¹Laboratoire de Biologie Moléculaire Eucaryote du CNRS, 118 route de Narbonne 31062, Toulouse cedex- France

²Departament de Biologia Cel.lular i Fisiologia, Facultat de Medicina, Universitat Autònoma de Barcelona, 08193 Bellaterra, Barcelona, Spain

³Unité Hormones et Cancer, U-148 INSERM, 60 rue de Navacelles 34090 Montpellier, France

^aAuthor for correspondence

We have compared the DNase I hypersensitivity of the regulatory region of two estrogen-regulated genes, pS2 and cathepsin D in hormone-dependent and -independent breast carcinoma cell lines. This strategy allowed the identification of two important control regions, one in pS2 and the other in cathepsin D genes. In the hormone-dependent MCF7 cell line, within the pS2 gene 5'-flanking region, we detected two major DNase I hypersensitive sites, induced by estrogens and/or IGFI: pS2-HS1, located in the proximal promoter and pS2-HS4, located -10.5 Kb from the CAP site, within a region that has not been cloned. The presence of these two DNase I hypersensitive sites correlates with pS2 expression. Interestingly in MCF7 cells, estrogens and IGFI induced indistinguishable chromatin structural changes over the pS2 regulatory region, suggesting that the two transduction-pathways converge to a unique chromatin target. In two cell lines that do not express pS2, MDA MB 231, a hormone-independent cell line that lacks the estrogen receptor , and HE5, a cell line derived from MDA MB 231 by transfection that expresses estrogen receptor , there was only one hormone-independent DNase I hypersensitive site. This site, pS2-HS2, was located immediately upstream of pS2-HS1. In MCF7 cells, two major DNase I hypersensitive sites were present in the 5'-flanking sequences of the cathepsin D gene, which is regulated by estrogens in these cells. These sites, catD-HS2 and catD-HS3, located at positions -2.3 Kb and -3.45 Kb, respectively, were both hormone-independent. A much weaker site, catD-HS1, covered the proximal promoter. In MDA MB 231 cells, that express cathepsin D constitutively, we detected an additional strong hormone-independent DNase I hypersensitive site, catD-HS4, located at position -4.3 Kb. This region might control the constitutive over-expression of cathepsin D in hormone-independent breast cancer cells. All together, these data demonstrate that a local reorganization of the chromatin structure over pS2 and cathepsin D promoters accompanies the establishment of the hormone-independent phenotype of the cells.

Breast cancer; chromatin; pS2; cathepsin D; estrogens; IGFI

Introduction

Proliferation of breast cancer cells is controlled by estrogens and by growth factors such as EGF or IGFI. A major problem in breast cancer biology concerns the changes in breast cancer cells that result in the progression from hormone dependence to hormone independence. This phenomenon limits the use of current therapy protocols, such as treatment with the non-steroid anti-estrogen, tamoxifen. While the mechanisms responsible for hormone dependence in human breast cancers are complex and in most cases unresolved, the hormonally dependent phenotype correlates with the presence of estrogen receptor (ER) in tumors. Although two closely related estrogen receptors, ER and ER have been described, for many cases the biological responses to estradiol are thought to be mediated through ER. In addition ER and ER are generally not co-expressed in breast cancer cell lines (Vladusic et al., 1998). Due to the presence of ER, approximately 60% of estrogen-receptor-positive human breast tumors respond to tamoxifen or other forms of endocrine therapies. Understanding the molecular mechanisms by which estrogens and anti-estrogens regulate the proliferation of hormone-dependent breast cancer is an essential prerequisite to understand how breast cancer develops resistance to endocrine therapies.

In hormone-dependent breast cancer cell lines, typified by MCF7 cells (that express ER), estrogens and growth factors stimulate transcription of pS2 and cathepsin D genes (Berry et al., 1989; Cavailles et al., 1989). In hormone-independent breast cancer cell lines such as MDA MB-231 (that do not express ER), pS2 is not expressed and cathepsin D is constitutively expressed. In the primary tumor of breast cancer patients these two markers have opposite significance, cathepsin D being associated with a poor prognosis (Rochefort, 1990), whereas pS2 is associated with a good prognosis (Foekens et al., 1993). One explanation for this difference is that cathepsin D, but not pS2, is over-expressed in ER negative breast cancers.

The 5' flanking region of pS2 gene contains a complex promoter/enhancer region responsive to estrogens, EGF, phorbol ester tumor-promoter, cHa-ras oncoprotein, and c-jun protein (Nunez et al., 1989) and a putative AP1 site (Gillesby et al., 1997). The estrogen responsive element (ERE) has been characterized (Berry et al., 1989). It has been proposed that methylation of CpGs located within the 5' flanking sequence of the pS2 gene is involved in the control of its expression (Martin et al., 1997). Recently, the high-resolution chromatin structure of the pS2 proximal promoter has been reported (Sewack and Hansen, 1997). As already described for other hormone-regulated promoters (Richard-Foy and Hager 1987; Carr and Richard-Foy, 1990), two nucleosomes are precisely positioned on the pS2 promoter, and hormone treatment, concomitantly with transcriptional induction, results in a change in chromatin structure of this region.

