Oncogene (2013) 32, 5233–5240 & 2013 Macmillan Publishers Limited All rights reserved 0950-9232/13 www.nature.com/onc

ORIGINAL ARTICLE overexpression induces malignant changes in receptor a negative MCF-10A cells

J Wang1, JJ Gildea2 and W Yue1

Estrogen is a risk factor of . Elevated expression of aromatase (estrogen synthase) in breast tissues increases local concentrations and is associated with breast cancer development, but the causal relationship between aromatase and breast cancer has not been identified. Accumulating data suggest that both (ER)-dependent and -independent effects are involved in estrogen carcinogenesis. We established a model by expressing aromatase in ERa À MCF-10A human breast epithelial cells to investigate ERa-independent effects of estrogen in the process of malignant transformation. Overexpression of aromatase significantly increased anchorage-independent growth. Parental- or vector-expressing MCF-10A cells did not form colonies under the same conditions. The anchorage-independent growth of MCF-10Aarom cells can be completely abolished by pre- treatment with the , . Neither MCF-10Aarom nor MCF-10Avector cells grown in monolayer were affected by short-term exposure to estradiol. Enhanced motility is another characteristic of cellular transformation. Motility of MCF-10Aarom cells was increased, which could be inhibited by letrozole. Increases in stem cell population in breast cancer tissues are associated with tumor recurrence and metastasis. CD44high/CD24low is a stem cell marker. We found that CD24 mRNA levels were reduced in MCF-10Aarom cells compared with those in parental- and vector-transfected cells. By examining individual clones of MCF-10Aarom with various aromatase activities, we found that the CD24 mRNA levels were inversely correlated with aromatase activity. The ability of MCF-10Aarom cells to form mammospheres in the absence of serum was increased. Our results suggest that overexpression of aromatase in MCF-10A cells causes malignant transformation. Estrogen metabolite-mediated genotoxicity and induction of a stem cell/progenitor cell population are possible mechanisms. These studies provide additional evidence for ERa-independent mechanism(s) in estrogen carcinogenesis and implicate superiority of aromatase inhibitors to antiestrogens for breast cancer prevention.

Oncogene (2013) 32, 5233–5240; doi:10.1038/onc.2012.558; published online 26 November 2012 Keywords: aromatase; MCF-10A; estrogen; breast cancer; stem cell

INTRODUCTION are characteristic of those in women, which predispose to later Long-term exposure to endogenous and exogenous estradiol development of breast cancer. increases the risk of breast cancer in women.1–3 A key piece of Overexpression of aromatase increases tissue estrogen levels. evidence is that bilateral oophorectomy before the age of 35 years The precise mechanism by which estrogen causes breast cancer is 4,5 reduces the lifetime incidence of breast cancer by 75%. Post- not fully understood. A widely accepted concept is that E2, acting menopausal women in the highest quintile of plasma free through estrogen receptor a (ERa), stimulates cell proliferation estradiol (E2) experienced at least a 2.58-fold higher rate of and initiates mutations that occur as a function of errors during breast cancer over the ensuing 10 years compared with those in DNA replication. The promotional effect of E2 then supports the 1,6 the lowest quintile. Inhibition of E2 synthesis with aromatase growth of cells harboring mutations, which accumulate until inhibitors or abrogation of its action with antiestrogens prevents cancer ultimately results. Our studies and those of others the development of contralateral breast cancer during adjuvant suggested that ERa-independent, metabolite-mediated mechan- therapy.7–9 isms are also involved in the process of estrogen carcinogen- 14–17 The aromatase catalyzes the rate-limiting step in the esis. Increased local E2 concentrations due to aromatase estrogen synthesis pathway. Overexpression of aromatase is expression would increase the levels of oxidative metabolites of associated with cell proliferation of both normal and malignant E2, which cause DNA damage and eventually initiate breast cancer. breast tissues. Increased aromatase expression has been reported Stem cells are defined by their capacity for self-renewal and in breast epithelium, fibroblasts and peripheral lymphocytes differentiation into cell lineages present in a specific tissue. The in patients with familial or non-familial .10,11 common characteristics of stem cells and tumor cells imply that Aromatase overexpression occurs in three independent mouse cancer may be originated and sustained by a small population of hyperplastic alveolar nodule (HAN) models of breast cancer,12 as a stem-like cells. The long lifespan of a stem cell would allow the result of an integration mutation in the aromatase . accumulation of mutations and epigenetic changes required for Overexpression of aromatase in the mammary glands of neoplastic transformation. Their self-renewal properties would transgenic mice induces pre-malignant lesions.13 These lesions allow expansion of the cells carrying mutant . It has been

1Department of Medicine, University of Virginia Health System, Charlottesville, VA, USA and 2Department of Pathology, University of Virginia Health System, Charlottesville, VA, USA. Correspondence: Dr W Yue, Department of Medicine, University of Virginia Health System, Fountaine Research Park, PO Box 801416, Charlottesville, VA 22908, USA. E-mail: [email protected] Received 29 June 2012; revised 2 October 2012; accepted 4 October 2012; published online 26 November 2012 Aromatase and breast epithelial transformation J Wang et al 5234 12 7 0.8 10 0.7 6 8 0.6 5

mg/h) 0.5 4 6 -1 0.4 3 4 0.3 (pmol/mg/h) (pmol/mg/h) 0.2 2 2 1/v (pmol Aromatase activity Aromatase activity 0.1 1 0 0.0 0 0 5 10 15 20 25 30 35 -0.2 0.0 0.2 0.4 0.6 0.8 1.0 1.2 0 10-1010-910-810-710-6 Δ4A (nM) 1/s (nM-1) Letrozole (M)

10 70 70 E1+E2 H2O T+A 60 60 8 50 50 6 40 40

4 30 30 20 20 2 10 10 % radioactivity recovered % radioactivity recovered % radioactivity recovered 0 0 0

Figure 1. Aromatase activity in MCF-10Aarom cells determined by tritiated water release assay (a–c) and product isolation (d–f). (a) Substrate saturation curve of aromatase activity. (b) Lineweave–Burk reciprocal plot. (c) Dose-dependent inhibition of aromatase activity by letrozole. (d) Conversion of to E1 and E2 (E1 þ E2). (e) Percent radioactivity recovered from water phase. (f) Percentage of aromatase substrate testosterone and (T þA) left in the cultures after 2-h incubation in the presence or absence of letrozole.

