; Estrogen receptor

1. Introduction The estrogen receptor (ER) is a nuclear receptor for the steroid hormone estrogen, which regulates development and maintenance of female phenotype and behavior (Norman and Litwack, 1997), and is a member of the steroid-thyroid hormone receptor superfamily (Evans, 1988; Beato, 1989). The receptor is expressed in many tissues, not only in the reproductive tract but also in the neuroendocrine system and visceral organs. There are at least two types of ER, ERa and ERb, but their functional roles are not completely understood. ERa is located on chromosome 6 and has 8 exons encompassing more than 140 kb (Ponglikitmongkol et al., 1988). There are at least six transcriptional initiation sites in ERa, and the study of their usage in various tissues and tumor cell lines is of great importance to understand the Abbreviations: ER, estrogen receptor; ERE, estrogen response element; TK, thymidine kinase * Corresponding author. Tel.: 181-298-616189; fax: 181-298-616190. E-mail address: [email protected] (R. Kiyama). 0378-1119/02/$ - see front matter. Published by Elsevier Science B.V. PII: S 0378-111 9(02)00803-X

functions in non-reproductive tracts (Flouriot et al., 1998). Among the alternative promoters, the best characterized P0, or promoter C, is located 2 kb upstream of the canonical promoter, promoter A (Keaveney et al., 1992; Grandien et al., 1995). The tissue-specific transcription from these promoters is regulated by transcription factors in a tissuespecific manner but also requires ligand-dependent regulation to show a response to estrogen in these tissues. Expression of the ERa gene is also regulated by estrogen and several DNA sequence elements including the estrogen response element (ERE) as well as DNase I-hypersensitive sites (Grandien et al., 1993). Although the consensus ERE sequence is GGTCANNNTGACC (Evans, 1988; Beato, 1989), which is composed of two half GGTCA motifs with a 3 bp stuffer forming a pair of inverted repeats, a half motif and its variants are also active when they are placed in specific environments. Specific transcriptional modulators found in the promoter regions of eukaryotic genes can be categorized into two types according to the mechanisms involved: one that has

280

X.-M. Li et al. / Gene 294 (2002) 279–290

more emphasis on recognition of sequences by protein factors, and thereby has rather strict sequence specificity, and another that has more emphasis on the structural arrangement for the transcription machinery. While the ERa promoter region contains general transcription factor binding sites such as Sp1 as well as ER-specific motifs giving basic views as to the transcription complexes, more extensive research is required to characterize the structural part arranged by the transcription factors, the transcription machinery and the chromatin structure to fully understand the mechanism. A number of functional arrangements of ERE motifs have been reported such as inverted repeats (Evans, 1988; Beato, 1989; Schwabe et al., 1993), direct repeats (Treilleux et al., 1997; Klinge et al., 1997), repeats with mutations (Hyder et al., 1992; Teng et al., 1992; Murdoch et al., 1995; Nardulli et al., 1996; Driscoll et al., 1998) or those with Sp1 sites (Batistuzzo de Medeiros et al., 1997; Wang et al., 1998). However, the functional mechanisms involved in these arrangements are mostly unknown. One of the structural components that influences transcriptional modulation is bent DNA. The structure has a direct influence on transcription by modulation of the local protein structures and the locations of active sites. Bent DNA also has an indirect influence by arranging the chromatin structure (Kiyama and Trifonov, 2002). We previously reported a class of bent DNA, which appears mostly periodically in the several loci of higher eukaryotes and which functions as signals for chromatin folding or nucleosome phasing. Detailed analysis of DNase I-hypersensitive site 2 (HS2) in the human b-globin locus control region (Kiyama, 1998; Wada-Kiyama et al., 1999b) revealed that these sites determine the nucleosomal phases by having higher affinity to core histones, and thus should play a key role in initiation of nucleosome formation (Onishi et al., 1998). Therefore, it is quite reasonable that these sites are present universally and nearly periodically in eukaryotic genomic DNA. Previously, we mapped five DNA bend sites within 3 kb of the promoter region of ERa, one of which, ERB-2, located -850 bp from promoter A was close to the predicted site of higher nucleosome formation (Wada-Kiyama et al., 1999a). The bend center of the site was actually shown to be located close to the dyad axis of the nucleosome and located between the two half ERE motifs.

2. Materials and methods 2.1. Cell culture and plasmid construction MCF-7 breast carcinoma cells were obtained from American Type Culture Collection (Rockville, MD) and maintained in minimal essential medium (Gibco-BRL) supplemented with heat-inactivated 10% (v/v) fetal bovine serum, 1 mM sodium pyruvate and 2.2 g/l of NaHCO3. Reporter plasmids expressing luciferase cDNA were

