Comparison of increased aromatase versus ERα in the generation of mammary hyperplasia and cancer

Abstract

Factors associated with increased estrogen synthesis increase breast cancer risk. Increased aromatase and estrogen receptor α (ERα) in both normal epithelium and ductal carcinoma in situ lesions are found in conjunction with breast cancer, leading to the idea that altered estrogen signaling pathways predispose the mammary gland to cancer development. Here, we developed a transgenic mouse that conditionally expresses aromatase in the mammary gland, and used it along with a deregulated ERα expression model to investigate the molecular pathways involved in the development of mammary gland preneoplasia and carcinoma. Both increased ERα and aromatase expression led to the development of preneoplasia, but increased preneoplasia, in addition to carcinoma, was found in aromatase over-expressing mice. Increased prevalence of mammary pathological changes in mice expressing aromatase correlated with increased Cyclin E and Cyclin-dependent kinase 2 expression. Gain of both ERα and aromatase increased expression of ERα and progesterone receptor, but aromatase produced a higher increase than ERα, accompanied by higher levels of downstream target genes cyclin D1, c-Myc and RANKL. In summary, while gain of both ERα and aromatase activate abnormal growth pathways in the mammary gland, aromatase induced a wider range of abnormalities that was associated with a higher prevalence of mammary preneoplasia and cancer progression.

Introduction

Breast cancer develops as genetic changes accumulate in the ductal epithelium giving rise to precursor lesions which may progress to ductal carcinoma in situ (DCIS) and eventually invasive breast cancer (1). Factors associated with increased estrogen synthesis, such as obesity and increased dietary fat, increase the risk for breast cancer. Estrogens have been considered one of the most important factors in driving this process (2). Estrogens are biosynthesized from androgens by aromatase, the product of the CYP19 gene and a member of the cytochrome P450 enzyme superfamily (3). In postmenopausal women, estrogens are synthesized in peripheral tissues through the aromatization of androgen precursors produced by the adrenal glands (4). The concentration of estradiol, the most potent endogenous estrogen, in breast tumors can be significantly higher than plasma concentrations in postmenopausal women (5) and higher in DCIS lesions and tumor tissues than normal breast areas (6, 7). Aromatase expression (7, 8), and activity (9) in DCIS lesions and tumors is greater than in normal breast tissue.

Hormone receptor status is one of the main differentiating characteristics of human breast cancers and modifies response to therapy. About 60–70% of human breast cancers are estrogen receptor α (ERα) positive and estrogen-dependent (10). In humans, estrogen receptors are expressed in the mammary gland epithelial cell compartment (11). Increased expression of ERα in normal breast epithelium has been found in conjunction with breast cancer supporting the concept that loss of normal regulatory mechanisms controlling ERα expression levels in normal breast epithelium increases breast cancer risk (12). Intratumoral estradiol levels showed a strong positive correlation with ERα expression in pre- and post-menopausal breast cancer patients (13).

Deregulated estrogen and progesterone signaling contributes to breast cancer pathophysiology. Estrogen and progesterone combination hormone replacement therapy increases the risk of development of breast cancer in post-menopausal women (14, 15). The effects of progesterone are mediated by two isoforms of the progesterone receptor, PR-A and PR-B. PR-B plays a role in alveologenesis, while PR-A is involved in ductal development and side branching (16). Cyclin D1 and c-Myc are both ERα and PR downstream genes, while amphiregulin, nuclear factor kappa B ligand (RANKL), and Wnt4 are more specifically PR downstream genes (17–21). Different observations point to potential interactions between ERα and the signal transducer and activator of transcription (STAT) proteins. Both STAT3 and STAT5 are activated in a significant proportion of breast cancers, both ERα positive and negative (22–24). Crosstalk between ERα and STAT5a has been shown in both normal and breast cancer cells (25). Changes in Stat5a expression levels can modify progression of ERα-initiated mammary preneoplasia (26).

Normally, aromatase is not expressed in the mouse mammary gland. In the first mouse model of aromatase over-expression, MCF-7 cells transfected with the human placental aromatase gene (MCF-7Ca cells) were inoculated into nude mice (27). In the first transgenic mouse model of mammary-targeted aromatase, murine aromatase produced mammary gland hyperplasia (28).