reported that ovarian stimulate the stem cell individual spots representing testosterone, androstenedione, population in MCF-10F cells as well as in mouse mammary glands. (E1) and E2 were measured separately. As MCF-10A cells To investigate the role of aromatase in estrogen carcinogenesis, express 17b-steroid dehydrogenase that catalyzes the conversions we established a model by overexpressing aromatase in MCF-10A of testosterone to androstenedione and E1 to E2, radioactivity human breast epithelial cells lacking ERa. The unique design of presented in Figure 1d is the sum of E1 and E2 and in Figure 1f the the studies was to culture aromatase expressing MCF-10A cells for sum of testosterone and androstenedione. After a 1-h incubation months with or without the aromatase inhibitor, letrozole. of MCF-10Aarom cells with [1,2,6,7-3H]testosterone, the production of from the horse serum in the medium of E1 and E2 (Figure 1d) was observed, as well as the by-product, 3 was the only source of . This model provides a setting of H2O (Figure 1e). Inclusion of letrozole in the culture reduced the long-term exposure of breast epithelial cells to physiological levels of E1 and E2 (Figure 1d). In contrast to estrogen products, concentrations of estrogens, which allows us to investigate ERa- the levels of the substrates, testosterone and andros- independent estrogen carcinogenesis. In this study, we demon- tenedione, were reduced in MCF-10Aarom cells treated with the strated that enhanced aromatase expression can induce transfor- vehicle due to aromatization to estrogens. This reduction was mation of benign breast epithelial cells. An increase in the blocked by letrozole (Figure 1f). population of stem-like cells might have a significant role in the overall carcinogenic process. Aromatase expression promotes anchorage-independent growth of MCF-10A cells RESULTS MCF-10A cells do not express ERa. Any ERa-mediated acute effects Aromatase activity in MCF-10A cells with stable expression of of E2 on proliferation and other biological effects are not expected. aromatase cDNA Therefore, the experiments described in this and the following More than 20 clones of MCF-10A cells were selected after trans- sections were carried out in MCF-10Aarom and MCF-10Avector cells fection with plasmid pHbArm and 2-week treatment with G418. that had been cultured in the medium containing 5% horse serum Aromatase activities measured by tritiated water release assay for at least 3 months. This procedure ascertains that the cells were ranged between 0.37 and 10 pmol/mg protein/h (Supplementary exposed to E2 long enough for E2 to exert mutagenic effects. No Figure 1). androgen was added to the culture of MCF-10Aarom cells. Horse In-depth analysis of aromatase activity was carried out in two serum contains testosterone at nanomolar levels.19 As a clones selected because of high aromatase activity in the initial conservative estimate, 2 nmol/l in horse serum would provide À 10 screening. As expected from the known kinetic properties of 10 M testosterone in the culture. aromatase, estrogen production increased as a function of We first examined the rate of anchorage-dependent growth of substrate concentration and times (data not shown) in the MCF- MCF-10Aarom cells. Compared with MCF-10Avector cells, MCF- 10Aarom cell clones (Figure 1a) with saturation-mediated flattening 10Aarom cells grew faster as reflected by a 3-h reduction in of the curve when androstenedione concentrations exceed 10 nM doubling time (Po0.05; Supplementary Figure 2). Accelerated arom (Figure 1a). The calculated Km of 9.4 nM was comparable to that of growth of MCF-10A cells after long-term culture was not due 18 other cell lines transfected with the same vector. As evidence of to promotional growth stimulation of E2 as observed in ER þ arom specificity, aromatase activity of MCF-10A was inhibitable with breast cancer cells. Indeed, acute treatment with exogenous E2, the specific aromatase inhibitor, letrozole, in a dose-dependent androstenedione or letrozole did not alter the growth rate of MCF- manner (Figure 1c). 10Aarom cells (data not shown). Aromatase activity in MCF-10Aarom cells was confirmed by In contrast, the ability of anchorage-independent growth of product isolation assay after incubation with [1,2,6,7-3H]testoster- MCF-10Aarom and MCF-10Avector cells differed strikingly. The ave- one as a substrate. Levels of radioactivity recovered from rage colony number of MCF-10Aarom cells was 68±5 per 1500

Oncogene (2013) 5233 – 5240 & 2013 Macmillan Publishers Limited Aromatase and breast epithelial transformation J Wang et al 5235 80

