Estrogen receptor function can drive cyclin D1 expression and proliferation in human breast cancer cells (MCF-7). Recent studies showing that estrogen receptor-positive epithelial cells in the human mammary gland are nonproliferative suggest that the direct mitogenic effect of estrogen on mammary epithelial cells may be acquired during breast cancer development. Because estrogen-dependent cyclin D1 expression has been linked to its mitogenicity, we characterized the ability of estrogen to regulate cyclin D1 expression in estrogen receptor-negative breast cancer cells (MDA-MB-231) and nontransformed human keratinocytes (HaCaT) stably expressing the estrogen receptor. In both cases, estrogen receptor function did not induce cyclin D1 expression. Although MCF-7 cells respond to estrogen by inducing the AP-1 family components c-Fos and c-Jun, HaCaT cells expressing estrogen receptor do not. These results may explain the lack of estrogen-dependent cyclin D1 expression and proliferation in cells ectopically expressing the estrogen receptor. Therefore, estrogen receptor function alone is not sufficient for estrogen-dependent cyclin D1 expression and proliferation. Other transcriptional cofactors that allow estrogen receptor to induce expression of AP-1 may be required for estrogen to act as a mitogen.
Estrogen is an essential hormone that controls the normal physiology of the mammary gland and breast cancer development. To determine how estrogen regulates growth of breast cancer cells, we and others have characterized the effects of estrogen on the cell cycle of MCF-7 breast cancer cells (1, 2, 3, 4). Such studies have indicated that the induction of cyclin D1 by estrogen may be a key to understanding estrogen-dependent proliferation. This estrogen-dependent expression of cyclin D1 is essential for estrogen-induced proliferation of MCF-7 cells (5) and is the earliest estrogen-mediated effect on the cell cycle machinery (3, 4). Ectopic expression of cyclin D1 in MCF-7 cells mimics estrogen effects on the cell cycle (6). In addition, inducible overexpression of cyclin D1 in these cells reverses the growth-inhibitory effects of antiestrogen (7). Together, these results suggest that the ability of estrogen to drive cyclin D1 expression is crucial for the proliferation of ER3-positive breast tumors.
The precise molecular mechanism by which estrogen and its receptor control cyclin D1 expression is at present poorly defined. Cyclin D1 does not represent a classical ER target gene, because the cyclin D1 promoter lacks an ERE. Altucci et al. (1) mapped the estrogen-responsive region to a fragment between −944 and −136 of the cyclin D1 promoter. Several potential binding sites for known transcription factors can be found in this region of the promoter, such as a site for the AP-1 transcription factor. This suggested the possibility that estrogen regulates cyclin D1 expression through modulation of AP-1 activity. However, no conclusive evidence exists at present to validate this hypothesis. It is unclear whether the presence of ER is sufficient to confer estrogen-mediated cyclin D1 expression.
Several investigators have stably introduced ER into ER-negative cells and have demonstrated estrogen-dependent expression of classical ERE-containing genes such as cathepsin D and transforming growth factor-α (8, 9, 10). However, estrogen did not stimulate proliferation of these cells, and in many cases, estrogen caused growth inhibition (10, 11). In addition, a recent histopathological study demonstrated that normal ER-positive cells in the human mammary gland in vivo are nonproliferating, whereas ER-positive breast cancer cells are actively proliferating (12). These two observations together suggest that the presence of ER per se is not sufficient for estrogen-induced proliferation of the ER-positive cell.
A possible explanation for the lack of estrogen-stimulated proliferation in cells ectopically expressing ER is that estrogen is unable to induce cyclin D1 expression in these cells. To test this hypothesis, we determined whether MDA-MB-231 and HaCaT cells engineered to express ER can up-regulate cyclin D1 in response to estrogen. Our results suggest that expression of ER alone is not sufficient to confer estrogen-inducible cyclin D1 expression. The absence of cyclin D1 induction by estrogen and the ER may be related to the inability of this receptor to regulate the expression of components of the AP-1 transcription factor complex.
Materials and Methods
Cell Culture, DNA, and Transfection.
S30 cells were generously provided by Dr. V. Craig Jordan (Northwestern University Medical School, Chicago, IL). The plasmid pCMV-ER was a gift of Dr. Myles Brown (Dana-Farber Cancer Institute, Boston, MA). The ERE-driven luciferase constructs ERE-SV40-luc and ERE2-109-A3-luc were kindly given to us by Dr. Barry Gehm (Northwestern University Medical School, Chicago, IL). ER-expressing clones were obtained after calcium phosphate-mediated transfection of MDA-MB-231 and HaCaT cells with pCMV-ER and selection with G-418. Transient transfections of HaCaT cells were performed using Tfx-50 reagent (Promega Corp., Madison, WI), following the manufacturer’s guidelines.