Here we characterized a novel conditional transgenic mouse model of mammary-targeted human aromatase expression and then compared the impact of aromatase expression to ERα over-expression. Altered estrogen signaling pathways predispose the mammary gland to cancer development; however, few studies have systematically examined similarities and differences between substrate excess versus increased receptor. The objective of this study was to investigate whether development of mammary preneoplasia and carcinoma caused by ERα over-expression (26, 29, 30) results from the same or different aberrant molecular pathways than that induced by increased local estrogen production through mammary-targeted aromatase expression. Results showed that aromatase by itself mediated increased ERα and PR expression levels accompanied by increased p-AKT and increased Cyclin E and CDK2 expression that were not found with over-expression of ERα alone.

Materials and Methods

Conditional aromatase transgenic mice

A 2.4-kb full-length human aromatase cDNA was cloned downstream of the tet-op promoter in pPF43 (31). Integrity was verified by nucleotide sequencing. The tet-op-Arom construct was restriction digested with NotI to remove bacterial DNA sequences. Transgenic mice were generated by standard procedures. Potential tet-op-Arom founders were identified by PCR of tail DNA (5’-CGAGCTCGGTACCCGGGTCG-3’ (forward) and 5’-CAGGCATGGCTTCAGGCACGA -3’ (reverse) and bred with MMTV-rtTA transgenic mice (32) to generate double transgenic "Arom" mice.

Mouse models, genotyping and necropsy

Cohorts of experimental female nulliparous MMTV-rtTA/tet-op-Arom (Arom) and MMTV-rtTA/tet-op- ERα (CERM) (29) mice were maintained on a C57Bl/6 genetic background on diet containing 200 mg doxycycline (dox) per kilogram food (Bio-Serv, NJ) from weaning until euthanasia at age 12 months (m) to test for development of preneoplasia and cancer (CERM (n = 34), and Arom (n = 40). A cohort of Arom mice (n=3) were exposed at age 10m to letrozole (2.5 mg/60-day release subcutaneous letrozole pellet, Innovative Research of America, FL) and euthanized at age 12m to test the impact. Cohorts of CERM and Arom mice (n=4–5) were exposed at age 5–6m to mifepristone (Mife) and ICI 182,780 (ICI) (17.5 mg/60-day and 7.5 mg/60 day release subcutaneous pellets respectively), and euthanized at age 6–7m. Additional cohorts of Arom mice were fed dox diet (n=3) or regular diet (n=3) for 3 months and euthanized at age 4m to test dox responsiveness. Wild type (WT) C57Bl/6 mice served as controls (n = 42). To test the effect of exogenous estrogen, cohorts of WT and CERM mice (n=5) were exposed at age 10m to 17β-estradiol (0.72 mg/60-day release subcutaneous pellets) and euthanized at age 12m. At the time of necropsy, mice were anesthetized by isoflurane inhalation and blood samples were obtained by cardiac puncture for measurement of serum 17β-estradiol levels by radioimmunoassay (Cayman Chemical Company). All mice were maintained and euthanized in accordance with institutional and federal guidelines approved by the Georgetown University Animal Care and Use Committee.

RNA expression analyses

Total RNA was isolated using TRIzol reagent (Invitrogen) as described previously (30). RT-PCR primers; Arom: 5’-CCTTGCACCCAGATGAGACT-3’ (forward); 5’-GACAGCACAACAACCAGCAC-3’ (reverse). Real-Time PCR Taqman Gene Expression Assays (ABI Prism 7700): COX-2 (Ptgs2) (Mm01307329_m1), PR (Pgr) (Mn00435625_m1), Pgr-b isoform (forward: TGGACTCAGGTCCCTTCCAA, reverse: CGTCCAGGGAGATCGGTATAGG, reporter: ACGTGTCGTCTGTAGTCTC), RANKL [TNF ligand superfamily member 11 (Tnfsf11)](Mm01313944_g1), wingless-related MMTV integration site 4 (Wnt4) (Mm01194003_m1), amphiregulin (Areg) (Mm00437583_m1), c-myc (myelocytomatosis oncogene) (Myc) (Mm00487804_m1), cyclin D1 (Ccnd1) (Mm00432360_m1), Erbb2 (Mm0065854_m1), and eukaryotic 18s rRNA (Hs99999901_s1). Relative mRNA gene expression normalized against WT control mice (2^-ΔΔ (Ct); where Δ(Ct) = Ct (target gene) - Ct (18s rRNA). Four independent samples were randomly selected from each group for RNA analysis.

