Estrogen promotes estrogen receptor negative BRCA1-deficient tumor initiation and progression

The generation of p18^−/−, p18^−/−;Brca1^MGKO (p18^−/−; Brca1^f/f;MMTV-Cre or p18^−/−;Brca1^f/−;MMTV-Cre) and p16^−/−;Brca1^MGKO (p16^−/−;Brca1^f/f;MMTV-Cre or p16^−/−; Brca1^f/−;MMTV-Cre) mice has been described previously [23, 37, 39]. The Institutional Animal Care and Use Committee at the University of Miami approved all animal procedures. Histopathologic analysis and immunohistochemical analysis (IHC) were performed as described previously [19, 23, 37]. The primary antibodies used were Ck14 (Thermal Scientific), ERα (Santa Cruz), p-Akt (Ser473), p-4E-bp1 (Thr37/46), vimentin (Vim), E-cadherin (E-cad), p-Fra1 (Ser265) (Cell Signaling), Ki67 and fibronectin (Fn) (Abcam). Immunocomplexes were detected using the Vectastain ABC DAB kit according to the manufacturer’s instructions (Vector Laboratories) or by using Alexa Fluor 488-conjugated or Alexa Fluor 594-conjugated secondary antibodies (Biolegend). The positive results of IHC were quantified by the H score, as previously described [41].

Mammary glands and tumors were dissected from female mice and cell suspensions were prepared as previously described [19, 23, 37]. For surface marker analysis; mammary and tumor cells were isolated, stained, and analyzed as previously described [23, 37]. Briefly, cells were stained with anti-CD24-PE (BD Pharmingen), anti-CD29-FITC (Biolegend, San Diego, CA, USA), biotinylated and allophycocyanin (APC)-conjugated CD45, CD31 and TER119 antibodies (BD Pharmingen), and Violet dye (Dead Cell Stain, ThermoFisher Scientific). For intracellular staining of Vim, 1 × 10⁶ cells were fixed with 4% paraformaldehyde and permeabilized with 90% methanol. Anti-Vim was added as primary antibody and Alexa Fluor 594 anti-rabbit (Invitrogen) as secondary antibody. For intracellular staining of ERα, cells were fixed and permeabilized with the Cytofix/Cytoperm fixation/permeabilization kit (BD Pharmingen). ERα staining was performed according to the manufacturer’s instructions. For bromodeoxyuridine (BrdU) incorporation, tumor cells were labeled with 10 μM BrdU (Sigma) for 1.5 h, fixed with 75% ethanol, and permeabilized with 2 M HCl. The cells were then stained with fluorescein isothiocyanate (FITC)-conjugated anti-BrdU (Cell Signaling) and propidium iodide (PI) and analyzed by flow cytometry. FACS was performed using the LSR-Fortessa machine (BD Pharmingen). Data analysis was performed using Kaluza software (Beckman Coulter). To isolate tumor cells depleted of hematopoietic and endothelial cells, we stained mammary tumor cells with biotinylated and APC-conjugated CD45, CD31 and TER119 antibodies and Violet dye. Lin⁻ (CD45⁻, CD31⁻, TER119⁻) living (Violet dye negative) cells were sorted on a BD FACS SORP Aria-IIu machine. For tumorsphere formation assay, 30,000 p18^−/−; Brca1^MGKO mammary tumor cells were plated in triplicate ultra-low attachment plates with or without addition of 17β-estradiol (E2, Sigma) in serum-free DMEM-F12 supplemented with B27, epidermal growth factor (EGF), and basic fibroblast growth factor (bFGF) as described [23, 37]. The number and size of tumorspheres formed were calculated after 7 days.

MCF7 and SUM149 cells were cultured per American Type Culture Collection (ATCC) recommendations. Primary murine mammary tumor cells were cultured in phenol-free DMEM/F12 (Gibco), with 10% charcoal-stripped FBS (Gibco), 10 μg/ml insulin, and 10 ng/ml EGF. For treatment of estrogen, AZD5363, and 4OHT, tumor cells were cultured in 10% charcoal-stripped FBS in the presence of E2, AZD5363 (ApexBio Technology), 4OHT (Sigma), or dimethyl sulfoxide (DMSO) for the indicated times and then collected for further analysis. To determine cell viability, 50,000 cells were plated in 24-well plates and treated with DMSO or drugs at the indicated concentrations for 5 days. Viable cell numbers were determined on day 1, day 3, and day 5 by an automatic cell counter (Bio-rad) with trypan blue exclusion. For western blot, tissue and cell lysates were prepared as previously reported [19, 23, 37]. Primary antibodies used are as follows: HSP90 (Santa Cruz), Gapdh (Ambion), ERα (Santa cruz), Brca1, p-Akt (Ser473), p-4E-bp1 (Thr37/46), p-mTor (Ser2248), p-Gsk3β (Ser9), E-cad, Vim, Snail, Slug, p-Fra1 (Ser 265), p-Rb (ser780) (Cell Signaling) and Fn (Abcam).