60 SE) ± 40

(mean 20 per 1500 cells

Number of colonies 0 MCF-10Aarom MCF-10Avector MCF-10A MCF-vector Vehicle Letrozole 10A MCF-10Aarom Figure 2. Anchorage-independent growth of MCF-10A cells. (a) Images of colonies formed in Methocel after 21 days of culture. (b) Quantification of colonies of MCF-10Avector cells and MCF-10Aarom cells with or without letrozole pre-treatment. cells (Figure 2), indicating anchorage independence, a property of Aromatase expression induces stem/progenitor-like cells early malignant transformation. These effects occurred in the Cancer stem cells represent a small population of self- absence of ERa, indicating a receptor-independent mechanism of renewable cells, which are more resistant to conventional E2 or its metabolites. Parental or vector-expressing MCF-10A cells therapies and responsible for disease recurrence and metastasis. did not form colonies at all under the same conditions. Stem-like cells have been isolated from human breast cancer arom Anchorage-independent growth of MCF-10A cells was com- tissues that are enriched with CD44 and low for CD24 surface pletely abolished by long-term treatment with letrozole, which markers.20 To determine whether aromatase overexpression blocked the formation of estrogen from androgen precursors. increases stem-like cell population, we measured the expression Under these circumstances, the genotoxic metabolites of estrogen of CD24 mRNA by semiquantitated reverse transcription– would not be formed and DNA damage from these metabolites polymerase chain rection (RT–PCR). would not occur. CD44 mRNA was expressed in aromatase- and vector-trans- fected MCF-10A cells at a similar level. In contrast, CD24 mRNA levels were reduced in MCF-10Aarom cells compared with those in Aromatase expression increases motility of MCF-10A cells the parental- and vector-transfected cells. Further studies were Migration is one of the hallmarks of malignancy. The migratory carried out on several clones of aromatase-transfected cells with arom ability of MCF-10A cells was compared with that of MCF- various aromatase activities. We found a reverse correlation vector 10A cells after more than 3 months culture in the absence of between the CD24 mRNA levels and aromatase activity exogenous androgens. Cell migration was first evaluated using a (Figure 5a). To confirm that reduction of CD24 mRNA in MCF- wound healing assay. Figure 3a shows the images of MCF- 10Aarom was due to estrogen production, MCF-10Avector cells were vector arom 10A and MCF-10A at 0 and 24 h after generating the treated with E2 for 1 year. Similar to aromatase overexpression, wound. Aromatase expression significantly increased the motility long-term estrogen exposure resulted in reduction of CD24 mRNA as shown by enhancement of the number of cells migrating across levels (Supplementary Figure 3A). the initial lines (Figure 3a). We then used Boyden chamber assays Growth of mammospheres in suspension culture without serum to assess another parameter of cell motility. As shown in Figure 3b, represents a functional test of stem/progenitor cells. Compared arom enhanced motility of MCF-10A cells was significantly reduced with the vector-transfected MCF-10A cells, MCF-10Aarom cells vector by long-term letrozole treatment. In contrast, MCF-10A cells formed more (eightfold over the MCF-10Avector) and larger vector were less migratory. Exposure of MCF-10A cells to E2 for 3 mammospheres (Figure 5b). At 5 months of treatment, letrozole À 7 months significantly increased the number of cells passing (10 M) significantly reduced mammosphere number to the through the pores in a dose-dependent manner (Figure 3c). These levels found with MCF-10Avector cells (Figure 5c). Long-term arom vector data suggest that stimulation of migration of MCF-10A cells treatment with E2 of MCF-10A cells increased the number requires production of estrogens and is mediated by a non-ERa of mammospheres, confirming that the effect of aromatase mechanism. overexpression on induction of stem cell population is mediated by E2 (Supplementary Figure 3B). Aromatase expression increases the levels of catecholestrogen- depurinating adducts ERa expression and function in MCF-10A cells To determine whether the required for estrogen To confirm the absence of ERa in parental MCF-10A and in stable existed in MCF-10A cells and aromatase overexpres- lines expressing empty vector and aromatase, expression and sion led to an elevation of harmful estrogen metabolites, we function of ERa was examined. MCF-7 breast cancer cells served as treated the cells with different precursors of key enzymes in the a positive control. Neither full-length nor 46 kDa ERa was detected metabolic pathways and measured the products (Figure 4a). The by western blot analysis in any of the three types of MCF-10A cells results of pilot studies demonstrated the presence of the enzymes (Figure 5a). ERa mRNA in these cells as determined by quantitative for estrogen metabolism in MCF-10A cells (data not shown). As an RT-PCR method were 20 000- to 50 000-fold lower than that in end point of DNA damage, we measured the levels of 4-OHE2-1- MCF-7 cells (data not shown). To determine whether such a low arom N7guanine (4-OHE2-1-N7Gua) in MCF-10A cells treated with level of ERa could have a potential function, we quantified different precursors. As shown in Figure 4b, the level of 4-OHE2-1- messenger RNA of three ERa-regulated genes in response to E2 À 10 N7Gua in vehicle control was undetectable. The cells treated (10 M). Progesterone receptor (PR) mRNA in MCF-7 cells with E2 (10 mM), 4-OHE2 (10 mM) or androstenedione (10 mM) increased 40-fold after E2 stimulation, which was inhibited by significantly increased the levels of 4-OHE2-1-N7Gua. Concomitant concomitant treatment with ICI 182780. In contrast, PR mRNA was À 7 arom incubation of letrozole (10 M) could completely block the undetectable in MCF-10A and MCF-10A cells. An extremely vector formation of 4-OHE2-1-N7Gua. These results indicate that over- low level of PR mRNA was detected in MCF-10A cells (at least expression of aromatase in breast epithelial cells could lead to 100-fold lower than MCF-7 cells), which was not stimulated by E2 enhancement in the levels of carcinogenic estrogen metabolites (Figure 6b). The basal levels of amphiregulin (AREG) expression in and DNA damage. MCF-10A cells were comparable to that in MCF-7 cells. Treatment

& 2013 Macmillan Publishers Limited Oncogene (2013) 5233 – 5240 Aromatase and breast epithelial transformation J Wang et al 5236 MCF-10Avector MCF-10Aarom 250 vehicle vehicle letrozole * 200

SE) 150 0 h ± † 100 (Mean 50

Number of cells migrated 0 24 h MCF- vector vehicle letrozole 10A MCF-10Aarom

MCF-10Aarom 50 vehicle letrozole 40 30 * 20 10 Number of cells migrated per field 0 vehicle letrozole

vector MCF-10A 50 -10 -8 vehicle E2(10 M) E2(10 M) 40 * 30 20 * 10 Number of cells migrated per field 0 vehicle E2 E2 10-10M 10-8M Figure 3. Effect of aromatase expression on motility of MCF-10A cells. (a) Wound healing assay. The left panel is the photo taken at 0 and 24 h of wound making. The white lines are the marker starting line. Cells that migrated and crossed the line after 24 h were counted. The right panel depicts the quantification of cells that migrated. The bars represent average numbers of 5–10 fields. *P 0.0001 compared with MCF- w o 10Avector; Po0.0001 compared with MCF-10Aarom vehicle control. The experiment was repeated for three times with similar results. (b) Boyden chamber assay for MCF-10Aarom cells with or without letrozole pre-treatment. The cells migrated through the pores were counted. The bars represent the average of migrated cells in nine microscopic fields. *Po0.0001 compared with the vehicle control. (c) Boyden chamber vector assay for MCF-10A cells pre-treated with E2. The bars represent the average of nine microscopic fields counted in each cell type or treatment. *Po0.0001 compared with the vehicle control. Boyden chamber assay was repeated twice for the results shown in (b) and (c).