constructed as follows. The promoter region (PvuII–HincII fragment) of the HSV-TK gene was cloned upstream of the luciferase cDNA in pGEM3 (Promega). This reporter plasmid, pTK-Luc, and the pGEM3 plasmid containing the bgalactosidase gene were kindly supplied by Dr. T. Tsuchiya (Tokyo Medical and Dental University, Japan) (Oda et al., 1996). The double-stranded DNA containing T3A, T3B, T3A 1 T3B or XVI was inserted into the BamHI site immediately upstream of the TK promoter of pTK-Luc. For the assays using the canonical and alternative promoters (PA, PB and PC, see Fig. 1) in the ERa gene, the regions between 2100 bp and 1100 bp relative to each cap site (position 21955 for the cap site C and position 2319 for the cap site B relative to the canonical cap site A at position 11) (Flouriot et al., 1998) were inserted at the promoter cloning site in pGL3-Basic Vector (Promega). The plasmid PA 1 E contains the region between 2100 bp and 1210 bp relative to the cap site A, which includes a binding site for ERF-1. 2.2. Bending assay with oligonucleotides The 20-mer oligonucleotide, TGGTATGAAAAGGTCACATT (for T3B), and oligonucleotides containing mutations in T3B were annealed with the respective 20-mer complementary oligonucleotides with two-base 5 0 -protruding ends at the final concentration of 0.2 mg/ml. The size standard A20 1 T20 was prepared by annealing (dA)20 with (dT)20. The annealed duplex DNA were treated with T4 polynucleotide kinase in the presence of 1 mM ATP at 37 8C for 30 min, followed by treatment with T4 DNA ligase at 4 8C overnight. The ligation products were resolved by electrophoresis through an 8% polyacrylamide (29 : 1 ¼ mono : bis) gel in 45 mM Tris-borate, 1 mM EDTA buffer at 4 8C for 23 h. 2.3. Transcription assay The transient expression assay with the luciferase reporter gene was performed essentially according to Oda et al. (1996). Cells were collected by trypsinization, resuspended in minimal essential medium and then seeded in 60-mm petri dishes. After incubation at 37 8C for 24 h, cells were washed twice with OPTI-MEM I (Gibco-BRL) without fetal bovine serum. The reporter plasmid DNA (2.0 mg/1 £ 105 cells), mixed with plasmid DNA containing the b-galactosidase gene (2.0 mg/1 £ 105 cells) was transfected into MCF-7 cells with Lipofectin (Gibco-BRL) according to the supplier’s instructions. After incubation for 6 h in 1 ml of OPTIMEM I, cells were washed and resuspended in minimal essential medium with 10% (v/v) fetal bovine serum and incubated for 24 h. 17b-Estradiol (Sigma) dissolved in DMSO was then added to the dishes at a final concentration of 1 £ 1029 or 1 £ 1028 M, and the cells were incubated for a further 24 h. The same volume of DMSO was added to the control dishes. Cells were then washed twice with phosphatebuffered saline and suspended in 500 ml of lysis buffer (Toyo Ink). The luciferase activity of the cell lysates (100 ml) was

X.-M. Li et al. / Gene 294 (2002) 279–290

281

Fig. 1. The 5 0 -region of the human ERa gene. The locations of ERB-2 site, and the canonical exon 1 (with cap site A) and two alternative exons 1 (with cap sites B and C) are shown as open boxes. The regions used as promoters for PA, PB, PC and PA 1 E are indicated as horizontal lines. The locations of DNA bend center of ERB-2 site and half ERE motifs (solid boxes) and the regions of 20 nt sequences used here (T3A, T3B, T3C, T3AB and T3(A 1 B)) are shown below.

assayed with PicaGene (Toyo Ink) using a luminometer (Lumat LB 9501, Berthold) (de Wet et al., 1987). The bgalactosidase activity was assayed with Galacto-Light Plus (Tropix Inc.). Turner’s light units were measured for 10 s for both activities and the luciferase activity was normalized relative to the b-galactosidase activity. 2.4. Gel-shift assay The gel-shift assay was performed according to the previously described procedure using a 32P-labeled DNA fragment (L2R4; nucleotides 2975 to 2779) as a probe (Wada-Kiyama et al., 1999a). Ten-microliter aliquots of the DNA probe were suspended in 20 ml of 16 mM HEPES (pH 7.5), 150 mM KCl, 16% (v/v) glycerol, 1.6 mM MgCl2, 0.8 mM dithiothreitol, 0.4 mM PMSF, 1 mM EDTA, 0.8 mg/ ml BSA, 0.06 mg/ml poly(dI-dC) and 0.01% (v/v) NP-40 in the presence or absence of competitor oligonucleotides, and after incubation with human ERa protein (Santa Cruz Biotechnology) on ice for 5 min, mixed with 5 ml of 30% (v/v) glycerol dye. The samples were electrophoresed through a 4% polyacrylamide (40 : 1 ¼ mono : bis) gel in 40 mM Tris–acetate, 1 mM EDTA and 5% (v/v) glycerol at

30 mA for 2 h at room temperature. The gel was fixed with 10% (w/v) trichloroacetic acid for 30 min, dried and autoradiographed. Competitor duplex oligonucleotides containing T3A, T3AB, T3B, T3, T3C and T3B mutants were prepared by ligation of the annealed complementary oligonucleotides as described above. 2.5. Methylation interference The radioactively labeled DNA fragment (L2R4) was first methylated with dimethylsulfate as in the Maxam–Gilbert method. After gel-shift assay, both the free and ERa-bound DNAs were recovered form the gel, cleaved with piperidine, and then applied to an 8% polyacrylamide–7 M urea gel under denaturing conditions. 2.6. Southwestern blotting Aliquots of approximately 60 mg of nuclear extracts or 1 mg of ERa protein were resolved by sodium dodecyl sulfate–polyacrylamide gel electrophoresis followed by electroblotting onto Hybond P membranes (Amersham Pharmacia Biotech). Membranes were incubated in 10 mM Tris–HCl (pH 8.0), 0.15 M NaCl, 0.05% Tween, and

