Activation of the p38 Mitogen-activated Protein Kinase Pathway by Estrogen or by 4-Hydroxytamoxifen Is Coupled to Estrogen Receptor-induced Apoptosis*

17b-Estradiol (E2) or the antiestrogen, 4-hydroxytamoxifen (OHT), induce apoptosis in stably transfected estrogen receptor (ER)-positive HeLa-ER5 cells. p38 mitogen-activated protein kinase is implicated in cellular processes involving apoptosis. The p38 kinase inhibitor, SB203580, partially protects HeLa-ER5 cells against apoptosis induced by E2 or by OHT. E2 induces the p38 pathway 12–36-fold in ER-positive cell lines, while OHT induces p38 activity 2–5-fold. In an ER-positive cell line selected for resistance to E2-induced apoptosis, E2 no longer induced p38, and the ER no longer bound to the estrogen response element, while OHT induced both p38 and apoptosis. In cells selected for resistance to OHTinduced apoptosis, OHT no longer induced p38, while E2 induced p38 and apoptosis, and transactivated an estrogen response element-containing reporter gene. In MCF-7 cells, whose growth is stimulated by estrogen, E2 did not induce p38 or apoptosis, while OHT induced both p38 and apoptosis, and SB203580 protected against OHT-induced apoptosis. This work shows that E2 and OHT activate the p38 pathway, suggests that they use different pathways for p38 activation, and links activation of the p38 pathway to apoptosis induced by E2 and by OHT.

Several mitogen-activated protein (MAP) 1 kinase pathways have been described in mammalian cells, including the extracellular signal-regulated kinases (ERKs) (1), the 3 Jun N-terminal kinases (JNKs) (2), and the p38 MAP kinases (3) (also termed stress-activated protein kinase 2). The p38 MAP kinase pathway is activated by proinflammatory cytokines and by environmental stress (4). While the ERK kinases are activated by mitogenic stimuli, activation of the p38 pathway is strongly correlated with apoptosis (3). Although the relationship between p38 activation and apoptosis is not well understood, a recent report links activation of p38 and the induction of apoptosis. p38 kinase, activated by UV light, was shown to phosphorylate the p53 tumor suppresser (5). Phosphorylation of p53 is coupled to its ability to activate transcription of genes involved in control of cell cycle progression and apoptosis (6,7).
The estrogen, 17␤-estradiol (E 2 ), can stimulate proliferation, or induce differentiation or death depending on the cell context. In estrogen receptor (ER)-positive MCF-7 human breast cancer cells, E 2 strongly stimulates cell proliferation and does not induce cell death. In contrast, in several types of stably transfected cell lines expressing significant levels of ER, E 2 is highly toxic and induces cell death (8,9). The clinically important antiestrogen, tamoxifen, or its active metabolite, 4-hydroxytamoxifen (OHT), also induces cell death in ER-positive cell lines (10 -13). While most studies of ER, and of the other steroid/ nuclear receptors, have centered on their ability to act as ligand-regulated nuclear transcription factors, recent studies demonstrate that ER also exerts non-genomic effects (14 -17). Some non-genomic effects of ER may be related to its ability to activate signaling pathways, such as nitric-oxide synthase in endothelial cells (18) and the p21 ras /MAP kinase pathway (19).
We recently described several lines of HeLa cells stably transfected to express ER. These HeLa-ER cell lines are killed by E 2 or by OHT (9). In this work, we show that the specific inhibitor of the p38 MAP kinase pathway SB203580 (20) partially protects the HeLa-ER cells against E 2 or OHT-induced apoptosis. We show that E 2 and OHT induce the p38 pathway. To analyze the role of the p38 pathway in E 2 -or OHT-induced apoptosis, we used human breast cancer cell lines and cell lines selected for resistance to killing by E 2 or by OHT. In each of these cell lines, we observed an excellent correlation between the ability to activate the p38 pathway and the ability of E 2 or OHT to induce apoptosis. Activation of the p38 signal transduction pathway represents a novel ER-dependent activity of E 2 and OHT.