DISCUSSION 4A 0.007 E2 has long been identified as a risk factor of breast cancer. It has T 0.006 been well documented that E2 stimulates proliferation of breast aromatase 0.005 cancer cells, but the precise mechanisms whereby E2 initiates E 0.004 2 -1-N7Gua breast cancer development are not well understood. A number of 2 CYP 1B1 0.003 studies suggest that ER-independent mechanisms are involved in 0.002 4-OHE2 4-OHE 0.001 estrogen carcinogenesis. To specifically identify ER-independent (fmol/million cells) 0.000 effects of E2, we established a model characterized by over- E -3,4-Q E 2 2 4 A 4 A 2   expression of aromatase in benign breast epithelial cells that do vehicle 4-OHE not express ERa. The results reported here demonstrate that 4-OHE -1-N3Ade 4-OHE -1-N7Gua 2 2 Letrozole overexpression of aromatase in MCF-10A breast epithelial cells Precursor promotes early signs of malignant transformation in the absence Figure 4. (a) Pathways of estrogen synthesis and metabolism. Boxed of ERs. This is evidenced by increased cellular mobility and were the precursors used in measurement of estrogen anchorage-independent growth. Aromatase expression increased arom metabolites. (b) Levels of 4-OHE2-1-N7Gua in MCF-10A cells the levels of 4-OHE2-1-N7Gua. Measurement of this compound in treated with various precursors (10 mM each) for 24 h. Bars represent its DNA-free form indicates formation of the 4-OHE2 adduct and s.e.m. of three replicates. that depurination has taken place, as evidence of DNA damage. Aromatase expression increased the stem-like cell population as indicated by reduction of CD24 mRNA levels in MCF-10Aarom cells. with E2 did not increase AREG expression in these cells. In MCF-7 Production of genotoxic metabolites and induction of stem cell cells, the same concentration of E2 induced a fivefold increase in population are potential ER-independent mechanisms underlying AREG mRNA (Figure 6b). The expression pattern for pS2 was estrogen carcinogenesis. similar to AREG (data not shown). These data confirm that there is Several lines of experimental and clinical evidence suggest that no functional ERa expression in MCF-10A cells and stable upregulation of aromatase has a critical role in breast cancer expression of aromatase does not alter ERa levels. development. Our previous studies showed that in situ aromatiza- The levels of ERb expression were low in MCF-10A cells and tion of androgens increases tissue E2 concentrations more were not changed by stable expression of aromatase (Figure 6a). efficiently than uptake from the circulation.21 Increased

Oncogene (2013) 5233 – 5240 & 2013 Macmillan Publishers Limited Aromatase and breast epithelial transformation J Wang et al 5237

V 7 30 70 14 49 95 A 140 1.4 Arom activity 120 CD24 100x 1.2

CD24 100 1.0

80 0.8

60 0.6

40 0.4 GAPDH 20 0.2 Aromatase activity (pmol/mg/h) 0 0 Levels of CD24 mRNA (CD24/GAPDH) #7 #70 #14 #30 #84 #49 #95 vector MCF-10Aarom

MCF-10Aarom MCF-10Avector

100 m 100 m 100 m VehicleVehicle Letrozole

100

* 80

SE) 60 ±

40 (mean 20 † Number of mammosphere 0

vector vehicle Letrozole arom MCF-10A MCF-10A Figure 5. Effect of aromatase expression on stem/progenitor cell marker and function. (a) Inverse correlation of aromatase activity and CD24 mRNA in MCF-10Aarom cells. (b) Images of the largest mammospheres of MCF-10Avector and MCF-10Aarom cells treated with or without lectrozole after 21 days of culture. (c) Quantification of mammospheres. The bars represent the average number of mammospheres of five experiments for MCF-10Avector and vehicle-treated MCF-10Aarom cells and of two experiments for letrozole-treated MCF-10Aarom cells. *P 0.05 w o (one tail) compared with MCF-10Avector; Po0.05 (one tail) compared with MCF-10Aarom vehicle control.

27 concentrations of E2 lead to increased activities that are mediated was confirmed in a women study reported by Chand et al. They by ERs or independent of ER. found that the levels of aromatase expression in normal breast Earlier studies found that aromatase expression was elevated in tissues of BRCA1 mutation carriers are significantly higher than in the adipose fibroblasts from the quadrants bearing breast the control subjects. These data implicate that aromatase over- cancers.22,23 Studies on BRCA1 (breast cancer 1, early onset) expression due to the loss of the inhibitory function of BRCA1 is a mutation and aromatase expression provide supportive evidence critical factor responsible for breast cancer development. of causal relationship between aromatase expression and breast As an evidence of the biological effects of aromatase over- cancer development. It is known that women harboring a expression, Diaz-Cruz et al.28 compared the effect of aromatase germline BRCA1 mutation incur an 85% lifetime risk of breast expression versus ERa expression in transgenic mouse models on cancer and a greatly elevated risk of . BRCA1 is the generation of mammary hyperplasia and cancer. They found ubiquitously expressed. Selective increases in the incidence of that mammary-targeted aromatase led to more diffuse ductal breast and ovarian cancers in germline BRCA1 mutation carriers disease and a higher prevalence of hyperplastic alveolar nodules, cannot be explained by its general functions, such as DNA repair, ductal hyperplasia, and invasive adenocarcinomas when chromatin remodeling and regulation of tumor suppressor gene, compared with mammary-targeted ERa overexpression. Notably, p53. Hu and co-workers24 have reported that BRCA1 exerts a invasive adenocarcinomas only occurred in aromatase-expressing repressive function on aromatase expression in an ovarian animals. Although this study could not distinguish ER-dependent 25 granulosa cell line. Later, Lu et al. found that BRCA1 negatively and -independent effects of E2, it demonstrated that increases in regulated aromatase expression in breast adipose fibroblasts tissue E2 concentrations resulting from aromatase expression obtained from women undergoing reduction mammoplasty. The caused more lesions in mammary glands and induced carcinomas. negative regulation of aromatase transcripts was through direct The involvement of ER-independent mechanisms in breast interaction of BRCA1 with the aromatase promoters I.3 and II that cancer development has been suggested by clinical observations are activated aberrantly in tumor-bearing breast tissues.26 The and animal studies. Clinical evidence of ER-independent carcino- negative regulatory function of aromatase expression by BRCA1 genic effect of estrogens came from the study involving BRCA1

& 2013 Macmillan Publishers Limited Oncogene (2013) 5233 – 5240 Aromatase and breast epithelial transformation J Wang et al 5238 50 PR antigenic phenotype and are highly tumorigenic when inoculated to immunodeficient mice.20 In addition to the tumorigenic 40 property, CD44high/CD24low breast cancer cells are highly 37 30 invasive. Two recent reports indicated that adult mouse MW mammary stem cells are highly responsive to steroid 20 38,39

PR/GAPDH signaling. Mammary stem cell number fluctuates during the * 75 ER 10 reproductive cycle with highest levels when progesterone reaches * 50 the maximum at the luteal diestrus phase.39 Ovariectomy 37 0 markedly diminished mammary stem cells, and estrogen and 6 AREG progesterone treatment increased mammary stem cell activity. 75 5 ER Intriguingly, treatment with the aromatase inhibitor letrozole for 3 * 50 4 weeks was sufficient to reduce the mammary stem cell pool. KO of -actin 3 aromatase resulted in a dramatic reduction in mammary stem cell numbers and the capability of reconstituting the mammary 2 gland.38,39 These studies suggest that although both E and