282

X.-M. Li et al. / Gene 294 (2002) 279–290

then twice in HEPES binding buffer (20 mM HEPES (pH 7.9), 50 mM NaCl, 3.9 mM MgCl2, 1 mM EDTA, 0.1 mM ZnSO4 and 1 mM DTT) in the presence of 6 M guanidine HCl for 20 min each at 4 8C. Membranes were incubated in HEPES binding buffer and 3 M guanidine HCl for 10 min at 4 8C, which was repeated six times, each time using HEPES binding buffer containing a twofold dilution of guanidine HCl from the previous wash. Non-specific binding was blocked by incubating the membranes in 5% (w/v) skim milk in HEPES binding buffer for 1 h in an orbital shaker at room temperature. The membranes were washed once with HEPES binding buffer and then incubated with 0.25% skim milk, HEPES binding buffer, 2 mg/ml poly(dI-dC), 1 mg/ml herring sperm DNA, 3 mg/ml competitor and the 32P-labeled L2R4 probe for 1 h at room temperature with gentle shaking. The membrane was washed with a 1/20 dilution of HEPES binding buffer. Membranes were dried and analysed by BAS-5000 (FUJIX). Anti-ERa antibody (Geneka Biotechnology) was an affinity-purified rabbit polyclonal antibody raised against a peptide within the hinge region of the human ERa protein. The nuclear extracts were prepared from MCF-7 cells according to Dignam et al. (Dignam et al., 1983) as described previously (Kiyama and Camerini-Otero, 1991).

3. Results 3.1. Transcriptional activation by ERE motifs at ERB-2 We previously mapped a DNA bend site, ERB-2, between the cap sites B and C (Fig. 1). As the site contained a pair of ERE motifs which are separated by a 23 bp sequence containing the bend center but not by a 3 bp sequence as seen in many cases (see Section 1), we examined the effect of these ERE motifs on transcription and, if so, its mechanism. We first examined the effect of ERB-2 site on transcription using the transient expression assay with the luciferase gene (pTK-Luc) as a reporter (Fig. 2). It has been shown that an ERE motif introduced into the thymidine kinase (TK) promoter responded to natural estrogen (17b-estradiol) at physiological concentrations (10 29–10 27 M) (Oda et al., 1996). This was reproduced in our assay system (Fig. 2A). Transcriptional enhancement was observed with the ERE motif derived from the Xenopus vitellogenin gene in a dose-dependent manner (see the results with 10 29 and 10 28 M estrogen). We later used a concentration of 10 29 M estrogen for our assay. When the 20 bp sequence T3A or T3B, or the 40 bp T3(A 1 B) were assayed in the absence of estrogen, only T3B showed repression of transcription (Fig. 2B), while T3A and T3(A 1 B) showed equivalent degrees of transcription as in the assay with control pTK-Luc. When this assay was performed in the presence of estrogen, ligand-dependent transcriptional enhancement was observed with T3B and T3(A 1 B). In both cases, 3.4-fold (T3B) or 2.3-fold

(T3(A 1 B)) enhancement was observed, suggesting that although T3A contributed to enhancement of the basal transcription level as seen in T3(A 1 B), estrogen responsiveness was likely to be attributable to the T3B sequence. To understand this estrogen responsiveness, several base changes were introduced into either the half ERE motif (T3B1, T3B2, T3B3 and T3B5) or the A4 tract located next to the motif (T3B4). When T3B and these mutants were assayed in both the presence and absence of estrogen, T3B5 responded most to estrogen (Fig. 2C). Note that the degree of overall estrogen responsiveness (the ratio of the transcription activity of each construct in the presence to that in the absence of estrogen) of T3B5 was 7.2-fold. Transcription rates were significantly decreased when the mutations were introduced into the ERE motif at the second and fifth positions (T3B2 and T3B3). Mutation at the fourth position (T3B1) retained both transcription efficiency and estrogen responsiveness. Mutation introduced outside of the motif did not change the efficiency (T3B4). We further analysed the effects of base-pair changes using the constructs derived from T3B3, which showed significant decreases both in basal transcription levels and estrogen responsiveness (Fig. 2D). Mutations were introduced at the fifth position, by changing A to C (T3B31) or A to T (T3B32), or at the position next to the fifth by one base-pair insertion (T3B33) or deletion (T3B34). The results indicated that transcription activities were decreased more by changing the fifth A to pyrimidines (T3B31 and T3B32) than to G (T3B3). In contrast, changing the rotational positions of the motif around the double helix did not change both basal transcription levels and estrogen responsiveness (T3B33 and T3B34). The mutations used in Fig. 2C,D are summarized in Fig. 2E. No such responsiveness was observed for any of the constructs in Fig. 2B and XVI in HeLa cells (Fig. 2F). Ligand-dependent transcriptional enhancement by T3B was investigated with the canonical and alternative promoters in the ERa gene (Fig. 3). There were three transcription initiation sites (cap sites A to C, see Fig. 1) including the canonical site (cap site A) in the 2 kb region of the ERa gene promoter (Flouriot et al., 1998). The regions containing 2100 bp to 1100 bp to each cap site (PA to PC) were fused to a promoter-less vector pGL3 and assayed for luciferase activity in the presence or absence of estrogen (Fig. 3A–C). The estrogen responsiveness was also examined with the constructs containing T3B located in front of the promoters (T3B/PA, T3B/PB, and T3B/PC). The results indicated that although the promoter activities of these regions were generally low, the response to estrogen was observed only when T3B was present. In contrast, when the transcription factor ERF-1 binding site located at the 5 0 noncoding region (DeConinck et al., 1995) was included in the construct (PA 1 E), there was approximately 25-fold activation of transcription. While a moderate estrogen-responsiveness was observed with PA 1 E alone, there was no additional influence of T3B (T3B/PA 1 E). These results indicate that T3B adds estrogen responsiveness irrespective