EXPERIMENTAL PROCEDURES
Cell Cultures-All HeLa-derived cell lines were grown in Dulbecco's modified Eagle's medium, supplemented with 10% charcoal-dextrantreated fetal bovine serum (CD-FBS). These cell lines include: HeLa, the HeLa-ER5 and HeLa-ER3 cell lines that were stably transfected to express hER␣ (9); T R HeLa-ER5, which is a HeLa-ER5-derived cell line selected for resistance to killing by OHT, and E R HeLa-ER5 cells, which is a HeLa-ER5-derived cell line selected for resistance to killing by E 2 . MCF-7 and MCF-10F cells were maintained in MEM supplemented with 5% FBS (containing 10 g/ml phenol red). Three days before transfection, MCF-7 and MCF-10F cells were transferred to phenol red-free Dulbecco's modified Eagle's medium/F-12 medium supplemented with 10% CD-FBS, and they remained in this medium until harvesting.
Plasmids-pFA-CHOP and pFR-luc were from Stratagene, La Jolla, CA. pRL-SV40 was from Promega, Madison, WI. 4ERE-luc was constructed by Dr. G. de Haan in this laboratory. MKK3(Ala) was a kind gift of Prof. R. Davis, University of Massachusetts, Worcester, MA.
Establishment of the Estrogen-resistant and OHT-resistant HeLa-ER5 Cell Lines-HeLa-ER5 cells were grown in Dulbecco's modified Eagle's medium/10% CD-FBS with 10 nM E 2 or 10 nM OHT, respectively. Colonies that grew out and were therefore resistant to cell death * This work was supported by National Institutes of Health Grant HD16720. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
were isolated and reseeded into 24-well culture plates in selection media. These cell lines, named E R HeLa-ER5 and T R HeLa-ER5, were further expanded to confluence in T-175 flasks. For long term growth, the cells were maintained under selection in medium containing 10 nM E 2 or 10 nM OHT, respectively.
Cell Viability Assays-The cells were treated with the specific p38 inhibitor SB203580 (20) or a specific MEK1 inhibitor PD098059 (Calbiochem, San Diego, CA) as described in the figure legends. Cell viability was assayed by exclusion of trypan blue.
Apoptosis Assays-DNA fragmentation assays were performed using the method described by McGahon et al. (21). Annexin V assays were performed using an apoptosis detection kit (R&D Systems, Minneapolis, MN). Briefly, one million cells were plated in T-25 flasks, treated for 12 h with the indicated concentrations of ligands, harvested, washed in cold phosphate-buffered saline, incubated for 15 min with fluoresceinconjugated annexin V and propidium iodide, and analyzed by flow cytometry.
Transient Transfections-Transient transfections of HeLa, HeLa-ER5, HeLa-ER3, T R HeLa-ER5, and E R HeLa-ER5 cells were performed using Tfx™-20 reagent (Promega, Madison, WI), using the manufacturer's protocol. Transient transfections of MCF-7 and MCF-10F cells were performed using Tfx™-50 reagent (Promega). Transfections were done in 12-well plates, containing the indicated amounts of pFA-CHOP, pFR-luc, or 4ERE-luc, MKK3(Ala), 2.5 ng of PRL-SV40 as internal standard, and PTZ18 U (or pcDNA3 if MKK3(Ala) was used) to bring the total amount of DNA to 600 ng in each well. The indicated concentrations of hormone were added immediately after transfection. After 36 or 48 h, the cells were harvested and dual luciferase assays (Promega) were performed using the manufacturer's protocol.
Western Blots-10 g of nuclear extract containing ER (for determining ER status) or 100 g of cell lysates (for p38 MAPK assays) was analyzed by electrophoresis on a 12.5% glycine-SDS-polyacrylamide gel, and the proteins were electroblotted onto a nitrocellulose membrane. The membrane was probed with ER-specific primary antibody for 1 h at room temperature (D547, generously provided by Dr. G. Greene, University of Chicago, Chicago, IL) or with antiphospho-p38 overnight at 4°C, incubated with horseradish peroxidase-conjugated secondary antibody (at 1:2000 dilution) and detected by chemiluminescence with an ECL™ kit (Amersham Pharmacia Biotech).