Histological analyses

Hyperplastic alveolar nodules (HANs) (33) were detected using whole mounts of inguinal mammary glands (30). Ductal morphology was analyzed by reflectance confocal microscopy (RCM) using VivaCell 5000 (VivaCell-TiBa, NY) with a 30× water immersion lens as described previously (34). Ductal hyperplasia (DH) and carcinoma was detected on hematoxylin and eosin (H&E) stained formalin fixed sections of the second inguinal gland. DH was defined as mammary gland ductal epithelium consisting of at least four epithelial cell layers. Immunohistochemistry was performed on unstained tissue sections using the Vectastain ABC kit or the Mouse On Mouse (M.O.M) Peroxidase kit (Vector Laboratories, Inc., CA) as appropriate by standard procedures. Primary antibodies: ERα (sc-542); PR (sc-538); Her2/Neu (sc-284); E-Cadherin (sc-7870); Cyclin E (sc-481); CDK2 (sc-163); E2F-1 (sc-193); Retinoblastoma (Rb) (sc-50) (Santa Cruz Biotechnology, Inc.), CA; p-STAT3 (D3A7); STAT3 (#9132); p-STAT5 (C11C5); STAT5 (#9363); p-Rb (Ser807/811) (Cell Signaling Technology, Inc., MA); aromatase antibody Ab677: a gift from Dr. Dean B. Evans, Novartis, Basel, Switzerland; KI-67 (RTU-Ki67-MM1) (Novocastra, UK); Cyclin D1 (RM-9104-S) (Thermo Scientific, CA); COX-2 (#160126) (Cayman Chemical, MI). The proliferative index (PI) was calculated as the percentage of Ki-67 positive cells in a total of >1000 cells per section. Five sections were randomly selected from each experimental and control group for evaluation. Negative control slides in which primary antibody was omitted, were analyzed in parallel. Digital photographs were taken using a Nikon Eclipse E800M microscope with Nikon DMX1200 software (Nikon Instruments, Inc., NY).

Aromatase activity

Aromatase enzyme activity was measured using a tritium water-release assay. Dissected thoracic mammary gland was homogenized in 0.1M sodium phosphate buffer (pH=7.4). 1.5 mg protein was used for each assay, and each sample was assayed in triplicate. Protein solution was pre-incubated at 37°C for 3 min followed by the substrate addition (androstenedione including [1β-³H]androst-4-ene-3,17-dione) and a NADPH regeneration system (containing glucose-6-phosphate, glucose-6-phosphate dehydrogenase, and NAPD⁺) to initiate enzyme reaction in a 37°C water bath with constant shaking for 1 hr 45 min. Reaction was terminated with ether. After 3 cycles of this extraction step, 200-ml of dextran-treated charcoal solution was added and incubated at 42°C for 10 min. After centrifugation, a 600-ml aliquot was counted in 5 ml of liquid scintillation mixture. Results were corrected for blanks and protein contents per sample, and expressed as picomoles of ³H₂O formed per milligrams of protein per hour incubation time (pmol/mg protein/hr).

Immunoblotting

Dissected thoracic mammary gland was homogenized in RIPA buffer (Cell Signaling Technology) containing 1mM PMSF plus protease and phosphatase inhibitors (Roche Diagnostics, Indianapolis, IN). Primary antibodies: p-Extracellular signal-regulated kinase (ERK1/2) (E10), ERK1/2, p-AKT (D9E), AKT (C67E7), p-JNK (98F2), JNK (56G8), p-STAT3 (D3A7), STAT3 (#9132), p-STAT5 (C11C5), STAT5 (#9363), p-STAT1 (Y701), STAT1 (#9172), p-IGF-1R (19H7), IGF-1R (#3027), Bcl-2 (#2870), Bax (#2172) (Cell Signaling Technologies); CDK4 (C-22), CDK2 (M2), p27 (C-19), Cyclin E (M-20) and β-Actin (I-19) (Santa Cruz Biotechnology). Four independent samples were randomly selected from each experimental and control group for western blot analysis. Protein bank intensity was quantified by densitometry using Adobe Photoshop CS4 software. Fold changes were normalized to total β-actin, in addition to total protein for phosphorylation.

Statistical analyses

Statistical differences among groups were analyzed using Fisher’s exact for HANS, DH, and adenocarcinoma prevalence; t tests for percentages of mammary epithelial cells expressing nuclear-localized staining, and fold changes in mRNA and protein expression using GraphPad Prism version 4.03 for Windows (GraphPad Software, San Diego, CA). Significance was assigned at P ≤ 0.05.