Antibodies and Western Analysis.
Preparation of cell extracts and Western blot analysis was carried out as described previously (3). Monoclonal antibody against cyclin D1 (HD-45) was a gift from Ed Harlow (Massachusetts General Hospital, Charlestown, MA). Rabbit polyclonal antibodies used to detect ER (SC-543), c-Fos (SC-52), and c-Jun (SC-44) were obtained from Santa Cruz Biotechnology (Santa Cruz, CA).
Cyclin D1 Regulation in ER-containing MDA-MB-231 Cells.
Several studies have shown that estrogen responsiveness of target genes can be obtained by stably expressing ER in ER-negative cells (8, 9, 10). However, estrogen does not have the ability to induce proliferation of these ER-containing cells. We wanted to understand this phenomenon in greater detail by determining whether the lack of proliferation was attributable to the inability of estrogen to induce cyclin D1. For this purpose, we used S30 cells, a derivative of MDA-MB-231, an ER-negative breast cancer cell line, which stably express ER and show estrogen-dependent expression of the progesterone receptor (10).
We wished to determine the ability of the ectopically expressed ER to induce cyclin D1 protein in the ER-expressing S30 cells. For this analysis, we first reduced the basal level of cyclin D1 protein in S30 cells by culturing these cells in phenol red-free medium containing 0.1% CSS. In this way, we also minimized exposure to residual ER-activating agents. After 48 h of serum and estrogen deprivation, S30 cells and MCF-7 cells received either 5 nm 17β-estradiol or 5% FBS as a positive control. Another set of dishes remained untreated to determine the basal levels of cyclin D1. All of the samples were harvested after 5 h of treatment and processed for Western blot analysis.
Estrogen was unable to induce cyclin D1 expression significantly, whereas exposure of these cells to FBS resulted in strong induction of cyclin D1 (Fig. 1,A, top). In contrast, in MCF-7 cells, both estrogen and FBS were capable of strongly inducing cyclin D1 expression (Fig. 1 A, top). This suggested that the presence of the ER in mammary carcinoma cells, although necessary for the estrogen-mediated stimulation of cyclin D1 synthesis in MCF-7 cells, was not sufficient in other cell types.
To confirm that the ectopically expressed ER was functional in our assay, the same Western blot was analyzed for changes in ER protein levels after estrogen treatment. Estrogen can down-regulate expression of its own receptor in MCF-7 cells or in other cells when ER is ectopically expressed (13, 14). This decrease in ER occurs at the protein and mRNA levels and depends on a functional ER. Indeed, addition of estrogen to serum-starved MCF-7 cells caused down-regulation of ER protein expression (Fig. 1,A, bottom), as has been reported previously (13). In a similar way, S30 cells were able to modulate ER expression after estrogen addition (Fig. 1 A, bottom), indicating that the ER is functional in these experiments.
To rule out the possibility that the results obtained were specific to the S30 cell clone, we generated a new series of ER-expressing MDA-MB-231 cells. Eighteen cell clones were isolated after transfection of MDA-MB-231 with pCMV-ER. Twelve of the clones obtained, termed 231-ER, expressed detectable levels of ER as assayed by Western immunoblot analysis (Fig. 1,B). MCF-7 cells (left lane) were used as positive control. Four clones that displayed different levels of ER were further characterized to assess ER function and estrogen-dependent cyclin D1 expression. The results obtained with one of these clones (231-ER-8) are shown in Fig. 1,C. Expression of ER protein was down-regulated to 30% of the initial level after 2.5 h of estrogen addition to estrogen-deprived 231-ER-8 cells (Fig. 1,C, top). This indicated that ER was functional in these cells. However, cyclin D1 expression was not affected when estrogen was given to serum-starved 231-ER-8 cells (Fig. 1 C, bottom). The other three clones analyzed were also not able to induce cyclin D1 expression after treatment with estrogen (data not shown). We concluded that the inability of estrogen to up-regulate cyclin D1 levels may be due to some specific features of these breast cancer cells (MDA-MB-231) that are extraneous to the ER itself.
Cyclin D1 Regulation in ER-containing HaCaT Cells.
