Regulation by Estrogen through the 5-Flanking Region of the Transforming Growth Factor a Gene

Toshiaki Saeki, Audrey Cristiano, Mark J. Lynch, Michael Brattain, Nancy Kim, Nicola Normanno, Nicholas Kenney, Fortunato Ciardiello*, and David S. Salomon Tumor Growth Factor Section (T.S., A.C., N.K., N.K., N.N., F.C., D.S.S) Laboratory of Tumor Immunology and Biology National Cancer Institute National Institutes of Health Bethesda, Maryland 20892 Department of Cellular and Molecular Biology (M.J.L., M.B.) Bristol-Myers Squibb Company Bristol-Myers Pharmaceutical Research and Development Division Wallingford, Connecticut 06492-7660

Expression of transforming growth factor a (TGFa) mRNA and protein can be stimulated by estrogens such as 17/9-estradiol (E2) in estrogen-responsive rodent and human breast cancer cells. To ascertain if E2 can directly regulate TGFa expression through the 5'-flanking region of the human TGFa gene, E2responsive MCF-7 or ZR-75-1 human breast cancer cells or E2-nonresponsive MDA-MB-231 breast cancer cells were transiently transfected with a plasmid containing an 1140-base pair (bp) Sac-I fragment of the TGFa 5'-flanking region ligated to the chloramphenicol acetyltransferase (CAT) gene. Cells that were transfected and subsequently treated with physiological concentrations of E2 (10~11-10~8 M) for 24 h exhibited a 2- to 10-fold increase in CAT activity. The E2 stimulation of CAT activity was dose-dependent with an increase first found at 10~10 M E2. The increase in CAT activity could be detected within 24-36 h after the addition of E2. There was no significant change in CAT activity in transiently transfected MDA-MB-231 cells as mediated through the TGFa 5'-flanking region after E2 treatment. MCF7 cells were also transiently transfected with different fragments of the TGFa 5'-flanking region ligated to the luciferase gene. In the absence of E2 treatment, no detectable luciferase activity was found. E2 was able to stimulate, in a dose-dependent manner, a 30- to 300-fold increase in luciferase activity in MCF-7 cells, which have been transfected with either a 2813-bp PsM, a 1565-bp Spe-I, a 1140-bp Sac-I, or a 370-bp Bam-HI TGFa-luciferase fragment. However, a loss in E2 responsiveness occurred when MCF-7 cells were transfected with a 77-bp Sac-ll

TGFa-luciferase plasmid. The induction of luciferase activity by E2 could be effectively blocked by simultaneous treatment of the cells with the antiestrogens tamoxifen or droloxifene. In addition, the increase in luciferase activity was specific for E2 since progesterone or dexamethasone were ineffective in modifying luciferase activity through the TGFa 5-flanking sequence. The results show that the TGFa 5'-flanking region contains a potential estrogen-responsive element(s) and that this region is upstream of the Sac-ll restriction site. (Molecular Endocrinology 5: 1955-1963, 1991)

INTRODUCTION Transforming growth factor a (TGFa) is a peptide that exhibits structural and functional homology to epidermal growth factor (EGF) and that is able to bind to the EGF receptor (1). TGFa is initially synthesized as a high molecular weight, membrane-associated precursor that is biologically active (2). Processing of the precursor to the mature, low molecular weight TGFa peptide may differ between normal and malignant tissues (3). TGFa and the EGF receptor are coexpressed in many human tumors and tumor cell lines, suggesting that TGFa may be functioning through an autocrine mechanism to regulate the growth of tumor cells (1, 4, 5). In fact, synthesis of high levels of TGFa using expression vectors that contain a TGFa cDNA in established cell lines, such as in rodent fibroblasts and in mouse and human mammary epithelial cells that possess a sufficient threshhold level of functional EGF receptors, can lead to transformation of these cells in vitro and in some cases to tumorigenicity (6-9). Moreover, TGFa transgenic mice

0888-8809/91 /1955-1963$03.00/0 Molecular Endocrinology Copyright © 1991 by The Endocrine Society

1955

The Endocrine Society. Downloaded from press.endocrine.org by [${individualUser.displayName}] on 22 November 2015. at 06:15 For personal use only. No other uses without permission. . All rights reserved.

Vol5No. 12

MOL ENDO-1991 1956

have been generated. Founder animals that express the transgene develop epithelial hyperplasia, premalignant lesions, and frank carcinomas in the colon, liver, and mammary gland, which demonstrates that TGFa can function as an oncogene in several tissues when inappropriately expressed in vivo (10-12). TGFa is also produced by several adult tissues including the skin (13), colon (14), liver (15), mammary gland (16, 17), ovary (18), and brain (19), suggesting that this growth factor may be involved in regulating certain aspects of normal growth and differentiation. This may be possible because in the mouse mammary gland exogenous TGFa can stimulate lobuloalveolar development in vivo, whereas endogenous TGFa may function as an autocrine growth factor (16, 17, 20, 21). In addition, overexpression of TGFa in mouse HC11 mammary epithelial cells blocks differentiation in response to lactogenic hormones in that induction of (S-casein fails to occur (22). Regulation of TGFa expression is complex because a variety of disparate agents including retroviral oncogenes such as ras (9, 22, 23), phorbol esters such as 12-O-tetradecanoylphorbol-13-acetate (TPA) (24, 25), sodium butyrate (25), interferon-7 (26), EGF or TGFa (24, 27-29), and progesterone (30) can increase the levels of TGFa mRNA. TGFa is synthesized and secreted by many human breast cancer cell lines (31). In several human breast cancer cell lines and in rat mammary adenocarcinoma cells that possess estrogen receptors and that respond mitogenically to estrogens, 17/3-estradiol (E2) can increase the expression of TGFa mRNA and can stimulate the production of TGFa protein (28,32-34). E2-induced expression of TGFa mRNA can be blocked by antiestrogens such as tamoxifen, demonstrating that this response operates through an estrogen receptor-mediated pathway (30,34). Although E2 can enhance the level of TGFa mRNA in estrogenresponsive breast cancer cells, it is unknown whether this is due to a direct effect of estrogen on stimulating transcription of the TGFa gene or whether estrogen is able to regulate the turnover of TGFa mRNA or is able to control other posttranslational events. The rat and human TGFa genes have been cloned and sequenced (35, 36). Furthermore, the 5'-flanking region of these genes has been characterized and found to be GC rich, to lack a conventional TATA box, and to contain multiple Sp1 and AP-2 binding sites. In addition, when this region is ligated to the chloramphenicol acetyl transferase (CAT) gene, it is able to facilitate expression of CAT activity in transient transfection assays when assessed in several cell types, demonstrating that there is promoter activity in a 300- to 400-base pair (bp) region of the 5'-flanking sequence (35, 36). The present study was designed to determine whether the 5'-flanking region of the human TGFa gene could direct expression of a reporter gene in human breast cancer cells and to ascertain if plasmids containing various size fragments of the TGFa 5'-flanking region could confer estrogen responsiveness to CAT or luciferase reporter genes. The results of this study

show that in estrogen-responsive human breast cancer cells, E2 is able to induce CAT or luciferase activity through the 5'-flanking region of the human TGFa gene in transient transfection assays, and that the TGFa 5'flanking region probably contains a potential c/s-acting regulatory element(s) that can respond to E2.