Results

MMTV-rtTA/tet-op-Aromatase transgenic mice express mammary targeted aromatase

Aromatase transgene RNA was identified in Arom mice exposed to doxycycline (Fig. 1A,B). Significant levels of aromatase activity were measured in mammary tissue of Arom but not WT mice (Fig. 1C). HANs and multiple foci of abnormal ductal epithelial cell growth appeared along secondary and tertiary branches in Arom mice exposed to doxycycline (Fig. 1D). Letrozole exposure for 2 months resolved these abnormalities (Fig. 1D). Aromatase and increased COX-2 protein expression were documented by IHC in Arom mice (Fig. 1E). Increased COX-2 was paralleled by a corresponding increase in RNA (Ptgs2) levels (Arom (1.7±0.3) as compared to WT (1.00±0.02) mice (P<0.05 unpaired t-test, data not shown).

Figure 1.

Generation, characterization, and validation of conditional MMTV-rtTA/tet-op-Arom (Arom) mice. A, Tetracycline-responsive reverse transactivator (rtTA) binds to tetracycline operator (tet-op) sequences upstream of the full-length human aromatase cDNA and activates transcription when doxycycline (dox) is present. Aromatase over-expression targeted to mammary epithelial cells using the mouse mammary tumor virus-long terminal repeat (MMTV-LTR). B, Comparison of aromatase expression levels in WT and Arom mice in the absence and presence of dox using RT-PCR. β-actin control RT-PCR illustrated. C. Bar graph comparing aromatase activity in WT and Arom mice. Mean ± SEM shown. *, P<0.01. D, Representative reflectance confocal microscopy (RCM) and mammary gland whole mounts (WM) images from 4m WT and Arom mice in absence/presence of dox, and a 12m Arom mouse letrozole-treated for 2 months. Note HANs in mammary glands with aromatase expression (white and black arrows) that were not treated with letrozole. Magnification=4X; Size bar WM=100µM. E, IHC demonstrating aromatase and COX-2 expression in mammary epithelial cells. Black arrows indicate representative cells with protein expression. Magnification=40X; Size bar IHC=20µM.

Mammary-targeted aromatase led to more diffuse ductal disease and a higher prevalence of HANs, DH and invasive adenocarcinomas as compared to mammary-targeted ERα over-expression

The impact of aromatase vs. ERα over-expression was compared in nulliparous WT, CERM and Arom mice at age 12m. DH prevalence was significantly higher in aromatase and ERα over-expressing mice as compared to control WT mice (48%, 41%, 2%, respectively, P<0.0001, Fig. 2A,B,C). Diffuse disease was defined as multifocal HANs appearing along multiple secondary and tertiary branches. Aromatase expression resulted in a higher prevalence of diffuse ductal disease (47%) as compared to ERα over-expression (6%, P<0.0001) (Fig. 2B). In the cohorts studied, invasive ductal adenocarcinomas appeared in 7.5% of the Arom mice but none of the CERM or WT mice (Fig. 2D). While DH were ERα/PR positive, carcinomas ERα/PR negative (Fig. 2D). Cyclin D1 and E-cadherin were expressed in both with low levels of HER2 expression (Fig. 2D). To determine if exogenous estrogen exposure could mimic the effect of mammary-targeted aromatase over-expression, WT and CERM mice were exposed to 17β-estradiol. Mammary glands from both genotypes showed lobulo-alveolar development following estrogen exposure but the prevalence of HANs or DH was not changed (Fig. 3A,B). A normal two-layered mammary epithelium was observed in the WT mice while the CERM mice demonstrated ductal hyperplasia. As reported previously (30), CERM mice demonstrated significantly higher levels of nuclear-localized ERα and PR (10±1% and 15±1%, respectively) as compared to WT mice (5±1% and 8±2%, respectively) and this was unchanged by exposure to exogenous 17β-estradiol (P<0.05) (Fig. 3C). Serum 17β-estradiol levels were increased over three-fold in the mice exposed to 17β-estradiol while no significant differences were found between non-exposed WT, CERM and Arom mice (Fig. 3D).

Figure 2.