A mixed promoter controls the cathepsin D gene, and estrogens stimulate only the TATA-dependent transcription in breast cancer cells (Cavailles et al., 1993). The promoter region of the cathepsin D gene does not contain a canonical ERE palindrome (Augereau et al., 1994). To be active, the hormone-response element requires other elements located upstream and/or downstream of the non-consensus central ERE. The promoter contains several GC-rich boxes that can bind the transcription factor Sp1 and putative AP1, AP2 and PEA3 sites. The ERE/Sp1 sites are functional enhancer elements responsible for estrogen induced transcription in vivo (Krishnan et al., 1994).

Transcriptional activation of a gene by a steroid, via its cognate nuclear receptor, or by growth factors, via their tyrosine-kinase membrane receptors, both result in changes in composition and/or activity of large multi-protein complexe(s) containing the CREB binding protein (p300/CBP) (Montminy, 1997). CBP may stimulate the activity of target genes through its intrinsic histone acetylase activity (Bannister and Kouzarides, 1996) and through its association with functional RNA polymerase II complexes (Nakajima et al., 1997). This process is accompanied by changes in chromatin structure over the proximal promoter and regulatory regions. In eukaryotes, packaging of the DNA in chromatin plays a key role in regulating gene expression either during development and cell differentiation or in response to extracellular signals. Changes in chromatin structure within regions regulating gene expression result in either sequestration of cis-regulating elements, or conversely allow them to be accessible to regulatory factors. A convenient way to investigate such changes, and as a consequence to identify cis-regulatory elements, is to map the DNase I hypersensitive sites. We have analysed the chromatin organization of regulatory regions of two genes activated by estrogens, pS2 and cathepsin D in breast cancer cell lines dependent (MCF7) or not (MDA MB 231) on estrogens for their growth. Here we demonstrate differences in the chromatin structure of pS2 and cathepsin D promoters in these cell lines, which can be correlated with the changes in expression of the two genes.

Results

Measurement of pS2 and cathepsin D mRNA levels in MCF7, MDA MB 231 and HE5 cells

We compared first the effects of hormone and anti-hormone treatments on pS2 and cathepsin D mRNAs levels in the three cell lines, MCF7 (ER+), MDA MB 231 (ER-) and HE5 (MDA MB 231 cells, expressing ER). Figure 1 presents the results of a representative experiment in which mRNAs levels were quantified from a Northern blot. Cells were either untreated, or treated for 24 h with 10 nM estradiol (E2) or 100 nM of anti-estrogen ICI 182 780 (ICI). For pS2, in MDA MB 231 cells there was no detectable amount of mRNA, whatever the treatment was. In contrast in the ER+MCF7 cells, pS2 mRNA level was induced approximately threefold by E2 and ICI decreased the basal level of pS2 mRNA. In HE5 cells, although ER was present and functional (Touitou et al., 1991), pS2 mRNA remained undetectable, even after estrogen treatment. These results are in agreement with previously published data (Chalbos et al., 1993). For cathepsin D, in MDA MB 231 cells, neither E2, nor the anti-estrogen increased significantly in the mRNA level. In MCF7 cells, E2-treatment resulted in a twofold induction of cathepsin D mRNA and ICI decreased its basal level of expression. In HE5 cells, contrasting with pS2 gene, expression of ER restored the estrogen regulation of cathepsin D mRNA expression (threefold induction). As in MCF7 cells, ICI decreased its basal level of expression. These results indicate that re-expression of ER in ER- cells was not sufficient to restore the expression of a gene that was extinguished (i.e. pS2). In contrast it allowed re-establishing the hormone-dependent expression of a gene that was constitutively expressed in the ER- cells.

Chromatin structure analysis of the regulatory regions of pS2 and cathepsin D genes in MCF7, MDA MB 231 and HE5 cells

We have analysed the chromatin structure of the regulatory regions of pS2 and cathepsin D genes in the three cell lines. Cells were either treated for 2 h with 10 nM E2, or untreated. Nuclei were isolated and digested with the indicated amounts of DNase I. DNAs were purified and cut with XbaI (pS2 gene) or NcoI (cathepsin D gene). The fragments were separated by agarose gel electrophoresis and the resulting blots were hybridized with either pS2 or cathepsin D probes.

Figure 2a shows the pS2 gene DNase I hypersensitive sites in the three cell lines. XbaI cleavage generated the band present in all lanes on top of the blot.

In MCF7 cells there were three DNase I hypersensitive sites. Hormone treatment induced two strong DNase I hypersensitive sites, pS2-HS1 and pS2-HS4 (compare the intensity of the bands in lanes 6, 12, 18 and 7, 13, 19, respectively). The third site (pS2-HS3) was weak and present in both control and estrogen-treated cells (compare lanes 18 and 19). Results are summarized in Figure 2b. The pS2-HS1 region spans from approximately -0.4 Kb up to almost the transcription start site. It includes the ERE (-393/-405); pS2-HS4 is located at positon -10.5 Kb in a region that has not been cloned yet. There is no information available on the role of this region on the control of pS2 gene expression. Both sites are hormone-dependent. The pS2-HS3 region is centered on position -1.2 Kb. In MDA MB 231, a unique DNase I hypersensitive site (pS2-HS2), located around position -0.5 Kb, a region upstream of the ERE that contains repeated sequences in its most distal part (Figure 2b), was present. Contrasting with pS2-HS1 and pS2-HS4, its intensity was unchanged whether or not the cells were treated with E2 (Figure 2a, compare lanes 10, 11 and 16, 17). In HE5, although ER was present and functional, the DNase I hypersensitivity pattern was identical to that in MDA MB 231 cells (lanes 8, 14, 20 and 9, 15, 21).