RESULTS

MCF-7 cells have proven to be a useful system for defining c/s-acting estrogen-responsive elements in the 5'-flanking region of different genes because they possess estrogen receptors and are responsive to E2 (37, 38). However, several MCF-7 sublines differ in their degree of responsiveness to E2 (39). MCF-7 clone E3 cells, therefore, were tested for anchorage-dependent growth in the absence or presence of E2. MCF-7 clone E3 cells were grown for 1 week in phenol red-free medium containing 5% charcoal-stripped, sulfatasetreated calf serum (estrogen-depleted medium) to sensitize the cells to the effects of E2 (40). Cells were then treated with 10 nM E2 for an additional 7 days. Under these conditions, E2 was able to produce a 50-300% stimulation in cell growth (data not shown). Addition of different concentrations of the antiestrogens tamoxifen or droloxifene (3-hydroxytamoxifen), to Entreated MCF-7 clone E3 cells was able to inhibit E2-stimulated proliferation in a dose-dependent manner (data not shown). Droloxifene was approximately 10-fold more active than tamoxifen in this assay (41). To ascertain if E2 could also stimulate the production of TGFa in this particular clone, MCF-7 clone E3 cells were maintained for 1 week in estrogen-depleted medium and then changed to and grown in PC-I serum-free medium containing 10 nM E2 or E2 (10 nM) with 100 nM tamoxifen or droloxifene for an additional 4 days. Media were conditioned during the last 48 h, and conditioned medium (CM) samples were then analyzed for the presence of immunoreactive TGFa using a specific TGFa RIA. E2 produced a 2.8-fold increase in TGFa levels in the CM (50 ng/108 cells) as compared with the levels of TGFa found in the control CM (18 ng/108 cells) sample. This response could be attenuated by the simultaneous addition of either tamoxifen or droloxifene (data not shown). Because E2 can enhance the production of TGFa in these cells, this suggests that there may be hormonal regulation through a c/s-acting element within the 5'flanking region of the TGFa gene. To address this possibility, an 1140-bp Sac-I TGFa 5'-fragment relative to the translation start site of the human TGFa gene that was ligated to the CAT gene (S/A-TGFa-CAT) was used to determine whether CAT activity was estrogen responsive. MCF-7 clone E3 cells that had been grown in estrogen-depleted medium for 7 days were transiently transfected with the S/A-TGFa-CAT vector or with an equivalent amount of a pSV2-CAT plasmid. Twenty-four hours after transfection, the cells were

The Endocrine Society. Downloaded from press.endocrine.org by [${individualUser.displayName}] on 22 November 2015. at 06:15 For personal use only. No other uses without permission. . All rights reserved.

1957

Estrogen Regulation of TGFa

treated with either 10 nM E2 and/or tamoxifen (10~6 M) for an additional 24 h before assaying cell lysates for CAT activity (Fig. 1, A and B). In this experiment, there was a relatively low basal level of CAT activity that could be measured in the cells, which had been transfected with the S/A-TGFa-CAT plasmid (Fig. 1A). In contrast, in MCF-7 cells that had been exposed to E2 for 24 h, there was a 36-fold increase in CAT activity. The magnitude of E2 stimulation of CAT activity in MCF7 clone E3 cells, which had been transfected with the S/A-TGFa-CAT plasmid, varied between individual experiments, as did the basal level of CAT activity. These differences are probably due to variations in the ability to adequately deplete the cells of any residual serumderived estrogen during the presensitization stage of the experiments. Tamoxifen alone was able to function as a weak estrogen agonist with respect to enhancing

CAT activity (Fig. 1 A). However, tamoxifen was able to significantly inhibit E2-stimulated CAT activity (Fig. 1B). The effect of E2 and/or tamoxifen was specific for the TGFa 5'-flanking region since cells that had been transfected with a pSV2-CAT vector, which contains an SV40 early promoter, showed no change in CAT activity after E2 and/or tamoxifen treatment (Fig. 1 A). A 40-fold increase in CAT activity in S/A-TGFa-CAT-transfected MCF-7 cells could first be detected within 24-36 h after E2 treatment as compared with cells that had not been exposed to E2 (Fig. 2, A and B). Although in this particular experiment, induction of CAT activity by E2 in MCF-7 cells was not maximal by 24 h, in the majority of other kinetic experiments that were conducted, maximum E2 induction was observed at this time. E2 responsiveness of CAT activity through the TGFa 5'flanking region was not limited to MCF-7 cells. The S/ A-TGFa-CAT plasmid was transiently transfected into ZR-75-1 cells, another estrogen-responsive human breast cancer cell line. A17-fold increase in CAT activity could be detected in these cells within 24 h after the addition of E2 (Fig. 2, C and D). Physiological concentrations of E2 in the range of 10~11-10~8 M can induce a 2- to 3-fold increase in the growth of MCF-7 clone E3 cells. To ascertain if estrogen induction through the 1140-bp TGFa 5'-flanking fragment was dependent upon the concentration of estrogen that would be minimally sufficient to saturate estro-

E2

Com

TAM

E2 + TAM Com

PSV2-CAT B

E2

TAM

E2 + TAM

TGFa-CAT

TTTTT

1.2 1.0

S

0.8

Time (hr)

0.6 0.4

TtT

0.2 0.0

Control

E2

TAM

E2+TAM

Fig. 1. Effect of E2 and Tamoxifen (TAM) on CAT Activity in Transiently Transfected MCF-7 Cells MCF-7 clone E3 cells were grown for 7 days in phenol redfree medium containing 5% CSCS. Cells were then transfected with 10 Mg of either a pSV2-CAT plasmid or S/A-TGFa-CAT plasmid by the calcium phosphate-glycerol shock procedure as described in Materials and Methods. After 24 h, cells were treated with either E2 (10~8 M) and/or TAM (10~6 M) for 24 h before assaying cell extracts (100 ^g protein/assay) for CAT activity. A, Autoradiograph of thin-layer chromatogram illustrating the nonacetylated substrate and the mono- and diacetylated products. B, Quantification of CAT activity was performed by scraping the appropriate regions from the thin-layer plate and counting the radioactivity.

