Estrogen response elements can mediate agonist activity of anti-estrogens in human endometrial Ishikawa cells.

Anti-estrogens like hydroxytamoxifen (OHT) have mixed agonist/antagonist activities, leading to tissue-specific stimulation of cellular proliferation. Partial agonist activity of OHT can be observed in vitro in endometrial carcinoma cells like Ishikawa. Here, we have compared several anti-estrogens (including extensively characterized OHT and pure anti-estrogens such as ICI164, 384 and RU58,668, which are devoid of uterotrophic activity) for their capacity to stimulate promoters containing estrogen response elements (EREs) or AP1-binding sites (12-O-tetradecanoylphorbol-13-acetate response elements, TREs), the two types of DNA motifs known to mediate transcriptional stimulation by estrogen receptors. Assays were performed in Ishikawa cells either by transient transfection or by using cell lines with stably propagated reporter vectors. In transient transfection experiments, none of the anti-estrogens displayed agonist activity on the promoters tested. In contrast, significant transcriptional stimulation was observed with low concentrations of OHT and RU39,411 in Ishikawa cells stably propagating reporter constructs containing a minimal ERE3-TATA promoter. In addition, micromolar concentrations of OHT, but not of RU39,411, stimulated stably propagated AP1-responsive reporter constructs. No transcriptional stimulation of ERE- or TRE-containing promoters was observed with the pure anti-estrogens ICI164,384 and RU58,668. These results indicate that the presence of estrogen response elements in promoters is sufficient to mediate cell-specific agonism of anti-estrogens at the transcriptional level, and that stimulation of AP1 activity may be restricted to a subset of anti-estrogens possessing agonist activity on EREs. In addition, our results suggest that transient transfections do not fully recapitulate in vivo conditions required to observe agonist activity of anti-estrogens.

The estrogen 17␤-estradiol (E2) 1 regulates gene transcrip-tion by binding to the estrogen receptor (ER), which interacts with specific target DNA sequences known as estrogen response elements (EREs). When bound to DNA the ER stimulates transcription via two transcription activation domains, AF-1 and AF-2 (1-3). AF-1 is located in the poorly conserved N-terminal A/B domain of the ER (2,4), whereas AF-2 is found in the C-terminus of region E, the hormone-binding region (5)(6)(7)(8). Binding of E2 to the ER is thought to induce a conformational change in the hormone binding domain, stimulating its transactivation properties.
Different types of synthetic compounds have been developed that are capable of antagonizing ER action in reproductive tissues and, in particular, of blocking estradiol stimulation of cellular growth in breast and uterine tissues. These anti-estrogens act by competing with E2 for binding to the ER and block ER-mediated activation of transcription when co-administered with hormone (9). However, tamoxifen, one of the most widely used anti-estrogens in breast cancer treatment (10), can induce uterine cell growth in vivo in animal models (11) and in humans (12,13). Hydroxytamoxifen also induces cellular proliferation (14,15) and transcription of endogenous estrogen target genes such as the progesterone receptor (PR) gene in human endometria and cultured human endometrial carcinoma cells (15,16). Other anti-estrogens, like ICI164,384 (17) or the more recently developed RU58,668 (18), were reported not to stimulate uterine cell growth and may therefore prove more appropriate for breast cancer therapy (17).
To better understand the mechanisms of tissue-specific estrogenic activity of anti-estrogens, we compared transcriptional activation of ERE-containing reporter constructs by full or partial anti-estrogens in the estrogen-dependent breast carcinoma cell line MCF7 and in the endometrial carcinoma cell line Ishikawa. We also assessed whether anti-estrogens may regulate transcription of target genes through AP1-binding sites (TPA response elements, TREs) rather than, or as well as, through EREs. Indeed, estrogenic stimulation of promoters containing TRE sites has been documented (19 -22), a possible mechanism being direct interaction between ER and AP1 components (22). Assays used for investigating the contribution of EREs or TREs in transcriptional stimulation by anti-estrogens with partial agonist activity included, in addition to transient transfection assays in MCF7 or Ishikawa cells, direct hormonal stimulation of reporter vectors stably propagated as episomes in these cell lines. The latter assay was selected because of increasing evidence for mechanistic links between transcriptional stimulation and reorganization of chromatin structure.
Comparison of the two types of assays and implications for the mechanism of cell-specific agonism by tamoxifen are discussed.