Electrophoretic Mobility Shift Assays-Electrophoretic mobility shift assays were carried out as we recently described (9).

SB203580 Protects Cells from E 2 -or OHT-induced Cell
Death-We recently described the isolation and characterization of several HeLa cell lines stably transfected to express human ER␣ (hER␣). We found that addition of E 2 or OHT to the culture medium blocks the growth of these ER-positive cells (9). Our subsequent studies showed that concentrations of E 2 or OHT as low as 10 Ϫ13 M inhibit growth of these cells (data not shown). To examine the effects of E 2 or OHT on the growth of HeLa-ER5 cells, we examined the time course of cell growth and of cell death when the cells were maintained in culture medium containing or lacking E 2 or OHT, or inhibitors of MAP kinase pathways. E 2 or OHT killed virtually all of the cells. The specific inhibitor of the ERK pathway, PD098059, had no effect on cell death mediated by E 2 or by OHT. In contrast, the specific inhibitor of the p38 MAP kinase pathway SB203580 partially protected against cell death induced by E 2 or by OHT (Fig. 1, A and B). Because we found ( Fig. 1B and data not shown), as have others (19), that high concentrations of SB203580 alone also induced cell death, we used a range of concentrations of SB203580. 1 M SB203580 largely protected the cells against cell death induced by E 2 or by OHT for 3 days (Fig. 1, A and B). The effects of lower (0.1 M) and higher (10 M) concentrations of SB203580 were different in the cells treated with E 2 or OHT, suggesting that there are differences in the induction of cell death by E 2 and by OHT. The observation that SB203580 partially protects HeLa-ER5 cells from cell death suggests that the p38 pathway is involved in the E 2 -or OHT-induced death of the cells. E 2 or OHT Induce Apoptosis in HeLa-ER5 Cells-While our data showed that E 2 or OHT induced death of the cells, it was unclear whether cell death was due to necrosis, or to apoptosis. To determine whether E 2 or OHT induced apoptosis, we treated the HeLa-ER5 cells with either 10 Ϫ7 M E 2 or 10 Ϫ7 M OHT. Genomic DNA was isolated from the treated cells and fractionated on an agarose gel. The DNA samples from the E 2 -or OHT-treated cells both exhibited the DNA laddering pattern, which is widely used as a marker for apoptosis ( Fig. 2A). We also performed cell cycle analyses and found that E 2 or OHT treatment resulted in G 1 /S arrest (data not shown). To more accurately quantitate the fraction of the cells undergoing apoptosis, we used the annexin V assay, which provides a way to determine the proportion of cells in the early stages of apoptosis, and the proportion of cells in the late stages of apoptosis, or undergoing necrosis. We found that treating the cells with E 2 or OHT increased the proportion of cells in the early stages of apoptosis by Ͼ10-fold. Perhaps because of the extended time of exposure to E 2 and OHT, only a fraction of the cells were in the early stages of apoptosis at a given time. Treating the cells with the specific inhibitor of the p38 signaling pathway, SB203580, partially suppressed E 2 -or OHT-induced cell death (Fig. 2B).
Apoptosis Is Related to p38 Activity in Vivo-To measure the activity of the p38 signaling pathway, we used a recently developed reporter gene system (Fig. 3A). This system is based on the fact that activated p38 directly phosphorylates the CHOP activation domain (CHOP-AD), enabling the CHOP-AD to activate transcription (20). In this system, the CHOP activation domain is fused to the Gal4 DNA binding domain (pFA-CHOP). p38 activity is measured by the ability of the Gal4 DNA binding domain-CHOP-AD to activate transcription of a luciferase reporter gene controlled by a synthetic promoter that contains five copies of the GAL4 DNA binding site (pFR-luc). We evaluated the level of CHOP activity in response to E 2 and OHT, and the pure antiestrogen ICI 182, 780. E 2 strongly induced CHOP activity (Fig. 3B), while OHT elicited a much smaller induction. The pure antiestrogen ICI 182,780 did not induce CHOP activity, and usually suppressed basal CHOP activity (Fig. 3B, and data not shown). This suggests that part of the basal activity of CHOP in the HeLa-ER5 cells may be due to the presence of traces of estrogen in the medium. The -fold induction of CHOP activity by E 2 and by OHT may therefore be somewhat greater than what we report using no added ligand as our base line. The inability of ICI 182,780 to induce CHOP activity is consistent with our previous observation that ICI 182,780 does not induce death of HeLa-ER cells, and actually stimulates their growth (9). Evidence that the reporter gene system measures p38 activity in the HeLa-ER5 cells is shown by the ability of the p38 pathway inhibitor SB203580 to block induction of luciferase activity by E 2 or by OHT. 1 M SB203580 largely blocked the induction of CHOP activity, and 10 M SB203580 completely blocked the induction (Fig. 3B).
