Tamoxifen Induction of CCAAT Enhancer-binding Protein α Is Required for Tamoxifen-induced Apoptosis*

Low concentrations of tamoxifen or its active metabolite 4-hydroxytamoxifen (OHT) induce estrogen receptor α (ERα)-dependent apoptosis. To analyze the pathway of OHT-ERα-induced apoptosis, we developed stably transfected lines of HeLa cells expressing wild-type ER and an inactive mutant ERα unable to bind estrogen response elements. HeLa cells expressing the mutant ERα and HeLa cells expressing wild-type ERα in which the ER was knocked down with an ER-specific small interfering RNA were not killed by Tam or OHT, suggesting that estrogen response element-mediated transcription is required for Tam- and OHT-induced apoptosis. Microarray analysis to identify a gene(s) whose expression is important in OHT-ER-mediated apoptosis identified 19 mRNAs that OHT up-regulated by >1.6-fold and 15 down-regulated mRNAs. Gene function and the time course of induction by OHT-ERα led us to further investigate CCAAT enhancer-binding protein α (C/EBPα), which has roles in cell cycle progression and apoptosis, and p21. Quantitative reverse transcription-PCR, Western blot analysis, and RNA interference knockdown suggest that cell cycle arrest resulting from OHT-ERα induction of p21 may facilitate apoptosis. OHT-ERα, but not E2-ERα, induced C/EBPα mRNA and protein. RNA interference knockdown of C/EBPα nearly abolished OHT-ERα-induced apoptosis. We isolated stable cell lines that were resistant to OHT-induced apoptosis, contain full-length functional ERα, and undergo apoptosis in response to etoposide. In these OHT-resistant cell lines both before and after OHT treatment, C/EBPα levels are much lower than in OHT-sensitive cells. These studies establish a novel molecular site responsible for Tam- and OHT-ERα-induced apoptosis of cancer cells.

tems. Binding of a potent estrogen such as 17␤-estradiol (E 2 ) to the ER may induce dissociation of ER from a heat shock protein/chaperone complex (1) and enable the ER to dimerize and bind to specific DNA sequences termed estrogen response elements (EREs) (2). Transcription activation by ERs is mediated by two interacting activation functions, AF1 and AF2. AF1 mediates ligand-independent transactivation and is thought to be especially important in transcription by the clinically important selective estrogen receptor modulator, tamoxifen (Tam), and by its active metabolite, 4-hydroxytamoxifen (OHT). AF2 activity depends on estrogen binding to the ER ligand binding domain (LBD) (3)(4)(5)(6)(7). When bound to estrogen, the ER LBD assumes a conformation that enables the recruitment of coactivators. The bound coactivators help assemble a multiprotein complex that facilitates both chromatin remodeling and formation of an active transcription complex (8). When bound to Tam or OHT, the ligand binding domain of ER assumes a different conformation that interferes with coactivator binding and appears to facilitate corepressor binding (9).
By competing with E 2 and other estrogens for binding to the ligand binding site of ER, Tam and OHT may induce formation of an ER complex that is unable to effectively activate transcription of estrogen-regulated genes important in the growth and development of estrogen-dependent tumors. Tam and OHT can both arrest cell growth by preventing the growth promoting activities of E 2 and can induce death of both ER-positive and ER-negative cells (10 -12). Tam triggers apoptosis in vitro and shrinks some tumors in vivo. Several mechanisms have been proposed to explain tamoxifen-induced cell death including transcriptional regulation of Bcl-2 family proteins (13), activating mitogen-activated protein kinase and other kinases through nongenomic pathways (11), and triggering an increase of intracellular Ca 2ϩ (14). However, in many of these experiments, very high M concentrations of Tam or OHT were used, and some of the studies were carried out in cells that lack ER. To analyze these processes in cells that were otherwise identical and differed only in the presence or absence of ER, we analyzed Tamand OHT-induced cell death in ER-negative HeLa cells stably transfected to express hER␣ (HeLaER6 (15)). We identified two pathways by which OHT induces cell death. When ER-negative HeLa cells are maintained in medium containing 10 -20 M Tam, OHT, E 2 , or raloxifene, the cells die within 24 h by a reactive oxygen-based pathway that triggers classical caspasedependent apoptosis (16). Low (nM) concentrations of OHT and submicromolar concentrations of Tam do not kill or damage ER-negative HeLa cells (Fig. 1). When ER positive HeLaER cells are maintained in medium containing 1-100 nM OHT, they undergo apoptosis over several days. The antagonist ICI 182,780, the selective estrogen receptor modulator, raloxifene, and E 2 protect the cells against OHT-ER-mediated cell death, suggesting a role for ER␣ (16). However, the most widely accepted current standard of proof for the involvement of a protein in a response, demonstrating an effect using RNAi knockdown was not performed. The mechanism(s) by which low concentrations of OHT induce cell death remains largely unknown. The long time course of OHT-ER␣-induced cell death suggested that genomic processes based on ER-regulated transcription were likely to be important.