For mammary tumor cell transplantation, cells were suspended in a 50% solution of Matrigel (BD) and then inoculated into the left and/or right inguinal mammary fat pads (MFPs) of 6–8-week-old female NSG mice (Jackson Laboratory) respectively. At 6 or 7 weeks after transplantation, animals were euthanized and mammary tumors were dissected for histopathological, immunohistochemical, and biochemical analyses. For PDX tumor tissue transplantation, 4 mm × 2 mm tissue fragments prepared from BRCA1 mutant PDX tumors (TM00091, Jax lab) were transplanted into MFPs of 4-week-old female NSG mice. At 4 weeks after transplantation, or when tumor volume reached the maximal size that the Institutional Animal Care and Use Committee (IACUC) allowed, animals were euthanized and tumors were analyzed. For estrogen treatment in vivo, 0.72 mg E2 (SE-121, IRA, Sarasota, FL, USA) or placebo beeswax pellet was implanted subcutaneously in mice receiving tumor cell or tissue transplants. To analyze the estrogen-induced metastasis from mammary tumors, metastatic p18^−/−;Brca1^MGKO tumor cells were inoculated into the MFPs of 4-week-old female NSG mice in which either E2 or placebo beeswax pellet was implanted subcutaneously. When newly generated tumors reached the maximum size (1.3 cm³) allowed by the IACUC in 3–6 weeks, or the mice became moribund, the lungs and other major organs were examined for detection of metastasis. For quantification of the number of metastatic nodules in the lungs, fixed lung tissues of all five lobes were sagittally sectioned at 200-μm intervals. At least three sections for each lobe were prepared and stained with H.E. The metastatic nodules in each lobe of lung tissue were confirmed by H.E. staining, counted under a microscope, and averaged. The number of nodules in all lobes was then calculated. For AZD5363 treatment of pre-existing mammary tumors, p18^−/−;Brca1^MGKO tumor cells were transplanted into MFPs of NSG mice and allowed to reach ~ 250 mm³ in size. Mice were then treated with AZD5363 ((150 mg/kg solubilized in a 10% DMSO 25% w/v (2-Hydroxypropyl)-β-cyclodextrin buffer (Sigma)) or vehicle by oral gavage once a day. The tumor size was measured daily with a caliper. Tumor volumes were calculated as:

V  =  a  ×  b²/2

All data are presented as the mean ± SD for at least three repeated individual experiments for each group. Quantitative results were analyzed by the two tailed Fisher exact test or two-tailed Student’s t test. P < 0.05 was considered statistically significant.

We previously discovered that the expression of p16 and p18 along with senescence markers was increased in MECs of Brca1^+/− (heterozygous germline deletion of Brca1) and Brca1^MGKO (specific deletion of Brca1 in epithelia) mice [19, 23, 39]. p18^−/− mice in the Balb/c background developed ER-positive luminal-type mammary tumors [37] whereas p18^−/−;Brca1^+/− double mutant mice formed ER-negative basal-like mammary tumors [19]. We generated p16^−/−;Brca1^MGKO and p18^−/−;Brca1^MGKO double-mutant mice in the Balb/c-B6 mixed background. We found that 47% of p18^−/−, 73% of p18^−/−;Brca1^MGKO, 60% of p16^−/−;Brca1^MGKO, and 8% of Brca1^MGKO mice in the Balb/c-B6 mixed background developed mammary tumors, yet no p16^−/− mice developed mammary tumors at similar ages (Table 1 and Additional file 1: Figure S1). Though the mammary tumor incidence of p18^−/− mice in the Balb/c-B6 mixed background was lower than those in the Balb/c background (47% vs. 83%) (Table 1 and Reference [19, 37]), p18^−/− tumors in the Balb/c-B6 mixed background were also predominantly ERα-positive, Ck5/CK14 negative tumors (Table 1 and Additional file 1: Figure S1). These results indicate that loss of p18 induces ER-positive mammary tumors independent of mouse genetic background. Further analysis revealed that only 18% (n = 11) of p18^−/−;Brca1^MGKO and none (n = 6) of p16^−/−;Brca1^MGKO tumors were positive for ERα. These rates were significantly lower than that of p18^−/− tumors (71%, n = 7). Consistent with these results, 82% of p18^−/−;Brca1^MGKO and all of p16^−/−;Brca1^MGKO tumors were positive for Ck5 or Ck14, whereas only 29% of p18^−/− tumors were positive for Ck5 or Ck14 (Table 1 and Additional file 1: Figure S1). These data confirm the role of Brca1 in the suppression of ER-negative, basal-like tumorigenesis in an epithelium autonomous manner.