Plasmid Recombinants-Expression vectors pSG5, pSG5-HEG0, and reporter recombinants Vit-tk-CAT (23) and STR-CAT (equivalent to construct 84-CAT in Ref. 24) were kindly provided by Dr. P. Chambon (Illkirch, France). ERE3-tk-CAT was constructed by insertion of three copies of double-stranded oligonucleotides containing the 15-bp Xenopus vitellogenin A2 ERE sequence (25) flanked by HindIII and XbaI sites between the HindIII and XbaI sites of pBLCAT8ϩ. ERE3-TATA-CAT was constructed in several steps from GRE5-CAT (26). First, the BglII site upstream of the CAT gene in GRE5-CAT was deleted by filling-in with Klenow, creating GRE5-CAT[-BglII]. A fragment containing the ERE3-TATA promoter was then excised from the vector ERE3-pAL10 (27) by digestion with Asp-718, end-filling with Klenow fragment, and digestion with BamHI; this fragment was inserted into GRE5-CAT[-BglII] which had been digested with SacI, treated with Klenow fragment, and then digested with BamHI to remove the GRE5-TATA promoter. By taking advantage of the unique XhoI and BglII sites upstream and downstream from the three EREs, respectively, these motifs were removed and replaced by multimerized oligonucleotides containing a consensus TPA response element (TRE, Fig. 1A), creating TRE2-TATA-CAT and TRE6-TATA-CAT. ERE3-TATA-CAT/ EBV, TRE2-TATA-CAT/EBV, and TRE6-TATA-CAT/EBV were obtained by removal of XbaI fragments containing the whole minimal promoter-CAT gene transcriptional unit from the parental vectors and insertion into GRE5-CAT/EBV also digested by XbaI (28).
Cell Culture and Transfections-MCF7 and Ishikawa cells were grown in ␣-minimum Eagle's medium supplemented with 10 and 5% fetal bovine serum (FBS), respectively, and switched to phenol red-free (29) DMEM supplemented with 5% charcoal-treated FBS 72 h before plating for transient transfections. Cells were divided into 10-cm plates (1.5 million cells/plate) and transfected using the calcium-phosphate coprecipitation method (30) with 15 g of DNA (1 g of expression vector where applicable, 2 g of CAT reporter vector, 2 g of internal standard vector RSV-LacZ, and Bluescribe M13ϩ to 15 g). After 20 h, medium was changed twice to remove precipitate, and hormones were added for a further 24 h (as indicated in figure legends). Cells were harvested by scraping in 1 ml of phosphate-buffered saline 1ϫ, followed by centrifugation at 2,500 rpm for 10 min. Extracts were prepared by three cycles of freeze-thawing in 0.25 M Tris-HCl, pH 8.0, and standardized for ␤-galactosidase activity. CAT activity was determined by incubation of protein samples for 1 h with 0.25 Ci of [ 3 H]chloramphenicol and N-butyryl-CoA (0.2 mg/ml), followed by extraction with xylene, and liquid scintillation counting (31). To test for regulation of progesterone receptor expression, cells maintained in phenol red-free medium supplemented with charcoal-treated serum were further incubated for 5 days in the absence or presence of estradiol or of anti-estrogens. Medium was changed every 2nd day, and hormones were added every day. On day 5, cells were plated (1.8 million cells in 10-cm dishes) and transfected in duplicate with 2 g of GRE5-TATA-CAT plasmid (26), 1 g of internal control RSV-LacZ, and Bluescribe M13ϩ as carrier DNA (total 15 g). Progesterone was added 1 day later, after removal of calcium-phosphate precipitates by two consecutive washes. Cells were harvested 24 h later, and CAT activity was assayed as described above. All CAT assays were reproduced a minimum of three times.
COS-1 cells were grown in DMEM supplemented with 5% FBS. Cells were plated at a confluency of 1.5 ϫ 10 6 cells/10-cm plate and transfected with 15 g of pSG5 or pSG5-HEG0 expression vectors. Cells were harvested 36 h later, and extracts were prepared in gel retardation buffer 4ϫ (20 mM Tris-HCl, pH 7.5, 20% glycerol, 400 mM KCl, 0.1 mM EDTA, pH 8.0, 2 mM dithiothreitol, supplemented with protease inhibitor mixture) by three cycles of freeze-thawing on ice. For gel retardation assays, extracts were incubated with 2 g of poly(dI-dC) in gel retardation buffer 1ϫ for 20 min on ice, followed by further incubation in the presence of hormone (10 Ϫ8 M E2 or 10 Ϫ7 M anti-estrogens) and of labeled ERE (20,000 cpm) for 1 h on ice, and then for 30 min at room temperature. Complexes were then resolved by 5% polyacrylamide gel electrophoresis (120 V, 4°C for 4 h).