X.-M. Li et al. / Gene 294 (2002) 279–290

283

Fig. 2. Transcription assay with constructs derived from pTK-Luc. Constructs containing 20 bp sequences (T3A, T3B, T3B derivatives and XVI) or 40 bp sequences (T3(A 1 B)), or the control pTK-Luc (TK) were used for luciferase assay. XVI (TCAGGTCACAGTGACCTGAT) contained the ERE motifs (underlined) from the Xenopus vitellogenin A2 gene (Walker et al., 1984). Plasmid DNA was transfected into MCF-7 breast carcinoma cells (panels A–D) or into HeLa cells (panel F), and the indicated concentration of estrogen (17b-estradiol) was added to the culture after incubation for 24 h. No DNA was transfected as a control (CB). pRSV-b-Gal DNA was co-transfected with the pTK-Luc or its derivatives as an internal control. Cells were harvested after 24 h and subjected to luciferase assay. The averages and standard deviations were calculated for each plasmid sample from two or three dishes and are shown in the graphs. Samples from each dish were assayed twice for luciferase and b-galactosidase activities, and the luciferase activity was normalized relative to the bgalactosidase activity. (A) The normalized luciferase activities of TK or XVI in the absence (2) or presence of 10 29 M (29) or 10 28 M (28) estrogen. Relative ratios of the activity to that of TK(29) are shown. (B) The luciferase activities of the constructs containing T3A, T3B or T3A 1 T3B or the control TK in the absence or the presence of 10 29 M estrogen. Data were normalized relative to the activity of TK. (C) The luciferase activities of the constructs containing T3B and its derivatives. Data were normalized to the activity of the control TK as in panel (B). (D) The luciferase activities of the constructs containing T3B, T3B3 and its derivatives. (E) The nucleotide sequences of T3B and its derivatives. The A4 tract is underlined and the ERE motif is boxed. The locations of mutations are highlighted. (F) The normalized luciferase activities of the constructs shown in panel (B) and XVI assayed in HeLa cells.

284

X.-M. Li et al. / Gene 294 (2002) 279–290

of the ERa promoters. Furthermore, although ERF-1 confers high transcription rates, the degree of estrogen responsiveness was equivalent to that by T3B. 3.2. DNA curvature at ERB-2 site We then investigated the relationship between the DNA bend center at T3B and the nearby ERE motifs with mutations introduced into the T3B sequence (T3B1 to T3B5) (Fig. 4A,B). When the bendability of oligonucleotides containing these mutations was assayed, the mutations introduced into the first base (G to A; T3B5) and the fourth base (C to T; T3B1) of the motif caused notable changes in migration: introduction of A at the first base added an additional A base to the A4 tract, thereby increasing bendability, while a C to T change at the fourth base caused a decrease in bending. Note that the latter change created a GGTTA motif, which is also located in the next sequence T3A and forms an imperfect palindrome with another ERE motif in

T3A. Mutation at the second base (G to C; T3B2) caused a slight increase in bending. Other mutations did not markedly affect bendability. The DNA curvature profile of this region was further examined using the computer software, TRIF1.00 (Fig. 4C) (Shpigelman et al., 1993; Wada-Kiyama et al., 1999a). The projection of the DNA structure that gave the highest degree of distortion of the axis (on the right) indicated that the direct distance between the two half ERE motifs was reduced by this curvature. The C to T change in the fourth base (T3B1) changed the direction of the helical axis of DNA while maintaining the wedge angle, resulting in a reduction of overall curvature (data not shown). 3.3. Interaction of ERa protein with the ERE motifs at ERB2 The bend center at ERB-2 has two half palindromic motifs of the ERE (see Fig. 1). We next examined binding

Fig. 3. Transcription assay with constructs containing the canonical or alternative promoters of the ERa gene. Constructs containing 200 bp sequences from the canonical (promoter A: PA) or alternative (promoters B and C: PB and PC, see Fig. 1) promoters or a construct containing promoter A and an enhancer (PA 1 E) were cloned into the promoter-less pGL3-Basic Vector (pGL3). Plasmid DNA was transfected into MCF-7 cells and estrogen (10 29 M 17b-estradiol) was added to the culture and transcription assays were performed as described in Fig. 2. The averages and standard deviations were calculated for each construct from two or three dishes and are shown in the graphs. Samples from each dish were assayed twice for luciferase and b-galactosidase activities, and the luciferase activity was normalized relative to the b-galactosidase activity. Ratios of the normalized luciferase activities of PC (A), PB (B), PA (C) and PA 1 E (D) constructs with or without T3B relative to those of pGL3 in the absence of estrogen are shown.

X.-M. Li et al. / Gene 294 (2002) 279–290

of ERa protein to the region including these motifs by gelshift assay to understand how the interaction of ERa protein to the ERE motifs at ERB-2 is influenced by the mutations introduced in or around the motifs (Fig. 5). ERa protein showed two bands (indicated by arrows) in the assay with a 197 bp fragment containing ERB-2 site, both of which were competed by oligonucleotides containing T3A and T3B (Fig. 5A, lanes 6 and 8) but not by those containing A20 1 T20, T3 or T3C (Fig. 5A, lanes 3, 4, 9, 10–12). To examine whether binding of ERa protein was due to the

285

recognition of EREs in this region, we performed the same assay with T3AB (see Fig. 1), which did not contain ERE motifs, and sequences containing ERE motifs from the chicken ovalbumin gene (OVA) and the Xenopus vitellogenin gene (XVI) (Geiser et al., 1983; Walker et al., 1984; Evans, 1988) (Fig. 5B, lanes 7 and 8, 11 and 12, 13 and 14, respectively). T3AB and XVI showed competition in binding, suggesting that they share a common feature for ERa protein recognition, while OVA did not. Note that T3AB contained a modified ERE motif (GGTTA) and no other