I 6 12 Time (hr)

24

Fig. 2. Time Course of CAT Induction by E2 in Transiently Transfected MCF-7 and ZR-75-1 Cells MCF-7 (A and B) or ZR-75-1 (C and D) cells were presensitized to estrogen by growth for 7 days in phenol red-free medium containing 5% CSCS before being transfected with 10 Mg of the S/A-TGFa-CAT plasmid as described in Fig. 1. Twenty-four hours after transfection, cells were maintained for varying periods either in the absence or presence of E2 (10"8 M) before CAT assay. A and C, Autoradiographs of thin-layer chromatograms. B and D, Quantification of CAT activity in cell lysates expressed as percent conversion of substrate per 100 Mg protein.

The Endocrine Society. Downloaded from press.endocrine.org by [${individualUser.displayName}] on 22 November 2015. at 06:15 For personal use only. No other uses without permission. . All rights reserved.

Vol5No. 12

MOL ENDO-1991 1958

gen receptors, MCF-7 clone E3 cells, which had been maintained in estrogen-depleted medium, were transiently transfected with the S/A-TGFa-CAT plasmid and subsequently exposed to different concentrations of E2 for 24 h (Fig. 3). A dose-dependent increase in CAT activity was observed with a maximum 5- to 6-fold increase found at 10"8-10~7 M E2. An increase in CAT activity could first be detected at 10~10 M E2. The luciferase gene is generally more sensitive than the CAT gene in recombinant constructs designed to measure the activity of relatively weak promoters (42). In addition, this reporter gene already has been demonstrated to be effective in assessing EGF and TPA responsiveness through a 1100-bp human TGFa 5'flanking sequence in MDA-MB-468 breast cancer cells and in monitoring estrogen response elements in MCF7 cells (29,43). Therefore, MCF-7 cells were transiently transfected with a Sac-11140-bp TGFa-luciferase plasmid (pTGFa1140l_uc) and treated with E2. In the absence of E2, there were very low levels of luciferase activity in MCF-7 cells as mediated through the 1140bp TGFa 5'-flanking region. However, 24 h after the addition of various concentrations of E2, there was a 150- to 250-fold increase in luciferase activity that was maximum at IO~10 M E2, and that was similar to the dose-response for E2 induction of CAT activity (data not shown). The stimulatory effect on luciferase activity was specific for estrogens, since other steroid hormones at a concentration of 10~8 M such as progesterone or dexamethasone were without effect (Fig. 4). Two growth factors that are mitogenic for MCF-7 cells, TGFa and insulin-like growth factor-l (IGF-I) (33, 34), were also tested for their ability to modulate luciferase activity through the TGFa 5'-flanking region. TGFa (10

0)

c o CJ

10,-7

10"" 10~ lu 10~y 10 E2 C o n c e n t r a t i o n

(M)

Fig. 3. Effect of Different Concentrations of E2 on CAT Activity in Transiently Transfected MCF-7 Cells Estrogen-presensitized MCF-7 cells were transfected with 10 fig of the S/A-TGFa-CAT plasmid. Twenty-four hours after transfection, cells were maintained for 24 h in the indicated concentration of E2 before harvest and assay for CAT activity as described in Materials and Methods.

100000 P

TGFa

Dex.

Prog. IGF-1

Cont.

Fig. 4. Effect of Different Hormones and Growth Factors on Luciferase Activity in Transiently Transfected MCF-7 Cells Estrogen-presensitized MCF-7 cells were transfected with 15 ng of the Sacl-TGFa-luciferase plasmid. Twenty-four hours after transfection, cells were maintained for 24 h in medium containing TGFa (10 ng/ml), IGF-I (10 ng/ml), E2 (10~8 M), dexamethasone (Dex; 10" 8 M), or progesterone (Prog: 10"8 M) before assaying for luciferase activity, which is expressed in relative light units per 400 ng protein.

ng/ml) was able to induce an increase in luciferase activity after a 24-h incubation in transiently transfected MCF-7 cells, whereas a comparable concentration of IGF-I was ineffective. The stimulatory effect of E2 through the TGFa: promoter region on CAT or luciferase activity is dependent upon the presence of functional estrogen receptors. Estrogen receptor-negative MDA-MB-231 human breast cancer cells were transiently transfected with 10 ng of either the S/A-TGFa-CAT plasmid or with an equivalent amount of the pSV2-CAT plasmid (data not shown). The level of CAT activity with either the pSV2CAT plasmid or with the S/A-TGFa-CAT plasmid was found to be approximately 10- to 20-fold higher in these cells as compared with comparably transfected MCF-7 cells. This may reflect a higher transfection efficiency of the MDA-MB-231 cells compared with MCF-7 cells. In addition, since MDA-MB-231 cells produce and secrete 2- to 5-fold more TGFa than MCF-7 cells (28, 31, 32), and since TGFa can up-regulate its own expression (24, 27, 28), then the higher basal levels of TGFa in these cells may contribute to the enhanced level of CAT activity after using the S/A-TGFa-CAT plasmid. Nevertheless, treatment of MDA-MB-231 cells with 10 nivi E2 for 24 h had no significant effect on modulating the level of CAT activity after transfection of the S/A-TGFa-CAT plasmid. Antiestrogens such as tamoxifen and droloxifene can antagonize the response to estrogens by directly competing with these hormones for binding to estrogen receptors (41). Therefore, MCF-7 cells were transfected with the pTGFa-1140Luc plasmid and treated with 10 nM E2 for 24 h in the absence or presence of varying concentrations of tamoxifen or droloxifene (Fig. 5). Both antiestrogens are effective in

The Endocrine Society. Downloaded from press.endocrine.org by [${individualUser.displayName}] on 22 November 2015. at 06:15 For personal use only. No other uses without permission. . All rights reserved.

