Using novel murine models of claudin-low and basal-like breast cancer, we tested the hypothesis that diet-induced obesity (DIO) and calorie restriction (CR) differentially modulate progression of these aggressive breast cancer subtypes. For model development, we characterized two cell lines, “mesenchymal (M)-Wnt” and “epithelial (E)-Wnt,” derived from MMTV-Wnt-1 transgenic mouse mammary tumors. M-Wnt, relative to E-Wnt, cells were tumor-initiating cell (TIC)-enriched (62% vs. 2.4% CD44high/CD24low) and displayed enhanced ALDEFLUOR positivity, epithelial-to-mesenchymal transition (EMT) marker expression, mammosphere-forming ability, migration, invasion, and tumorigenicity (P < 0.001; each parameter). M-Wnt and E-Wnt cells clustered with claudin-low and basal-like breast tumors, respectively, in gene expression profiles and recapitulated these tumors when orthotopically transplanted into ovariectomized C57BL/6 mice. To assess the effects of energy balance interventions on tumor progression and EMT, mice were administered DIO, control, or CR diets for 8 weeks before orthotopic transplantation of M-Wnt or E-Wnt cells (for each cell line, n = 20 mice per diet) and continued on their diets for 6 weeks while tumor growth was monitored. Relative to control, DIO enhanced M-Wnt (P = 0.01), but not E-Wnt, tumor progression; upregulated EMT- and TIC-associated markers including N-cadherin,fibronectin, TGFβ, Snail, FOXC2, and Oct4 (P < 0.05, each); and increased intratumoral adipocytes. Conversely, CR suppressed M-Wnt and E-Wnt tumor progression (P < 0.02, each) and inhibited EMT and intratumoral adipocyte accumulation. Thus, dietary energy balance interventions differentially modulate EMT and progression of claudin-low and basal-like tumors. EMT pathway components may represent targets for breaking the obesity–breast cancer link, particularly for preventing and/or controlling TIC-enriched subtypes such as claudin-low breast cancer. Cancer Prev Res; 5(7); 930–42. ©2012 AACR.

Obesity increases risk and progression of breast cancer in postmenopausal women (1), whereas intentional weight loss or calorie restriction (CR) moderately protects against breast cancer (2, 3). Breast cancers are multiple distinct diseases, with intrinsic molecular subtypes categorized as basal-like, luminal A, luminal B, triple-negative, claudin-low, or Her-2-positive breast cancer (4, 5). Epidemiologic data suggest that the strength of the obesity–breast cancer link varies by intrinsic breast cancer subtype and differentiation status (6). While obesity is a well-established risk and prognostic factor for the luminal A breast cancer subtype in postmenopausal women (6, 7), the relationships between dietary energy balance and claudin-low or basal-like breast cancers are not well established, due partially to a paucity of relevant experimental model systems.

Features of claudin-low, and to a lesser extent basal-like, breast tumors include Wnt/β-catenin pathway activation, stem cell–like gene expression, and poor morphologic differentiation (4, 8). The Wnt/β-catenin pathway is associated with the epithelial-to-mesenchymal transition (EMT), which is a key developmental program and regulator of tumor-initiating cells (TIC) possessing stem/progenitor cell properties (8). Links between dietary energy balance and EMT in breast cancer are plausible but remain unclear. TGFβ, a master regulator of EMT, is overexpressed in multiple tissues from obese humans and rodents (9, 10). Snail and Twist, transcription factors that regulate E-cadherin and trigger EMT, are upregulated in melanoma cells treated in vitro with serum from obese mice (11). EMT-driven TIC enrichment correlates with hallmarks of cancer progression, including proliferation, angiogenesis, invasion, metastasis, wound healing, and therapeutic resistance (12). Given the poor prognosis conferred by claudin-low and basal-like breast cancers (6), identification of mechanistic targets and strategies to prevent or control these breast cancer subtypes is critical.

Here, we describe 2 cell lines that recapitulate claudin-low and basal-like breast tumors, respectively, when orthotopically transplanted into ovariectomized C57BL/6 mice. Ovariectomy was used to mimic the postmenopausal state, which is associated with increased susceptibility to weight gain and breast cancer in women. Both cell lines were derived from spontaneous mammary tumors from mouse mammary tumor virus (MMTV)-Wnt-1 transgenic mice. Using these cells in vitro and in vivo, we tested the hypothesis that alterations in dietary energy balance, specifically diet-induced obesity (DIO) and CR, modulate claudin-low and basal-like mammary tumor progression, possibly through EMT-related pathways.