Fig. 4. Oligonucleotide bending assay of T3B mutants from ERB-2 site. (A) Bending assay with concatenated oligonucleotides containing the mutant sequences from T3B. Mutations were introduced into the ERE motif of the T3B sequence (T3B1, T3B2, T3B3 and T3B5) or into the A4 tract next to the motif (T3B4). T3B4 and T3B5 changed the length of the A4 tract. (B) Summary of the bending assays in panel (A). Relative degrees of migration (RL) were calculated as the ratios of the migration distance of each band to that of A20 1 T20 and plotted against the lengths of the oligonucleotides. (C) Profiles of DNA curvature predicted by computer analysis. DNA curvature of the T3B sequence was predicted using TRIF1.00 software, and the phosphate backbones and the axes of the double helix are projected from two different directions 908 to each other to show the direction of curvature. The structure on the right shows a profile with the most distorted axis.

286

X.-M. Li et al. / Gene 294 (2002) 279–290

Fig. 5. Binding of ERa to the ERE motifs of ERB-2. (A) Gel-shift assay with ERa protein using the 197 bp 32P-labeled L2R4 (nucleotides 2975 to 2779) as a probe. The assay was performed in the presence (lanes 3–12) or absence (lanes 1 and 2) of the indicated amounts of the concatenated competitor oligonucleotides containing A20 1 T20 (AT), T3A, T3B, T3 or T3C sequences. The positions of the generated bands and the probe L2R4 are indicated by arrows. (B,C) Gelshift assay with ERa protein and L2R4 in the presence (lanes 3–14) or absence (lanes 1 and 2) of the indicated amounts of competitors shown above (see Figs. 1 and 2 for competitor sequences). OVA (TCAGGTAACAATGTGTTTTC) contained the ERE motifs (underlined) from the chicken ovalbumin gene (Evans, 1988). (D) Gel-shift assay with the ERa protein and L2R4 in the presence (lane 3) or absence (lane 2) of an antibody against ERa protein.

common features were present among T3A, T3AB and T3B sequences. When either this modified motif or the perfect ERE motif was absent (T3 for example), binding of nuclear factors did not occur. On the other hand, although OVA was reported as an ERE-containing sequence, it does not share common motifs GGTCA or GGTTA with others. Therefore, it seems that ERa protein recognized a little broader sequence variation of ERE motifs. We further investigated the specificity of binding of ERa protein to the ERE motif located in T3B (Fig. 5C). Oligonucleotides with a mutation at position 5 (T3B3; Fig. 5C, lanes 9 and 10) showed an equivalent level of competition as T3B (Fig. 5C, lanes 3 and 4), while T3B1 (Fig. 5C, lanes 5 and 6), T3B4 (Fig. 5C, lanes 11 and 12) and T3B5 (Fig. 5C, lanes 13 and 14) showed moderate to weak competition. T3B2 containing a mutation at the second position (Fig. 5C, lanes 7 and 8) showed no competition. Note that the GGTTA motif (T3B1) was recognized by ERa protein. These bands were supershifted with anti-ERa antibody (Fig. 5D) with

the same probe containing the T3 region, confirming that ERa protein interacts with this region. The binding region of ERa in ERB-2 was detected by methylation interference assay using the ERa bound or unbound L2R4 probe (Fig. 6). The two half ERE motifs in this region were both interfered with binding of ERa when the G nucleotides were methylated. Furthermore, the region in between the motifs (shown by hatched lines) also showed the effect of methylation, indicating some interaction of ERa in this region.

3.4. Proteins interacting with the ERE motifs at ERB-2 The nuclear extracts from MCF-7 were examined by Southwestern blotting to identify the protein species that bound to the ERE motifs in ERB-2 (Fig. 7). A number of proteins were detected in the assay with the probe L2R4 (Fig. 7, lanes 4, 6, 8), although only a few were competed

X.-M. Li et al. / Gene 294 (2002) 279–290

287

important not only for nucleosome alignment but also for arranging transcription-factor binding sites (Onishi et al., 1998). The effects of nucleosome alignment reach the enhancer of b-LCR located in the second next nucleosome from the bend sites, suggesting that a long-range interaction is mediated by this type of bent DNA (Onishi and Kiyama, 2001). As we discussed previously, placing transcription factor binding sites at the dyad axis of nucleosomes would be crucial for chromatin remodeling to initiate the transcription process (Wada-Kiyama et al., 1999a). The far upstream region (position 2724 and further upstream), which includes ERB-2 site, was previously shown to contribute to tissue-specific transcriptional enhancement of the ER gene (DeConinck et al., 1995). This ERB-2 site, as also observed in other sites, was shown to contain nucleosome dyad axis both by computer-based and experimental analyses (Wada-Kiyama et al., 1999a). Fine mapping of the bend site indicated that the bend center was located in T3B, and the A4 tract in T3B was likely to be the bend center. We here examined the effects of bendability with mutations introduced into this A4 tract or into the neighboring half-ERE motif (Fig. 4). The results indicated that the length of the A4 tract is important for bendability, although the fourth and fifth bases in the

Fig. 6. Methylation interference assay for detecting the interacting site of ERa in ERB-2. The labeled L2R4 fragment was incubated with ERa protein after methylation using dimethylsulfate. The ERa-bound (1) and unbound (2) fragments were recovered from a 4% polyacrylamide gel and cleaved with piperidine. Both the sense and antisense sequences of the region between the positions 2890 and 2858 are shown. Horizontal arrows show the positions of the half ERE motifs. The G residues in the half ERE motifs are shown by closed circles and the regions of methylation interference with binding of ERa are shown by hatched lines.

by T3B (Fig. 7, lane 8). ERa protein was among those proteins which were competed by T3B (Fig. 7, lanes 3, 5, 7).