1959

Estrogen Regulation of TGFa

30000 r

5000

4000 en 'c Z>

c

3000 -

20000

10000

.1 2000 a)

DC

1000 pSV2 Sac-ll Bam HI Sac-I

10

10"

,-6

10

C o n c e n t r a t i o n (M) Fig. 5. Effect of Antiestrogens on E2-lnduced Luciferase Activity in Transiently Transfected MCF-7 Cells Estrogen-presensitized MCF-7 cells were transfected with 15 ng of the Sacl-TGFa-luciferase plasmid. Twenty-four hours after transfection, cells were maintained for an additional 24 h in medium with or without E2 (10~8 M) in the absence or presence of different concentrations of either tamoxifen (•— • ) or droloxifene (O—O) before assaying for luciferase activity, which is expressed in relative light units per 400 ^g protein.

producing a concentration-dependent inhibition of the

E2-induced increase of luciferase activity. Specifically, a 10-fold excess of either tamoxifen or droloxifene is sufficient to completely inhibit this response to E2. The data also show that the lower concentrations (10~910r8 M) of droloxifene are more potent than tamoxifen in blocking the E2-induced increase in luciferase activity. The ability of the 1140-bp fragment of the 5'-flanking region of the human TGFa gene to render the CAT or luciferase genes sensitive to estrogen regulation suggests that there are potential c/s-acting regulatory elements that are present within this sequence. To more accurately define other potential regions within the 5'flanking sequence that may confer estrogen responsiveness or that may modify the response to E2, luciferase plasmids that contained varying size fragments of the TGFa 5'-flanking sequence were generated. The plasmids were derived from a 2800-bp Pst-\ fragment (pTGFa-2813Luc) of the human TGFa promoter region isolated from a leukocyte genomic library. The other luciferase reporter plasmids were subsequently generated from pTGFa-2813Luc by restriction endonuclease digestion. These plasmids and a control pSV2-luciferase plasmid were then used in transient transfection assays in MCF-7 cells that had been treated with or without 10 nM E2 for 24 h (Fig. 6). Equivalent levels of E2 stimulation in the range of 200-fold over basal levels of luciferase activity are found after transfection with either the pTGFa-2813Luc (Psf-I), the pTGFa-1565Luc (Spe-I), or the pTGFa-1140Luc (Sac-I) plasmids. How-

Spe-I

Pst-1

Fig. 6. Effect of E2 on Luciferase Activity in Different TGFaLuciferase Plasmids Estrogen-presensitized MCF-7 cells were transfected with 15 Mg of either a pSV2-luciferase plasmid or with an equivalent amount of different TGFa-luciferase plasmids containing various segments of the 5'-flanking region of the TGFa gene. Twenty-four hours after transfection, cells were maintained for an additional 24 h in medium with or without E2 (10"8 M) before assaying for luciferase activity, which is expressed in relative light units per 400 ^g protein.

ever, transfection of a 370-bp Sam-HI TGFa fragment (pTGFa-370Luc) resulted in a 50-fold reduction in the absolute level of E2-stimulated luciferase activity (Fig. 6, inset). Furthermore, there was a complete loss of E2induced luciferase activity when a Sac-ll construct (pTGFa-77Luc) containing only a 77-bp insert of the TGFa 5'-flanking region was used in MCF-7 cells in transient transfection assays. These results suggest that potential c/s-acting estrogen response sequences are located upstream of the Sac-ll restriction site and within a 370-bp Sam-HI fragment. Additional estrogenenhancing sequences are also present upstream of the Bam-HI site, since the stimulatory effect of E2 can be significantly augmented in these larger 5'-flanking fragments.

DISCUSSION

Although a number of different agents such as oncogenes, hormones, growth factors, and other biological response modifiers have been shown to enhance the expression of TGFa (23-30), little information is available regarding the mechanism(s) by which regulation of this gene might occur. In addition, there is coordinate regulation of mRNA expression for TGFa, the EGF receptor, and other growth factors such as TGF/31 after treatment of cells with TPA or after transformation with mos or an activated ras gene (9, 23, 24, 44). Nevertheless, it is unclear whether these agents are directly

The Endocrine Society. Downloaded from press.endocrine.org by [${individualUser.displayName}] on 22 November 2015. at 06:15 For personal use only. No other uses without permission. . All rights reserved.

MOL ENDO-1991 1960

affecting the transcription of the TGFa gene or whether they are indirectly modifying TGFa mRNA stability or turnover. In estrogen-responsive rat and human breast cancer cells, physiological concentrations of E2 can rapidly increase the levels of TGFa mRNA (32-34). This is reflected by a corresponding increase in the amount of secreted TGFa protein that can be detected in the culture medium from these cells. The stimulatory effect upon TGFa mRNA expression can first be detected within 6 h after E2 treatment, suggesting that E2 may be directly stimulating the transcription of this gene. In this study we have clearly demonstrated that the 5'flanking region of the human TGFa gene can confer estrogen responsiveness to the CAT or luciferase reporter genes using chimeric constructs in transient transfection assays. This effect is specific for the TGFa promoter, since the SV40 early promoter is insensitive to E2 stimulation of CAT or luciferase activity. In addition, E2 stimulation through the TGFa 5'-flanking sequence can only be demonstrated in estrogen-responsive breast cancer cells such as MCF-7 and ZR-75-1 cells that possess functional estrogen receptors but not in MDA-MB-231 estrogen-nonresponsive, estrogen receptor-negative human breast cancer cells. These results suggest that the effect of E2 on stimulating CAT and luciferase activity are mediated through an estrogen receptor-mediated pathway. This conclusion is supported further by the observation that the stimulatory effects of E2 on TGFa protein production and on CAT (data not shown) or luciferase activity can be blocked completely by concentrations of two antiestrogens, tamoxifen and droloxifene, that are known to be effective in competitively inhibiting the binding of E2 to the estrogen receptor in these cells (41). Moreover, the greater potency of droloxifene as compared with tamoxifen in inhibiting E2-induced luciferase activity is in reasonable accord with the differential activity that these two antiestrogens have on inhibiting E2-stimulated growth, since droloxifene is more than 10-fold more potent than tamoxifen in blocking E2-stimulated growth in MCF-7 cells and in antagonizing the binding of [3H]17/3-estradiol to the estrogen receptor in these cells (41). Although the binding affinity of droloxifene is approximately 100-fold lower than that of 17|8-estradiol, it is about 10-fold higher than that of tamoxifen. Low concentrations of E2 maximally stimulate the activity of CAT and luciferase. These results suggest that the stimulatory effects of E2 through the TGFa 5'flanking region are physiologically relevant and are not due to some nonspecific pharmacological response to the steroid. The stimulation of CAT or luciferase activity by E2 occurs in a dose-dependent manner over a concentration range that is similar to concentrations of E2 that are able to bind to estrogen receptors and that are capable of stimulating growth in MCF-7 cells (39). Moreover, the stimulation of CAT activity can first be observed within 12-24 h after E2 addition, suggesting that this response precedes any effect that E2 might have upon cell growth (39, 41, 45). Although MCF-7 cells possess progesterone and glucocorticoid receptors