One explanation for the inability of the ectopic ER to regulate cyclin D1 expression in the MDA-MB-231 cells is that these cells are derived from tumors, and that some of the changes that they have undergone during tumor progression prevent the ER from interacting functionally with the cyclin D1 promoter. To address this possibility, we expressed the ER in a quite distinct type of human epithelial cell, the HaCaT nontransformed human keratinocyte cell.
Thirty-eight HaCaT clones stably transfected with the pCMV-ER vector were obtained and analyzed for expression of ER. Twelve of these clones expressed ER at levels comparable with those seen in MCF-7 cells. Eight of them, termed HaCaT-ER clones, expressed functional ER as determined by the ability of estrogen to down-regulate its own receptor (data not shown).
To obtain a more quantitative estimate of ER function in these HaCaT-ER cells, we also determined their ability to mediate estrogen-dependent transcription. For this purpose, we measured estrogen-dependent luciferase expression after transient transfection with two distinct promoter constructs that contained either one (ERE-SV40-luc) or two copies (ERE2-109-A3-luc) of the ERE (15). Fig. 2 A shows the results obtained with the cell clone (HaCaT-ER-38) that gave consistently the highest estrogen-dependent transcriptional induction. The amount of induction obtained (5-fold with ERE-SV40-luc and 10-fold with ERE2-109-A3-luc) is comparable with the response seen with MCF-7 cells (data not shown), thereby confirming the presence of functional ER in HaCaT-ER-38 cells.
We then proceeded to evaluate the ability of the ectopically expressed ER in HaCaT-ER-38 cells to activate cyclin D1 gene expression. To do so, we starved these cells for 48 h to reduce the basal level of cyclin D1 expression. Similar to what we observed with 231-ER cells, estrogen addition was unable to up-regulate cyclin D1 expression, whereas 5% FBS was able to induce cyclin D1 levels dramatically (Fig. 2 B). Comparable analyses were carried out with eight other ER-expressing HaCaT cell clones, and none of these was able to up-regulate cyclin D1 expression in response to added estrogen (data not shown). Thus, as was seen previously with the MDA-MB-231 cells, the presence of a functional ER did not suffice to allow estrogen to induce expression of cyclin D1.
Changes in AP-1 Components after Estrogen Treatment.
These data suggested that the regulation of cyclin D1 expression by the ER is complex and may involve the mediation of other proteins that convey signals between the ER and the cyclin D1 promoter. Such proteins might hypothetically be absent or functionally inactive in the MDA-MB-231 cells and the HaCaT cells, explaining the inability of the ER to activate cyclin D1 expression in these cells.
As mentioned earlier, an attractive candidate for regulating cyclin D1 expression is the AP-1 transcription factor. Estrogen can affect AP-1 either by regulating the synthesis of AP-1 family members (16, 17, 18) or by modulating its transcriptional activity (19). Moreover, both the Fos and Jun proteins, two common component subunits of the AP-1 factor, are known to be involved in the regulation of cyclin D1 gene expression (20, 21). For these reasons, we decided to evaluate whether there was a differential regulation of these genes by estrogen in MCF-7 and HaCaT-ER cells.
To approach this question, we tested whether estrogen could modulate expression of c-Fos and c-Jun levels in either MCF-7 or HaCaT-ER-38 cells. Asynchronously growing MCF-7 and HaCaT-ER-38 cells were serum-starved in 0.1% CSS for 48 h. Cells were treated with either 5 nm 17β -estradiol or 5% FBS and harvested for Western analysis at 1 or 3 h. To begin, we determined the expression of the cyclin D1 protein. Cyclin D1 levels were induced in MCF-7 cells after 3 h of either estrogen or FBS treatment. As before, we observed that although estrogen and serum each were able to induce cyclin D1 expression in MCF-7 cells, only serum succeeded in doing so in HaCaT-ER-38 cells (Fig. 3, top).
The same Western blot used to detect cyclin D1 above was subsequently probed for the presence of Fos. The antibody used by us is specific for c-Fos and does not detect FosB, Fra-1, or Fra-2. Because of changes in phosphorylation, c-Fos is present as several distinctly migrating electrophoretic species (22). In MCF-7 cells, the levels of most of these isoforms increased by 3–4-fold after addition of either estrogen or FBS (Fig. 3, middle). HaCaT-ER-38 cells showed high basal levels of Fos expression when compared with MCF-7, and no induction of Fos occurred after addition of estrogen. However, FBS was able to augment Fos expression in these cells, suggesting that the lack of response to estrogen is not due to the presence of a c-fos gene in these cells that is refractory to further induction. Thus, the inability of the ER to induce cyclin D1 in HaCaT-ER-38 cells was paralleled by its inability to induce Fos synthesis.