4. Discussion 4.1. DNA curvature and estrogen-dependent transcriptional enhancement We previously mapped DNA bend sites in the promoter region of the human ERa gene as a first step to understand the estrogen-dependent transcriptional regulation of the gene (Wada-Kiyama et al., 1999a). DNA bend sites have been shown to appear nearly periodically at intervals of roughly 680 bp in many loci and to regulate chromatin structure by providing anchorage sites for nucleosome formation (Wada-Kiyama et al., 1999b; Wada-Kiyama and Kiyama, 1994, 1995, 1996). These anchorage sites are

Fig. 7. Detection of nuclear proteins bound to the ERE motifs of ERB-2. (A) Results of Southwestern blotting using ERa protein (lanes 3, 5, 7) and the nuclear extracts from MCF-7 cells (lanes 4, 6, 8) in the presence or absence of competitors (AT or T3B; see Fig. 5). Coomassie-staining patterns of ERa protein (arrowed) and MCF-7 extracts are shown in lanes 1 and 2.

288

X.-M. Li et al. / Gene 294 (2002) 279–290

half ERE motif (T3B1 and T3B3) also moderately affected the bendability. Although T3B alone did not show enhancement of the basal transcription level, it added estrogen-dependent transcriptional enhancement (positive estrogen responsiveness) to the TK promoter and the canonical and alternative promoters of the ERa gene (Figs. 2 and 3). Interestingly, the estrogen responsiveness conferred by T3B was equivalent to that by the region containing ERF-1 binding site located in the 5 0 non-coding region of the gene. ERF-1 is a member of the AP2 family (McPherson et al., 1997) and might work with other transcription factors by heterodimer formation to confer activation of the basal level. Estrogen responsiveness, on the other hand, could be attributable to some ERE motif in the nearby region. There are several GGTCA and GGTTA motifs in the region, which were excluded in the PA construct. 4.2. Scheme of estrogen responsiveness There are two half-ERE motifs in ERB-2 site which are separated by 23 bp, facing each other as a pair of inverted repeats. The gel-shift assay indicated that the oligonucleotides containing these ERE motifs and a modified motif (GGTTA) located between them were most likely the sites of interaction between ERB-2 site and ERa protein (Fig. 5). The interaction of ERa protein with these motifs was confirmed by methylation interference assay (Fig. 6). Bending of DNA between the two motifs might facilitate interaction between the ERE-binding proteins in the nuclear extracts including ERa protein, which were revealed by Southwestern blotting (Fig. 7). To understand the relationships between bendability, estrogen-dependent transcriptional enhancement and binding of ERa protein, we used five mutations introduced into

either the half ERE motif in T3B (T3B1, T3B2, T3B3 and T3B5) or into the region containing the A4 tract next to the motif (T3B4). The results are summarized in Table 1. When the ERE motif was modified to GGTTA (T3B1), the sequence retained the estrogen responsiveness although the overall but not the local bendability was markedly decreased. This explains the affinity of ERa protein for another sequence T3AB (Fig. 5B), which completely lacked the ERE motif but contained a GGTTA motif. The mutation introduced at the second position (T3B2) decreased both transcriptional level and ERa protein binding, confirming that binding of ERa protein to this motif is required for the basal transcription. Note that higher bendability in T3B2 gave higher estrogen responsiveness although the basal transcription level was low. Interestingly, T3B3 mutation decreased the basal transcriptional activity although it retained the affinity of ERa protein for this motif. This was not instantly explained by the affinity of ERa protein alone and probably involved a fine structure of DNA-protein interaction. Mutation introduced at the fifth position to pyrimidines (T3B31 and T3B32, Fig. 2D) caused more negative effects on the basal transcription while overall rotational shifting did not (T3B33 and T3B34). Therefore, base transition at this position did not reduce the basal transcription significantly compared with base transversion. On the other hand, the dyad axis of the double helix is distorted by the base change (overall 4.68 wedge angle of the base pairs for CAC trinucleotides in T3B compared with 12.78 for CGC at the identical positions in T3B3) (Bolshoy et al., 1991), probably disturbing RNA polymerase recruitment or transcriptional complex formation in both transcription and estrogen responsiveness. This is also true for other mutations where mutations having more distortion (3.28 for GGT in T3B compared with 13.48 for GCT in T3B2) affected more in transcription than those having less (10.68 for TCA in T3B

Table 1 Summary of bendability, affinity and transcriptional activity of EREs a Sequence

Bendability

T3A T3AB T3(A 1 B) T3B (A4GGTCA) T3B1 (A4GGTTA) T3B2 (A4GCTCA) T3B3 (A4GGTCG) T3B4 (A3TGGTCA) T3B5 (A4AGTCA) ERE (XVI)

2c 1c ND 11 1 111 11 11 111 1 ND

a

(Value for 10mer)

(1.44) e (1.35) (1.48) (1.41) (1.44) (1.63)

Affinity for ERa

Transcriptional activity (basal/TK ^ SD)