Vol5No. 12

(45), progesterone or dexamethasone are unable to stimulate luciferase activity through the TGFa 5'-flanking sequence. In contrast, exogenous TGFa is able to produce an equivalent level of stimulation of luciferase activity as E2 in transiently transfected MCF-7 cells, demonstrating that EGF/TGFa-responsive c/s-acting element(s) are also present within the TGFa 5'-flanking region. These results agree with a recent report defining a 313-bp EGF/TGFa responsive element in the human TGFa 5'-flanking region at positions -373 to - 5 9 using MDA-MB-468 human breast cancer cells that had been transiently transfected with different TGFa-luciferase constructs (29). Using deletion plasmids that contained different-sized fragments of the TGFa 5'-flanking region ligated to the luciferase gene, it was possible to grossly map various domains within a 2813-bp Pst-\ TGFa fragment that are responsible for E2 stimulation and that might contain potential c/s-acting elements that are responsive to E2. E2 inducibility was essentially lost when a 77-bp Sac IIluciferase plasmid was used in transient transfection assays suggesting that positive estrogen regulatory element(s) are upstream from this restriction site. In this respect, E2 stimulation was partially reacquired in a 370-bp Bam-HI TGFa 5'-fragment and fully restored in the 1140-bp Sac-I TGFa fragment of the 5'-flanking region and in the larger 1565-bp Spe-I TGFa 5'-segment. These data suggest that a potential estrogenresponsive element(s) (ERE) may be present within the 5'-flanking sequence of the human TGFa gene. The sequence for a canonical perfect palindromic ERE is 5'-GGTCANNNTGACC-3' (37, 38, 46). Two imperfect 13-bp palindromic ERE-like sequences are present between positions -215 to -203 (5'-GGTCAGCTGTGCC-3') and between positions -249 to -237 (5'-GGTGACGGTAGCC-3') within the human TGFa 5'flanking sequence. These two sequences are nearly homologous to other imperfect EREs that have been identified within the 5'-flanking sequences of the genes for Xenopus and chicken vitellogenin, chicken ovalbumin, and human pS2 (37, 38, 46, 47). In fact, a 58-bp oligonucleotide that is identical to the sequence between -260 to -203 of the TGFa 5'-flanking region and that contains these two imperfect palindromic sequences can function as a bona fide ERE. This 58-bp oligonucleotide, when ligated to a thymidine kinase promoter-CAT plasmid, can confer estrogen regulation to the thymidine kinase promoter of the herpes simplex virus when this construct is used in transient transfection assays in MCF-7 cells or in COS cells that have been cotransfected with an estrogen receptor expression vector (El-Ashry, D., and F. Kern, personal communication). The inability to fully restore E2 induction of luciferase activity in the 370-bp Bam-HI TGFa fragment, which contains these two EREs, suggests that additional c/s-acting elements are present upstream of these EREs, which can amplify the effects of E2. In this respect, there are seven AP-2 and seven Sp1 consensus binding sites within the 5'-flanking region of the human TGFa gene, which may subserve this function

The Endocrine Society. Downloaded from press.endocrine.org by [${individualUser.displayName}] on 22 November 2015. at 06:15 For personal use only. No other uses without permission. . All rights reserved.

1961

Estrogen Regulation of TGFa

(35). All of the AP-2 binding sequences and two of the Sp1 binding sequences are found 5' to the Bam-Y\\ restriction site. Moreover, in the estrogen-responsive chicken vitellogenin gene there is an additional regulatory domain upstream of the ERE that can be activated by several different steroid hormones including E2 (38). It is possible that a similar region(s) may also be present in the human TGFa 5'-flanking region. Nevertheless, the ability of the 5'-flanking sequence to confer estrogen responsiveness to heterologous reporter genes demonstrates that transcription of the TGFa gene can be directly regulated by estrogens and that this effect may account in part for the increase in TGFa mRNA expression and TGFa protein production, which is observed after estrogen treatment of some hormoneresponsive human breast cancer cells (32-34).

MATERIALS AND METHODS Isolation of Genomic Clones and Plasmid Construction An SV40 early promoter (pSV2)-CAT plasmid was obtained from Dr. Bruce Howard (National Cancer Institute, NIH, Bethesda, MD) (48), and a corresponding pSV2-luciferase plasmid was provided by Dr. Allan Brasier, (Massachusetts General Hospital, Harvard Medical School, Boston, MA) (49). The S/ATGFa-CAT vector containing an 1140-bp 5'-flanking sequence of the human TGFa gene was generously provided by Dr. Arthur Levinson and Dr. Rik Derynck (Genetech Inc., South San Francisco, CA) (35). The TGFa promoter region was isolated from a human leukocyte genomic library (Clontech, Palo Alto, CA) using an oligonucleotide probe, 5'-CGCCCATAAAATGGTCCCCTCGGCTG-3' that is complementary to the first exon of the human TGFa gene. From a positive clone a 2800-bp Psf-1 to Sna-BI fragment was subcloned into pBLUESCRIPT (Stratagene, La Jolla, CA). The clone was sequenced and contained the first exon for TGFa and 2818bp of the promoter region. The first TGFa exon was removed using 8a/-31 digestion from the Sna-BI end generating a fragment which ended five nucleotides upstream of the ATG codon. This fragment was cloned into the EcoRV site of pBLUESCRIPT. From this plasmid a Pst-\ to Hind\\\ fragment containing the TGFa promoter region was cloned into the luciferase vector pXP1, which was kindly provided by Dr. Stephen Nordeen (University of Colorado Health Science Center, Denver, CO) (50) to generate pTGFa-2813Luc. The other plasmids used in this study were generated from pTGFa2813Luc by removing a Pst-\ to Spe-I fragment to generate pTGFa-1565Luc, a Psf-I to Sac-I fragment to generate pTGFa1140Luc, a Psf-I to 8am-HI fragment to generate pTGFa370Luc and a Psf-I to Sac-ll fragment to generate pTGFa77Luc. Cell Culture and Transfection MDA-MB-231 and ZR-75-1 cells were procured from the American Type Culture Collection (Rockville, MD). MCF-7 clone E3 cells were obtained from Dr. Samuel Brooks (Michigan Cancer Foundation, Detroit, Ml) (39). E2, tamoxifen, progesterone, testosterone, and dexamethasone were obtained from Sigma Chemical Co. (St. Louis, MO). Droloxifene (3hydroxytamoxifen) was generously provided by Dr. Max Hasmann (Klinge Pharma GmBH, Munich, FRG). Human TGFa was purchased from Bachem (Torrance, CA), and IGF-I was obtained from Collaborative Research (Waltham, MA). MDAMB-231 cells were maintained in a 1:1 mixture of Dulbecco's modified Eagle's medium and Ham's F-12 medium (vol/vol)