Generation and Hormonal Treatment of Stably-transfected Cell Lines Derived from Ishikawa and MCF7 Cells-Ishikawa or MCF7 cells were transfected with 15 g of ERE3-TATA-CAT/EBV, TRE2-TATA-CAT/ EBV, or TRE6-TATA-CAT/EBV (10-cm plates, 1.5 million cells). Forty eight hours after transfection, cells were passaged into 15-cm plates using medium containing 150 g/ml hygromycin B and maintained in this medium for about 2 weeks until disappearance of all cells in control non-transfected plates (28). Surviving cells in each 15-cm plate were then pooled, propagated, and tested for estrogen or TPA induction of CAT activity. Different pools of cells carrying the same reporter plasmid were found to behave similarly. For generation of stable cell lines containing non-episomal TRE-based reporter vectors, Ishikawa cells were cotransfected with 15 g of TRE6-TATA-CAT vectors and 1.5 g of neomycin resistance gene expression vector Rc/RSV (Invitrogen). 48 h after transfection, cells were trypsinized and replated into selection medium (␣-minimum Eagle's medium containing 5% FBS and 1 mg/ml G418). Two weeks later, individual clones were selected, expanded, and tested for stimulation of CAT activity by incubation with TPA (100 ng/ml) for 24 h. Established cell lines were subsequently maintained in medium containing half the concentration of antibiotic used for selection. For hormonal treatment, cells were preincubated for 72 h in medium without phenol red, supplemented with charcoal-treated serum. Incubation with estrogen or anti-estrogens was then carried out for 24 h except when indicated otherwise.
Reverse-transcription-PCR Amplification of Actin and CAT mRNAs-Ishikawa-ERE3/EBV cells maintained in phenol red-free DMEM, 5% charcoal-treated FBS were incubated with E2 (25 nM), OHT (100 nM), or ethanol for 8 h before harvesting and isolation of total RNA by CsCl gradient centrifugation. 3 g of total RNA were precipitated, resuspended in 6 l of 1ϫ DNase I digestion buffer (Promega) containing 0.5 units of DNase I, incubated for 15 min at 37°C and for 10 min at 75°C, and then transferred on ice. DNase I-treated RNAs (2 l) were reversed-transcribed using Superscript II RNase H reverse transcriptase (Life Technologies, Inc.) and 0.5 M random hexamer in a 20-l final volume of RT buffer (50 mM Tris-HCl, pH 8.3, 75 mM KCl, 3 mM MgCl 2 , 5 mM dithiothreitol supplemented with 0.5 mM dNTPs) at 37°C for 1 h, followed by 75°C for 10 min. Aliquots of resulting cDNAs (2 l) were amplified by PCR using Taq DNA polymerase (Amersham Pharmacia Biotech) in a 50-l final volume of 1ϫ Taq buffer supplemented with 0.2 mM dNTPs and 0.5 M forward and reverse CAT or ␤-actin primers as follows: CAT forward, 5Ј-CCGCCTGATGAATGCT-CATCCG-3Ј, and CAT reverse, 5Ј-GCATTCTGCCGACATGGAAGCC-3Ј; ␤-actin forward, 5Ј-GCTGTGCTATCCCTGTACGC-3Ј, and ␤-actin reverse, 5Ј-GCCATGGTGATGACCGGC-3Ј. 24 cycles of PCR (95°C for 30 s, 56°C for 1 min, and 72°C for 25 s) were performed for amplification of ␤-actin cDNAs and 28 cycles were performed (95°C for 30 s, 58°C for 1 min, and 72°C for 30 s) for amplification of CAT sequences, followed by a final elongation step (72°C for 10 min). The amplified products were then resolved on a 1.5% agarose gel.

Differential Effects of Anti-estrogens on the Electrophoretic
Migration of Estrogen Receptor-DNA Complexes-Partial antiestrogens such as hydroxytamoxifen are thought to induce a specific conformation of the estrogen receptor, which differs from both those of unliganded and estrogen-liganded ER. As a result, interaction between ER and estrogen or OHT differentially affects electrophoretic mobilities of ER⅐ERE complexes in gel shift assays. Full anti-estrogens like ICI164,384 induce yet another conformation, resulting in a migration that is closer to that of unliganded ER (32).