To analyze the effect of E 2 and OHT on p38 activity in more detail, we carried out a series of cotransfections. To determine if the induction of p38 activity by E 2 or by OHT requires ER, we performed transfections using pFA-CHOP and pFR-luc in HeLa cells, which are ER-negative, and are not killed by E 2 or OHT (9). Neither E 2 nor OHT induced p38 activity in the ER-negative HeLa cells, indicating that ER is required for induction of p38 by E 2 and by OHT (Fig. 3C, HeLa). We then carried out a series of transfections in HeLa-ER5 cells using increasing amounts of pFA-CHOP transactivator and a fixed concentration of the pFR-luc reporter plasmid. E 2 and OHT each elicited a CHOP-dependent increase in reporter gene activity. At 1.25 ng of CHOP transactivator, 10 nM E 2 induced a 36-fold increase in luciferase activity and 10 nM OHT induced a 5-fold increase in activity (Fig. 3C). Since there is no induction of p38 in HeLa cells, which are not killed by E 2 or by OHT, and ICI 182,780, which does not kill the ER-positive HeLa-ER5 cells, does not induce p38, the stimulation of p38 activity by E 2 and by OHT is correlated with apoptosis induced by these ligands. However, OHT, which induces apoptosis at least as well as E 2 , is much less effective in inducing p38 activity. To be certain that the induction of p38 activity by E 2 and by OHT was not due to a unique property of the HeLa-ER5 cell line, we carried out similar experiments in a second ER-positive HeLa cell line, HeLa-ER3, which expresses less ER (9). The induction of p38 activity by E 2 or by OHT was similar in the HeLa-ER5 and HeLa-ER3 cells (see Table I below).
These studies of the activity of the p38 pathway employed a transactivation assay that is dependent on phosphorylation of the p38 MAP kinase target, CHOP. Since this is a relatively new assay, and the subcellular site of CHOP phosphorylation is not precisely defined, we also carried out Western blot analysis with an antibody that directly recognizes phosphorylated p38 MAP kinase. Phospho-p38 was undetectable at time 0 and was clearly induced by both E 2 and OHT at 4 h, with the induction persisting for at least 8 h. Consistent with the data obtained using the CHOP phosphorylation assay, E 2 was a much stronger inducer of p38 activity than OHT (Fig. 3D). Since Western blot analysis with antibodies to p38 MAP kinase indicated that levels of p38 MAP kinase were unchanged from 0 -16 h in cells treated with E 2 or with OHT (data not shown), the induction of phospho-p38 by E 2 and by OHT was not due to an increase in the level of p38 MAP kinase.
Effect of a Dominant Negative MKK3 on ER-dependent Induction of p38 -Activation of MKK3 and related proteins is often the first committed step in the p38 pathway (24). Cotransfection of HeLa cells with the dominant negative mutant MKK3(Ala) and the pFR-luc and pFA-CHOP reporter system decreased basal p38 activity approximately 2-fold (Fig. 4A). In contrast, cotransfection of the dominant negative MKK3(Ala) with pFR-luc and pFA-CHOP into HeLa-ER5 cells did not affect the ability of E 2 or OHT to induce p38 (Fig. 4B). Since MKK3(Ala) only suppressed p38 activity 2-fold in the ERnegative HeLa cells, and p38 activation by E 2 and by OHT may involve other members of the MKK family, MKK4 and MKK6 (25), a role for these MKK proteins in activation of p38 by E 2 or by OHT remains to be established.