In this study we analyzed the role of ER␣-mediated gene transcription in OHT-ER-induced cell death. We initially used RNAi and stable transfection of mutant ERs to examine the requirement for ER␣ and for ER␣-mediated gene transcription. Because these experiments supported a role for OHT-ER␣-mediated gene transcription in cell death, we carried out microarray analysis.
The microarray data, more detailed regulatory studies, and what was known about their function led us to focus on CCAAT enhancer-binding protein ␣ (C/EBP␣) and p21/waf1/CDKN1A/ Cip1. C/EBP␣ and p21 encoded proteins whose roles in cell growth and cell death and expression patterns were consistent with a possible role in OHT-ER␣-dependent apoptosis.
CEBP/␣ was originally reported as a transcriptional factor containing a basic leucine zipper DNA binding motif (17)(18)(19). C/EBP␣ is exported from the nucleus to the cytoplasm in response to tumor necrosis factor ␣ (20 -22). Roles for C/EBP␣ in signal transduction, inhibition of cell cycle progression based on protein-protein interactions through C-terminal interaction with p21 and cyclin-dependent kinase have been proposed (23)(24)(25)(26)(27)(28). Rat C/EBP␣ increases apoptosis of NCI-H358 cells (29). We analyzed the roles of C/EBP␣ and p21 in OHT-ERinduced apoptosis and conclude that although p21 likely contributes to OHT-ER-induced apoptosis, the induction of C/EBP␣ by OHT-ER plays the central role in OHT-induced apoptosis.
Establishment of the HeLaER LBD and HeLaER M-DBD Cell Lines-HeLa cells were plated in 100-mm culture dishes at a concentration of 1.6 ϫ 10 6 cells per dish. HeLa cells obtained from ATCC were used to construct the HeLaER M-DBD and HeLaER LBD cell lines. The original ER␣ M-DBD mutant was a gift from Prof. L. Jameson and was recloned into the pIE plasmid (15). The cells were transfected with 6 g of SspI-linearized pIE-LBD, pIE-M-DBD using Polyfect (Qiagen, Valencia, CA) liposome following the manufacturer's instructions. The medium was replaced with selection media (50% Dulbecco's modified Eagle's medium, 10% charcoal-dextran-fetal bovine serum, 50% conditioned medium, 500 g/ml G418 in the HeLaER LBD and HeLaER M-DBD selections) ϳ24 h posttransfection. After sufficient growth, colonies were isolated and reseeded into 24-well culture plates in selection medium. Cell lines were further expanded to confluence in 6-well plates, then T-75 flasks. For long-term growth, the cells were maintained under selection in medium containing 150ϳ200 g/ml G418. ER expression was detected by Western blotting. H222 anti-ER antibody (NeoMarkers) was used to detect ER in the HeLaER LBD cells, and NCL-ER-6F11 anti-ER antibody was used to detect ER in the HeLaER M-DBD cells.
Isolation of OHT-resistant Cell Lines-HeLaER cells in Phenol Red-free Dulbecco's modified Eagle's medium (Sigma) supplemented with 10% charcoal-dextran-treated fetal bovine serum and 150 g/ml G418 were plated in 96-well plates at 0.5 cell/well in medium containing 100 nM OHT. After several weeks, colonies from wells containing a single colony were isolated as OHT-resistant clonal cell lines.