We used a gel shift assay to examine ER conformational changes induced by two other anti-estrogens, RU39,411 (RUp) and RU58,668 (RUf), which are both 11-␤ derivatives of estradiol and were previously reported to behave as partial and full anti-estrogens, respectively (18,33). Extracts from COS-1 cells transiently transfected with an expression vector for wild-type estrogen receptor (pSG5-HEG0) were incubated with estrogen, anti-estrogens, or vehicle together with a labeled consensus estrogen response element. ER-containing complexes were then separated from free probe by 5% polyacrylamide gel electrophoresis. A specific complex could be detected in the absence of ligand (Fig. 2, 2nd lane). Indeed, this complex was absent when using extracts from cells transfected with the parental expression vector pSG5 (Fig. 2, 1st lane) or when HEG0-containing extracts were incubated with a glucocorticoid-response element instead of an ERE (data not shown). Estrogen treatment of the cellular extracts increased the electrophoretic mobility of the ER⅐ERE complex (Fig. 2, E2, arrow 3). OHT treatment, on the other hand, resulted in a slower migrating complex (Fig. 2, OHT, arrow 1), and RUp treatment generated a complex migrating at a position similar to that of OHT (Fig.  2, RUp, arrow 1). Complexes between ER and the anti-estrogen RUf migrated at a position intermediary between those of ER⅐E2 and ER⅐OHT (Fig. 2, RUf, arrow 2), and indistinguishable from that of the ER⅐ICI164,384 complexes (Fig. 2, ICI, arrow 2). Similar results were obtained using the ER mutant HE0, except that complexes were only observed in the presence of estrogen or of the various anti-estrogens and not in the absence of ligand (data not shown), consistent with the destabilizing effect of the G400V mutation on unbound ER (34,35).
In conclusion, the electrophoretic mobilities of ER liganded with RUp or RUf are consistent with their identification as partial and full anti-estrogens, respectively.
Differential Induction of Progesterone Receptor Gene Expression by Anti-estrogens in Ishikawa Cells-Hydroxytamoxifen was previously reported to induce expression of the estrogen target gene human progesterone receptor in Ishikawa cells (15,16). We decided to investigate whether RUp, which similarly affects ER migration in gel shift assays, can also induce expression of PR in Ishikawa and MCF7 cells. Cells were pretreated with estrogen or anti-estrogens prior to transient transfection with GRE5-TATA-CAT/EBV (Fig. 1), followed by addition of progesterone for 24 h (see "Experimental Procedures"). No stimulation of CAT activity by progesterone could be observed in extracts of cells in the absence of treatment with estrogen or anti-estrogens (Fig. 3A, lane 2). In cells treated with estrogen or with the ER agonist moxestrol, the 11␤-methoxy derivative of ethynyl estradiol (36,37), CAT activity was induced over 20-fold in the presence of progesterone (Fig. 3A, lanes 4 and 12). In cells pretreated with OHT, progesterone-induced CAT expression levels were 12% of those obtained with estrogen pretreatment (Fig. 3A, lane 6). The anti-estrogen RUp induced PR expression to comparable levels (Fig. 3A, lane 10), whereas no stimulation could be detected following incubation with RUf (Fig. 3A, lane 8). Contrary to what was observed with Ishikawa cells, none of the anti-estrogens assayed detectably stimulated PR transcriptional activity in MCF7 cells using this assay (Fig.  3B, compare lanes 5-10 to lanes 1 and 2). These observations confirm that partial agonist activity of anti-estrogens on expression levels of endogenous estrogen target genes can be observed in Ishikawa cells (15,16).
Lack of Agonist Activity of Anti-estrogens in Transient Transfection Assays of ERE-containing Reporter Vectors-In order to analyze the mechanisms of transcriptional regulation by antiestrogens with partial agonist activity in Ishikawa cells, we examined whether synthetic estrogen-responsive promoters can be stimulated by these anti-estrogens in transient transfection assays (Fig. 4). Three estrogen-sensitive reporter recombinants, ERE3-TATA-CAT/EBV, ERE3-tk-CAT, or Vit-tk-CAT (Fig. 1), were transiently transfected into Ishikawa cells in the absence of cotransfected estrogen receptor expression vector (Fig. 4A). Anti-estrogens did not detectably stimulate CAT expression from ERE3-TATA-CAT/EBV (Fig. 4A, lanes  3-6), ERE3-tk-CAT (Fig. 4A, lanes 10 -13), or Vit-tk-CAT (Fig.  4A, lanes 17-20) under conditions where estradiol stimulated these reporter constructs 35-, 10-, and 25-fold, respectively (Fig. 4A, lanes 2, 9, and 16). Similar results were obtained when CAT assays were repeated with 10 times more extract than in Fig. 4A to confirm that none of the anti-estrogens tested stimulated CAT activity higher than vehicle (data not shown).