p38 Induction in Cell Lines Selected for Resistance to Killing by E 2 or by OHT-To further analyze the induction of apoptosis by OHT and by E 2 , we isolated HeLa-ER5 cell lines resistant to killing by OHT or by E 2 , and analyzed the properties of their ERs and their ability to induce the p38 pathway and apoptosis. OHT has lost the ability to induce apoptosis (data not shown), in the cell line selected for resistance to killing by OHT (designated T R HeLa-ER5, and referred to as T R in Fig. 5). We analyzed the ability of OHT and of E 2 to induce p38 and the affect of MKK3(Ala) in the T R HeLa-ER5 cells (Fig. 5A, T R , left panel). OHT completely lost the ability to induce p38 in the T R HeLa-ER5 cells. When OHT was present, p38 activity in the T R HeLa-ER5 cells was at the same basal level as in the absence of ligand, and was reduced about 2-fold at 50 ng of cotransfected MKK3(Ala) (Fig. 5A, T R , left panel). However, E 2 retained full ability to induce p38 activity in the T R HeLa-ER5 cells and the induction was not blocked by MKK3(Ala) (Fig. 5A, T R , left panel). E 2 also retained full ability to induce apoptosis in the T R HeLa-ER5 cells (data not shown).
The cell line selected for resistance to killing by E 2 was designated E R HeLa-ER5, and is referred to as E R in Fig. 5. In these cells E 2 has lost the ability to induce the p38 pathway (Fig. 5A, E R , left panel) and to induce apoptosis (data not shown). In the E R HeLa-ER5 cell line, OHT retains the ability to induce both the p38 pathway (Fig. 5A, E R , left panel) and apoptosis (data not shown).
To determine whether resistance to E 2 or OHT was based on altered transactivation by ER, we analyzed the ability of the ER in these cell lines to activate transcription of a transfected estrogen response element (ERE)-containing reporter gene. The ER in the T R HeLa-ER5 cells retained the ability to activate the ERE-containing reporter gene, while the ER in the E R -HeLa-ER5 cells completely lost the ability to activate transcription of the ERE-containing reporter gene (Fig. 5A, right panel, E R ). Since it was possible that expression of ER protein was lost in the E R HeLa-ER5 cells, we carried out Western blot analysis on extracts from the T R HeLa-ER5 and the E R HeLa-ER5 cells. Levels of ER expression were similar in the two cell lines (Fig.  5B), indicating that the loss of ability to transactivate the ERE-containing reporter gene was not due to the loss of ER in  (1-101), driven by the cytomegalovirus promoter. The reporter plasmid, pFR-luc, contains the luciferase gene downstream of a basic TATA box joined to five copies of the GAL4 DNA binding site. In panels B and C when ligand was present, the cells were maintained in 10 nM E 2 , 10 nM OHT, or 10 nM ICI 182,780 for 48 h before harvesting. An equivalent volume of the ethanol vehicle was added to control (no ligand) wells. The -fold induction in panels B and C represents a normalized value with the ϩ EtOH control taken as 1, and the data represent the mean Ϯ S.E. for three separate experiments. B, the specific inhibitor of the p38 pathway, SB203580, inhibits induction of the CHOP reporter by E 2 or by OHT. HeLa-ER5 cells were transiently transfected with 2.5 ng of pFA-CHOP, 550 ng of pFR-luc reporter gene, and 2.5 ng of PRL-SV40 internal standard as described under "Experimental Procedures." The cells were maintained in the presence or absence of the indicated ligands and the indicated concentration (0, 1, or 10 M) of SB203580 for 48 h before harvesting. An equivalent volume of the ethanol vehicle was added to control wells. C, E 2 or OHT induces p38 activity in HeLa-ER5 cells. HeLa (HeLa) or HeLa-ER5 cells (ER5) were transiently transfected with the indicated amount of pFA-CHOP, 550 ng of pFR-luc reporter gene, and 2.5 ng of PRL-SV40 internal standard as described under "Experimental Procedures." D, time course of E 2 or OHT activation of p38 MAP kinase in HeLa-ER5 cells. p38 MAPK activation was assayed as described under "Experimental Procedures." Phospho-p38 levels were determined by Western blot analysis at the indicated times using anti-phospho-p38 MAPK antibody. the E R HeLa-ER5 cells.