All together, the data are consistent with the lack of expression of pS2 in both MDA MB 231 and HE5 cells and with its hormone-induction in MCF7 cells.

Figure 3a shows the cathepsin D gene DNase I hypersensitive sites in the three cell lines. NcoI cleavage generated the 6.7 Kb band, present in all lanes on top of the blot. In MCF7 there are three DNase I hypersensitive sites (lanes 10, 11, 16, 17, 20, 21). CatD-HS1 was only present in the samples treated with the highest DNase I concentrations (lanes 20, 21). In all the experiments the signal was weak making it difficult to determine if it is increased upon E2-treatment. It covers the cathepsin D proximal promoter. It is centered at -0.2 Kb and includes the non-canonical hormone response elements of the promoter (Figure 3b). CatD-HS2 and catD-HS3 sites were unaffected by E2 treatment. They are located at -2.3 Kb and -3.5 Kb from the transcription start site, respectively, upstream of a large AT-rich region (Figure 3b). In MDA MB 231 cells catD-HS1 is also very weak and detected only in samples digested with the highest DNase I concentrations. CatD-HS2 and catD-HS3 were present along with an additional DNase I hypersensitive site, catD-HS4, located at position -4.3 Kb from the transcription start site (Figure 3b). In HE5 cells the DNase I hypersensitivity pattern was identical to that in MDA MB 231 cells. Comparison of DNase I hypersensitivity of the cathepsin D gene in MDA MB 231 and HE5 cells demonstrates that re-expression of ER has no significant effect on the DNase I hypersensitivity pattern of the cathepsin D gene. This indicates that regulation of the cathepsin D promoter by estrogens in cell line HE5 is probably achieved through an interaction of the receptor with its target on the DNA within a region whose chromatin structure is already opened. Such a region could be involved in the constitutive over-expression of cathepsin D in the hormone-independent cells.

Influence of anti-estrogen treatment on chromatin structure of the regulatory regions of pS2 and cathepsin D genes in MCF7 cells

Proliferation of MCF7 cells is activated by estradiol and inhibited by antiestrogens. In MCF7 cells, pS2 transcription is activated by estradiol but not by anti-estrogens. As shown here this activation is accompanied by localized changes of the chromatin structure over pS2 regulatory regions. We have investigated the effect of two anti-estrogens, ICI and hydroxy tamoxifen (OH-Tam) on the DNase I hypersensitive sites located in the 5'-flanking regions of pS2 (Figure 4). The top two panels (E2) show the effect of estradiol treatment. As expected, the amount of the XbaI fragment decreased upon hormone-treatment and concomitantly the intensity of the hormone-dependent sites pS2-HS1 and pS2-HS4 was increased. The hormone independent site pS2-HS3 was present in untreated and E2-treated cells. The four bottom panels (ICI and OH-Tam) show the effect of anti-hormone treatment. The DNase I hypersensitivity patterns of pS2 gene in MCF7 cells untreated or treated with antiestrogens were undistinguishable. These results demonstrate that in MCF7 cells, the anti-estrogens alone do not induce changes in DNase I hypersensitivity of the pS2 promoter. The chromatin structure of the cathepsin D gene, that does not change upon treatment with estrogens was unaffected by anti-estrogen treatment (not shown).

The same experiment was performed in MDA MB 231 and HE5 cells. In these two cell lines, as expected for hormone-independent DNase I hypersensitive sites, anti-estrogen treatment had no influence on the DNase I hypersensitivity pattern of pS2 and cathepsin D genes (not shown).

Influence of insulin like growth factor I on chromatin structure of the regulatory regions of pS2 and cathepsin D genes in MCF7 cells

The growth of metastatic human breast cancer cell lines containing ER is also stimulated by growth factors. Activation of pS2 and cathepsin D transcription by growth factors has been reported (Cavailles et al., 1989; Chalbos et al., 1993). To investigate the effect of growth factors on chromatin structure it is necessary to grow the cells in serum-deprived medium to avoid uncontrolled effects due to the presence of growth factors in the serum.

We investigated the influence of the serum withdrawal on the DNase I hypersensitivity pattern of the two genes in MCF7 cells. Figure 5 shows the results obtained for the pS2 gene. The two hormone-induced DNase I hypersensitive sites (pS2-HS1 and pS2-HS4) were present in the estrogen treated cells grown in the presence of serum (lane 9) or in a serum deprived medium (lane 7). This demonstrates that serum withdrawal does not prevent the hormone-induced chromatin structural changes that occur within the pS2 gene regulatory regions. For the cathepsin D gene neither estrogen treatment (as shown in Figure 3), nor the amount of serum in the tissue culture medium resulted in a change in its DNase I hypersensitivity pattern (not shown).

The effect of growth factor on transcription of pS2 and cathepsin D genes in MCF7 cells has been previously described (Chalbos et al., 1993). Since the degree of response of MCF7 cells may vary, depending on passage number and tissue culture conditions, we evaluated the respective effects of E2 and insulin like growth factor I (IGFI) on regulation of the pS2 mRNA levels, in our tissue culture conditions (Figure 6). We showed that both IGFI and E2 induced the pS2 gene expression 3 - 4-fold. OH-TAM alone was not able to induce the pS2 gene expression and along with IGFI it antagonized the growth factor effect. In the experiment reported here, induction of pS2 mRNA level in cells treated simultaneously with estradiol and IGFI was 12-fold. This value is higher than that expected for an additive process (sevenfold induction). In this experiment OH-Tam, although it antagonized IGFI effect, was unable to antagonize the estradiol effect. This was clearly visible for samples treated either with estradiol and OH-Tam, in which pS2 mRNA level was induced fourfold, or with estradiol, OH-Tam and IGFI, in which pS2 mRNA level was induced eightfold. Since these results were comparable to those previously described in MCF7 cells (Chalbos et al., 1992), we did not investigate the effect of the different treatments on cathepsin D gene expression.