AREG/GAPDH 2 1 progesterone are required for enhancement of mammary stem 0 cells, estrogen may act as a primary factor because inhibition of 2 2 2 2 E E E E estrogen synthesis alone can reduce the stem cell population and

+ ICI + ICI + ICI + ICI 40 2 2 2 2 vehicle vehicle vehicle vehicle its activity. The studies by Fillmore et al. showed that breast E E E E cancer stem-like cell pool can be expanded by estrogen 40 vect arom stimulation in ER þ breast cancer cell lines. MCF-7 10A 10A MCF-10A We demonstrated here that overexpression of aromatase Figure 6. ERa expression and function. (a) Western blot of ERa and increases the number and the function of stem-like cells in MCF- 10A cells in the absence of ERa. CD24low is the marker of stem ERb.(b) Expression of PR and AREG. The cells were shifted to arom red-free medium containing charcoal-stripped serum and treated cells. The levels of CD24 mRNA in MCF-10A cells inversely À 10 À 8 for 24 h with vehicle, E2 (10 M)±ICI 182780 (10 M) before correlate with aromatase activity. Furthermore, inhibition of harvest for quantitative PCR analysis. Expressions of PR and AREG in aromatase activity by letrozole could completely abrogate response to treatments were expressed as relative levels to the mammosphere formation of MCF-10Aarom cells. Taken together, vehicle control of each cell line. our results demonstrate that aromatase overexpression causes early-stage transformation of MCF-10A benign breast epithelial cells. This effect is mediated by estrogen produced in the cells mutant carriers. Bilateral oophorecomy reduces the risk of breast and does not require ERa. The ultimate demonstration of the cancer by 53% in BRCA1 mutant carriers.29 As only 10–24% of role of aromatase in breast cancer development would be breast tumors in patients carrying BRCA1 mutations are ER þ ,30,31 tumorigenesis of aromatase-expressing MCF-10A cells in immuno- the carcinogenic effect of estrogens cannot be explained by an deficient mouse. ERa-mediated proliferative function. Menopausal hormone therapy is associated with increased incidence of both ER À and 17 ER þ breast cancer. Our laboratory recently reported that E2 MATERIALS AND METHODS administration to ERa knockout (ERKO) mice bearing the Wnt-1 Materials 14 oncogene significantly increased mammary tumor incidence. MCF-10A cells were obtained from ATCC (Manassas, VA, USA) and cultured ERKO mice exhibit high levels of circulating E2 because they lack in Dulbecco’s modified Eagle’s medium/F12 medium containing 5% horse feedback regulation of estrogen synthesis. The use of the serum, 40 ng/ml epidermal growth factor, 10 mg/ml insulin, 500 ng/ml aromatase inhibitor, letrozole, in the mice with intact ovaries hydrocortisone and 100 ng/ml cholera toxin. E2 and androstenedione were effectively reduced tumor formation in ERKO mice. Russo et al.32 obtained from Steraloids (Newport, RI, USA). [1b-3H]androstenedione, 3 14 14 found that treatment of ER À MCF-10F cells with physiological [1,2,6,7- H]testosterone, [6,7- C]E2 and [6,7- C]E2 were purchased from Perkin-Elmer (Waltham, MA, USA). Dulbecco’s modified Eagle’s medium/ concentrations of E2 for 2 weeks increased the ability of colony formation on soft agar/Methocel and reduced the ability of F12, horse serum and G418 were from Invitrogen (Carlsbad, CA, USA). 32 Insulin, hydrocortisone, cholera toxin, epidermal growth factor and Nile red differentiation. More importantly, some of the transformed MCF- were obtained from Sigma-Aldrich (St Louis, MO, USA). 10F cells that could migrate through the Matrigel-coated invasion chamber formed tumors in severe combined immunodeficiency mice.33 One possible mechanism underlying estrogen Stable transfection of MCF-10A cells carcinogenic effect is the production of genotoxic metabolites MCF-10A cells, grown in 100 mm dishes until 50% confluence, were of estrogens via P450 enzymes and oxidase.16 These transfected with 5 mgpHbArm plasmid or the empty vector in serum-free metabolites can bind directly to DNA to form depurinating medium for 6–8 h using Qiagen Effectene kit (Valencia, CA, USA). The adducts15 or cause DNA damage indirectly due to the production culture medium was then aspirated and regular medium containing G418 34 (150 mg/ml) added. Growing colonies were then selected 2 weeks later. of oxygen-free radicals during cycling. In support of the MCF-10Aarom and MCF-10Avector cells were maintained in the same depurination mechanism, two reports indicate that women with culture medium for non-transfected MCF-10A cells. It should be mentioned breast cancer or at high risk for the disease have significantly that there was no androgen supplement in the culture of MCF-10Aarom higher levels of depurinating estrogen–DNA adducts in their urine cells. The culture medium contained 5% horse serum, which was prepared and serum than women at normal risk for breast cancer.35,36 Our from adult horses of both male and female. Serum testosterone levels of proof-of-principle studies provide direct evidence that aromatase the horse vary from o0.69 nmol/l in mares/geldings to 3.5–14 nmol/l in stallions.19 A conservative estimate of 2 nmol/l in horse serum would overexpression can increase genotoxic metabolites of estrogens in À 10 benign breast epithelial cells. provide 10 M testosterone in the culture. The results of current studies indicate that additional mechan- ism might be involved. The theory that cancer may be originated Aromatase assays from a small proportion of stem-like, self-renewing cells has Tritiated water release assays were carried out in MCF-10Aarom cells as gained support in recent years. Breast cancer stem cells have been described previously.41 For aromatase assays, the cells were incubated with high low 3 À 6 isolated from tumor tissues. These cells exhibit a CD44 /CD24 [1,2,6,7- H]testosterone in the presence or absence of letrozole (10 M)