To further analyze the functional activity of the ER in the T R HeLa-ER5 and the E R HeLa-ER5 cells, we prepared nuclear extracts from the two cell lines and examined the ability of the ER in the extracts to bind to a labeled ERE in an in vitro electrophoretic mobility shift assay (Fig. 5C). Consistent with its ability to transactivate an ERE-containing reporter gene in vivo, the ER in nuclear extracts from the T R HeLa-ER5 cells exhibited concentration-dependent binding to the ERE (Fig.  5C, T R NE). While the Western blot (Fig. 5B) showed that the ER is present in E R HeLa-ER5 cells, it had lost the ability to bind to the ERE (Fig. 5C, E R NE). The inability of the ER in the E R HeLa-ER5 cells to bind to the ERE makes it impossible for the ER in these cells to activate transcription of the EREcontaining reporter gene.
These data demonstrate a linkage between the ability of E 2 or OHT to induce apoptosis and their ability to induce the p38 pathway. The data also suggest that the there are significant differences in the induction of apoptosis by OHT and by E 2 . Although the p38 pathway was not examined, differences in apoptosis induced by OHT and by E 2 were also found in a study of apoptosis in human mammary epithelial cells stably transfected to express ER (26).
OHT, but Not E 2 , Induces p38 and Apoptosis in MCF-7 Human Breast Cancer Cells-MCF-7 human breast cancer cells have been widely used as a model for estrogen-dependent growth of breast cancer cells. E 2 stimulates the growth of MCF-7 cells and does not induce apoptosis (27). Interestingly, transient transfections with the CHOP reporter system show that E 2 fails to induce p38 activity in MCF-7 cells (Fig. 6). In contrast, in MCF-7 cells, OHT induces both p38 activity (Fig.  6A) and apoptosis (Fig. 6B). Consistent with the data from HeLa cells, OHT does not induce p38 in an ER-negative human mammary epithelial cell line, MCF-10F (Fig. 6A, MCF-10F), indicating that ER is required for the OHT induction of p38 activity.
Because of their large size and autofluorescence, quantitation of apoptosis in MCF-7 cells can be difficult. Using the annexin V assay system, we find that SB203580 largely protects MCF-7 cells from OHT-induced apoptosis (Fig. 6B). DISCUSSION E 2 and OHT Induce p38 Activity-While the ability of ER to activate transcription of nuclear genes has received the most attention, recent studies indicate that in some cell contexts E 2 can activate an ERK kinase signaling pathway (15,16,19). Under our conditions, addition of E 2 to the HeLa-ER culture medium resulted in a robust induction of the p38 signaling pathway, while OHT addition resulted in a much smaller, but readily detectable, induction of p38 activity. We observed induction of the p38 pathway both with a reporter gene assay based on p38 phosphorylation of the CHOP activation domain (Fig. 3, B and C), and by direct detection of phospho-p38 by Western blot analysis. Although precise quantitation is difficult with the Western blot assay for phospho-p38, E 2 clearly induces a much higher level of phospho-p38 than OHT (Fig.  3D). The specific inhibitor of the p38 pathway, SB203580, blocked the induction, providing additional evidence that E 2 and OHT induce p38 activity. These data provide the first evidence that estrogens or antiestrogens activate the p38 signaling pathway. In many contexts, activation of JNKs parallels TABLE I Induction of p38 activity and apoptosis is linked in ER-positive cell lines Summary data are for the several cell lines we analyzed, and are derived from the data presented in Figs. 2-7 and from data not shown. ϩϩϩϩ represents a maximal response. The symbols denote relative not absolute values, and ϩϩϩϩ does not represent twice as much apoptosis or p38 activity as ϩϩ.  activation of p38. Our preliminary data indicate that E 2 and OHT also induce JNK (data not shown). In contrast to the p38 pathway, a specific inhibitor of the