All animal studies and procedures were approved and monitored by the University of Texas (Austin, TX) Institutional Animal Care and Use Committee.

Generation and authentication of M-Wnt and E-Wnt tumor cell lines

Twenty clonal cell lines were derived from spontaneous mammary tumors from MMTV-Wnt-1 transgenic mice, as previously described (13). In brief, excised tumors were dissected, mechanically dissociated, and forced through 40-μm mesh. Viable cells were plated at low density in 100-mm plates, grown at 37°C in 5% CO2 in RPMI-1640 media, 10% FBS, penicillin/streptomycin and glutamine (complete media; all components from HyClone). One clone (denoted “M-Wnt”) had mesenchymal morphology by microscopic evaluation and was selected for further characterization. Of the remaining 19 clones, all had virtually identical epithelial morphology, and a randomly selected subset (4 clones) had similar cell surface marker expression (see below). From that subset, one clone (denoted “E-Wnt”) was randomly selected for further characterization.

The M-Wnt and E-Wnt cell lines were tested for species verification, karyotyping, and genomic instability and were authenticated by the Molecular Cytogenetics Core facility at the University of Texas MD Anderson Cancer Center (Houston, TX). Before in vivo use, M-Wnt and E-Wnt cells were trypsinized, washed with PBS, and viable cells were quantified by trypan blue exclusion using a hemocytometer (Fisher Scientific).

In vitro characterization of M-Wnt and E-Wnt cell lines

Flow cytometric analysis of CD44 and CD24 cell surface markers.

Cells were grown to 70% confluence in complete media, scraped or trypsinized, and stained with allophycocyanin anti-mouse CD44 and/or phycoerythrin anti-mouse CD24 (BD Biosciences; refs. 14, 15). Rat IgG (BD Biosciences) was used as isotype control (14). Cell staining was evaluated (4 biologic replicates conducted on different days) using flow cytometry (FACSCalibur, BD Biosciences), with data from 1 × 104 or more cells acquired and analyzed (CellQuest Pro software, BD Biosciences; ref. 14).

Flow cytometric analysis of aldehyde dehydrogenase activity.

The ALDEFLUOR Kit (StemCell Technologies) was used per manufacturer's instructions to identify and enumerate by flow cytometry (Accuri C6 Flow Cytometer, Accuri) the population of cells with high aldehyde dehydrogenase (ALDH) enzymatic activity (16). In brief, cells were incubated with an ALDH fluorescent substrate either with the ALDH inhibitor, diethylaminobenzaldehyde (DEAB; to establish baseline fluorescence), or without DEAB (to define the ALDEFLUOR-positive region) before sorting. Gates were set to include viable, 7-aminoactinomycin D–negative cells (7 biologic replicates conducted on different days).

Mammosphere-forming capacity.

Cells were trypsinized, washed in PBS, and plated in a serial dilution (from 60 to 1 cell per well) in a 96-well plate (6 replicates per dilution). Cells were cultured under serum-free conditions (Dulbecco's Modified Eagle's Medium:Nutrient Mixture F12 supplemented with insulin, B27, N-2, EGF, and fibroblast growth factor; BD Biosciences) in low-adherence plates (Corning) and fed weekly. At 14 days, mammospheres with more than 5 cells per sphere were counted (×20 magnification), photomicrographs of representative mammospheres taken (n = 5 per cell line), and diameters measured using SPOT camera software.

Scratch assay of cell migration.

Cells were grown to confluency at 37°C in 5% CO2 in complete media, wounded using a 10-μL pipette tip, and allowed to migrate. Photomicrographs of the wound were taken immediately and 12 hours later (3 separate experiments, each in triplicate).

Invasion assay.

Cells were plated (2.5 × 104 cells per chamber) in serum-free RPMI-1640 media in Matrigel-coated invasion chambers (BD Biosciences) and allowed to invade (10% FBS as chemoattractant) for 8, 18, or 30 hours. Noninvasive cells were wiped from the upper portion of the membrane. Invasive cells (lower side of the membrane) were stained with 1% crystal violet in methanol for 30 minutes, washed twice in water, mounted on slides, and counted at ×20 magnification (3 separate experiments, each in triplicate).