A very similar outcome was noticed when c-Jun expression was evaluated (Fig. 3, bottom). This molecule is also subject to modification and can be seen as a series of bands. In the absence of serum, MCF-7 cells expressed low levels of the Jun protein. However, addition of estrogen or FBS to serum-starved MCF-7 cells caused a transient induction of Jun protein that parallels the changes detected for Fos. Again, as was the case with Fos, HaCaT-ER-38 cells had high basal levels of Jun that could be superinduced by FBS but not by estrogen. If anything, estrogen caused a decrease in Jun expression. Thus, the ability of either estrogen or FBS to induce cyclin D1 correlated well with their ability to increase Fos and Jun expression. Because induction of AP-1 members occurs within 1 h of addition of estrogen, before any cyclin D1 is observed, this suggests that induction of these genes anticipates and may be required for the proper regulation of cyclin D1 synthesis. Together, these results suggest that the lack of cyclin D1 induction by ectopically expressed ER may be explained by its inability to modulate expression of the c-fos and c-jun genes.
An elucidation of the molecular mechanism of cyclin D1 regulation by the ER is of central importance to our understanding of the molecular pathogenesis of human breast cancers. If estrogen-dependent cyclin D1 expression is necessary for tumor growth, antiestrogens may be able to inhibit cancer growth by preventing estrogen-mediated cyclin D1 expression. Thus, development of antiestrogen resistance may be associated with changes in hormonal regulation of cyclin D1. It is plausible that alterations in the expression or function of specific molecules will allow cyclin D1 expression in the presence of tamoxifen. It is also possible that the ability of estrogen to induce cyclin D1 and proliferation is acquired as a pathological trait during the course of breast cancer development. For this reason, it is important to determine the mechanisms by which cancer cells display ER-dependent cyclin D1 transcription.
In the studies presented here, we have investigated the ability of ER to drive expression of cyclin D1 after its stable introduction into two distinct, previously ER-negative human cell types. Our aim was to evaluate whether the inability of ectopically expressed ER to confer estrogen-dependent mitogenesis was related to the lack of cyclin D1 induction in ER-negative cells forced to express ER. Our findings indicate that in all of the clones ectopically expressing ER, cyclin D1 expression cannot be augmented by addition of estrogen. Nevertheless, these clones were competent to down-regulate ER expression or drive transcription of an ERE-luciferase construct after addition of estrogen. Moreover, we found that the inability of ER to drive cyclin D1 may be related to the lack of induction of AP-1 components by estrogen. Consequently, these data support the idea that estrogen may control proliferation of breast cancer cells by its ability to induce AP-1 members and cyclin D1.
Cyclin D1 as a Mediator of ER-stimulated Proliferation.
The primary cell cycle target of estrogen in MCF-7 cells appears to be cyclin D1. This is supported by the fact that inducible expression of cyclin D1 overcomes growth-arrest mediated by antiestrogens (7). More recently, an extension of this earlier work demonstrated that high levels of cyclin D1 expression led to p21 redistribution, cyclin E-cdk2 activation, and retinoblastoma hyperphosphorylation of the pRB, the retinoblastoma protein, in antiestrogen-arrested MCF-7 cells (6). Thus, the increases of cyclin D1 protein by estrogen may be necessary and sufficient for proliferation of breast cancer cells.
In breast cancer tumors, the expression of cyclin D1 has been correlated with the expression of ER (23). The amplification of cyclin D1 occurs preferentially in ER-positive tumors, and the levels of cyclin D1 parallel in many cases the levels of ER present in the tumors (23). These observations suggest that the functional connections between the ER and cyclin D1 observed in MCF-7 cells in vitro correctly model what is seen in breast carcinomas in vivo. If this is the case, why does ER expression in ER-negative cells not confer estrogen sensitivity to the cyclin D1 gene?
In many cases where ER has been stably introduced into ER-negative cells, different endogenous genes can be turned on, depending on the cell type used. One recent report, which appeared while our studies were ongoing, analyzed cyclin D1 expression in cells engineered to ectopically express ER (24). These authors analyzed cell proliferation and cyclin D1 after estrogen treatment of ER-transfected MCF-10AEwt5. Similar to our results, they could not detect any increases in cyclin D1 expression after estrogen treatment of the transfected cells. They did not determine whether ER was transcriptionally active in MCF-10AEwt5 cells. However, they found that ER from MCF-10AEwt5 had altered ligand-binding affinity when compared with ER derived from MCF-7 cells by sucrose gradient sedimentation.