Aromatase expression resulted in more diffuse mammary ductal epithelial growth abnormalities and a higher incidence of invasive mammary ductal adenocarcinomas than ERα over-expression. A, Representative mammary gland WM images of WT, CERM and Arom mice. Note focal HAN in CERM WM (insert, black arrow) and diffuse HANs in Arom WM (insert, white arrow). B, Representive RCM, WM, and H&E-stained images of normal ductal structures from WT as compared to HANs (white and black arrows) and DH (arrowheads) found in CERM and Arom mice. Note more diffuse HAN development along ducts of Arom mice. C, Bar graph summarizing mean percentages of WT, CERM and Arom mice with DH, HANs, and carcinoma at 12 months of age. *, P=0.003 vs WT, **, P<0.0001 vs WT, #, P<0.0001 vs CERM. D, H&E and IHC images of representative DH and invasive adenocarcinoma developing in Arom mice. DH but not adenocarcinomas were ERα/PR positive. E-cadherin expression classified lesions as ductal. Both lesions showed prominent staining for Cyclin D1 and low levels of Her2. Black arrows indicate mammary epithelial cells with representative staining patterns. Magnification = 4X for WM and 40X for H&E and IHC. Size bar WM = 100µM, Size bar 40X images = 20µM.

Figure 3.

Exogenous 17β-estradiol (E2) did not induce an increase in the prevalence of mammary ductal hyperplasia in WT and CERM mice. A, Representative mammary gland WM, H&E-stained, and ERα and PR IHC images of E2-exposed mice. B, Bar graph summarizing mean percentages of WT and CERM mice with HANs and DH. C, Bar graph summarizing percentages of ERα and PR-positive mammary epithelial cells. D, Bar graph summarizing serum concentrations of 17β-estradiol in WT, CERM and Arom mice and E2-exposed WT and CERM mice. Mean ± SEM shown. Magnification = 4X for WM and 40X for H&E and IHC. Size bar WM = 100µM, Size bar 40X images = 20µM.

Expression of ERα, PR, and downstream genes linked to mammary cancer progression were significantly increased in the mammary glands of Arom mice

Percentages of mammary epithelial cells expressing nuclear-localized ERα and PR were compared in Arom, CERM and WT mice (Fig. 4A). The percentage of mammary epithelial cells showing nuclear-localized ERα was significantly increased in Arom (33±2%) compared to both CERM (12.5±0.9%) and WT (8.0±0.6%) mice (P<0.0001) (Fig. 4B). Percentages of mammary epithelial cells expressing PR were evaluated as a measure of downstream ERα signaling. The percentage of mammary epithelial cells showing nuclear-localized PR was highest in Arom mice (30±2%) followed by CERM (15.1±0.8%) and WT (8.4±0.5%) mice (P<0.0001, P<0.0005, respectively) (Fig. 4B). Cyclin D1 lies downstream of both ERα and PR. Arom mice showed the highest percentage of mammary epithelial cells with nuclear-localized Cyclin D1 (68±2) compared to CERM (43±6) and WT (27.3±0.9, P<0.0001) mice. To determine if increases in protein expression of ERα downstream genes were correlated with increases in steady state RNA levels, real-time PCR was used to quantify expression levels of Pgr, Pgr(b), Ccnd1, Myc, Tnfsf11, Areg, and Wnt4 (Fig. 4C). Arom mice demonstrated two-fold higher levels of total Pgr (PR) expression as compared to WT mice (P<0.005). Both CERM and Arom mice expressed significantly higher levels of Pgr(b) (PR-B isoform) as compared to WT mice (P<0.005). Tnfsf11 (RANKL) showed the largest magnitude of difference in expression with an 8-fold increase in Arom and and a 5-fold increase in CERM mice (P<0.0005, compared to WT mice). Ccnd1 (Cyclin D1) and Myc (c-Myc) were expressed at significantly higher levels in Arom but not CERM mice (P<0.05). No significant changes in expression levels of Areg and Wnt4 were found in either Arom or CERM mice. Expression levels of Erbb2 (ErbB2/Her2) and Egfr (epidermal growth factor receptor) were examined because of their roles in the initiation and progression of both ERα positive and negative breast cancers (35). Erbb2 mRNA expression was significantly increased in the mammary gland of both CERM and Arom mice as compared to WT mice (P<0.005) while only CERM mice showed a significant increase in Egfr expression levels (Fig. 4D).

Figure 4.

Significantly higher expression levels of ERα, PR, c-myc, Cyclin D1 and RANKL were found in Arom as compared to CERM and WT mice. A, Patterns of nuclear-localized ERα, PR and Cyclin D1 expression in WT, CERM and Arom mice by IHC. Black arrows indicate representative cells showing nuclear-localized protein. Magnification = 40X. Size bar 20µM. B, Bar graphs summarizing percentages of ERα and PR-positive mammary epithelial cells in WT, CERM and Arom mice. Mean ± SEM shown. **, P<0.005 vs WT; ***, P<0.0001 vs WT; ###, P<0.0005 vs CERM. C, Bar graphs comparing relative RNA expression levels of total Pgr (PR), Pgr-b (PR-B isoform), Ccnd1 (Cyclin D1), Myc (c-Myc), Tnfsf11 (RANKL), Tnfsf11 (RANKL), Areg (Amphiregulin), and Wnt4 (WNT4) in WT, CERM and Arom mice. *, P<0.05 vs WT; **, P<0.005 vs WT; ***, P<0.0005 vs WT; #, P<0.05 vs CERM; ##, P<0.01 vs CERM. D. Bar graphs comparing relative RNA expression levels of Erbb2 and Egfr in mammary tissue from WT, CERM and Arom mice. **, P<0.005 vs WT; ***, P<0.0005 vs WT. Panels C and D; mean ± SEM normalized to 18S rRNA.