Cotransfection of an expression vector for wild-type ER (pSG5-HEG0) along with ERE3-TATA-CAT/EBV, ERE3-tk-CAT, or Vit-tk-CAT resulted in increased background levels in the absence of hormonal treatment (Fig. 4B, lanes 1, 8, and 15). Treatment with anti-estrogens did not increase CAT expression levels compared with vehicle alone (Fig. 4B, lanes 3-6,  10 -13, and 17-20), although CAT levels in the presence of OHT or RUp were slightly higher than with other anti-estrogens. Background CAT expression in the absence of hormonal treatment could result from binding of residual estrogens in media, which would be competed out by incubation with anti-estrogens. Alternatively, unliganded ER may be a weak transcriptional activator under the conditions of this assay.
Results similar to those observed in Ishikawa cells were obtained by transient transfection of MCF7 cells (data not shown). In both cell lines, the concentrations of anti-estrogens used (100 nM) were sufficient to totally repress stimulation by 1 nM estrogen, demonstrating that estrogen receptors are fully saturated by these anti-estrogens under the conditions used to assay for agonist activity (data not shown). In conclusion, agonist activity of anti-estrogens could not be detected in transient transfection assays of ERE-containing reporter vectors either in Ishikawa or in MCF7 cells.
Detection of Cell-specific Agonist Activity of Anti-estrogens Using a Stably Propagated ERE3-TATA-CAT Episomal Reporter Vector-The failure to detect significant transcriptional activation by OHT or other anti-estrogens in transient transfections raised the possibility that in vivo conditions required for agonist activity of OHT are not fully reconstituted in this assay. Therefore, we established Ishikawa and MCF7 cell lines stably propagating the ERE3-TATA-CAT/EBV episomal plasmids, selecting pools of transfected cells by addition of hygromycin B in cell culture medium (see "Experimental Procedures"; note that for brevity cell lines stably propagating ERE3-TATA-CAT/EBV episomal plasmids will be indicated by the suffix -ERE3/EBV). Stimulation with estrogen (25 nM, 24 h) of Ishikawa-ERE3/EBV cells led to a marked stimulation of CAT expression levels (ϳ17-fold, Fig. 5A, lane 2). No stimulation was observed with the pure anti-estrogens RUf or ICI (100 nM, Fig. 5A, lanes 5 and 6). In contrast, OHT and RUp stimulated CAT activity to ϳ22 and 30% of the levels obtained with estrogen, respectively (Fig. 5A, lane 3 and 4). These results were obtained with different pools of Ishikawa-ERE3/EBV cells generated by two independent rounds of selection. On the other hand, neither the partial anti-estrogens OHT and RUp nor the full anti-estrogens ICI or RUf stimulated CAT expression more than background in MCF7-ERE3/EBV cells; note, however, that the fold stimulation by estrogen was lower in stably transfected MCF7 cells than in Ishikawa-derived cell lines (Fig. 5B,  lanes 1-7).
Agonist activity of OHT was dose-dependent, and stimulation of CAT activity could be detected with concentrations as low as 1 nM, whereas ICI or RUf did not induce levels of CAT activity above basal levels at any of the concentrations tested (Fig. 5C). Note that none of the anti-estrogens were capable of inducing CAT expression in a cell line derived from Ishikawa cells by stable transfection of an episomal vector containing five glucocorticoid response elements (GRE5) instead of three EREs (GRE5-TATA-CAT/EBV, Ref. 28), demonstrating that the presence of EREs is required for transcriptional activation by antiestrogens (data not shown).
The levels of CAT mRNAs after stimulation by estrogen and OHT for 8 h were measured by semi-quantitative RT-PCR in order to confirm that the anti-estrogen increased CAT mRNA levels. CAT mRNA levels obtained after OHT stimulation were ϳ40% of those obtained after incubation with estrogen, whereas actin mRNA levels did not vary between the different samples (Fig. 5D). These results are in good agreement with those obtained by measuring levels of CAT enzyme activity. The agonist effect of OHT and RUp persisted when incubations were performed for longer periods than 24 h. Treatment for 2-5 days with OHT or RUp consistently generated increased CAT expression compared with treatment with full anti-estrogens or vehicle (Fig. 5E).