Oncogene (2013) 5233 – 5240 & 2013 Macmillan Publishers Limited Aromatase and breast epithelial transformation J Wang et al 5239 for 1 h. The cells were then scraped out and collected together with the dehydrogenase (GAPDH), 50-ACGGATTTGGTCGTATTGGGCG-30 (forward) medium. Steroids in the culture were extracted with ether and separated and 50-CTCCTGGAAGATGGTGATGG-30 (reverse). The PCR conditions are 14 by thin-layer chromatography in ether/hexane solution (3:1). C-labeled E2 as follows: 94 1C for 1 min, 60 1C for 1 min, 72 1C for 2 min, 20 cycles for and E1 (B5000 d.p.m.) tracers were added before extraction for calculation CD24, CD44 and 30 cycles for GAPDH. The PCR products were resolved in of recovery. Unlabeled steroid standards used for visualizing steroid spots 1.2% agarose gels, scanned with Typhoon 9400 Imager (Amersham were added before thin-layer chromatography. Steroid spots were scraped Biosciences, Piscataway, NJ, USA), and quantified using the ImageQuant 3 off the plate, eluted with ether and H-labeled E2,E1, testosterone and software (Molecular Dynamics, Sunnyvale, CA, USA). androstenedione were measured in Beckman LS6500 scintillation counter (Beckman Instruments, Inc., Fullerton, CA, USA). Mammosphere culture MCF-10Aarom and MCF-10Avector cells were plated at the density of 400 cells Anchorage-dependent and -independent growth assays per well into Corning Ultra-low attachment 96-well plate. The cells were For the anchorage-dependent growth assays, cells were plated in six-well cultured in serum-free mammary epithelial growth medium (Lonza, plates at a density of 30 000 cells per well. On each day of the 6-day culture Walkersville, MD, USA) supplemented with B27 serum-free supplement period, nuclei were prepared and counted using the method described (Invitrogen), epidermal growth factor (20 ng/ml; BD), basic fibroblast previously.42 The doubling time was calculated by the formula T2-T1/ growth factor (20 ng/ml; BD Biosciences, San Jose, CA, USA) and heparin log2 N2 À log2 N1, where N2 and N1 are cell numbers at time points T2 and (4 mg/ml; Sigma), 20 wells per cell line or treatment. Fresh medium of 50 ml T1, respectively. was added to each well every week. Mammospheres were counted on day Anchorage-independent growth assays were carried out using the 24 of culture. method of Russo et al.,43 with modifications. In all, 1% base agar was prepared by mixing 2% agar and 2 Â Dulbecco’s modified Eagle’s medium ERa expression and function analysis containing 40% horse serum (1:1) and distributed to 24-well plates, 500 ml ERa expression was examined by western blot analysis as described per well. An aliquot of 0.8 ml of the mixture of pre-warmed methyl 46 cellulose (Sigma-Aldrich) solution and 2 Â Dulbecco’s modified Eagle’s previously. The primary antibodies against ERa (HC-20) and ER b (H150) medium/F12 containing 1500 cells was added on the top of base agar. The were obtained from Santa Cruz Biotechnology (Santa Cruz, CA, USA) and b- final concentration of methyl cellulose was 0.8%. The wells at the edge of actin antibody was from Sigma-Aldrich. Secondary antibodies conjugated the plates were filled with sterile distilled water to maintain adequate with IRDye 800CW and IRDye 680 infrared fluorescence were purchased from LI-COR Biosciences (Lincoln, NE, USA). After incubation with moisture. The cells were cultured at 37 1C with 5% CO2 for 21 days. The colonies were stained with Nile Red solution (2.5 mg/ml). The plates were secondary antibody, the membrane was scanned on an Odyssey infrared scanned using an Olympus IX81 microscope (Olymus America Corporate, imaging scanner and specific protein bands quantitated. Transcription of Center Valley, PA, USA) and SlideBook software (Intelligent Imaging three E2-regulated genes, PR, AREG and pS2, was evaluated by quantitative Innovations, Inc., Denver, CO, USA). Colonies with a diameter X50 mm were RT-PCR method. Total RNA was extracted using Qiagen RNeasy Mini kit. In counted. all, 1 mg total RNA from each sample was reversed transcribed to cDNA using iScript cDNA Synthesis kit (Bio-Rad, Hercules, CA). An aliquot of cNDA equivalent to100 ng RNA was used for quantitative PCR analysis using iQ Cell migration assay SYBR Green Supermix (Bio-Rad). The primers used are: PR,50- Wound healing assay. Cells were plated onto glass cover slips in six-well TCAGTGGGCAGATGCTGTATTT-30(forward) and 50-GCCACATGGTAAGGCAT plates. As growth rates of MCF-10Aarom and MCF-10Avector are different, the AATGA-30 (reverse); AREG,50-GGGAGTGAGATTTCCCCTGT-30 (forward) and initial plating numbers were adjusted so that the two cell lines reached 50-AGCCAGGTATTTGTGGTTCG-30 (reverse); pS2,50-ACGACACCGTTCGT similar numbers at the time of wound assay. At 95% confluence, the cells GGGGTC-30 (forward) and 50-ACGGCACCGCGTCAGGATG-30 (reverse); and 0 0 0 were pre-incubated with the cell cycle inhibitor, cytosine b-D-arabinofur- GAPDH, 5 -ACCCACTCCTCCACCTTTG-3 (forward) and 5 -CTCTTGTGCTC 0 anoside hydrochloride (10 mM; Sigma), for 1 h to prevent cell proliferation. A TTGCTGGG-3 (reverse). Relative expression of E2-regulated genes was wound of B3 mm in width was then made by a plastic cell lifter and calculated using the comparative Ct method with GAPDH as the reference photographs were taken as 0 h records. After 24-h incubation at 37 1C with gene and with the 2 À DDCt formula. 5% CO2, photographs of the same areas were taken. Photographs at 0 and 24 h were printed out, overlapped to each other and initial lines were Statistical analysis drawn on the 24-h pictures. Cells migrating across the initial lines were ± counted. In all, 5–10 photos of different fields were counted for each cell All quantitative data are expressed as the mean s.e. Statistical line. The results were expressed as mean±s.e.m. significance was determined by Student’s t-test.