One reason for the inability of estrogen to drive cyclin D1 in ER-transfected cells could be that the cyclin D1 locus is not responsive to transcriptional activation in these cells. However, the ability of FBS to rapidly yield increases in cyclin D1 levels argues against this possibility. Thus, the lack of estrogen-mediated cyclin D1 expression is most likely due to the presence or absence of other factors that mediate estrogen-dependent cyclin D1 transcription.
The recent reports that normal estrogen receptor-positive human mammary epithelial cells are nonproliferative suggest that estrogen is unlikely to act as a direct mitogen of normal ER-positive cells in the mammary gland (12, 25). Accordingly, the ability of estrogen to act as a direct mitogen and induce expression of cyclin D1 would appear to represent a pathological aberration acquired during the process of breast cancer progression. This aberration in signaling may occur early during tumor progression, because overexpression of cyclin D1 mRNA has been observed at early stages of breast cancer (26).
AP-1 Members as Potential Mediators of Estrogen-dependent Cyclin D1 Expression.
In the search for possible molecular intermediaries between estrogen and cyclin D1 transcription, components of the AP-1 transcription factors emerge as highly attractive candidates. Studies of cells from mice bearing germ-line inactivations of the c-jun, c-fos, or c-fosB genes have revealed that c-Jun and either c-Fos or FosB are necessary for normal transcription of cyclin D1 (20, 21). Moreover, previous mapping of an estrogen-responsive region in the cyclin D1 promoter identified a region of the promoter that contained an AP-1 site but no ERE elements (1). Together, these observations suggest that the ability of estrogen to modulate AP-1 activity may be required for cyclin D1 expression and therefore proliferation.
Several possible scenarios can be considered to explain how estrogen interacts with AP-1: (a) the ER protein, like other steroid receptors, may bind physically to the AP-1 factor to activate transcription from AP-1 sites; or (b) alternatively, estrogen may act indirectly to drive synthesis of the component subunits of the AP-1 factor, which then proceed to assemble and drive the transcriptional activation of the cyclin D1 gene locus (27); or (c) a third scenario could be that both mechanisms may operate in MCF-7 cells. Our results revealed that there is a differential regulation by estrogen in the synthesis of AP-1 members between MCF-7 cells and HaCaT-ER cells. There is a clear induction of c-Jun and c-Fos by estrogen in MCF-7 but not in HaCaT-ER. However, both gene products are induced by FBS, indicating that, as was the case with the cyclin D1 promoter, the defect in estrogen signaling is not due to promoter silencing. This suggests that the lack of cyclin D1 induction by estrogen may be traced to the inability of estrogen to modulate c-Jun and c-Fos expression in HaCaT-ER cells.
The mechanisms by which estrogen promotes synthesis of AP-1 components in MCF-7 cells are not clear. Several investigators have tried to map estrogen-inducible sequences in the c-fos gene (28, 29, 30) or c-jun gene (31). These studies have led to the identification of a variety of candidate regulatory sequences in the promoters of these genes. For example, a recent study using MCF-7 was able to map the estrogen-responsive region to an imperfect Sp1-binding site (28). The authors claim that induction of c-fos expression depends on the formation of a transcriptionally active ER/Sp1 complex. Thus, the inability of estrogen to drive cyclin D1 expression in HaCaT-ER cells may be linked to a deficient interaction between ER and Sp1 in these cells.
In summary, we propose that the ability of ER to drive ERE-mediated transcription can be dissociated from its role as a mitogen. The competence of estrogen to drive proliferation may be linked to the regulation of cyclin D1 transcription, which in turn may be mediated by the actions of the AP-1 transcription factor. This points to the importance of determining how estrogen modulates AP-1 function and what the specific changes in breast cancer cells are that allow estrogen to affect AP-1, cyclin D1, and proliferation. Understanding these mechanisms may provide clues on how ER-positive breast tumors develop and how they become refractory to antiestrogen treatment.
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.
This work was supported by Grant DAMD 17-96-1-6285 (to R. A. W.).
The abbreviations used are: ER, estrogen receptor; ERE, estrogen receptor element, CSS, charcoal-stripped fetal bovine serum; FBS, fetal bovine serum.
We are grateful to Drs. Barry Gehm and Myles Brown for providing ERE-luciferase constructs and pCMV-ER, respectively. We also thank Dr. Craig Jordan for providing S30 cells and Dr. Ed Harlow for antibodies. We thank Brian Elenbaas and Lisa Spirio for critical reading of the manuscript.