The effect of IGFI treatment on DNase I hypersensitivity of the pS2 and cathepsin D genes 5'-flanking sequences in MCF7 cells, grown in serum-deprived medium was then investigated. Figure 7a presents the results obtained for the pS2 gene. The two hormone-dependent sites pS2-HS1 and pS2-HS4 are clearly visible in the lanes corresponding to cells treated with either E2 (lanes 7, 11) or IGFI (lanes 8, 12) or both (lanes 9, 13). In this experiment, pS2-HS3 signal is again weak and the site present in all lanes. The modest variations in its intensity result from slight variations in the degree of digestion and amount of material loaded in the lanes. The hormone-induced change in the intensity of the band corresponding to pS2-HS4 was not clear as in the experiment presented in Figure 2a. Quantification of the bands using a phosphor-imager confirmed the increased DNase I hypersensitivity in this region. This is not clearly visible on the autoradiogram, due to the strong signal leading to a saturation of the film. This experiment demonstrates that treatment of the cells with IGFI results in a chromatin remodeling over the regulatory regions of the pS2 gene indistinguishable from that induced by estrogens. Although IGFI and E2 displayed a synergetic effect on pS2 mRNA induction, the intensity of the bands, corresponding to both pS2-HS1 and pS2-HS4 were similar whatever the treatment was. DNase I hypersensitivity of the cathepsin D gene 5'-flanking sequences in IGF-I treated MCF7 cells grown in serum-deprived medium are shown in Figure 7b. CatD-HS1 was almost undetectable because of a digestion with DNase I less extensive here than in the experiment presented in Figure 3. CatD-HS2 and catD-HS3 were present, whatever the treatment of the cells was. The intensity of the bands corresponding to these DNase I hypersensitive sites was identical for all samples. This indicates that neither hormone nor growth factor induction of cathepsin D mRNA transcription requires the opening of a new chromatin region.

Discussion

We have analysed the chromatin structure of the regulatory regions of two estrogen-regulated genes, pS2 and cathepsin D, in human breast tumor cell lines, which are either estrogen-dependent, or estrogen-independent for growth.

In MCF7 cells we identified two hormone-dependent DNase I hypersensitive sites in the pS2 gene promoter, located over the proximal promoter (pS2-HS1) and at -10.5 Kb from the CAP site (pS2-HS4). PS2-HS1 covers a region whose structure has been recently analysed in detail at a high resolution (Sewack and Hansen, 1997). This region contains the ERE and is assembled in two positioned nucleosomes in both MCF7 and MDA MB 231 cells. The high-resolution analysis revealed only weak changes over that region, without nucleosome displacement, contrasting with the clear chromatin reorganization, unambiguously demonstrated by the strong induction of pS2-HS1, observed at a low resolution. This might reflect the fact the chromatin structural changes result in overall increased nuclease accessibility over an approximately 200-bp region, not detectable by genomic footprinting, rather than from dramatic changes over a few nucleotides. Our study revealed the presence of another region that might be important in controlling pS2 gene expression since its chromatin structure is altered concomitantly with transcriptional activation. It is located at -10.5 Kb from the CAP site, within a region that has not yet been cloned. The isolation of this region is in progress. The role of chromatin structure in controlling hormone-induced expression of pS2 gene in MCF7 cells appears similar to what has been described for other steroid-regulated genes (Richard-Foy and Hager, 1987; Carr and Richard-Foy, 1990). In MDA MB 231 cells that do not express pS2 and that lack ER, the two hormone-induced DNase I hypersensitive sites were missing. Instead there was a constitutive DNase I hypersensitive site (pS2-HS2) located immediately upstream of pS2-HS1, in a region containing an AP1 site (Gillesby et al., 1997). Elevated levels of AP1 binding activity accompany progression of MCF7 cells towards anti-estrogen-resistant phenotype (Dumont et al., 1996). PS2-HS2 could result either from an increased non-productive binding of AP1 in this region, or from the binding of a repressor not yet identified. In HE5 cells that are MDA MB 231 cells that express a functional ER (Touitou et al., 1991), the DNase I hypersensitivity profile of pS2 regulatory regions was indistinguishable from the one in MDA MB 231. This observation is in agreement with the lack of expression of the pS2 gene in HE5 cells. The absence of the two hormone-dependent DNase I hypersensitive sites in both MDA MB 231 and HE5 cells demonstrates that pS2 gene extinction is accompanied with changes in chromatin organization of the pS2 gene. This indicates that although the estrogen receptor is necessary, ER is not able by itself to promote the chromatin structural changes that accompany pS2 transcription. In the cell, anti-estrogen-occupied ER binds estrogen response DNA (Reese and Katzenellenbogen, 1992); treatment of MCF7 cells with either OH-Tam or ICI did not induce pS2-HS1 and pS2-HS4. This demonstrates that binding of the estrogen receptor to its DNA target is not sufficient to promote chromatin structural changes. The formation of a transcriptionally competent conformation of the transactivation domain AF2 of ER (Brzozowski et al., 1997) is probably part of the mechanism involved in chromatin remodeling over DNase I hypersensitive sites.