Microarray Analysis-HeLaER cells were treated with 10 nM OHT or ethanol vehicle for 34 h. RNAs were purified with the RNeasy kit (Qiagen). The mRNAs were reverse-transcribed to cDNAs by using Moloney murine leukemia reverse transcriptase (Invitrogen) and hybridized to Affymetrix GeneChip Human Genome U133A Arrays (Affymetrix, Santa Clara, CA). The array data were initially processed by the W. M. Keck Center for Comparative and Functional Genomics at the University of Illinois using the GeneChip Robust Multichip Average package in R/Bioconductor. A false discovery rate of Ͻ0.05, as deter- mined by Significance Analysis of Microarray software (31), was used to identify genes significantly regulated by ligand treatment. Expression data were then loaded into GeneSpring software and normalized to the control (ethanol-treated) samples. Microarray analysis was carried out in triplicate for OHTtreated samples and for ethanol-treated samples.
Quantitative Real Time PCR-RNA levels in ethanol-treated control samples were set equal to 1. Genes up-regulated or down-regulated by OHT at least 1.6-fold were selected for quantitative RT-PCR analysis for further confirmation. RNAs from HeLaER cells treated with OHT or E 2 for 4, 24, or 34 h were extracted with TRIzol reagent (Invitrogen) and purified with the RNeasy mini kit (Qiagen). One g of RNA was reverse-transcribed using Moloney murine leukemia reverse transcriptase (Invitrogen), and 1% of the cDNA product was used in quantitative RT-PCR. The forward and reverse primer mix (1 l of 10 M) and 12.5 l of SYBR green 2ϫ PCR master mix (Applied Biosystems, Carlsbad, CA) in a total volume of 25 l were put into each well of a 96-welliCycler iQ TM PCR plate and assayed using a Bio-Rad iCycler TM optical system (Bio-Rad). PCR was as follows: 95°C for 2 min; 45 repeats (95°C for 25 s, 55°C for 25 s, 72°C for 15 s); melting from 55 to 95°C with 0.5°C increases per 10 s in each cycle for 80 cycles. The internal standard was 36B4 mRNA.
Apoptosis Analysis-Apoptosis was assayed essentially as we described previously (16).
Transient Transfections-Transient transfections of HeLaER, HeLaER LBD, and HeLaER M-DBD cells were performed using Lipofectin Plus reagent (Invitrogen) using the manufacturer's protocol. 100,000 cells per well were seeded in 12-well plates. After 24 h the cells were transfected with 0.4 g of either (ATL) 4 -luc or AP1-luc reporter and 50 ng of pRLSV40 per well. E 2 or OHT was added to a final concentration of 10 nM after transfection. After 48 h the cells were harvested, and dual luciferase assays (Promega) were performed using the manufacturer's protocol.
Flow Cytometry and Light Phase Microscopy-Flow cytometry assays were performed as previously described (16). Briefly, cells were plated in 6-well plates and incubated 24 h. After treatment with 10 Ϫ8 M OHT or EtOH vehicle for 3 days, cells were washed with phosphate-buffered saline (PBS) and harvested using PBS-EDTA. Cells were pelleted by centrifugation at 600 rpm at 4°C for 5 min and resuspended in 1 ml of phosphate-buffered saline. 40 nM of 3,3Ј-dihexylocarbocyanine iodide was added to the resuspended pellet, and the samples were incubated at 37°C for 15 min, after which propidium iodide was added to 5 g/ml. Relative fluorescence intensities were measured using a Coulter XL benchtop flow cytometer with excitation at 488 nm and emission at 520 nm. The cell sorter was also used to ascertain the light scattering properties of the cells. Cells were visualized by light phase microscopy using a TE 300 Micron Eclipse inverted fluorescence microscope.
Statistical Analysis-p values were calculated using the unpaired, two-tailed Student's t test. Data are presented as the mean Ϯ S.E. Significance was established when p Ͻ 0.05.