supplemented with 10% heat-inactivated fetal calf serum, 20 mM HEPES (pH 7.4), penicillin (100 U/ml), and streptomycin (100 Mg/ml). MCF-7 clone E3 cells were grown in the same basal medium, but they were supplemented with 5% heatinactivated calf serum (CS). One week before transfection, cells were changed to phenol red-free improved modified Eagle's medium (IMEM) supplemented with 5% charcoalstripped, sulfatase-treated calf serum (CSCS) to deplete the cells of exogenous estrogens (40). Cells (106 per dish) were then seeded into 100-mm tissue culture dishes (Costar, Cambridge, MA) 24 h before transfection. Duplicate dishes were transfected with 10-15 ^g/dish plasmid DNA in the presence of 10 Mg/dish carrier calf thymus DNA by the calcium phosphate precipitation method and glycerol shocked as previously described (51). Twenty-four hours after transfection, the cells were washed several times with phenol red-free IMEM medium containing 5% CSCS and treated with various hormones or growth factors for the indicated times. Anchorage-Dependent Growth MCF-7 cell clone E3 cells were grown for 7 days in phenol red-free IMEM supplemented with 5% CSCS. Cells (5 x 104) were then seeded into 60-mm dishes in the same medium and maintained for 7 days in the presence of 10 nM E2 in the absence or presence of different concentrations of either tamoxifen or droloxifene. Cells were then trypsinized and counted using a model ZBI Coulter counter (Coulter Electronics, Hialeah, FL). Preparation of Conditioned Medium and RIA of TGFa MCF-7 clone E3 cells were grown in phenol red-free IMEM supplemented with 5% CSCS in three 175-cm2 tissue culture flasks until they were 80% confluent. The medium from each flask was then changed, and cells were cultured in the absence or presence of 10 nM E2 with or without 50 nM tamoxifen or droloxifene for 4 days. During the last 2 days of the conditioning period, cells were maintained in 35 ml PC-1 serum-free medium (Ventrex, Portland, ME). Conditioned medium was removed, concentrated, and applied to a reverse-phase C18 Sep-pak column (Waters Instruments, Rochester, MN) to selectively absorb TGFa as previously described (9). TGFa protein was then determined using a specific TGFa RIA from Biotope Inc. (Redmond, WA) or as described (9). CAT Assay Cells were placed on ice and washed twice with cold PBS. Cells were scraped from the plate after the addition of 1.5 ml TEN buffer containing 40 mM Tris-HCI (pH 7.5), 1 mM EDTA, and 150 mM NaCI. Cells were then centrifuged, resuspended in 0.15 ml 250 mM Tris-HCI (pH 7.8), and sonicated on ice. Cells were subsequently freeze-thawed and centrifuged at 10,000 rpm for 5 min at 4 C. Clarified cell lysates were assayed for total protein using the Bradford assay as previously described (52) and for CAT activity according to the method of Gorman ef a/. (48). Reaction products were separated on plastic silica gel thin-layer plates in a chlorofomrmethanol (95:5 vol/vol) mixture. The plates were dried, sprayed with Enhance7 (NEN, Boston, MA) and exposed to x-ray film for 5 days. Spots were then identified, cut out, and counted. Data are expressed as the percent chloramphenicol acetylated per 100 ^g cell extract as determined from duplicate assays obtained from duplicate plates in two separate experiments. Variation between duplicate assays was less than 10%. Luciferase Assay Cells were washed three times with cold PBS, scraped in 5 ml cold PBS, and centrifuged. Cell pellets were then lysed in 0.2 ml of a 1 % Triton X-100 buffer containing 25 mM glycyl-glycine

The Endocrine Society. Downloaded from press.endocrine.org by [${individualUser.displayName}] on 22 November 2015. at 06:15 For personal use only. No other uses without permission. . All rights reserved.

Vol5No. 12

MOL ENDO-1991 1962

(pH 7.8), 15 ITIM MgSO4, 4 mM EGTA, and 1 IDM dithiothreitol. Lysates were clarified by centrifugation at 10,000 rpm at 4 C for 15 sec in a Sorvall microfuge (Sorvall Instruments, Wilmington, DE). Cell lysates were assayed for total protein using the Bradford assay. Luciferase activity was determined according to the method of Brasier et al. (49) using a 1251 LKB (Rockville, MD) luminometer. The intensity of light emission was assessed by integration over a 20-sec interval after the injection of 2 mM ATP and 0.4 mM luciferin. Each assay was internally controlled by monitoring the relative light-forming units generated over a linear concentration range for purified firefly luciferase (Sigma Chemical Co.). Activity is expressed in relative light units per 400 ^g of cell extract and represents the average of duplicate assays obtained from duplicate plates for each experimental group from two separate experiments. Variation between duplicate assays was less than 10%.

10.

11.

12.

13.

Acknowledgments 14. Received August 2, 1991. Revision received October 1, 1991. Accepted October 8,1991. Address requests for reprints to: Dr. David S. Salomon, Laboratory of Tumor Immunology and Biology, Building 10, Room 5B39, National Cancer Institute, National Institutes of Health, Bethesda, Maryland 20892. * Present Address: Cattedra di Oncologia Medica, II Facolta di Medicina e Chirurgia, Universita deli Studi di Napoli, Via S. Pansini 5, 80131 Napoli, Italy.