Relative tumorigenicity of M-Wnt and E-Wnt cells

Animals and study design.

Upon arrival after purchase (Charles River), 110 ovariectomized, 6- to 8-week-old female C57BL/6 mice were singly housed and fed modified AIN-76A diet (catalog no. #D12450B, Research Diets, Inc.) ad libitum. Following a 4-week adaptation period, mice were orthotopically transplanted in the ninth mammary fat pad (17), with either M-Wnt cells (1 × 107, 1 × 106, 5 × 105, 2 × 105, 5 × 104, 5 × 103, 500, or 50 cells per mouse) or E-Wnt cells (5 × 107, 1 × 107, 5 × 106, 5 × 105, 5 × 104, 5 × 103, 500, or 50 cells per mouse). For each cell line, n = 10 mice per group for the first 3 groups listed; otherwise, n = 5 mice per group.

Once detected, tumors were measured twice weekly in 2 perpendicular dimensions using electronic calipers; cross-sectional area was calculated (maximal length × width; mm2). When tumor diameter reached 1.0 cm in either dimension (or at 16 weeks, whichever occurred first), mice were euthanized by cervical dislocation following isoflurane anesthesia; also for each group with 10 mice per group, 5 mice were randomly selected at 3 weeks and euthanized. Tumors were excised, measured, and weighed.

Tumor processing and storage.

Tumors were equally divided into 3 portions that were (i) fixed in 10% neutral-buffered formalin for 24 hours, transferred to 70% ethanol for 24 hours, embedded in paraffin, and cut into 4-μm thick sections for hematoxylin and eosin (H&E) staining or immunohistochemical analysis; (ii) placed in a cryotube, flash-frozen in liquid nitrogen, and stored at −80°C for subsequent molecular analyses; or (iii) flash-frozen in optimal cutting temperature medium (Tissue-Tek) and stored at −80°C for subsequent immunofluorescence analysis.

Molecular characterization of M-Wnt and E-Wnt cell lines or tumors

Microarray gene expression analysis.

RNA was purified from M-Wnt and E-Wnt cells using RNeasy Mini Kit (Qiagen). Microarray hybridizations were conducted as previously described (5), except that the new arrays reported here were hybridized to custom 180K Agilent microarrays (BARCODE25503) and scanned using an Agilent Technologies Scanner G2505C with Feature Extraction software. Microarray hybridization data from the study of Herschkowitz and colleagues (ref. 5; GSE3165) were clustered along with data from the new arrays reported here (GSM652368-GSM652393). The intrinsic gene list used to cluster the arrays was pared from 866 genes to the 666 genes common across both array platforms. A cross-platform normalization factor was computed on the basis of mouse models common to both platforms, including MMTV-Neu and C3(1)/SV40 large T-antigen (C3-Tag) transgenic mice. Arrays were then mean-adjusted and median-centered.

Real-time quantitative reverse transcription-PCR of EMT panel.

Total RNA was extracted from cell lines using RNeasy Mini kit (Qiagen) and from tumor samples using FastRNA Pro Green Kit (MP Biomedicals). RNA was reverse transcribed with Multiscribe RT (Applied Biosystems). The resulting cDNAs were assayed in triplicate for PCR using TaqMan Gene Expression Assays for an EMT panel that includes E-cadherin, N-cadherin, fibronectin, vimentin, Snail, Twist, Slug, forkhead box C2 (FOXC2), TGFβ, Oct4, and Wnt-1 (Applied Biosystems). PCR and data collection were conducted on a Mastercycler RealPlex 4 (Eppendorf). Gene expression data were normalized to β-actin.

Immunostaining

Immunohistochemistry.

Immunohistochemical staining was conducted as previously described (18) using primary antibodies for Ki67 (dilution = 1:200; catalog #M7249, DAKO Cytomation), CD31 (1:400; Clone MEC 13.3, BD Biosciences), phospho-histone H3 (pHH3, Ser10; 1:1,000; catalog #06-570, Upstate Biotechnology), estrogen receptor (ER)-α (1:500; catalog #sc542, Santa Cruz Biotechnology), and progesterone receptor (PR; 1:100; catalog #ab2764, Abcam). Secondary antibody was horseradish peroxidase–labeled anti-rabbit antibody (DAKO Cytomation).

Immunofluorescence.