Boyden chamber assay. Cells were plated on the insert of 24-well plate at a density of 4 Â 105 cells per well and incubated at 37 1C for 16 h in the CONFLICT OF INTEREST presence of the anti-mitotic agent, cytosine b-D-arabinofuranoside hydro- The authors declare no conflict of interest. chloride (10 mM). Cells were fixed with formalin and stained with 0.5% crystal violet. Cells on the top surface of the insert were removed by a swab. Cell on the bottom surface were counted. Nine fields per membrane ACKNOWLEDGEMENTS were evaluated. This study is supported by DOD Grants: DOD DAMD17-01-1-0444 and DAMD17-01-1- 0445. Human aromatase cDNA plasmid was kindly provided by Dr Shiuan Chen Measurement of catecholestrogen-depurinating adducts (Beckman Institute at City of Hope, Duarte, CA, USA). Measurement of estrogen arom metabolites was performed by Dr Cavalieri at the Laboratory of Eppley Institute for MCF-10A cells were cultured in 150 mm dishes and treated with E2,4- Research in Cancer and Allied Diseases (University of Nebraska Medical Center D4 D4 OHE2, Aor A þ letrozole for 24 h. The media and cells were collected Omaha, NE, USA). and 2 mg/ml ascorbic acid added to protect E2 metabolites from oxidative degradation. The media were extracted, concentrated and partially purified in Certify II Sep-Pak cartridge. The extracts were subjected to high- performance liquid chromatography analysis as described previously.44,45 REFERENCES 1 Key T, Appleby P, Barnes I, Reeves G. Endogenous Hormones and Breast Cancer Collaborative Group. Endogenous sex hormones and breast cancer in post- Semiquantification of CD44 and CD24 mRNA menopausal women: reanalysis of nine prospective studies. J Natl Cancer Inst Total RNA was extracted using Qiagen RNeasy Mini Kit (Valencia, CA, USA). 2002; 94: 606–616. cDNA was synthesized by reverse transcription using Bio-Rad iScript cDNA 2 Kotsopoulos J, Chen WY, Gates MA, Tworoger SS, Hankinson SE, Rosner BA. Risk synthesis kit (Hercules, CA, USA). An aliquot of cDNA solution equivalent to factors for ductal and lobular breast cancer: results from the nurses’ health study. 37–296 ng RNA was used for PCR reaction. The primers used are: CD24, 50- Breast Cancer Res 2010; 12: R106. GCCAGTCTCTTCGTGGTCTC-30(forward), 50-CTCCATTCCACAATCCCATC-30 3 Chen WY, Manson JE, Hankinson SE, Rosner B, Holmes MD, Willett WC et al. (reverse); CD44, 50-GAGCATCGGATTTGAGACCTG-30 (forward), 50-AGCTC Unopposed estrogen therapy and the risk of invasive breast cancer. Arch Intern CATTGCCACTGTTGAT-30 (reverse); and glyceraldehyde-3-phosphate Med 2006; 166: 1027–1032.