In mammary and tumor cells ERα expression is restricted to the CD24⁺CD29^low luminal population while the CD24⁺CD29^high population is ERα negative while also enriched with basal and stem cells [42, 43]. Consistently, we previously reported that 40–60% of the cells in most of the ER-positive p18^−/− tumors and less than 10% of the cells in most of the ER-negative p18^−/−;Brca1^MGKO tumors were CD24⁺CD29^low [23, 37]. We detected a predominant CD24⁺CD29^high population containing 60–81% of the cells in most of the p18^−/−;Brca1^MGKO and p16^−/−;Brca1^MGKO tumors (Additional file 2: Figure S2A) [23]. We chose and further characterized two individual p18^−/−;Brca1^MGKO tumors in addition to two individual p16^−/−;Brca1^MGKO tumors in which ERα was undetectable by western blot and IHC (Table 1, Additional file 1: Figure S1, and Fig. 3c); there were more than 75% CD24⁺CD29^high cells and less than 2% CD24⁺CD29^low cells (Additional file 2: Figure S2A and data not shown). Taking advantage of the tumors predominantly composed of CD24⁺CD29^high cells, we transplanted 6 × 10⁴ FACS-sorted Lin⁻ cells (depleted of hematopoietic and endothelial cells) from a p18^−/−;Brca1^MGKO tumor and a p16^−/−;Brca1^MGKO tumor into eight MFPs of NSG mice (four for each cell type). After 1–2 months all recipient mice produced mammary tumors that were ERα negative (placebo group in Table 3, Additional file 2: Figure S2B, C, and data not shown). We cultured these tumor cells and found that most (> 90%) of the cells maintained their CD24⁺CD29^high feature and less than 2% of the cells were CD24⁺CD29^low, whereas, as a control, about half of the p18^−/− tumor cells were CD24⁺CD29^low (Additional file 2: Figure S2D). We determined expression of ERα by FACS and found that all cells derived from four individual p18^−/−;Brca1^MGKO and p16^−/−;Brca1^MGKO tumors were ERα negative whereas, as controls, 59% of MCF7 and 43% of p18^−/− tumor cells were ERα positive, respectively (Additional file 2: Figure S2E). We transplanted 1 × 10⁶ cultured cells from each of the two individual p18^−/−;Brca1^MGKO and p16^−/−;Brca1^MGKO tumors into MFPs of NSG mice. We found all recipient mice also generated ERα-negative mammary tumors, which were indistinguishable from tumors generated by Lin⁻ primary tumor cells in respect to ERα negativity (placebo group in Table 3, Additional file 2: Figure S2F, G, and data not shown). In summary these results demonstrate that transplantation of representative p18^−/−;Brca1^MGKO and p16^−/−;Brca1^MGKO tumor cells into MFPs of NSG mice generate reproducible ER-negative tumors, which build a model system to study BRCA1-deficient mammary tumor development and progression.

Estrogen has been found to activate EMT and stimulate the function of ER-negative breast CSCs through paracrine signaling produced in ER-positive cells in response to estrogen [7, 8]. However, clinically hormone therapy prevents development and suppresses the progression of BRCA1-deficient, ER-negative breast cancers [3, 4]. Inspired by the finding that a fraction of BRCA1-deficient human and mouse tumor cells harbor EMT features [16‐19, 22‐24, 39, 44], and the finding that E2 enhanced the heterogeneity of p18^−/−;Brca1^MGKO and p16^−/−;Brca1^MGKO tumors with elevated squamous metaplasia and spindle-shaped cells (typical morphological characteristics of mesenchymal cells), we hypothesized that estrogen promotes EMT and stimulates BRCA1-deficient, ER-negative tumorigenesis independent of ER. We found by western blot analysis that E2-treated p18^−/−;Brca1^MGKO tumors expressed a higher level of EMT marker (Vim) and EMT-inducing transcription factors (EMT-TFs) including p-Fra1 and Snail, than placebo-treated tumors. Importantly, we noticed by IHC that the number of cells that are positive for EMT markers and EMT-TFs was drastically enhanced in E2-treated p18^−/−;Brca1^MGKO tumors than in placebo-treated counterparts (Fig. 3a and Additional file 4: Figure S4A). We confirmed, as expected, that metastasized mammary tumors in lung were positive for Vim and p-Fra1 (Fig. 2d). These data suggest that estrogen activates EMT in a subset of ER-negative Brca1-deficient epithelial tumor (carcinoma) cells that have not undergone EMT, leading to an increase in the fraction of mesenchymal-like cells in Brca1-deficient tumors.