RNAi Knockdown of ER␣ Blocks OHT-mediated Apoptosis-
Our observations that low concentrations of Tam and OHT do not induce apoptosis of wild-type ER-negative HeLa cells and that pretreatment with an excess of ICI 182,780/Faslodex (ICI), the selective estrogen receptor modulator raloxifene (RAL), or E 2 largely protected HeLaER cells from OHT-and TAM-induced death (16) strongly suggested a role for ER␣ in Tam-and OHT-induced cell death. To test the role of ER␣ more directly, we carried out an RNAi knockdown of ER␣ in the HeLaER cells. The cells were transfected with either the control pGL3 (luciferase) siRNA or the ER-specific siRNA. Western blot analysis showed that the ER␣ knockdown lasted at least 5 days (Fig. 2A). The cells were transfected with the control pGL3 siRNA or with the ER␣ siRNA and then treated with either EtOH vehicle or 10 Ϫ8 M OHT (Fig. 2B). Cells transfected with the pGL3 siRNA and treated with OHT became sparse and exhibited the elongated morphology we previously described (16). When displayed using fluorescence-activated cell sorting, the OHTtreated cells transfected with the pGL3 siRNA exhibited a shoulder showing a substantial population of cells undergoing apoptosis. In contrast, in the cells in which ER␣ was knocked down with the ER␣-specific siRNA, OHT did not cause a reduction in cell number, a change in cell morphology, or the appear-ance of a peak of apoptotic cells (Fig. 2B). These data clearly demonstrate that ER␣ mediates OHT-and TAM-induced cell death in HeLaER cells.
OHT Does Not Induce Apoptosis in Stably Transfected Cell Lines Expressing ER Mutants That Do Not Bind to EREs-Although our data show that ER is required for OHT-induced cell death, the functions of ER that are essential were not known. ER␣ exerts its actions by rapid nongenomic actions based on modulation of signal transduction pathways (33-38) by regulating gene expression after being tethered to DNA through other transcription factors bound at AP1 (39 -41) and SP1 sites (42)(43)(44) and by direct binding to DNA at palindromic EREs and dispersed ERE half-sites (45). To identify the ER functions important in OHT-ER␣-induced apoptosis, we isolated stably transfected lines of HeLa cells expressing mutant ERs defective in specific ER functions. Because the ER LBD alone reportedly is sufficient for many non-genomic actions of ER, we isolated a stably transfected cell line expressing only the ER LBD. Previous reports indicated that OHT-ER␣ is far more effective than E 2 -ER␣ in activating transcription when ER␣ is tethered to DNA through AP1 sites (46). We, therefore, prepared a stably transfected cell line containing two mutations in the ER␣ DNA binding domain (E203A and G204A) previously shown to nearly abolish direct binding to EREs while retaining the ability to activate transcription when tethered to an AP1 site (47).
To produce the stably transfected cell lines, plasmids containing the constructs coding for the ER LBD or the ER with a mutated DNA binding domain (ER M-DBD) were constructed in the bicistronic expression plasmid that expresses the ER mutant and neomycin phosphotransferase as a bicistronic mRNA transcribed from a single promoter (15). After selection using medium containing G418, we showed that although the HeLa M-DBD did not efficiently activate transcription from an ERE-luciferase plasmid, it retained the capacity for ligand-de-pendent activation of an AP1-luciferase plasmid as previously reported (data not shown and Ref. 47).
To examine the role of ER-mediated transcription in OHT-ER␣-induced apoptosis, we examined at the ability of OHT to induce apoptosis of the cell lines expressing wild-type ER␣, the ER␣ LBD that cannot activate transcription by either direct binding to DNA or through tethering, and the ER␣ M-DBD that only activates transcription through tethering. Very high, Ͼ10 M, OHT should induce apoptosis through the ER-independent pathway (Fig. 1). Because all three cell lines were killed by 20 M OHT, they remained sensitive to caspasedependent apoptosis (Fig. 3, 2 ϫ 10 Ϫ5 M). OHT at 10 -1000 nM should induce apoptosis through the ER-dependent pathway (Fig. 1). The HeLaER, HeLaER LBD, and HeLaER M-DBD cell lines were maintained in medium containing either ethanol vehicle or the indicated concentrations of OHT (10 nM to 20 M) for 3 days, and cell death was determined (16). As expected, OHT at 10 -1000 nM induced apoptosis in the HeLaER cells. Neither the HeLaER LBD nor the HeLaER M-DBD cell lines were killed by 10 -1000 nM OHT (Fig. 3). Similar results were obtained with Tam (data not shown). These data support the idea that ERE-mediated transcription is required for OHT-ER␣-induced apoptosis.