Frozen tumor sections embedded in optimal cutting temperature media were cut (4-μm slices), washed with PBS, fixed in 4% paraformaldehyde, permeabilized (2 minutes) in 0.1% Triton X-100, and neutralized (5 minutes) in 100 mmol/L glycine. Immunofluorescence staining was conducted as previously described (19). Primary antibodies (1:300 dilution; BD Biosciences) included anti-E-cadherin (catalog #610181), anti-N-cadherin (catalog #610920), and anti-fibronectin (catalog #610077). Secondary antibody was fluorescein isothiocyanate (FITC)-conjugated, donkey anti-mouse IgG (1:600; catalog #715095150, Jackson ImmunoResearch Laboratories). Slides were counterstained with 4′,6-diamidino-2-phenylindole (DAPI; Invitrogen). Photomicrographs were taken with equal exposure on a Zeiss Axiovert 200M microscope (oil immersion objective; ×60 or ×100 magnification) with appropriate filters (Zeiss).

Effects of energy balance in models of claudin-low and basal-like breast cancer

Upon arrival, 120 ovariectomized, 6- to 8-week-old female C57BL/6 mice (Charles River) were singly housed and randomized (n = 40 per diet group) to receive 1 of 3 dietary regimens for 14 weeks (pelleted diets from Research Diets, Inc.; Supplementary Table S1): (i) Control diet, fed ad libitum (modified AIN-76A diet; catalog #D12450B), providing 3.8 kcal/g; (ii) 30% CR diet (catalog #D0302702); and (iii) DIO diet, fed ad libitum (catalog #D12492), providing 5.2 kcal/g with 60% kcal from fat. CR mice received a modified formulation of control diet providing 70% of the mean daily caloric consumption (and 100% of vitamins, minerals, fatty acids, and amino acids) of control mice (Supplementary Table S1). Mice were weighed weekly.

After 8 weeks on diet, mice were analyzed for percentage of body fat using quantitative magnetic resonance (Echo Medical Systems). Mice were orthotopically injected in the ninth mammary fat pad with 5 × 104 M-Wnt cells/mouse or 5 × 106 E-Wnt cells/mouse (n = 20 mice per diet for each cell line). The cell numbers used were based on their ability to generate comparably sized tumors in preliminary studies. Cross-sectional tumoral area was determined twice weekly (as described above).

Six weeks later, mice were fasted for 6 hours and then euthanized. Blood was collected by cardiac puncture, coagulated for 30 minutes (room temperature), and centrifuged at 10,000 × g for 5 minutes; serum was removed and stored at −80°C for subsequent analyses. Tumors were excised, measured, and processed and stored (as above) for immunohistochemical and immunofluorescence analyses. Serum from 12-to 13 randomly selected mice per group was analyzed for glucose (by Ascencia Elite Glucometer, Bayer); leptin, insulin, resistin, insulin-like growth factor (IGF)-1, interleukin (IL)-6, and adiponectin (by Luminex-based LINCOplex bead array assay, Millipore; read on multianalyte detection system, BioRad); and 17-β estradiol (by ELISA; Alphay Diagnostics).

Statistical analyses

Summarized data are reported as mean ± SEM. Differences were assessed using the unpaired Student t test (mammosphere size, cellular populations by flow cytometry, relative gene expression), one-way ANOVA followed by Tukey post hoc test (mammosphere counts, invasion capacity, body weight, percent body fat, serum analytes), one-way repeated measures ANOVA followed by Tukey post hoc test (serial cross-sectional tumor areas), or the Mann–Whitney U test (final tumor area). Significance was declared at P < 0.05.

Development of M-Wnt and E-Wnt mammary tumor models

In vitro characterization of M-Wnt and E-Wnt cell lines.

The M-Wnt and E-Wnt cell lines, each isolated from a MMTV-Wnt-1 mammary tumor, were characterized for morphology, mammosphere-forming capacity, cell surface expression of CD44 and CD24, ALDH activity, and migration and invasion capability (Fig. 1).

When grown in adherent culture in complete media, M-Wnt cells displayed a mesenchymal appearance, including elongated spindle-like cytoplasm, consistent with human claudin-low breast tumor cells. In contrast, E-Wnt cells had a rounded appearance typical of epithelium and more consistent with most murine and human basal-like breast cancer cell lines (Fig. 1A, top).