Estrogen responsiveness b ^ SD

11 d 11 ND 11 1 2 11 1 1 1

0.88 ^ 0.03 ND 1.36 ^ 0.15 0.49 ^ 0.14 0.42 ^ 0.05 0.25 ^ 0.04 0.21 ^ 0.04 0.39 ^ 0.17 0.41 ^ 0.03 0.99 ^ 0.32

1.19 ^ 0.15 ND 2.29 ^ 0.42 3.38 ^ 0.41 4.03 ^ 0.50 5.53 ^ 0.03 4.93 ^ 0.21 3.27 ^ 0.27 7.17 ^ 1.73 1.84 ^ 0.46

SD, standard deviation; ND, not done. Ratio of the transcription activities in the presence to those in the absence of estrogen. c See Wada-Kiyama et al. (1999a). ND, not done. d Relative degrees of affinity of ERa for each oligonucleotide as compared with the competition with T3B, estimated from gel-shift assay (Fig. 5): 1 1 , equivalent; 1, well below; 2, very little or none. e Values for 10mers obtained in Fig. 4B. b

X.-M. Li et al. / Gene 294 (2002) 279–290

compared with 8.18 for TTA in T3B1, and 10.58 for AGG in T3B compared with 15.68 for AAG in T3B5). Meanwhile, ERa protein retained some affinities for both T3B4 and T3B5, and the basal level of transcription was also retained in these mutants. Shortening the length of the A4 tract (T3B4) did not change bendability or estrogen-responsiveness whereas a longer tract increased both bendability and responsiveness. In conclusion, the half ERE motif (GGTCA) and GGTTA sequence are equally involved in binding of ERa protein and the basal transcriptional levels, while estrogen responsiveness is associated significantly with the DNA curvature at the A4 tract and the bendability of T3B. Details of these relationships remain to be future works but we currently hypothesize the mechanism as follows: ERa proteins mainly function as homodimers and the homodimer formation is influenced by estrogen binding. Therefore, estrogen responsiveness can be strongly associated with the state of interaction between the bound ERa proteins at the different ERE motifs and can be modulated by the strength and the way of the interaction. On the other hand, the basal transcription largely depends on the direct interaction of the bound ERa protein with the transcription machinery located at the promoter region and does not much depend on the interaction between the ERa proteins. It is unlikely that this scheme is true only for ERB-2 site, because all other bend sites are similarly accompanied by these motifs. Of the 11 motifs (four GGTCA and seven GGTTA) in the 3 kb region, nine were located within the bend sites and the other two were located within 100 bp from these sites. Therefore, although complete pairs of half EREs were not always found in close proximity, there are a number of combinations of GGTCA and GGTTA motifs in the DNA bend sites and some of them may modulate transcription in response to ligand binding. Acknowledgements We thank Dr. E. Trifonov for TRIF1.00, and Ms. M. Kameishi and Ms. K. Suzuki for technical assistance. This work has been supported by a Grant for Preventing Public Pollution from the Agency of Environment of Japan (to R.K.) and Grants-in-Aid for Priority Area from the Ministry of Education, Science, Sports and Culture of Japan and grants from Asahi Foundation (to Y.W.-K. and R.K.). Additional support was provided by the Foundation for the Promotion of Private Schools in Japan (to Y.S.). References Batistuzzo de Medeiros, S.R., Krey, G., Hihi, A.K., Wahli, W., 1997. Functional interactions between the estrogen receptor and the transcription activator Sp1 regulate the estrogen-dependent transcriptional activity of the vitellogenin A1 io promoter. J. Biol. Chem. 272, 18250– 18260. Beato, M., 1989. Gene regulation by steroid hormones. Cell 56, 335–344.

289

Bolshoy, A., McNamara, P., Harrington, R.E., Trifonov, E.N., 1991. Curved DNA without A-A: experimental estimation of all 16 DNA wedge angles. Proc. Natl. Acad. Sci. USA 88, 2312–2316. DeConinck, E.C., McPherson, L.A., Weigel, R.J., 1995. Transcriptional regulation of estrogen receptor in breast carcinomas. Mol. Cell. Biol. 15, 2191–2196. de Wet, J..R, Wood, K.V., DeLuca, M., Helinski, D.R., Subramani, S., 1987. Firefly luciferase gene: structure and expression in mammalian cells. Mol. Cell. Biol. 7, 725–737. Dignam, J.D., Lebovitz, R.M., Roeder, R.G., 1983. Accurate transcription initiation by RNA polymerase II in a soluble extract from isolated mammalian nuclei. Nucleic Acids Res. 11, 1475–1488. Driscoll, M.D., Sathya, G., Muyan, M., Klinge, C.M., Hilf, R., Bambara, R.A., 1998. Sequence requirements for estrogen receptor binding to estrogen response elements. J. Biol. Chem. 273, 29321–29330. Evans, R.M., 1988. The steroid and thyroid hormone receptor superfamily. Science 240, 889–895. Flouriot, G., Friffin, C., Kenealy, M., Sonntag-Buck, V., Gannon, F., 1998. Differentially expressed messenger RNA isoforms of the human estrogen receptor-alpha gene are generated by alternative splicing and promoter usage. Mol. Endocrinol. 12, 1939–1954. Geiser, M., Mattaj, I.W., Wilks, A.F., Seldran, M., Jost, J.-P., 1983. Structure and sequence of the promoter area and of a 5 0 upstream demethylation site of the estrogen-regulated chicken vitellogenin II gene. J. Biol. Chem. 258, 9024–9030. Grandien, K.F., Berkenstam, A., Nilsson, S., Gustafsson, J.A., 1993. Localization of DNase I hypersensitive sites in the human oestrogen receptor gene correlates with the transcriptional activity of two differentially used promoters. J. Mol. Endocrinol. 10, 269–277. Grandien, K., Backdahl, M., Ljunggren, O., Gustafsson, J.A., Berkenstam, A., 1995. Estrogen target tissue determines alternative promoter utilization of the human estrogen receptor gene in osteoblasts and tumor cell lines. Endocrinology 136, 2223–2229. Hyder, S.M., Stancel, G.M., Nawaz, Z., McDonnell, D.P., Loose-Mitchell, D.S., 1992. Identification of an estrogen response element in the 3 0 flanking region of the murine c-fos protooncogene. J. Biol. Chem. 267, 18047–18054. Keaveney, M., Klug, J., Gannon, F., 1992. Sequence analysis of the 5 0 flanking region of the human estrogen receptor gene. DNA Seq. 2, 347–358. Kiyama, R., 1998. Periodicity of DNA bend sites in eukaryotic genomes. Gene Ther. Mol. Biol. 1, 641–647. Kiyama, R., Camerini-Otero, R.D., 1991. A triplex DNA-binding protein from human cells: purification and characterization. Proc. Natl. Acad. Sci. USA 88, 10450–10454. Kiyama, R., Trifonov, E.N., 2002. What positions nucleosomes? – A model. FEBS Lett. 523, 7–11. Klinge, C.M., Bodenner, D.L., Desai, D., Niles, R.M., Traish, A.M., 1997. Binding of type II nuclear receptors and estrogen receptor to full and half-site estrogen response elements in vitro. Nucleic Acids Res. 25, 1903–1912. McPherson, L.A., Baichwal, V.R., Weigel, R.J., 1997. Identification of ERF-1 as a member of the AP2 transcription factor family. Proc. Natl. Acad. Sci. USA 94, 4342–4347. Murdoch, F.E., Byrne, L.M., Ariazi, E.A., Furlow, J.D., Meier, D.A., Gorski, J., 1995. Estrogen receptor binding to DNA: affinity for nonpalindromic elements from the rat prolactin gene. Biochemistry 34, 9144– 9150. Nardulli, A.M., Romine, L.E., Carpo, C., Greene, G.L., Rainish, B., 1996. Estrogen receptor affinity and location of consensus and imperfect estrogen response elements influence transcription activation of simplified promoters. Mol. Endocrinol. 10, 694–704. Norman, A.W., Litwack, G., 1997. Hormones, 2nd Edition. Academic Press, London, pp. 361–386. Oda, T., Tsuchiya, T., Sato, Y., Yasukochi, Y., 1996. Estrogen receptormediated transcriptional activation by 17 beta-estradiol and its analogs. Biol. Pharm. Bull. 19, 1018–1022.