In MCF7 cells, treatment with either IGFI or estradiol induced changes in pS2 gene chromatin structure over the same region (sites pS2-HS1 and pS2-HS4). This parallels the transcriptional activation of pS2 by IGFI. Simultaneous treatment of the cells with IGFI and E2 did not increase the intensity of the hypersensitive sites although the two compounds had a synergetic or at least additive effect on RNA synthesis (Chalbos et al., 1993). This indicates that both factors are able to promote a similar chromatin remodeling over pS2-HS1 and pS2-HS4, but once the remodeling is achieved, transcriptional transactivation could occur through distinct pathways that can cooperate. The cross talk between the two pathways may occur at different levels. IGFI may activate ER through phosphorylation of serine 118 by the MAPK kinase pathway (Kato et al., 1995). As a result, the interaction of the estrogen receptor with transcriptional coactivators might be changed. Although it cannot be totally excluded, a role of ER in this process seems unlikely. More recent studies have demonstrated the existence of a co-activator network, allowing a cross-talk between different signal transduction pathways (Montminy, 1997). One major link is CBP/p300 (CREB-Binding protein) that is a component of a larger complex critical for integration of several signal transduction pathways. CBP is associated with steroid-receptor co-activators (SRCs), which can also recognize ligand-bound nuclear receptors (Torchia et al., 1997). The increased accessibility of the chromatin structure of the pS2 gene after IGFI and estradiol treatment, as revealed by the presence of pS2-HS1 and pS2-HS4, may be due to the recruitment of the partners from different signal transduction pathways in the co-activators complex.

Contrasting with the pS2 gene, the cathepsin D gene is expressed in the three cell lines. The major difference is that in ER+ cells cathepsin D is estrogen-induced while in ER- cells it is expressed constitutively. In the three cell lines two strong and one much weaker DNase I hypersensitive sites lie within cathepsin D regulatory regions. These sites, catD-HS3, catD-HS2 and catD-HS1 are located at -3.45 Kb, -2.3 Kb, and over the proximal promoter, respectively. These regions might contain the targets for regulatory factors involved in hormonal control of the transcription, as it was shown for the proximal promoter (Augereau et al., 1994; Krishnan et al., 1994). An analysis at a high resolution of protein/DNA interaction within these regions will be the next step and should provide new insights on the regulation of cathepsin D gene expression. The most interesting observation is the presence of a site (catD-HS4), located at -4.2 Kb from the CAP in the hormone-independent cell lines HE5 and MDA MB 231 cells, that is absent in the hormone-dependent MCF7 cells line. This region could be a target for factors that modulate chromatin structure, whose expression or activity is altered during the establishment of the hormone-independent phenotype. The cloning of this region is in progress.

It has been established that factors other than ER are involved in the progression of the cell towards hormone-independent growth (Garcia et al., 1992; Levenson and Jordan, 1994). Such factors could also play a critical role in modulating chromatin structure. Our results demonstrate that, in breast cancer cells, the acquisition of the hormone-independent growth phenotype could be accompanied, in addition to the loss of ER, by alterations of the chromatin remodeling machinery. Conversely, acquisition of a constitutively high level of expression may result from the loss of a transcriptional repressor. Whatever the type of factor involved, we demonstrate here, that for cathepsin D gene, high level of expression is concomitant with the establishment of a new constitutive strong DNase I hypersensitive site. This links high level of expression with a local chromatin remodeling.

The analysis of the chromatin structure of pS2 and cathepsin D genes in hormone-dependent and independent breast cancer cells allowed us to identify potentially important sequences within pS2 and cathepsin D promoters. These sequences had not been identified previously, and their functional significance in breast cancer might be revealed only in the context of a chromatin structure. In addition we demonstrate that two signal transduction pathways, involved in transcriptional activation of pS2 in MCF7 cells remodel chromatin over the same regions, suggesting that chromatin remodeling takes place after the cross talk between the pathways.

Materials and methods

Cell lines and tissue culture

MDA MB 231 cells (estrogen-independent for growth and estrogen receptor negative (ER-)) were donated by R Cailleau (MD Anderson Center, Houston, TX, USA). MCF7 cells (estrogen-dependent for growth and estrogen receptor positive (ER+)) were a gift from M Lippman (Georgetown University, Washington, DC, USA). Both cell lines were isolated from metastatic pleural effusions of human breast carcinomas (Cailleau et al., 1974; Soule et al., 1973). HE5 is, a stable cell line, derived from MDA MB 231 cells, in which a vector expressing the HEG0 (Tora et al., 1989) estrogen receptor (ER) was established (Touitou et al., 1991).