Mammosphere-forming capacity, which typically identifies a more tumorigenic cell population that has undergone EMT (20), was more pronounced with M-Wnt than E-Wnt cells. When grown under nonadherent conditions for 14 days, M-Wnt cells, compared with E-Wnt cells, formed significantly larger mammospheres (mean ± SEM diameter: 378 ± 14.0 μm vs. 75.5 ± 10.4 μm, P < 0.001; Fig. 1A, middle). When plated in a limiting dilution (from 60 to 1 cell per well), M-Wnt cells, compared with E-Wnt cells, consistently formed significantly more mammospheres (P < 0.001 for all dilutions; Fig. 1A, bottom).

By flow cytometric analysis of unsorted cells, the M-Wnt cell line (relative to the E-Wnt cell line) was enriched (62.2% ± 7.8% vs. 2.4% ± 2.0%, P < 0.001; Fig. 1B) in CD44high/CD24low cells associated with murine mammary TICs (15).

Cells with increased ALDH activity exhibit stem cell properties (16). High ALDH activity, as assessed by flow cytometry using the ALDEFLUOR assay, occurred in a greater population of M-Wnt cells than E-Wnt cells (6.95% ± 1.01% vs. 0.53% ± 0.04%, P < 0.001; Fig. 1C). Consistently for each cell line, DEAB inhibition of ALDH significantly decreased ALDEFLUOR positivity (to 0.13% + 0.13% for M-Wnt, P < 0.001; and to 0.23% + 0.18% for E-Wnt, P < 0.05), and cell viability in assays (n = 7) was high (98.7% ± 0.13% for M-Wnt and 98.7% ± 0.09% for E-Wnt).

EMT is functionally characterized by a loss of cell–cell adhesion and an increase in migration and invasion (21). M-Wnt, but not E-Wnt, cells showed migratory capacity in a scratch assay for cell migration (Fig. 1D). In an invasion chamber assay, M-Wnt cells were significantly more (up to 226-fold) invasive than E-Wnt cells (P < 0.001 for assessments at both 18 and 30 hours; Fig. 1E).

Tumorigenicity of M-Wnt and E-Wnt cells and biologic features of resultant tumors.

To compare the tumorigenicity of (unsorted) M-Wnt and E-Wnt cell lines, cells were orthotopically transplanted into the mammary fat pads of female C57BL/6 mice in a limiting dilution analysis (1 × 107 to 50 cells for M-Wnt, and 5 × 107 to 50 cells for E-Wnt, per animal). Tumors were generated with as few as 50 M-Wnt cells, whereas at least 5 × 105 E-Wnt cells were required for tumor formation (P < 0.001; Fig. 2A, left). By 3 weeks after transplantation of 5 × 105 or more cells, M-Wnt tumors were greater in weight (Fig. 2A, right) and size (data not shown) than E-Wnt tumors, and tumors generated from the same number (1 × 107) of cells weighed 2.09 ± 0.30 g for M-Wnt and 0.03 ± 0.01 g for E-Wnt (P < 0.001). Gross necropsy examination found no evidence of metastatic disease in liver or lungs.

The M-Wnt tumors were poorly differentiated with metaplastic morphology, lacked ductal structures, and had intratumoral adipocytes and large areas of central necrosis; by comparison, E-Wnt tumors had clearly defined, albeit disorganized, ductal structures and basal-like morphology reminiscent of the spontaneous MMTV-Wnt-1 tumors from which the cell lines were derived (22) and lacked intratumoral adipocytes (Fig. 2B). The M-Wnt tumors, relative to E-Wnt tumors, had increased cellular proliferation (Ki67 staining), mitotic index (pHH3 staining), and microvascular proliferation (CD31 staining; Fig. 2C). E-Wnt, but not M-Wnt, tumors expressed ER-α (Fig. 2C); neither expressed PR (data not shown).

Molecular characterization of M-Wnt and E-Wnt cells and tumors.