ICI Protects HeLaER Cells from OHT-induced Cell Death up to 33 h Post-OHT Treatment-We previously reported that ICI 182,780 does not induce apoptosis. By competing with OHT for binding to ER␣, the addition of excess ICI 182,780 blocks OHT-ER␣-induced apoptosis (16). To determine whether ICI can protect HeLaER6 cells when it is added to the medium after OHT, HeLaER6 cells were treated with 10 Ϫ8 M OHT and 10 Ϫ6 M ICI 182,780 0, 2, 12, 24, 33, 36, and 48 h after OHT treatment. Consistent with our earlier report, the addition of an excess of ICI 182,780 at the same time as OHT blocked OHT-induced apoptosis. Surprisingly, the addition of ICI 182,780 up to 33 h after OHT also blocked OHT-induced apoptosis (Fig. 4). By 36 h the cells commit to OHT-ER-induced apoptosis, and ICI begins to lose its ability to block OHT-ER  induced apoptosis (Fig. 4). These data suggest that OHT-ER␣ mediates a key step in activation of the death pathway 33-36 h after OHT treatment.
Because our data indicated that ERE-mediated transcription was required for OHT-ER␣-induced apoptosis, we carried out microarray analysis to identify mRNAs whose expression is regulated by OHT-ER␣. We then analyzed likely mRNAs identified from the microarray analysis to determine whether or not they play a role in OHT-mediated apoptosis. Because the ICI data suggested that a key event in committing the cells to OHTmediated apoptosis occurs in the 33-36-h time frame, we looked at mRNA levels 34 h after OHT treatment.

Identification of OHT-ER␣-regulated Genes by Microarray Analysis-Microarray studies in ER␣ positive MCF-7 cells and
in MDA-MB-231 breast cancer cells stably transfected to express ER␣ indicated that Tam and OHT act as full or partial agonists on ϳ20% of E 2 -regulated genes and also regulate a group of genes not regulated by E 2 -ER (48). We, therefore, carried out microarray analysis to identify OHT-ER␣-regulated genes. HeLa-ER cells were maintained in medium containing 10 nM OHT for 34 h, the cells were harvested, and RNA was isolated, converted to cDNA, and analyzed on Affymetrix microarrays containing 18,000 human genes. We focused on the 19 genes whose mRNAs were up-regulated by Ͼ1.6-fold after OHT treatment and the 15 mRNAs whose levels were down-regulated by Ͼ1.5-fold after OHT treatment (Table 1). Approximately 50 additional mRNAs were up-regulated 1.5-1.6-fold after OHT treatment (data not shown). Because none of these mRNAs encoded proteins that seemed likely to mediate OHT-dependent apoptosis, the regulation of these mRNAs was not further characterized. The genes whose mRNA levels were regulated by OHT included two genes (C/EBP␣ and p21/ CDKN1A/waf1/Cip1) that code for proteins likely to be important in cell growth or cell death (Table 1). C/EBP␣ is a transcription factor implicated in cell death in terminally differentiated cells, and p21 is a widely studied cyclin-dependent kinase inhibitor (CDKN1A/waf1/Cip1).
OHT-ER Induction of C/EBP␣ and p21 mRNAs-We used quantitative RT-PCR to confirm the microarray data and to assay mRNA levels at selected time points. Consistent with other data using this system, in most cases the -fold induction was greater by quantitative RT-PCR than by microarray (48).
Because E 2 does not induce cell death, we concentrated on mRNAs that were regulated differently by OHT and by E 2 . Quantitative RT-PCR showed that p21 mRNA is induced by both OHT and by E 2 in the 24 -48-h post-OHT time frame important for OHT-induced apoptosis (Fig. 5A) and shows little or no induction by RAL and ICI 182,780. C/EBP␣ mRNA is strongly induced by OHT, is not induced by E 2 , and exhibits very weak induction by ICI 182,780 (Fig. 5B). We also tested the effects of E 2 and OHT on the levels of C/EBP␣ and p21 protein.