MDA MB 231 cells and HE5 cells were grown in Dulbecco's modified Eagle's medium containing 4.5 g/l glucose and MCF7 cells in DMEM/F12. Media contained 50 g/ml gentamicine and were supplemented with 2 mM L-glutamine and 10% (v/v) of heat inactivated fetal calf serum (Life technologies). Cells were grown at 37°C in a humidified atmosphere containing 5% CO₂. To study the effect of estrogens, cells were switched to media without phenol red, containing serum stripped of endogenous steroids by three successive incubations with dextran-coated charcoal. After at least 5 days of withdrawal, cells were treated or not with 10 nM estradiol (E2), 100 nM ICI 182 780 (ICI) (Zeneca Pharmaceuticals), 100 nM hydroxy-Tamoxifen (OH-Tam) (Zeneca Pharmaceuticals) for the indicated times. To study the effect of IGFI (Interchim), cells were switched to a medium without phenol red. The cells were grown for 2 days in the presence of 10% charcoal-stripped fetal calf serum. They were then progressively deprived in serum by growing them for 3 days in the presence of 3% serum, and then in 1% serum (Vignon et al., 1987). Cells were treated with 5 nM IGFI for the indicated times, 8 days after steroid withdrawal.

DNase I hypersensitivity studies

After 2 h of treatments, approximately 2´10⁶ subconfluent cells were scraped into phosphate-buffered saline solution (PBS), and the nuclei were isolated as described previously (Richard-Foy et al., 1987). Nuclei were diluted to an OD₂₆₀ of 10 - 20 in 10 mM Tris-HCl pH 7.4, 15 mM NaCl, 60 mM KCl, 0.15 mM Spermine, 0.5 mM Spermidine, adjusted to 1 mM MgCl₂, 0.5 mM CaCl₂ and held on ice. The reaction was initiated by addition of the enzyme (Worthington-Cooper 2000 - 2600 U/mg) and the tubes incubated at 37°C. After 10 min the reaction was stopped by addition of 1 vol of 25 mM EDTA, 2% SDS, 200 g/ml proteinase K.

DNA purification and Southern blot analysis

DNA was purified by two phenol extractions, three chloroform/isoamyl alcohol (24 : 1) extractions, ethanol precipitated and resuspended in TE (10 mM Tris-HCl pH 7.4, 1 mM EDTA) buffer. For Southern blot analysis, 50 g of purified DNA from each sample was digested to completion by the appropriate restriction enzymes under conditions specified by the supplier. Non-radioactive -DNA, cut with HindIII (5 ng/sample) was added to each sample and served as an internal size marker. The samples were run through a 1% agarose gel in TEA buffer (40 mM Tris acetate, 1 mM EDTA), transferred to 0.2 m Nytran membranes (Schleicher and Shuell) in 20´SSC (1´=0.15 M NaCl, 15 mM sodium citrate, pH 7.5) and cross-linked by UV treatment (Stratalinker, Stratagene). Membranes were prehybridized in 5´SSC, 5´Denhardt's solution, 0.1% SDS, 0.1 mg/ml herring sperm DNA for 2 h at 68°C and hybridized overnight at the same temperature and in the same medium containing the ³²P-labeled DNA probe. The pS2 probe was an exon 1 fragment (+46 to +403) amplified by PCR using as primers oligonucleotide 1: 5'-ACCATGGAGAACAAGGTGATC and oligonucleotide 2: 5'-TTGGGAGGATTGTATAGTCTT. The cathepsin D probe was an intron 1 fragment (+361 to +802) amplified using as primers oligonucleotide 1: 5'-TCTCCCCCATATGCCACCCTG and oligonucleotide 2: 5'-CATCCCATCCAGGCCCAATCG. They were labeled by repeated primer extension (5 ng and 25 ng of DNA fragment template for pS2 and cathepsin D, respectively) in 100 l containing 20 pmoles of oligonucleotide 2 as a single primer, 10 M each dATP, dGTP, dTTP and 100 Ci [³²P]dCTP (3000 Ci/mmol) and three units of Taq polymerase (Promega). Amplification was performed as follows: 30 cycles (denaturation for 1 min at 95°C, annealing for 2 min at 56°C and extension, for 3 min at 75°C for pS2 probe, and denaturation for 1 min at 95°C, annealing for 2 min at 63°C and extension, for 3 min at 75°C for cathepsin D probe). Membranes were washed five times at 65°C in 0.2´SSC, 0.2% SDS, analysed with the phosphorimager and autoradiographed. To dehybridize the probe, membranes were boiled for 5 min in 0.1% SDS. They were rehybridized with radioactive random primed--DNA cut with HindIII. The size of the bands was determined by comparison with the electrophoretic mobility of the internal standard markers.

Extraction of mRNA and Northern blot analysis

The mRNAs were isolated from subconfluent cells treated for 24 h with hormone and/or antihormone and/or growth factor, using the polyATract kit (Promega). Samples (approximately 10 g RNA) were run through 1% agarose gel in MOPS buffer (20 mM MOPS, 1 mM EDTA, 1 mM sodium acetate, pH 7.0) in presence of 2.2 M formaldehyde. The mRNAs were transferred on 0.2 m Nytran membranes (Schleicher and Shuell) in 10´SSC, cross-linked by UV treatment and hybridized with the radioactive probes as described for the Southern blots. The cathepsin D probe was an exon 2 fragment (+2456 to +2606) and the 36B4 probe was a 600 bp fragment corresponding to the PstI - PstI cDNA fragment. They were ³²P-labeled by random priming. The pS2 fragment used as a probe was amplified by PCR using the following primers: Oligonucleotide 1: 5'-GTGAATTTTAGACACTTCTGC and oligonucleotide 2: 5'-CTCTTTTAATTTTTAGGCCAA. It was labeled by repeated primer extension, as described for Southern blots, using 7.5 ng of the DNA fragment as template and oligonucleotide 2 as primer. Thirty extension cycles (denaturation for 1 min at 95°C, annealing for 2 min at 60°C and extension, for 3 min at 75°C) were performed. Membrane hybridization and washes were performed as described for Southern blot. The bands were quantified using a Phosphorimager (Fuji PC-Bas) and the Pcbas 20 software.