Aromatase expression increased expression of cell cycle proteins Cyclin E, CDK2, p-Rb, E2F-1, up-regulated Bcl-2, down-regulated Bax and p27, and increased relative levels of AKT phosphorylation

The proliferative index (PI) was highest in Arom (21 ± 1%) as compared to CERM (12.5 ± 0.9) and WT (2.2 ± 0.3) mice (Fig. 5A). Combined Western blot (Fig. 5B,C) and IHC (Fig. 5D) studies probed for significant differences in expression or activity of components of proteins that regulate the cell cycle (Cyclin E, CDK2, CDK4, p27, E2F-1, RB), proliferation (ERK1/2, IGFR, STAT5, STAT3, STAT1) and survival (JNK, AKT, Bcl-2, Bax) of mammary epithelial cells (22, 24, 26, 36–38). The percentage of mammary epithelial cells with CDK2 expression was significantly higher in Arom (26±1%) as compared to CERM (2.1±0.7%) and WT (1.7±0.1%) mice. Association of Cyclin E with CDK2 results in Rb phosphorylation. Arom mice also demonstrated a statistically significant increase in nuclear localized p-Rb protein (44 ± 5%) as compared to CERM (19.9±0.9%) and WT (8±2%) mice. Phosphorylation of Rb leads to release of E2F-1, which in turn facilitates the transcription of genes necessary for entry into the S phase of the cell cycle. Arom mice showed a significant increase in nuclear-localized E2F-1 expression (12±1%) as compared to CERM and WT mice (2.4±0.3% and 2.6±0.8% respectively). Total protein levels of ERK1/2, AKT, JNK, IGFR, STAT3, STAT5 and STAT1 were not altered in CERM and Arom mice as compared to WT mice. Relative levels of p-AKT were significantly increased in mammary tissue from Arom mice. In contrast, relative levels of p-ERK1/2, p-IGFR, p-STAT3 and p-STAT5 were increased in both Arom and CERM mice. Only Arom mice showed significantly higher expression levels of Cyclin E, CDK2, p-AKT, and Bcl-2 and decreased Bax and p27 expression. CERM but not Arom mice showed increased p-JNK. There was no evidence of STAT1 activation in any of the mice and CDK4 levels were equivalent in all models.

Figure 5.

Rates of mammary epithelial cell proliferation were highest in Arom as compared to CERM and WT mice. Aromatase, but not ERα over-expression, increased Cyclin E and CDK2 and Bcl-2 and down-regulated Bax and p27 expression, and increased relative levels of AKT phosphorylation. A, Bar graph comparing rates of mammary epithelial cell proliferation in WT, CERM and Arom mice. Mean ± SEM shown. ***, P<0.0001 vs WT; ###, P<0.0005 vs CERM. B, Western blot analysis of relative levels of phosphorylated and total ERK1/2, AKT, JNK, IGFR, STAT3, STAT5 and STAT1. C, Western blot analysis of relative expression levels of Cyclin E, Cdk2, Cdk4, Bcl-2, Bax and p27. Panels B and C; Fold changes in expression levels of total and phosphorylated proteins as compared to WT mice indicated. β-Actin shown as loading control. *, P<0.01 vs WT. D, IHC detection of KI67, Cyclin E, Cdk2, p-RB, total RB, E2F-1, p-STAT3, total STAT3, p-STAT5 and total STAT5 in mammary epithelial cells from WT, CERM and Arom mice. Black arrows indicate representative cells showing nuclear-localized protein. Magnification = 40X. Size bar 20µM.