Western blot analysis shows that E 2 does not induce C/EBP␣ protein (Fig. 5C, C/EBP, EtOH, and E 2 ) and moderately induces p21 (Fig. 5C, p21, EtOH, and E 2 ), whereas OHT strongly induces both C/EBP/␣ and p21 (Fig. 5C). The role of C/EBP␣ in apoptosis in other systems (29), the strong induction of C/EBP␣ mRNA in the important time period 24 -48 h after OHT treatment (Fig. 4), and the induction of C/EBP␣ mRNA by OHT, which induces apoptosis of the HeLaER cells, and the absence of a strong induction of C/EBP␣ by E 2 or ICI, which do not induce apoptosis, all are consistent with the idea that OHT induction of C/EBP␣ may be important in OHT-ER induced apoptosis.

TABLE 1 Genes whose expression is regulated by OHT in HeLaER cells
OHT (10 nM) was added at 0 time, the cells were harvested at 34 h, and mRNA was isolated and analyzed using microarrays as described under "Experimental Procedures." The lane labeled OHT shows mRNAs induced by OHT at least 1.6 fold (up-regulated genes) or down-regulated (down-regulated genes) by at least 1.5-fold. The first row shows the -fold change as seen on the microarray. The second row represents the -fold change of most of the genes as shown by quantitative RT-PCR. A few genes whose expression is unlikely to be related to cell death (such as complement C3) were not further analyzed. Regulation of the expression of some mRNAs by 10 nM E 2 and by 10 nM OHT at various times is shown for some mRNAs that were studied in more detail. TFF, transforming growth factor.  OCTOBER 19, 2007 • VOLUME 282 • NUMBER 42

JOURNAL OF BIOLOGICAL CHEMISTRY 30539
Regulation of C/EBP␣ and p21 Protein Levels and Their Knockdown Using RNAi-Because we were going to test the roles of p21 and of C/EBP␣ in OHT-ER␣-induced apoptosis using RNAi knockdown and we wanted to determine whether OHT and E 2 induce p21 and C/EBP␣ proteins, we carried out Western blot analysis. OHT induced p21 protein to a level higher than was seen with E 2 (Fig. 6). As expected, OHT, but not E 2 , induced C/EBP␣ protein. C/EBP␣ was knocked down by the C/EBP␣ siRNA and not by the p21 siRNA. Similarly, p21 was knocked down by the p21 siRNA but not by the C/EBP␣ siRNA (Fig. 6).
RNAi Knockdown of C/EBP␣ Blocks OHT-ER␣-induced Apoptosis-We then looked at the effect of knockdown of C/EBP␣ on apoptosis. To evaluate cell death, we used three fluorescence-activated cell sorter-based assays. These assays target distinct aspects of cell death. Release of a fluorescent dye evaluates loss of mitochondrial membrane potential, uptake of a fluorescent agent that intercalates into DNA targets the loss of plasma membrane integrity, and light scattering detects changes in cell morphology. In preliminary studies, we identified a time after OHT addition that resulted in substantial cell death as measured by all thee assays and showed that transfecting the C/EBP␣ and p21 siRNAs into the cells using oligofectamine did not cause cell death (data not shown). In three independent experiments the addition of OHT to the medium induced cell death as measured by all three assays (Fig. 7A,  OHT). Transfection with the control pGL3 luciferase siRNA had no effect on cell death (Fig. 7A, PGL3 siRNA ϩ OHT). RNAi knockdown of C/EBP␣ nearly abolished OHT-ER␣-induced cell death (Fig. 7A, CEBP␣ siRNA ϩ OHT). The difference between the level of cell death in the cells treated with OHT or treated with OHT and transfected with the pGL3 siRNA compared with the level of cell death in the cells transfected with the C/EBP␣ siRNA was significant using each of the three assays for cell death (p Ͻ 0.05). RNAi knockdown of p21 had a significant but smaller effect on OHT-ER␣-induced cell death (Fig. 7B). There was a statistically significant (p Ͻ 0.05) decrease in cell death in the cells treated with the p21 siRNA compared with the cells transfected with the pGL3 siRNA. A second control siRNA recommended by the supplier (Dharmacon) also had no effect on OHT-ER␣-induced cell death (Fig. 7B). These data show that induction of C/EBP␣ by OHT plays a key role in the ability of OHT to induce apoptosis.