JNK pathway was not available to us, preventing us from directly linking activation of the JNK pathway to induction of apoptosis by E 2 and by OHT. We therefore focused our studies on the p38 signaling pathway. E 2 and OHT Induction of p38 Activity May Involve Different Mechanisms-The close correlation between the induction of apoptosis by E 2 and by OHT, and the ability to induce p38, enabled us to select cell lines resistant to apoptosis, which clearly distinguish the different effects of E 2 and OHT on p38 activity. To simplify comparisons between the several cell lines we used, the properties of the cell lines we analyzed are summarized in Table I. In the E R HeLa-ER5 cell line, E 2 has lost the ability to induce p38 and apoptosis. The ER in E R HeLa-ER5 cells has also lost the ability to bind to the ERE in vitro and does not transactivate an ERE-containing reporter gene in the presence of E 2 . Since OHT retains the ability to induce the p38 pathway in these E R HeLa-ER5 cells, whose ER is non-functional for ERE binding and E 2 -dependent transactivation, EREdependent genomic effects of the OHT-ER complex are probably not important in the induction of p38 by OHT-ER. Additional evidence for distinct mechanisms of p38 induction by E 2 and by OHT comes from the T R HeLa-ER5 cells. In these cells, OHT has lost the ability to induce p38 and apoptosis, but the ER remains competent for ERE binding and E 2 -dependent transactivation, and for induction of p38 and apoptosis. The molecular basis for these intriguing differences in the interactions of OHT-ER and E 2 -ER with the p38 pathway is not yet known.

FIG. 4. A dominant negative MKK3 does not affect induction of p38 by E 2 or by OHT. HeLa cells (panel
The inability of the ER in the E R HeLa-ER5 cells to bind to FIG. 5. Analysis of cell lines selected for resistance to killing by OHT or by E 2 suggests a linkage between the induction of apoptosis and the induction of p38 activity. A, in the left panel, T R HeLa-ER5 or E R HeLa-ER5 cells (shown as T R and E R , respectively) were transiently transfected with 5 ng of pFA-CHOP, 550 ng of pFR-luc reporter gene, the indicated amount of MKK3(Ala), and 2.5 ng of PRL-SV40 internal standard. In the right panel, the cells were transfected with 550 ng of 4ERE-Luc reporter gene. Transfections were as described under "Experimental Procedures." When ligand was present, the cells were maintained in either 10 nM E 2 or 10 nM OHT for 48 h before harvesting. An equivalent volume of the ethanol vehicle was added to control (no ligand) wells. Luciferase activity represents a normalized value for each cell line, with no ligand and no MKK3(Ala) taken as 100%. Except for the 4ERE-Luc transfection of the T R cells, which is a single experiment, the data represent the mean Ϯ S.E. for three separate experiments. B, Western blot of 10 g of T R HeLa-ER5 (T R ) or E R HeLa-ER5 nuclear extract (E R ). The ER was visualized with the ER-specific monoclonal antibody D547. C, gel mobility shift assays of HeLa, T R HeLa-ER5, or E R HeLa-ER5 nuclear extracts (0.1 g, 0.2 g, 0.5 g, and 1 g). The same amount of ER (determined by Western blotting) was present in each of the T R HeLa-ER5 or E R HeLa-ER5 samples. the ERE raises interesting questions related to the activation of p38 by E 2 -ER. Although the x-ray structure of full-length ER has not yet been reported, the ER ligand binding domain (28) and DNA binding domain (9) can influence the structure of the AF2 activation domain in the ER hormone binding domain, and thereby alter the interaction of ER with coregulators. Mutations in the DNA binding domain and altered interaction of ER with the ERE could therefore potentially affect the ability of ER to activate the p38 pathway, either through direct proteinprotein interactions with the ER DNA binding domain or with other parts of the ER, or indirectly through altered ability of the ER to act at the DNA level.