Comparison of an ERα antagonist versus a PR antagonist on mammary ductal regression and expression of ERα and PR in CERM and Arom mice

The ERα antagonist ICI 182,780 resulted in a modest regression of the mammary gland ductal tree in both CERM and Arom mice with no changes in DH incidence (Fig. 6A,B). In contrast, the PR antagonist mifepristone resulted in more regression of the ductal tree in both CERM and Arom mice with a marked reduction in the prevalence of HANs and DH in Arom mice (Fig. 6A,B). The percentage of mammary epithelial cells showing nuclear-localized ERα was significantly decreased in both ICI (11±2%) and mifepristone (10±2%) treated Arom as compared to non-treated mice (33±2%) (both P<0.0001) (Fig. 6C). Similar results were also obtained for PR expression (12±1% and 12±2% compared to 30±2%) (P=0.0003 and P=0.002, respectively) (Fig. 6C). Tet-op-Arom and tet-op-ERα transgene expression was verified by RT-PCR following treatment and no significant changes were observed (Fig. S1)

Figure 6.

Impact of ICI 182,780 and mifepristone on mammary ductal structure and prevalence of mammary preneoplasia in CERM and Arom mice. A, Representative mammary gland WM, H&E-stained, and ERα and PR IHC images of ICI 182,780 and mifepristone-treated mice. Note HAN in ICI-treated Arom WM (insert). B, Bar graph summarizing mean percentages of CERM and Arom mice with HANs and DH. C, Bar graph summarizing percentages of ERα and PR-positive mammary epithelial cells. Mean ± SEM shown. Magnification = 4X for WM and 40X for H&E and IHC. Size bar WM = 100µM, Size bar 40X images = 20µM.

Discussion

This study compared the roles of increased local estrogen production versus ERα over-expression in mammary gland preneoplasia and cancer development in vivo. Although previous studies associated increased aromatase (28) and ERα expression (26, 29, 30) with mammary preneoplasia and cancer, this investigation compared the effects of each genetic lesion utilizing the same conditional system to target and regulate transgene expression levels. Differences in activation of downstream signaling pathways were correlated with a higher prevalence of preneoplasia and cancer in mice with mammary-targeted aromatase expression as compared to ERα over-expression.

CERM mice are an established tool to investigate the mechanisms involved in progression and regression of ERα-induced mammary hyperplasia, DCIS and invasive cancer (26, 29, 30). Expression of ERα is increased 2-fold in the mammary epithelial cells of these mice and is considered deregulated because it is not down-regulated by estrogen exposure. Here we introduce a companion model of mammary-targeted human aromatase expression with levels of aromatase activity comparable to those found in normal human breast samples and tumors (39, 40). In women, DH and DCIS lesions in the breast are associated with increased risk of invasive breast cancer development (1). HANs are considered a precancerous lesion in murine mammary gland (33). Our investigations identified local aromatase expression as a more potent inducer of DH, HANs, and invasive cancers than ERα over-expression. Control experiments were performed to determine if local aromatase over-expression could increase serum estradiol levels and to test if exogenous estradiol would produce effects similar to those found with local aromatase over-expression. Results demonstrated that aromatase over-expression in the Arom mice did not alter serum estradiol levels and that exogenous estradiol did not induce either pathological changes in the mammary gland or increase expression levels of ERα or PR. Significantly while increased aromatase activity stimulated the development of ERα/PR negative invasive cancers, the ductal hyperplastic lesions were ERα/PR positive. In women, a higher percentage of ER-negative breast tumors among samples with high aromatase activity has been reported (41) and development of ERα/PR negative disease from aberrant estrogen signaling is documented (42). While the exact mechanism is unclear, it is suggested that tumor heterogeneity, clonal selection of tumor cell subpopulations, or genetic instability of tumor cells may be some of the factors involved. Increased COX-2 expression has been detected in DCIS (43), and invasive breast carcinoma (44). Studies have shown a strong linear association between the sum of COX-1 and COX-2 expression and CYP19 expression in breast cancer specimens compared to normal breast tissue (45). The aromatase mouse model presented here also demonstrates an association between aromatase and COX-2 expression.

Increased PR expression in CERM mice from higher levels of ERα expression was anticipated (30). The significantly higher levels of ERα and PR expression following aromatase expression were not predicted but parallel findings in human breast disease (13). Progesterone has been linked to breast carcinogenesis due to its mitogenic effects (46). PR-A is reportedly up regulated by estradiol and down-regulated by progesterone, whereas PR-B is up-regulated by progesterone (47). In a randomized controlled trial, the Women’s Health Initiative linked the use of progestins to the onset and incidence of breast cancer (48). Overall, Arom mice showed higher expression levels of PR downstream genes than CERM mice and this was associated with a higher prevalence of preneoplasia and cancer. For example, both CERM and Arom models expressed significantly increased levels of RANKL mRNA, but levels in Arom mice were significantly higher than CERM mice. RANKL and its receptor (RANK) are expressed in primary breast cancers in humans and breast cancer cell lines (49) and and RANKL is reported to be a pivotal factor regulating incidence (20), development, and growth of murine mammary cancer (19). Arom mice also demonstrated higher mRNA expression levels of Cyclin D1 and c-Myc mRNA, consistent with the idea that ERα and PR downstream transcription was more highly activated in Arom as compared to CERM mice in correlation with the higher mammary disease prevalence.