In Cells Resistant to OHT-ER␣-induced Apoptosis, Expression of C/EBP␣ and OHT Induction of C/EBP␣ Are Lost-
The expression and RNAi knockdown data strongly supported a role for C/EBP␣ in OHT-ER␣-induced cell death and suggested a possible role for p21. If OHT induction of C/EBP␣ or p21 plays a key role in OHT-ER␣-induced apoptosis, HeLaER cells selected for resistance to OHT-ER␣-induced cell death might show altered expression or regulation of C/EBP␣ or p21. The overwhelming majority of HeLaER cells stop growing and die in medium containing 100 nM OHT, but a few cells survive. Based on the RNAi knockdown experiment (Fig. 2), we knew that HeLaER cells that lose ER␣ expression would continue to grow in OHT and would be represented among the clonal lines we isolated. Western blot analysis showed that about half of the clonal lines we isolated after growth in OHT-containing medium had lost all or most of their ER␣ expression or expressed an ER␣ fragment (data not shown). We identified four clonal cell lines that were resistant to OHT-ER␣-induced apoptosis and contained levels of full-length ER␣ comparable with two HeLaER cell lines that were sensitive to OHT-ER␣induced apoptosis (Fig. 8A). Data for all of the ER␣-containing, OHT-resistant clonal cell lines we isolated are shown. No cell lineswereexcluded.Etoposide,awidelyusedactivatorofcaspasedependent apoptosis, retained the ability to kill all 4 of the OHT-resistant cell lines in 24 h (data not shown). Therefore, the four OHT-resistant cell lines are specifically defective in OHT-induced apoptosis but retain functional machinery for caspase-dependent apoptosis activated by the mitochondrial pathway.
We used quantitative RT-PCR to determine the basal levels of C/EBP␣ and p21 in the OHT-sensitive and OHT-resistant cell lines and their regulation by OHT and by E 2 . The basal level of p21 mRNA in the OHT-resistant cell lines was equal to or   (59). Actin was used as the internal standard.
higher than the basal level of p21 mRNA in the two OHTsensitive cell lines. One of the OHT-resistant cell lines (Fig. 8B, clone 5) retained the ability to induce p21 in the presence of OHT. As expected, the two cell lines that are killed by OHT showed strong induction of C/EBP␣ mRNA by OHT and little or no induction by E 2 (Fig. 8C). In contrast to p21, in all four of the HeLaER lines selected for resistance to OHT-induced cell death, both the basal level of C/EBP␣ mRNA and the level of C/EBP␣ mRNA after OHT treatment were extremely low (Fig.  8C). Because this was an unbiased selection, the absence of an induced level of C/EBP␣ mRNA in all four ER-containing cell lines resistant to OHT-induced apoptosis provides compelling support for a central role of C/EBP␣ in cell death mediated by OHT-ER␣.

DISCUSSION
At least three lines of investigation demonstrate that OHT acts by binding to ER␣ to induce apoptosis. Wild-type HeLa cells which lack ER and HeLa cells that contain ER mutants are not killed by 10 -1000 nM OHT (Ref. 16 and Fig. 3). High concentrations E 2 , ICI 182,78, and RAL block OHT-induced apoptosis (16). RNAi knockdown of ER␣ abolishes OHT-induced cell death (Fig.  2). Therefore, loss of expression of ER␣ or mutational inactivation of ER␣ are likely to be observed in some cell lines resistant to OHT-induced cell death. Consistent with a key role for ER␣ in OHT-induced cell death, in several cell lines resistant to OHT-induced cell death we detected loss of ER␣ expression or production of a truncated ER␣ (data not shown).
A central role for C/EBP␣ in OHT-ER␣-induced apoptosis is supported by our observations that RNAi knockdown of C/EBP␣ nearly abolishes OHT-ER␣-induced apoptosis and by the loss of C/EBP␣ and of OHT induction of C/EBP␣ in cell lines selected by their ability to grow in concentrations of OHT that kill nearly all of the HeLaER cells. It is striking that all of the cell lines that we isolated that grew out in 100 nM OHT either exhibited reduced expression of full-length ER␣ or reduced expression and defective regulation of C/EBP␣.