In order to rule out the possibility that differences in the protocols used for transient transfection assays and experiments with stable cell lines might be the source of the discrepancy in the results observed with partial anti-estrogens (i.e. no stimulation of CAT activity in transient transfection versus stimulation in stable cell lines), we repeated these experiments in parallel using Ishikawa and Ishikawa-ERE3/EBV cells, with or without mock transfection of Ishikawa-ERE3/EBV cells (i.e. transfection with carrier DNA only). Estrogen stimulation of CAT activity was much higher in the transient transfection assay than with the stable cell line (55-and 10-fold, respectively), whereas background levels of CAT activity were similar (Fig. 5F, compare lanes 1 and 2, and 13 and 14). Despite the high levels of stimulation seen with E2, CAT activities observed with transient transfections performed in the presence of OHT or RUp were lower than those obtained with vehicle (compare lanes 15-17 with lane 13) and lower than those observed with OHT or RUp in Ishikawa-ERE3/EBV cells (compare lanes 15-17 with lanes 3-5; see also inset). Mock transfection of Ishikawa-ERE3/EBV cells slightly reduced stimulation with both estrogen and RUp while increasing background, thereby blunting the stimulation by anti-estrogens but not suppressing it. Therefore, differences in the results observed using the two assays seem to be due at least in part to the status of reporter vectors within cells, i.e. number of copies per cell and/or integration into chromatin.
Taken together, these results suggest that anti-estrogens with partial agonist activity, but not full anti-estrogens, can stimulate transcription directed by endogenous levels of estrogen receptors bound to minimal ERE-containing promoters in Ishikawa cells, even though this activity is undetectable in transient transfection assays using the same reporter vectors.
Minimal Promoters Containing AP1-binding Sites Are Not Activated by Estrogen or Anti-estrogens in Transient Transfection Assays of Ishikawa Cells-Previous reports have described induction of AP1 activity by estrogen treatment in several cell lines (19 -22) and by anti-estrogens in transient transfection assays of Ishikawa cells (22). To test whether TRE motifs are sufficient to mediate stimulation by estrogen and anti-estrogens in Ishikawa cells, we transiently transfected reporter vectors containing a minimal promoter composed of six TRE motifs upstream of a TATA box (Fig. 1) or a reporter vector containing the AP1-responsive rat stromelysin promoter, STR-CAT (Fig. 1, see also Ref. 24). Stimulation of expression from the STR-CAT reporter vector could be observed in the presence of estrogen (3-fold, compare lane 2 to lane 1 in Fig. 6A) but not in the presence of anti-estrogens. TPA-stimulated CAT expression levels 5-10-fold, and no additional increase was obtained in the presence of estrogen or anti-estrogens (Fig. 6A, lanes  5-8). No stimulation of the minimal TRE6-TATA-CAT reporter constructs could be observed after incubation of the transiently transfected cells with either estrogen or anti-estrogens (Fig.  6B). When the same promoter was incorporated into an episomal vector, a lower basal activity was observed (10-fold lower, data not shown) and a weak stimulation by estrogen (2.5-fold) could be detected in the absence but not in the presence of TPA (Fig. 6C, compare lanes 2 and 6 to lanes 1 and 5). Anti-estrogens had no effect on CAT expression levels directed from this reporter construct (Fig. 6C, lanes 3 and 4, and 7 and 8).
From these experiments, we conclude that stimulation of expression from promoters containing TRE motifs by estrogen is dependent on the promoter context. Although we did not observe transcriptional activation by anti-estrogens using these promoters, we cannot rule out the possible existence of promoter-specific effects.