We examined BRCA1 mutant PDX tumors and discovered that more than 80% of tumor cells expressed a high level of VIM without treatment (Fig. 3b). These data are supported by the findings in both mouse [23] and human mammary tumors [24, 44] that most BRCA1 mutant tumors are VIM positive and some of them are predominantly composed of VIM-positive cells. Though we failed to detect a further increase in the number of VIM-positive cells by E2-treatment in BRCA1 mutant PDX tumors (Fig. 3b), the number of tumor cells that are positive for fibronectin, another key marker of mesenchymal cells, and for SNAIL and SLUG was drastically enhanced by E2 relative to the placebo (Fig. 3b). These results indicate that estrogen also activates EMT in a subset of human BRCA1 mutant tumor cells.

As no tumor formed after the transplantation of p18^−/− tumor cells without an E2 supplement (Table 2), we determined EMT markers in E2-treated tumors and primary tumors. We found no significant change in the expression of E-cad, p-Fra1, and Snail in E2-treated p18^−/− tumors compared to p18^−/− primary tumors (Fig. 3c, Additional file 4: Figure S4C, and data not shown). Notably, except for a slight increase in Slug in a longer treatment, the expression of E-cad, Vim, Snail, and p-Fra1 in p18^−/− tumor cells in culture was not affected by E2 either, whereas, the expression of Vim, Snail, and Slug in p18^−/−;Brca1^MGKO tumor cells was clearly enhanced by E2 under the same culture conditions (Fig. 3e, g and Additional file 6: Figure S6A-D). Together, these results indicate that estrogen promotes EMT in Brca1-deficient, but not in Brca1-proficient, tumor cells.

Because the number of cells with mesenchymal features varies in different BRCA1-deficient tumors, we screened a panel of primary p18^−/−;Brca1^MGKO tumor cells by FACS for expression of Vim and identified at least two types of p18^−/−;Brca1^MGKO tumor cells, and in one of these, Vim-positive cells constituted 30–65% of the total tumor cells (type 1) and in the other, Vim-positive cells constituted > 90% of the total cells (type 2) (Fig. 3d-h and Additional file 6: Figure S6C). Due to the enrichment of EMT-promoting factors including transforming growth factors (TGFs) in serum, culture immortalized MECs in serum in vitro activates EMT [45]. Therefore, we cultured tumor cells in charcoal-stripped FBS with minimal hormones and cytokines (http://​www.​thermofisher.​com) and performed all of our in vitro experiments using either primary or early passaged (before passage 3) tumor cells.

We treated p18^−/−;Brca1^MGKO tumor cells with E2 in multiple time periods and observed by western blot that E2 stimulated expression of Vim and EMT-TFs including p-Fra1 and/or Snail/Slug in both type of cells (Fig. 3d-h, Additional file 6: Figure S6), indicating E2 promotes EMT in Brca1-deficient tumor cells in vitro. As the majority of BRCA1-deficient tumors contain a fraction of cells that are positive for VIM and/or other mesenchymal markers, and only a small number of human BRCA1-deficient tumors, i.e. metaplastic breast cancers, are predominantly composed of mesenchymal/VIM-positive cells [16‐19, 24, 39, 44, 46], we therefore focused on type 1 p18^−/−;Brca1^MGKO tumor cells. When type 1 p18^−/−;Brca1^MGKO tumor cells were examined by immunofluorescent staining and FACS, we found that E2 treatment enhanced the number of Vim-positive cells - converting Vim-positive cells from 57-62% to 80–90% after 72 h (Fig. 3d, f, h). Furthermore, the increase in Vim expression and conversion of cells from Vim-negative to positive by E2 in p18^−/−;Brca1^MGKO tumor cells was not blocked by 4OHT (Fig. 3g, h), whereas, 4OHT effectively blocked E2-enhanced cell proliferation and RB phosphorylation in ER-positive MCF7 cells, as expected (Additional file 7: Figure S7). These results demonstrated that estrogen activates EMT in a subset of Brca1-deficient tumor cells that have not undergone EMT, and further indicated that estrogen promotes EMT in ER-negative Brca1-deficient tumor cells independent of ER.