15.

16.

17.

REFERENCES 18. 1. Salomon DS, Kim N, Saeki T, Ciardiello F1990 Transforming growth factor-a: an oncodevelopmental growth factor. Cancer Cells 2:389-397 2. Brachmann R, Lindquist PB, Nagashima M, Kohr W, Lipari T, Napier M, Derynck R 1989 Transforming growth factored precursors activate EGF/TGF-a receptors. Cell 56:691700 3. Pandiella A, Massague J 1991 Cleavage of the membrane precursor for transforming growth factor a is a regulated process. Proc Natl Acad Sci USA 88:1726-1730 4. Derynck R 1988 Transforming growth factor-a. Cell 54:593-595 5. Di Marco E, Pierce JH, Fleming TP, Kraus MH, Molloy CJ, Aaronson SA, DiFiore PP 1989 Autocrine interaction between TGFa and the EGF receptor: quantitative requirements for induction of the malignant phenotype. Oncogene 4:831-838 6. Rosenthal A, Lindquist PB, Bringman TS, Goeddel DV, Derynck R 1986 Expression in rat fibroblasts of a human transforming growth factor-a cDNA results in transformation. Cell 46:301-309 7. Watanabe S, Lazar E, Sporn MB 1987 Transformation of normal rat kidney (NRK) cells by an infectious retrovirus carrying a synthetic rat type a transforming growth factor gene. Proc Natl Acad Sci USA 84:1258-1262 8. Shankar V, Ciardiello F, Kim N, Liscia DS, Merlo G, Langton BC, Sheer D, Callahan R, Bassin RH, Lippman ME, Hynes N, Salomon DS 1989 Transformation of an established mouse mammary epithelial cell line following transfection of a human transforming growth factor alpha cDNA. Mol Carcinog 2:1-11 9. Ciardiello F, McGeady ML, Kim N, Basolo F, Hynes N, Langton BC, Yokosaki H, Saeki T, Elliot JW, Masui H, Mendelsohn J, Soule H, Russo J, Salomon DS 1990 Transforming growth factor a expression is enhanced in human mammary epithelial cells transformed by an activated c-Ha-ras protooncogene but not by the c-neu pro-

19.

20.

21.

22.

23.

24.

25.

26.

tooncogene and overexpression of the transforming growth factor a cDNA leads to transformation. Cell Growth Differ 1:407-420 Sandgren EP, Luetteke NC, Palmiter RD, Brinster RL, Lee DC 1990 Overexpression of TGFa in transgenic mice: induction of epithelial hyperplasia, pancreatic metaplasia, and carcinoma of the breast. Cell 61:1121—1135 Jhappan C, Stahle C, Harkins RN, Fausto N, Smith GH, Merlino GT1990 TGFa overexpression in transgenic mice induces liver neoplasia and abnormal development of the mammary gland and pancreas. Cell 61:1137-1146 Matsui Y, Halter SA, Holt JT, Hogan BLM, Coffey RJ 1990 Development of mammary hyperplasia and neoplasia in MMTV-TGFa transgenic mice. Cell 61:1147-1155 Elder JT, Fisher GJ, Lindquist PB, Bennett GL, Pittelkow MR, Coffey RJ, Ellingeworth L, Derynck R, Voorhees JJ 1989 Overexpression of transforming growth factor a in psoriatic epidermis. Proc Natl Acad Sci USA 243:811814 Maiden LT, Novak U, Burgess AW 1989 Expression of transforming growth factor alpha messenger RNA in the normal and neoplastic gastrointestinal tract. Int J Cancer 43:380-384 Faust N, Mead J 1989 Regulation of liver growth: protooncogenes and transforming growth factors. Lab Invest 60:4-13 Liscia DS, Merlo G, Ciardiello F, Kim N, Smith GH, Callahan R, Salomon DS 1990 Transforming growth factor-a messenger RNA localization in the developing rat and human mammary gland by in situ hybridization. Dev Biol 140:123-131 Snedeker SM, Brown CF, DiAugustine RP 1991 Expression and functional properties of transforming growth factor a and epidermal growth factor during mouse mammary gland ductal morphogenesis. Proc Natl Acad Sci USA 88:276-280 Lobb DK, Korbin MS, Kudlow JE, Dorrington JH 1989 Transforming growth factor-alpha in the adult bovine ovary: identification in growth follicles. Biol Reprod 8:1087-1093 Kudlow JE, Leung AWC, Korbin MS, Paterson AJ, Asa SL 1989 Transforming growth factor-a in the mammalian brain: immunohistochemical detection in neurons and characterization of its mRNA. J Biol Chem 264:38803883 Vonderhaar BK 1987 Local effects of EGF, a-TGF, and EGF-like growth factors on lobuloalveolar development of the mouse mammary gland in vivo. J Cell Physiol 132:581-584 Bates SE, Valverius EM, Ennis BW, Bronzert DA, Sheridan JP, Stampfer MR, Mendelsohn J, Lippman ME, Dickson RB 1990 Expression of transforming growth factora/epidermal growth factor receptor pathway in normal human breast epithelial cells. Endocrinology 126:596-607 Hynes NE, Taverna D, Harwerth IM, Ciardiello F, Salomon DS, Yamamoto T, Groner B1990 Epidermal growth factor receptor, but not c-erbB-2, activation prevents lactogenic hormone induction of /?-casein gene in mouse mammary epithelial cells. Mol Cell Biol 10:4027-4034 Ciardiello F, Valverius EM, Colucci-D'Amato GL, Kim N, Bassin RH, Salomon DS 1990 Differential growth factor expression in transformed mouse NIH-3T3 cells. J Cell Biochem 42:45-57 Bjorge JD, Paterson AJ, Kudlow JE 1989 Phorbol ester or epidermal growth factor (EGF) stimulates the concurrent accumulation of mRNA for the EGF receptor and its ligand transforming growth factor-a in a breast cancer cell line. J Biol Chem 264:4021-4027 Murphy LD, Valverius EM, Tsokos M, Mickley LA, Rosen N, Bates SE 1990 Modulation of EGF receptor expression by differentiating agents in human colon carcinoma cell lines. Cancer Commun 2:345-355 Kumar R, Mendelsohn J 1990 Growth regulation of A431 cells. Modulation of expression of transforming growth

The Endocrine Society. Downloaded from press.endocrine.org by [${individualUser.displayName}] on 22 November 2015. at 06:15 For personal use only. No other uses without permission. . All rights reserved.