DNA sequencing

We have sequenced 790 bp and 3351 bp of 5'-flanking sequences for pS2 and cathepsin D genes, respectively. The regions are indicated in Figures 2b and 3b. Sequences have been deposited in GenBank. Accession numbers are AF048725 (pS2) and AF048726 (cathepsin D).

Acknowledgements

We thank MC Rio and P Chambon for pS2 probes. We are grateful to JC Faye for stimulating discussions and to Y Henry for critically reading the manuscript. CG was the recipient of fellowships from Association pour la Recherche sur le Cancer and Fondation pour la Recherche Médicale and MS was supported by a fellowship from Comissionat per a Universitats i Recerca. Generalitat de Catalunya. This work was partially supported by the Association pour la Recherche sur le Cancer, the Ligue Contre le Cancer, the Conseil de Region Midi Pyrénées and the European Economic Community, Biomed 2 contract PL95-0181.

Augereau P, Miralles F, Cavailles V, Gaudelet C, Parker M and Rochefort H. (1994). Mol. Endocrinol. 8, 693-703. MEDLINE

Bannister J and Kouzarides T. (1996). Nature 384, 641-643. MEDLINE

Berry M, Nunez AM and Chambon P. (1989). Proc. Natl. Acad. Sci. USA 86, 1218-1222. MEDLINE

Brzozowski AM, Pike AC, Dauter Z, Hubbard RE, Bonn T, Engstrom O, Ohman L, Greene GL, Gustafsson JA and Carlquist M. (1997). Nature 389, 753-758. Article MEDLINE

Cailleau R, Young R, Olive M and Reeves Jr WJ. (1974). J. Natl. Cancer Inst. 53, 661-674. MEDLINE

Carr KD and Richard-Foy H. (1990). Proc. Natl. Acad. Sci. USA 87, 9300-9304. MEDLINE

Cavailles V, Augereau P and Rochefort H. (1993). Proc. Natl. Acad. Sci. USA 90, 203-207. MEDLINE

Cavailles V, Garcia M and Rochefort H. (1989). Mol. Endocrinol. 3, 552-558. MEDLINE

Chalbos D, Joyeux C, Galtier F and Rochefort H. (1992). J. Steroid Biochem. Mol. Biol. 43, 223-228. MEDLINE

Chalbos D, Philips A, Galtier F and Rochefort H. (1993). Endocrinology 133, 571-576. MEDLINE

Dumont JA, Bitonti AJ, Wallace CD, Baumann RJ, Cashman EA and Cross-Doersen DE. (1996). Cell Growth Differ. 7, 351-359. MEDLINE

Foekens JA, van Putten WL, Portengen H, de Koning HY, Thirion B, Alexieva-Figusch J and Klijn JG. (1993). J. Clin. Oncol. 11, 899-908. MEDLINE

Garcia M, Derocq D, Freiss G and Rochefort H. (1992). Proc. Natl. Acad. Sci. USA 89, 11538-11542. MEDLINE

Gillesby BE, Stanostefano M, Porter W, Safe S, Wu ZF and Zacharewski TR. (1997). Biochemistry 36, 6080-6089. MEDLINE

Kato S, Endoh H, Masuhiro Y, Kitamoto T, Uchiyama S, Sasaki H, Masushige S, Gotoh Y, Nishida E, Kawashima H, Metzger D and Chambon P. (1995). Science 270, 1491-1494. MEDLINE

Krishnan V, Wang X and Safe S. (1994). J. Biol. Chem. 269, 15912-15917. MEDLINE

Levenson AS and Jordan VC. (1994). J. Steroid Biochem. Mol. Biol. 51, 229-239. MEDLINE

Martin V, Ribieras S, Song-Wang XG, Lasne Y, Frappart L, Rio MC and Dante R. (1997). J. Cell. Biochem. 65, 95-106. Article MEDLINE

Montminy M. (1997). Nature 387: , 654-655. MEDLINE

Nakajima T, Uchida C, Anderson SF, Lee CG, Hurwitz J, Parvin JD and Montminy M. (1997). Cell 90, 1107-1112. MEDLINE

Nunez AM, Berry M, Imler JL and Chambon P. (1989). EMBO J. 8, 823-829. MEDLINE

Reese JC and Katzenellenbogen BS. (1992). Mol. Cell. Biol. 12, 4531-4538. MEDLINE

Richard-Foy H and Hager GL. (1987). EMBO J. 6, 2321-2328. MEDLINE

Richard-Foy H, Sistare FD, Riegel AT, Simons Jr SS and Hager GL. (1987). Mol. Endocrinol. 1, 659-665. MEDLINE

Rochefort H. (1990). Breast Cancer Res. Treat. 16, 3-13. MEDLINE

Sewack GF and Hansen U. (1997). J. Biol. Chem. 272, 31118-31129. Article MEDLINE

Soule HD, Vazguez J, Long A, Albert S and Brennan M. (1973). J. Natl. Cancer Inst. 51, 1409-1416. MEDLINE