Micromolar Concentrations of OHT, but Not of Other Antiestrogens, Can Stimulate Transcription from Minimal Promoters Containing TRE Motifs in Stably Transfected Ishikawa
Cells-To investigate further the potential role of anti-estrogens in stimulation of the AP1 signaling pathway, cell lines were derived from Ishikawa cells by stable transfection of TRE6-TATA-CAT vectors (clonal selection by integration into the cellular genome) or of the episomal vectors TRE6-TATA-CAT/EBV or TRE2-TATA-CAT/EBV. Two clones obtained by selection for integration of the TRE6-TATA-CAT reporter vector responded to TPA stimulation (100 ng/ml) by an increase in CAT expression (3-7-fold; Fig. 7A, compare lane 5 to lane 1 for Ishikawa-TRE6 #29). Ishikawa-TRE6 clone 29 was further used to investigate whether estrogen and/or anti-estrogens can stimulate CAT activity under the same tissue culture conditions as used for the Ishikawa-ERE3/ EBV cell lines (i.e. phenol red-free DMEM supplemented with 5% charcoal-treated FBS). Treatment with estradiol or antiestrogens did not significantly modulate CAT expression in these cells, either in the presence or in the absence of TPA (Fig.  7A, lanes 2-4 and 6-8). Similar results were obtained with the other clone (data not shown). Note that estrogen did stimulate expression from an ERE3-hsp68-LacZ reporter vector transiently transfected in Ishikawa-TRE6 cells, demonstrating that lack of induction of the AP1 pathway in these cell lines was not due to loss of ER function (data not shown).
Pools of cells were selected for propagation of the TRE6-TATA-CAT/EBV vector, generating the Ishikawa-TRE6/EBV cell line (CAT activity was stimulated ϳ5-fold by TPA in these cells, data not shown). Treatment of Ishikawa-TRE6/EBV cells as well as of Ishikawa-TRE6 clone 29 cells with estrogen or anti-estrogen did not lead to detectable stimulation of CAT activity (Fig. 7B, lanes 1-10), whereas in the same assay strong agonist activity could be observed using the Ishikawa-ERE3/ EBV cells (Fig. 7B, lanes 11-15). Similar results were obtained in the presence of TPA, except that levels of CAT activity were higher in the presence of TPA in the cells containing AP1responsive promoters but not in the Ishikawa-ERE3/EBV cell lines (data not shown).
Because the conditions used for tissue culture in the abovedescribed experiments (phenol red-free DMEM containing 5% charcoal-treated FBS) may mask stimulation of AP1 activity by estrogen or anti-estrogens, we performed these assays again in the absence of serum. CAT activity was induced by estrogen (10-fold) and anti-estrogens (3-5-fold) in the Ishikawa-ERE3/ EBV cells (Fig. 7C, lanes 6 -10) but not in Ishikawa-TRE6/EBV cells (Fig. 7C, lanes 1-5) or in Ishikawa-TRE2/EBV cells (Fig.  7C, lanes 11-15), which propagate EBV episomal vectors containing only two TPA response elements. However, when higher concentrations of anti-estrogens were used (5 M instead of 0.1 M), small but reproducible stimulations of CAT activity were observed with OHT (3-fold) or with tamoxifen (2-fold) (Fig. 7D, compare lanes 3 and 4 to lane 1) but not with estrogen or with other anti-estrogens (Fig. 7D, lanes 2, 5 and 6). Similar results were observed using Ishikawa-TRE6 or Ishikawa-TRE2/EBV cells (data not shown).
These results indicate that TRE elements together with a TATA box can only mediate transcriptional stimulation by high concentrations of OHT (or TAM) in Ishikawa cells, whereas estrogen response elements can be activated at lower concentrations by both OHT and RUp under the same conditions. DISCUSSION Mechanisms underlying the partial agonist activity of antiestrogens are still poorly understood (9). In this report, we have investigated whether the partial agonist activity of anti-estrogens in Ishikawa cells can be mediated at the level of regulation of gene expression by typical estrogen response elements and/or by TPA response elements, which can mediate estrogen stimulation in some promoters (19 -22). Although transcriptional activation of ERE-containing promoters by OHT has been documented using transient transfection assays in a number of cell lines, including HeLa cells and chicken embryo fibroblasts (3), agonist activity of OHT and RU39,411 was not observed in transiently transfected Ishikawa cells using minimal promoters containing EREs (ERE3-TATA) or more complex promoters (ERE3-tk, Vit-tk). Failure to detect transcrip-  [11][12][13][14][15] were plated at a density of 1.5 ϫ 10 6 cells per 10-cm plate in DMEM containing 5% charcoal-treated FBS and supplemented with estrogen (25 nM) or anti-estrogens (100 nM) as indicated (a representative experiment is shown). C, Ishikawa-TRE6/EBV cells (lanes 1-5), Ishikawa-ERE3/EBV (lanes 6 -10), or Ishikawa cells stably transfected with TRE2-TATA-CAT/EBV (Ishikawa-TRE2/EBV; lanes [11][12][13][14][15] were plated in phenol red-free DMEM (0% FBS) and incubated with estrogen (25 nM) or anti-estrogens (100 nM) as indicated for 24 h. D, Ishikawa-TRE6/EBV cells plated in phenol red-free DMEM (0% FBS) were treated with estrogen (100 nM) or anti-estrogens (5 M) for 24 h. tional activation of ERE-containing promoters by OHT in transiently transfected Ishikawa cells is in agreement with previous observations (22). Contrary to results obtained with transient transfections, we observed significant levels of transcriptional activation by OHT and RUp in Ishikawa, but not in MCF7 cells, when the ERE3-TATA-CAT/EBV reporter vector was stably propagated as an episome. These results correlate well with the observed agonist effect of these two anti-estrogens on expression levels of the endogenous progesterone receptor (note that the human progesterone receptor upstream sequences contain a half-palindromic TGACC motif but no consensus EREs, Ref. 38). In addition, OHT and RUp, although structurally unrelated, induced similar shifts in mobility of ER⅐ERE complexes in gel retardation assays. The migration of these complexes was found to be distinct from those formed in the presence of the full antagonists RUf and ICI.