Estrogen Regulation of TGFa

27.

28.

29. 30.

31. 32.

33. 34.

35.

36.

37.

38.

factor-a mRNA and 2',5'-oligoadenylate synthetase activity. J Biol Chem 265:4578-4582 Coffey RJ, Derynck R, Wilcox JN, Bringman TS, Goustin AS, Moses HL, Pittelkow MR 1987 Production and autoinduction of transforming growth factor-alpha in human keratinocytes. Nature 328:817-820 Salomon DS, Kidwell WR, Kin N, Ciardiello F, Bates SE, Valverius EM, Lippman ME, Dickson RB, Stampfer MR 1989 Modulation by estrogen and growth factors of transforming growth factor a and epidermal growth factor receptor expression in normal and malignant human mammary epithelial cells. Recent Results Cancer Res 113:5769 Raja RH, Paterson AJ, Shin TH, Kudlow JE 1991 Transcriptional regulation of the human transforming growth factor-a gene. Mol Endocrinol 5:514-520 Murphy LC, Dotzlaw H 1989 Regulation of transforming growth factor a and transforming growth factor /3 messenger ribonucleic acid abundance in T-47D human breast cancer cells. Mol Endocrinol 3:611-617 Salomon DS, Ciardiello F, Valverius EM, Saeki T, Kim N 1989 Transforming growth factors in human breast cancer. Biomed Pharmacother 43:661-667 Bates SE, Davidson NE, Valverius EM, Freter CE, Dickson RB, Tarn JP, Kudlow JE, Lippman ME, Salomon DS 1988 Expression of transforming growth factor a and its messenger ribonucleic acid in human breast cancer: its regulation by estrogen and its possible functional significance. Mol Endocrinol 2:543-555 Lippman ME, Dickson RB 1989 Mechanisms of growth control in normal and malignant breast epithelium. Recent Prog Horm Res 45:383-440 Arteaga CL, Coronado E, Osborne CK 1988 Blockade of the epidermal growth factor receptor inhibits transforming growth factor «-induced but not estrogen-induced growth of hormone-dependent human breast cancer. Mol Endocrinol 2:1064-1069 Jakobovits EB, Schlokat U, Vannice JL, Derynck R, Levinson AD 1988 The human transforming growth factor alpha promoter directs transcription initiation from single site in the absence of a TATA sequence. Mol Cell Biol 8:5549-5554 Blasband AJ, Rogers KT, Chen X, Azizkhan JC, Lee DC 1990 Characterization of the rat transforming growth factor alpha gene and identification of promoter sequences. Mol Cell Biol 10:2111-2121 Klein-Hitpass L, Schorpp M, Wagner U, Ryffel GU 1986 An estrogen-responsive element derived from the 5'flanking region of the Xenopus vitellogenin A2 gene functions in transfected human cells. Cell 46:1053-1061 Slater EP, Redeuilh G, Beato M 1991 Hormonal regulation of vitellogenin genes: an estrogen-responsive element in the Xenopus A2 gene and a multihormonal regulatory region in the chicken 11 gene. Mol Endocrinol 5:386-396

1963

39. Butler WB, Berlinski PJ, Hillman RM, Kelsey WH, Toenniges MM 1986 Relation of in vitro properties to tumorigenicity for a series of sublines of the human breast cancer cell line MCF-7. Cancer Res 46:6339-6348 40. Berthois Y, Katzenellenbogen JA, Katzenellenbogen BS 1986 Phenol red in tissue culture medium. Proc Natl Acad Sci USA 83:2496-2500 41. Roos W, Oeze L, Loser R, Eppenberger U 1983 Antiestrogenic action of 3-hydroxytamoxifen in the human breast cancer cell line MCF-7. J Natl Cancer Inst 71:5559 42. Williams TM, Burlein JE, Ogden S, Kricka LJ, Kant JA 1989 Advantages of the firefly luciferase as a reporter gene: application to the interleukin-2 gene promoter. Anal Biochem 176:28-32 43. Pons M, Gagne D, Nicolas JC, Mehtali M 1990 A new cellular model of response to estrogens: a bioluminescent test to characterize (anti)estrogen molecules. Biotechnology 9:450-459 44. Jakowlew SB, Kondaiah P, Flanders KC, Thompson HL, Dillard PJ, Sporn MB, Roberts AB1988 Increased expression of growth factor mRNAs accompanies viral transformation of rodent cells. Oncogene Res 2:135-148 45. Lippman ME 1984 Definition of hormones and growth factors required for optimal proliferation and expression of phenotypic responses in human breast cancer cells. In: Barnes DW, Sirbasku DA, Sato GH (eds) Cell Culture Methods for Molecular and Cell Biology. Alan R Liss, New York, vol 2:183-200 46. Nunez A-M, Berry M, Imler J-L, Chambon P 1989 The 5'flanking region of the pS2 gene contains a complex enhancer region responsive to oestrogens, epidermal growth factor, a tumour promoter (TPA), the c-Ha-ras oncoprotein and the c-jun protein. EMBO J 8:823-829 47. Beato M 1989 Gene regulation by steroid hormones. Cell 56:335-344 48. Gorman CM, Moffat LF, Howard BH 1982 Recombinant genomes which express chloramphenicol acetyltransferase in mammalian cells. Mol Cell Biol 2:1044-1051 49. Brasier AR, Tate JE, Habener JF 1989 Optimized use of the firefly luciferase assay as a reporter gene in mammalian cell lines. Biotechnology 7:1116-1121 50. Nordeen SK 1988 Luciferase reporter gene vectors for analysis of promoters and enhancers. Biotechnology 6:454-457 51. Ciardiello F, Dono R, Kim N, Persico MG, Salomon DS 1991 Expression of cripto, a novel gene of the epidermal growth factor gene family, leads to in vitro transformation of a normal mouse mammary epithelial cell line. Cancer Res 51:1051-1054 52. Bradford MMA 1976 A rapid and sensitive method for the quantitation of microgram quantities of protein utilizing the principle of protein-dye binding. Anal Biochem 72:248-254

The Endocrine Society. Downloaded from press.endocrine.org by [${individualUser.displayName}] on 22 November 2015. at 06:15 For personal use only. No other uses without permission. . All rights reserved.