290

X.-M. Li et al. / Gene 294 (2002) 279–290

Onishi, Y., Kiyama, R., 2001. Enhancer activity of HS2 of the human bLCR is modulated by distance from the key nucleosome. Nucleic Acids Res. 29, 3448–3457. Onishi, Y., Wada-Kiyama, Y., Kiyama, R., 1998. Expression-dependent perturbation of nucleosomal phases at HS2 of the human beta-LCR: possible correlation with periodic bent DNA. J. Mol. Biol. 284, 989– 1004. Ponglikitmongkol, M., Green, S., Chambon, P., 1988. Genomic organization of the human oestrogen receptor gene. EMBO J. 7, 3385–3388. Schwabe, J.W., Chapman, L., Finch, J.T., Rhodes, D., 1993. The crystal structure of the estrogen receptor DNA-binding domain bound to DNA: how receptors discriminate between their response elements. Cell 75, 567–578. Shpigelman, E.S., Trifonov, E.N., Bolshoy, A., 1993. CURVATURE: software for the analysis of curved DNA. Comp. Appl. Biosci. 9, 435–440. Teng, C.T., Liu, Y., Yang, N., Walmer, D., Panella, T., 1992. Differential molecular mechanism of the estrogen action that regulates lactoferrin gene in human and mouse. Mol. Endocrinol. 6, 1969–1981. Treilleux, I., Peloux, N., Brown, M., Sergeant, A., 1997. Human estrogen receptor (ER) gene promoter-P1: estradiol-independent activity and estradiol inducibility in ER 1 and ER- cells. Mol. Endocrinol. 11, 1319–1331.

Wada-Kiyama, Y., Kiyama, R., 1994. Periodicity of DNA bend sites in human epsilon-globin gene region. Possibility of sequence-directed nucleosome phasing. J. Biol. Chem. 269, 22238–22244. Wada-Kiyama, Y., Kiyama, R., 1995. Conservation and periodicity of DNA bend sites in the human beta-globin gene locus. J. Biol. Chem. 270, 12439–12445. Wada-Kiyama, Y., Kiyama, R., 1996. An intrachromosomal repeating unit based on DNA bending. Mol. Cell. Biol. 16, 5664–5673. Wada-Kiyama, Y., Kuwabara, K., Sakuma, Y., Onishi, Y., Trifonov, E.N., Kiyama, R., 1999a. Localization of curved DNA and its association with nucleosome phasing in the promoter region of the human estrogen receptor alpha gene. FEBS Lett. 444, 117–124. Wada-Kiyama, Y., Suzuki, K., Kiyama, R., 1999b. DNA bend sites in the human beta-globin locus: evidence for a basic and universal structural component of genomic DNA. Mol. Biol. Evol. 16, 922–930. Walker, P., Germond, J.-E., Brown-Luedi, M., Givel, F., Wahli, W., 1984. Sequence homologies in the region preceding the transcription initiation site of the liver estrogen-responsive vitellogenin and apo-VLDLII genes. Nucleic Acids Res. 12, 8611–8626. Wang, F., Hoivik, D., Pollenz, R., Safe, S., 1998. Functional and physical interactions between the estrogen receptor Sp1 and nuclear aryl hydrocarbon receptor complexes. Nucleic Acids Res. 26, 3044–3052.