Rates of mammary epithelial cell proliferation were higher in Arom mice as compared to CERM mice. Expression and activation, as measured by phosphorylation, of key signaling pathways in mammary epithelial growth and survival (37) were measured to determine where similarities and differences could be found. Increased ERα and mammary-targeted aromatase resulted in similar levels of ERK1/2, IGFR, STAT3 and STAT5 activation but expression levels of Cyclin D1 and phosphorylated Rb were significantly higher in Arom as compared to CERM mice. Only Arom mice showed increased AKT activity and increased expression of Cyclin E, CDK2, E2F-1 and Bcl-2 and reduced expression levels of Bax and p27, all changes favoring cell proliferation and survival. E2F-1 deregulation may be involved in the progression of breast cancer because its expression levels are higher in DCIS and invasive cancers then in the normal breast (36). Both models demonstrated increased ErbB2 mRNA. CERM, but not Arom, mice showed significant increases in EGFR mRNA expression and phosphorylated JNK. Overall, however, Arom mice exhibited a higher number of alterations than CERM mice among the signaling pathways examined.

Signal transducers and activators of transcription (STAT) proteins are involved in the normal physiology of the mammary gland and also can regulate growth of breast cancer cells (22–26). It is possible that the abnormally high levels of phosphorylated STAT3 and STAT5 found in both CERM and Arom mice contributes to development of mammary disease. Increased expression of Cyclin D1 and c-Myc mRNA, known STAT3 and STAT5 target genes (50), were documented in the Arom but not CERM mice.

Introduction of aromatase expression in the mammary gland resulted in increased levels of the anti-apoptotic protein Bcl-2 and decreased expression of the anti-apoptotic protein Bax, changes that favor cell survival and are coincident with a previous report (38).

Mammary-targeted aromatase expression led to a different profile of molecular changes than ERα over-expression suggesting that the pathophysiological mechanisms triggered by aromatase over-expression (that include an up-regulation of ERα and PR expression levels) may be similar but not the same as those induced by primary ERα over-expression. This possibility is supported by the differential response of Arom and CERM mice to the PR antagonist mifepristone. In the Arom mice the PR antagonist resulted not only in regression of the ductal tree but also HANs and DH while in CERM mice, the DH persisted despite overall ductal regression. This is consistent with a model in which aromatase over-expression results in increased activity of the progesterone pathway that may then contribute to an increased risk of breast preneoplasia and cancer development. Because the impact of mifepristone was greater than ICI 182,780 on regression of pathophysiological changes in the Arom mice, it raises the possibility that a PR antagonist could have a role in reversing disease related to increased levels of aromatase expression. The absence of a significant reduction in either ERα or PR levels following ICI 182,780 or mifepristone treatment in the CERM mice may again reflect differences in molecular pathophysiology. This may be related to the fact that in the CERM model ERα expression is directly increased by a transgene while in the Arom model increased ERα expression is secondary to aromatase over-expression. The reduction of hormone receptor expression levels by both the ERα or PR antagonists in the Arom mice suggests that increased hormonal pathway activity contributed to the observed increase in ERα and PR expression levels.

All of the molecular changes identified following aromatase expression in the mammary gland correlate with alterations found in human disease. The deregulated signaling pathways identified here may represent candidate biomarkers for evaluating an interim response to aromatase inhibitors or tamoxifen. In the future, this conditional aromatase model can be used to determine how the time of mammary-targeted aromatase expression influences disease development, for example, duration of exposure to aromatase expression during a specific developmental stage such as puberty or menopause.

In conclusion, this study revealed that mammary-targeted aromatase expression correlated with higher ERα and PR expression levels, resulted in an increased number of aberrations in cell cycle regulation and signaling, and produced more extensive mammary disease than more modest levels of ERα over-expression alone. This suggests that changes in aromatase levels and increased local estrogen production may be more pathogenic than alterations in ERα expression, by itself, for breast cancer progression.