OHT Induces the p38 Pathway in MCF-7 Cells-It has been proposed that tamoxifen can exert a variety of effects in breast cancer cells that lack ER (11,13). Here we showed that in ER-negative HeLa cells and MCF-10F cells, OHT does not induce the p38 pathway (Figs. 3C and 6A). These data indicate that while OHT induction of p38 activity does not require that the ER be competent for ERE-dependent transactivation, intracellular ER must be present. While OHT elicits a lower level of activation of P38 than E 2 , this activation is unlikely to be based on the weak agonist activity of tamoxifen, as ER binding to ERE-containing genes is unrelated to OHT induction of P38. The inability of ICI 182,780 to activate the p38 pathway may arise from the ability of this class of pure antagonists to interfere with cytoplasmic-nuclear shuttling of ER and to induce the rapid degradation of ER (27,29).
Induction of p38 Activity Is Linked to the Induction of Apoptosis by OHT and by E 2 -Several types of data, which are summarized in Table I, indicate that the induction of p38 activity by E 2 and by OHT is strongly coupled to their ability to induce apoptosis. (i) In HeLa-ER5 cells, E 2 and OHT each induce p38 and each induce apoptosis. The specific inhibitor of the p38 signaling pathway, SB203580, abolishes induction of p38 by E 2 and by OHT and greatly reduces their ability to induce apoptosis. (ii) In three different cell lines, E R HeLa-ER5, T R HeLa-ER5, and MCF-7, the ER ligand that retains the ability to induce the p38 pathway also retains the ability to induce apoptosis, and the ER ligand that has lost the ability to induce p38 loses the ability to induce apoptosis.
The concentration of SB203580 most effective in blocking apoptosis was different for OHT and for E 2 (Fig. 1, 1 M and 10 M, respectively), suggesting that other pathways may also play a role in E 2 and OHT-induced apoptosis. Our observation that OHT is at least as effective an inducer of apoptosis as E 2 suggests that there is not a strict one-to-one relationship between the extent of p38 induction and the ability of an ER ligand to induce apoptosis.
A Potential Role for Loss of Apoptotic Regulation in Breast Cancer Formation and Progression-It is widely accepted that cell transformation can be achieved either by enhanced levels or activities of proteins which allow cells to transit the cell cycle (such as mitogens and cytokines), or by reduced levels or activities of proteins, such as p53, which serve as cell cycle checkpoints and inducers of apoptosis (30 -32). Mutations leading to inactivation of the p53 tumor suppresser are the most common mutations in human cancers (30). The recent report that activation of p38 stimulates phosphorylation and activation of p53 (5) provides another potential site for growth control in breast cancer cells. Several recent studies (3,33) have proposed that the balance of activities between the ERK mitogen-activated signaling pathway and the stress-activated p38 pathway may be an important factor in the decision of cells to proliferate, remain quiescent, or undergo apoptosis. Consistent with this hypothesis is our finding that in MCF-7 cells, in which E 2 stimulates cell growth, E 2 does not induce p38.
Antiestrogen Efficacy in Preventing the Growth of Breast Cancer Cells May Be Related to the Ability to Induce Apoptosis-Most studies of tamoxifen, and of other antiestrogens, have centered on their ability to compete with estrogens for binding to the ER, and thereby antagonize the ability of estrogens to induce the conformational change in the ER, which makes it competent to activate transcription. Several recent studies have described additional activities of tamoxifen, including the ability to induce antioxidant enzymes (34) and apoptosis (10 -13). It was known that tamoxifen could induce apoptosis in some cells, and clinical applications of tamoxifen in breast cancer therapy often result in tumor regression (35). However, there was no biological pathway or system associated with this action of tamoxifen, and it was suggested that this activity of tamoxifen might not require ER (11). In this work we demonstrate that, in the presence of ER, the active metabolite of tamoxifen, 4-hydroxytamoxifen, induces the p38 pathway and apoptosis. The close coupling of apoptosis with induction of p38 activity suggests that it may be possible to carry out preliminary screening of large numbers of putative antiestrogens and selective estrogen receptor modulators using the induction of p38 activity as a surrogate marker for the ability to induce apoptosis. This screen would be based on the ability of antiestrogens to increase luciferase activity using endogenous and stably transfected ER-positive cell lines, and the easily automated, luciferase-based, CHOP reporter gene system we used.