Tora L, Mullick A, Metzger D, Ponglikitmongkol M, Park I and Chambon P. (1989). EMBO J. 8, 1981-1986. MEDLINE

Torchia J, Rose DW, Inostroza J, Kamei Y, Westin S, Glass CK and Rosenfeld MG. (1997). Nature 387, 677-684. Article MEDLINE

Touitou I, Vignon F, Cavailles V and Rochefort H. (1991). J. Steroid Biochem. Mol. Biol. 40, 231-237. MEDLINE

Vignon F, Bouton MM and Rochefort H. (1987). Biochem. Biophys. Res. Commun. 146, 1502-1508. MEDLINE

Vladusic EA, Hornby AE, Guerra-Vladusic FK and Lupu R. (1998). Cancer Res. 58, 210-214. MEDLINE

Figure 1 Effect of estradiol and anti-estrogen on pS2 and cathepsin D mRNA levels in ER+ and ER- breast cancer cells. MCF7, MDA MB 231 and HE5 cells were withdrawn from hormone as described in Materials and methods. Cells were incubated for 24 h with 10 nM 17- estradiol (E2), 100 nM ICI 182 780 (ICI) or vehicle (C). The mRNAs were extracted and 10 g was separated by gel electrophoresis, blotted onto membranes and hybridized with 36B4, pS2 and cathepsin D probes. The bands corresponding to the three genes were quantified. Results were normalized using 36B4 as a standard and the mRNA induction fold was calculated

Figure 2 DNase I hypersensitive sites analysis of pS2 5'-flanking region in ER+ and ER- breast cancer cells. Nuclei isolated from MCF7 (MCF), MDA MB 231 (MDA) and HE5 cells treated for 2 h with (+) or without (-) 10 nM 17- estradiol (E2) were digested with the indicated amounts of DNase I. Purified DNA (50 g) was digested with XbaI. The fragments resulting from DNase I digestion were characterized by Southern blot analysis (a) using pS2 exon 1 probe (dashed arrow on the diagram on the right). The positions of the hypersensitive sites are labeled with arrows. (b) top shows a summary of the DNase I hypersensitive sites in the different cell lines (HD: hormone-dependent site, C: constitutive site). (b) middle and bottom, presents a map of the pS2 gene regions studied and a summary of the results. Middle: the dashed line represents uncloned regions. Continued line represents the cloned region. Solid bar is the region that has been sequenced. Closed boxes mark the location of repeated sequences. Open boxes mark the exons. Arrows indicate the positions of the DNase I hypersensitive sites (HS). Bottom: detailed view of the ERE region. Diamond indicates the ERE position. Hatched boxes indicate the positions of the hypersensitive sites

Figure 3 DNase I hypersensitive sites analysis of cathepsin D 5' region in ER+ and ER- breast cancer cells. Nuclei isolated from MCF7 (MCF), MDA MB 231 (MDA) and HE5 cells treated for 2 h with (+) or without (-) 10 nM 17- estradiol (E2) were digested with the indicated amounts of DNase I. Purified DNA (50 g) was digested with NcoI. The fragments resulting from DNase I digestion were characterized by Southern blot analysis (a) using cathepsin D intron 1 probe (dashed arrow on the diagram on the right). The positions of the DNase I hypersensitive sites are labeled with arrows. (b) shows a map of the cathepsin D gene regions studied and a summary of the results. Labeling is as in Figure 2b

Figure 4 Effect of anti-estrogens on pS2 DNase I hypersensi-tivity in MCF7 cells. Nuclei isolated from MCF7 cells untreated for 2 h with 10 nM 17- estradiol (E2), 100 nM ICI 182 780 (ICI), 100 nM hydroxy-Tamoxifen (OH-Tam) were digested with the indicated amounts of DNase I. DNA was analysed by Southern blotting as described in Figure 2. In each lane the radioactivity profile was determined with a phosphorimager. Dotted line: untreated cells, continued line: cells treated with E2, ICI or OH-Tam, as indicated in the panel. Closed boxes indicate the positions of the DNase I hypersensitive sites

Figure 5 Effect of serum withdrawal on DNase I hypersensitivity of the pS2 gene in MCF7 cells. MCF7 cells were grown for 8 days in presence of 10% serum (serum+) or progressively deprived in serum (-), as described in Materials and methods. Nuclei isolated from cells treated for 2 h with 10 nM 17- estradiol (E2+), or vehicle (E2-) were digested with 0.5 g/ml DNase I. DNA was analysed by Southern blotting as described in Figure 2

Figure 6 Effect of OH-Tam and IGFI on pS2 mRNA level in MCF7 cells. MCF7 cells were progressively deprived in serum. Cells were treated for 24 h with 10 nM 17- estradiol (E2), 100 nM hydroxy-Tamoxifen (OH-Tam), 5 nM IGFI, alone or in the indicated combinations. (a) Northern blot. (b) pS2 mRNA quantification, as described in Figure 1

Figure 7 >Effect of IGFI on DNase I hypersensitivity of pS2 and cathepsin D genes in MCF7 cells. MCF7 cells were progressively deprived in serum as described in Materials and methods. Nuclei were isolated from cells untreated or treated for 2 h with, either 10 nM 17- estradiol (E2) or 5 nM IGFI (IGFI), or both and digested with the indicated amounts of DNase I. DNA was analysed by Southern blotting as described in Figure 2 for pS2 or in Figure 3 for cathepsin D. (a) pS2 gene; (b) cathepsin D gene