& 2013 Macmillan Publishers Limited Oncogene (2013) 5233 – 5240 Aromatase and breast epithelial transformation J Wang et al 5240 4 Feinleib M. Breast cancer and artificial : a cohort study. J Natl Cancer 26 Agarwal VR, Bulun SE, Leitch M, Rohrich R, Simpson ER. Use of alternative pro- Inst 1968; 41: 315–329. moters to express the aromatase (CYP19) gene in breast adi- 5 Trichopoulos D, MacMahon B, Cole P. Menopause and breast cancer risk. J Natl pose tissues of cancer-free and breast cancer patients. J Clin Endocrinol Metab Cancer Inst 1972; 48: 605–613. 1996; 81: 3843–3849. 6 Kaaks R, Rinaldi S, Key TJ, Berrino F, Peeters PHM, Biessy C et al. Postmenopausal 27 Chand AL, kConFab, Simpson ER, Clyne CD. Aromatase expression is increased in serum androgens, oestrogens and breast cancer risk: the European prospective BRCA1 mutation carriers. BMC Cancer 2009; 9: 148. investigation into cancer and nutrition. Endocr Relat Cancer 2005; 12: 1071–1082. 28 Diaz-Cruz ES, Sugimoto Y, Gallicano GI, Brueggemeier RW, Furth PA. Comparison 7 Howell A, Cuzick J, Baum M, Buzdar A, Dowsett M, Forbes JF et al. Results of the of increased aromatase versus ERa in the generation of mammary hyperplasia ATAC (Arimidex, , Alone or in Combination) trial after completion of 5 and cancer. Cancer Res 2011; 71: 5477–5487. years’ adjuvant treatment for breast cancer. Lancet 2005; 365: 60–62. 29 Rebbeck TR, Kauff ND, Domchek SM. Meta-analysis of risk reduction estimates 8 Breast International Group (BIG) 1-98 Collaborative GroupThu¨rlimann B, Keshaviah associated with risk-reducing salpingo-oophorectomy in BRCA1 or BRCA2 muta- A, Coates AS, Mouridsen H, Mauriac L et al. A comparison of letrozole and tion carriers. J Natl Cancer Inst 2009; 101: 80–87. tamoxifen in postmenopausal women with early breast cancer. N Engl J Med 2005; 30 Lakhani SR, Van De Vijver MJ, Jacquemier J, Anderson TJ, Osin PP, McGuffog L et 353: 2747–2757. al. The pathology of familial breast cancer: predictive value of immunohisto- 9 Goss PE, Ingle JN, Ale´s-Martı´nez JE, Cheung AM, Chlebowski RT, Wactawski- chemical markers estrogen receptor, progesterone receptor, HER-2, and p53 in Wende J et al. for breast-cancer prevention in postmenopausal patients with mutations in BRCA1 and BRCA2. J Clin Oncol 2002; 20: 2310–2318. women. N Engl J Med 2011; 364: 2381–2391. 31 Foulkes WD, Metcalfe K, Sun P, Hanna WM, Lynch HT, Ghadirian P et al. Estrogen 10 Berkovitz GD, Guerami A, Brown TR, MacDonald PC, Migeon CJ. Familial gyne- receptor status in BRCA1- and BRCA2-related breast cancer: the influence of age, comastia with increased extraglandular aromatization of plasma carbon 19-ster- grade, and histological type. Clin Cancer Res 2004; 10: 2029–2034. oids. J Clin Inv 1985; 75: 1763–1769. 32 Russo J, Lareef HM, Tahin Q, Hu Y-F, Slater C, Ao X et al. 17b-Estradiol is 11 Bulard J, Mowszowicz I, Schaison G. Increased aromatase activity in pubic skin carcinogenic in human breast epithelial cells. J Steroid Biochem Mol Biol 2002; 80: fibroblasts from patients with isolated gynecomastia. J Clin Endocrinol Metab 149–162. 1987; 64: 618–623. 33 Russo J, Fernandez SV, Russo PA, Fernbaugh R, Sheriff FS, Lareef HM et al. 17-Beta- 12 DeOme KB, Faulkin LJ, Bern HA, Blair PB. Development of mammary tumors from estradiol induces transformation and tumorigenesis in human breast epithelial hyperplastic alveolar nodules transplanted into gland-free mammary fat pads of cells. FASEB J 2006; 20: 1622–1634. female C3H mice. Cancer Res 1959; 19: 515–520. 34 Shen Y-M, Troxel AB, Vedantam S, Penning TM, Field J. Comparison of p53 13 Tekmal RR, Ramachandra N, Gubba S, Durgam VR, Mantione J, Toda K et al. mutations induced by PAH o-quinones with those caused by anti-benzo[a]pyrene Overexpression of int-5/aromatase in mammary glands of transgenic mice results diol epoxide in vitro: role of reactive oxygen and biological selection. Chem Res in the induction of hyperplasia and nuclear abnormalities. Cancer Res 1996; 56: Toxicol 2006; 19: 1441–1450. 3180–3185. 35 Gaikwad NW, Yang L, Muti P, Meza JL, Pruthi S, Ingle JN et al. The molecular 14 Yue W, Wang J-P, Li Y, Fan P, Liu G, Zhang N et al. Effects of estrogen on breast etiology of breast cancer: evidence from biomarkers of risk. Int J Cancer 2008; 122: cancer development: role of estrogen receptor independent mechanisms. Int J 1949–1957. Cancer 2010; 127: 1748–1757. 36 Gaikwad NW, Yang L, Pruthi S, Ingle JN, Sandhu N, Rogan EG et al. Urine 15 Cavalieri E, Chakravarti D, Guttenplan J, Jankowiak R, Muti P, Rogan E et al. biomarkers of risk in the molecular etiology of breast cancer. Breast Cancer: Basic Catechol estrogen quinones as initiators of breast and other human cancers: Clin Res 2009; 3: 1–8. implications for biomarkers of susceptibility and cancer prevention. Biochim 37 Sheridan C, Kishimoto H, Fuchs RK, Mehrotra S, Bhat-Nakshatri P, Turner CH et al. Biophys Acta 2006; 1766: 63–78. CD44 þ /CD24 À breast cancer cells exhibit enhanced invasive properties: an early 16 Yager JD, Davidson NE. Estrogen carcinogenesis in breast cancer. N Engl J Med step necessary for metastasis. Breast Cancer Res 2006; 8: R59. 2006; 354: 270–282. 38 Asselin-Labat M-L, Vaillant F, Sheridan JM, Pal B, Wu D, Simpson ER et al. Control 17 Yager JD, Santen RJ. Mechanisms relating estrogens to breast cancer. Transl of mammary stem cell function by steroid hormone signalling. Nature 2010; 465: Endocrinol Metab 2012; 3: 1–19. 798–802. 18 Zhou D, Pompon D, Chen S. Stable expression of human aromatase com- 39 Joshi PA, Jackson HW, Beristain AG, Di Grappa MA, Mote PA, Clarke CL et al. plementary DNA in mammalian cells: a useful system for aromatase inhibitor Progesterone induces adult mammary stem cell expansion. Nature 2010; 465: screening. Cancer Res 1990; 50: 6949–6954. 803–807. 19 Maurice KT. Diagnosis and surgical removal of a granulosa–theca cell tumor in a 40 Fillmore CM, Gupta PB, Rudnick JA, Caballero S, Keller PJ, Lander ES et al. Estrogen mare. Can Vet J 2005; 46: 644–646. expands breast cancer stem-like cells through paracrine FGF/Tbx3 signaling. Proc 20 Al-Hajj M, Wicha MS, Benito-Hernandez A, Morrison SJ, Clarke MF. Prospective Natl Acad Sci USA 2010; 107: 21737–21742. identification of tumorigenic breast cancer cells. Proc Natl Acad Sci USA 2003; 100: 41 Yue W, Wang J-P, Conaway MR, Li Y, Santen RJ. Adaptive hypersensitivity 3983–3988. following long-term estrogen deprivation: involvement of multiple signal path- 21 Yue W, Wang J-P, Hamilton CJ, Demers LM, Santen RJ. In situ aromatization ways. J Steroid Biochem Mol Biol 2003; 86: 265–274. enhances breast tumor estradiol levels and cellular proliferation. Cancer Res 1998; 42 Yue W, Wang J-P, Conaway M, Masamura S, Li Y, Santen RJ. Activation of the 58: 927–932. MAPK pathway enhances sensitivity of MCF-7 breast cancer cells to the mitogenic 22 O’Neill JS, Elton RA, Miller WR. Aromatase activity in from breast effect of estradiol. Endocrinology 2002; 143: 3221–3229. quadrants: a link with tumor site. BMJ 1988; 296: 741–743. 43 Russo J, Hu YF, Yang X, Russo IH. Developmental, cellular, and molecular basis of 23 Bulun SE, Price TM, Aitken J, Mahendroo MS, Simpson ER. A link between breast human breast cancer. J Natl Cancer Inst Monogr 2000; 27: 17–37. cancer and local estrogen suggested by quantification of breast 44 Lu F, Zahid M, Saeed M, Cavalieri EL, Rogan EG. Estrogen metabolism and for- adipose tissue aromatase cytochrome P450 transcripts using competitive poly- mation of estrogen–DNA adducts in estradiol-treated MCF-10F cells: The effects merase chain reaction after reverse transcription. J Clin Endocrinol Metab 1993; 77: of 2,3,7,8-tetrachlorodibenzo-p-dioxin induction and catechol-O-methyltransfer- 1622–1628. ase inhibition. J Steroid Biochem Mol Biol 2007; 105: 150–158. 24 Hu Y, Ghosh S, Amleh A, Yue W, Lu Y, Katz A et al. Modulation of aromatase 45 Rogan EG, Badawi AF, Devanesan PD, Meza JL, Edney JA, West WW et al. Relative expression by BRCA1: a possible link to tissue-specific tumor suppression. imbalances in estrogen metabolism and conjugation in breast tissue of women Oncogene 2005; 24: 8343–8348. with carcinoma: potential biomarkers of susceptibility to cancer. Carcinogenesis 25 Lu M, Chen D, Lin Z, Reierstad S, Trauernicht AM, Boyer TG et al. BRCA1 negatively 2003; 24: 697–702. regulates the cancer-associated aromatase promoters I.3 and II in breast 46 Fan P, Wang J-P, Santen RJ, Yue W. Long-term treatment with tamoxifen facilitates adipose fibroblasts and malignant epithelial cells. J Clin Endocrinol Metab 2006; translocation of estrogen receptor a out of the nucleus and enhances its inter- 91: 4514–4519. action with EGFR in MCF-7 breast cancer cells. Cancer Res 2007; 67: 1352–1360.

Supplementary Information accompanies the paper on the Oncogene website (http://www.nature.com/onc)

Oncogene (2013) 5233 – 5240 & 2013 Macmillan Publishers Limited