Differences observed in transcriptional activity in the presence of anti-estrogens in transient transfection assays and using "reporter cell lines" may reflect the different status of the reporter vectors, which are present at a lower copy number when maintained as episomes (generally less than 50 copies per cell, Ref. 39) and are incorporated into chromatin to a higher degree (40) compared with transiently transfected reporter plasmids. These results suggest that stimulation of ERE-mediated transactivation by anti-estrogens requires cofactor(s) limiting in amounts or availability in transient transfection assays in Ishikawa cells, whereas estrogen-liganded ER may recruit other, non-limiting cofactors (41,42). Of interest is the fact that capacity to remodel chromatin structure via histone acetyltransferase or deacetylase activities has been attributed to an increasing number of nuclear receptor co-activators and co-repressors (42)(43)(44)(45)(46)(47)(48)(49), demonstrating that incorporation of target promoters into chromatin is an integral part of the mechanism of transcriptional activation by nuclear receptors. It is not clear at present whether ER can interact with these cofactors in vivo when bound by anti-estrogens with partial agonist activity. It is possible that cofactors specific to OHTbound ER, such as the newly described co-activator L7/SPA (50) may mediate the agonist activity of this anti-estrogen. Future functional characterization of cofactors interacting with ER in the presence of OHT and RUp should provide insights into the molecular mechanisms of action of anti-estrogens with partial agonist activity.
Low concentrations of OHT or RUp, which were sufficient for activation of ERE3-TATA promoters in stably propagated vectors, did not yield detectable transcriptional stimulation of promoters containing TRE sites inserted upstream of a TATA box either in transient transfection assays or using stably propagated vectors. Others have previously documented transcriptional activation of the TRE-containing collagenase promoter by anti-estrogens in transiently transfected Ishikawa cells (22). Discrepancy between these and our results could be due to the promoter context of TRE elements. Along the same line, our results indicate that promoter context influences stimulation of AP1-responsive promoters by estrogen in transient transfection assays. Alternatively, differences in cell lines or transfection methods could be the source of this discrepancy. Side-by-side comparison of cell lines carrying episomal reporter vectors whose promoters differed only by the response elements present in the minimal synthetic promoters confirmed that transcriptional stimulation by anti-estrogens like RUp or OHT could be mediated by estrogen response elements but not TPA response elements at low concentrations of anti-estrogens. Stimulation of AP1-responsive promoters could only be observed in serum-free medium using 5 M OHT or TAM but not estrogen or other anti-estrogens. Whether this effect is medi-ated by estrogen receptors or is initiated at the cellular membrane remains to be investigated.
In conclusion, our results suggest that transcriptional stimulation by anti-estrogens can be mediated by consensus EREs in Ishikawa cells, this observation being consistent with the absence of TRE sites in the promoters of genes whose expression can be induced by hydroxytamoxifen in the uterus (38,51). In addition, TREs are also capable of mediating transcriptional stimulation by anti-estrogens, although in a manner that is restricted by both the nature and the concentration of the anti-estrogen. Finally, while transient transfection has proven to be a powerful tool for analyzing intracellular signaling pathways, our study emphasizes that this assay only partially recapitulates the conditions required for initiation of transcription in vivo.