Because reduced expression of C/EBP␣ nearly abolished OHT-induced apoptosis, it seemed likely that expression of increased levels of C/EBP␣ enhances OHTinduced cell death. To test this idea, we attempted to transiently transfect wild-type C/EBP␣ and C/EBP␣ mutants defective in specific functions (28) into the HeLaER cells. However, we were unable to transfect more than ϳ30% of the cells, and the harsh conditions needed for high transfection efficiency damaged the cells. It was, therefore, not possible to rapidly test the effect of elevated levels of C/EBP␣ on OHT-mediated apoptosis.
C/EBP␣ is expressed in many cell types including breast and ovary tissues and localizes to the nucleus. C/EBP␣ mRNA gives rise to two isoforms of protein, the full-length 42-kDa and The data represent the average of three independent experiments Ϯ S.E. Comparing C/EBP␣ siRNA plus OHT treatment and pGL3 siRNA plus OHT treatment, the p value is 0.01 for the mitochondrial dysfunction assay, 0.01 for the plasma membrane integrity assay, and 0.02 for the cell morphology assay. Comparing p21 siRNA plus OHT treatment and pGL3 siRNA plus OHT treatment, the p value is 0.04 for the mitochondrial dysfunction assay, 0.03 for the plasma membrane integrity assay, and 0.03 for the cell morphology assay. Thus, all of the key data are significant at p Ͻ 0.05. 30-kDa with a truncated N terminus. The two isoforms of C/EBP␣ show contrasting functions in gene activation and cell proliferation, and the isoform ratio may be important in mediating proliferation and differentiation (49). Our Western blot data indicate that nearly all of the C/EBP␣ in these cells is the 42-kDa isoform, and this isoform is responsible for OHT-ER␣induced apoptosis. Many studies of C/EBP␣ focus on its role in lineage determination in hematopoiesis (50). C/EBP␣ is a key factor in driving the development of myeloid cells. Our data suggest a quite new but plausible role of C/EBP␣ as a critical factor in controlling an apoptosis pathway. Our finding of a role for C/EBP␣ in cell death, consistent with several reports describing diverse pathways including p21, and the E2F complex that are impacted by C/EBP␣ as it mediates cell cycle arrest and proliferation (51)(52)(53). Induction of C/EBP␣ expression in breast cancer cells resulted in growth inhibition accompanied by G 0 -G 1 cell cycle arrest. C/EBP␣ expression was associated with up-regulation of p21 and peroxisome proliferator-activated receptor ␥ and down-regulation of c-Myc (52). We suggest that p21 works with C/EBP␣ to help determine whether or not cells initiate apoptosis after cell cycle arrest.
Both C/EBP␣ and p21 are induced by OHT-ER␣, and it is likely that the interplay between these two proteins plays a role in OHT-ER-induced cell death. Although both E 2 and OHT induce P21 mRNA in HeLaER cells, induction of p21 protein by OHT is far more effective than induction of p21 by E 2 . The cyclin-dependent kinase inhibitor p21 exhibits a short half-life of 0.5 to Ͻ2 h (54,55) and is thought to be stabilized by binding to C/EBP␣ (56). Because OHT, but not E 2 , induces C/EBP␣, which stabilizes the p21 protein, OHT elicits a much larger induction of p21 protein than E 2 .
RNAi knockdown of p21 resulted in a moderate reduction in OHT-induced apoptosis. It is possible that the failure to isolate OHT-resistant cell lines that lack p21 stems from an essential role of p21 in regulating progression through the cell cycle.
Because p21 is a well known cyclin-dependent kinase inhibitor and participates in regulating the cell cycle and C/EBP␣ reportedly inhibits progression through the cell cycle (23)(24)(25)(26)(27) and contains binding sites for p21 and cyclin-dependent kinase (28), it is quite possible that OHT-induced apoptosis may require C/EBP␣ to associate with p21 and initiate downstream pathways leading to cell death. This might involve arresting cell growth. Consistent with this view, in MCF-7 cells tamoxifen acts mainly by arresting cells in the G 0 /G 1 phase of the cell cycle (57,58). Of course the ability of C/EBP␣ to mediate gene expression by acting as a transcriptional factor or to play a role in signal transduction pathway in the cytoplasm may also be important to its role in OHT-induced cell death.