We then selected a BRCA1 mutant, ER-negative human breast cancer cell line, SUM149, in which p16 is deleted and VIM is expressed in 10–30% of total cells [35, 47‐49]. This cell line shares many genetic (p16/p18 and Brca1 loss-of-function mutation) and phenotypical (ER negative with a subset of cells exhibiting EMT features) similarities with p16;Brca1 and p18;Brca1 mutant murine mammary tumor cells. We found that E2 treatment also drastically enhanced the number of Vim-positive cells - from 19% to 50% after 72 h (Fig. 3i). Notably, it has been reported that in the SUM149 cell line, VIM-positive, stem/basal cells are able to efficiently generate both basal and luminal cells, whereas, VIM-negative, luminal cells rarely generate VIM-positive, stem/basal cells under serum-containing culture condition [47]. In addition, we also detected by western blot the increase in VIM, p-FRA1, and SNAIL in SUM149 cells in response to E2 treatment (Additional file 6: Figure S6E), indicating the activation of EMT by estrogen.

Along with activation of the Akt pathway, estrogen promoted EMT and proliferation in Brca1-deficient mammary tumor cells, which prompted us to test whether pharmaceutical inhibition of Akt activity has any effect on Brca1-deficient cell proliferation and tumor progression. We treated p18^−/−;Brca1^MGKO tumor cells with E2 in the presence or absence of AZD5363, a well-studied preclinical Akt inhibitor [50]. We found that AZD5363 drastically inhibited E2-enhanced expression of p-4Ebp1, p-mTor, and p-Gsk3β, and that of Vim and p-Fra1 (Fig. 6a), suggesting that AZD5363 efficiently suppresses estrogen-enhanced Akt pathway and EMT program in Brca1 deficient tumor cells. Treatment of p18^−/−; Brca1^MGKO tumor cells including both type 1 and type 2 cells with AZD5363 significantly reduced the E2-enhanced number of cells and incorporation of BrdU (Fig. 6b, c and Additional file 8: Figure S8C), indicating that AZD5363 inhibits estrogen-enhanced proliferation of Brca1-deficient tumor cells.

We then determined if pharmaceutical inhibition of Akt activity suppresses the progression of pre-existing Brca1-deficient tumors. Transplanted p18^−/−;Brca1^MGKO tumors with E2 supplement were allowed to reach ~ 250 mm³ in size and then mice were treated with vehicle or AZD5363 daily. Three days after treatment, tumors from AZD5363-treated mice began to show a significant size reduction in comparison with the tumors from vehicle-treated animals. After 7-day treatment, tumors from vehicle-treated mice reached ~ 1475 mm³ in size, whereas those from AZD5363 -treated mice only reached ~ 188 mm³, which was comparable with that of the tumor size at the start of treatment (~ 224 mm³) (p = 0.59) (Fig. 6d), indicating that treatment with AZD5363 prevents estrogen-enhanced Brca1-deficient tumor progression.

Further analysis of tumors by IHC and western blot revealed that in tumors from mice with E2 supplement the expression of p-4Ebp1, p-mTor, and p-Gsk3β was clearly reduced by AZD5363 treatment relative to vehicle treatment (Fig. 6e and Additional file 8: Figure S8D), confirming the role of AZD5363 in inhibition of the estrogen-induced Akt pathway in Brca1-deficient tumor progression. Notably, the expression of Vim, Fn and Slug in tumors with E2 was drastically reduced and that of E-cad was enhanced in AZD5363-treated mice relative to control (Fig. 6e, f). These data demonstrate that AZD5363 efficiently suppresses estrogen-activated EMT in Brca1-deficient tumors in vivo. Consistent with the findings derived from p18^−/−;Brca1^MGKO tumor cells in vitro (Fig. 6b), we also detected that AZD5363 significantly reduced the number of Ki67 and Ck14 double-positive cells stimulated by E2 in p18^−/−;Brca1^MGKO tumors (Fig. 6g). Together, these results indicate that estrogen promotes EMT and proliferation of Brca1-deficinet tumor cells leading to tumor progression, which is dependent on the activation of the Akt pathway.