In microarray analyses (Fig. 3), M-Wnt cells clustered tightly with previously identified claudin-low mouse and human tumors (4), which show inconsistent expression of basal keratins (keratins 5, 14, and 17) and low expression of claudin 3, claudin 7, HER2, and luminal markers such as ER, PR, GATA3, and keratins 18 and 19. In contrast, E-Wnt cells clustered tightly with mammary tumors from C3-Tag transgenic mice, an established model of basal-like breast cancer (5, 23, 24). M-Wnt cells, relative to E-Wnt cells, had significantly lower expression of E-cadherin and higher expression of N-cadherin, fibronectin, and vimentin (P < 0.001, each gene; Fig. 4A, top). M-Wnt cells, relative to E-Wnt cells, consistently overexpressed the EMT- and TIC-associated genes Snail (P = 0.02), Twist (P < 0.001), Slug (P < 0.001), FOXC2 (P = 0.09) and TGFβ (P < 0.001). In the tumors, directional trends in gene expression mirrored those of the unsorted cells (Fig. 4A, bottom). No differences in Oct4 mRNA expression were detected between M-Wnt and E-Wnt cells or between their resultant tumors (data not shown). By immunofluorescent staining, E-cadherin was negligible in M-Wnt tumors, whereas E-Wnt tumors had E-cadherin–positive ductal structures (Fig. 4B).

Dietary energy balance interventions in orthotopic transplant models of claudin-low (M-Wnt) and basal-like (E-Wnt) breast cancer

Generation of DIO, control, and CR phenotypes.

Ovariectomized female C57BL/6 mice administered DIO, control, or CR diet regimens manifested 3 distinct phenotypes: obese, control, and lean. At 8 weeks, body weights were 32.2 ± 0.04 g in DIO mice (P < 0.001 vs. control), 25.6 ± 0.03 g in control, and 17.3 ± 0.03 g in CR mice (P < 0.001 vs. control). Similarly, percentage of body fat decreased from 42.0% ± 0.05% in DIO mice (P < 0.001 vs. control) to 26.1% ± 0.05% in control and to 19.1% ± 0.05% in CR mice (P < 0.001 vs. control). Significant differences in serum levels of energy balance–related hormones and cytokines after 14 weeks of diet included increased leptin, insulin, resistin, IGF-1, and IL-6 in DIO mice, relative to control (P < 0.05 for all); and decreased leptin, insulin, and 17-β estradiol, and increased adiponectin in CR mice, relative to control (P < 0.05, each; Fig. 5). Fasting serum glucose was reduced in CR mice, relative to control mice (P < 0.05).

Effects of DIO and CR, relative to control, on tumor progression.

The DIO, control, and CR mice were orthotopically transplanted at 8 weeks with M-Wnt and E-Wnt cells to model claudin-low and basal-like breast tumors, respectively. The effect of dietary energy balance interventions on tumor growth over a 6-week period differed by tumor type: DIO, relative to control, significantly enhanced progression of M-Wnt tumors (P = 0.011), but not E-Wnt tumors; and CR, relative to control, significantly diminished progression of both M-Wnt (P = 0.012) and E-Wnt (P = 0.001) tumors (Fig. 6A, top). The relative between-group differences in tumor area, as assessed in situ or ex vivo, mirrored each other (Fig. 6A, bottom). Relative to control diet, DIO increased, whereas CR decreased, necrosis and intratumoral adipocytes in both M-Wnt and E-Wnt tumors (Fig. 6B). Gross necropsy examination found no evidence in any diet group of metastatic disease in liver or lungs.

Effects of DIO and CR, relative to control, on EMT.

Consistently for both claudin-low (M-Wnt) and basal-like (E-Wnt) tumors, DIO (relative to control) promoted EMT, as evidenced by decreased E-cadherin, and increased fibronectin and N-cadherin protein expression; whereas CR (relative to control) promoted a shift toward an epithelial phenotype, as evidenced by increased E-cadherin, and decreased fibronectin and N-cadherin protein expression (Fig. 7A). Specifically, tumors from DIO mice had negligible E-cadherin, whereas tumors from CR mice had negligible fibronectin and N-cadherin. Furthermore, consistently for both M-Wnt and E-Wnt tumors, DIO (relative to control) promoted expression of various EMT- and TIC-associated genes, including TGFβ (P < 0.02 for both), Snail (P < 0.03 for both), FOXC2 (P < 0.001 for both), and Oct4 (P < 0.005 for both), whereas CR had no detectable effect on their expression (Fig. 7B). An effect of DIO or CR on Twist or Slug gene expression was not detected (data not shown). Diet had no significant effect on the level of Wnt-1 expression (data not shown).

In our novel murine syngeneic transplant models of claudin-low and basal-like breast cancers, dietary energy balance interventions (DIO or CR) differentially modulate EMT and tumor progression. The models involve 2 cell lines, M-Wnt and E-Wnt (each derived from a spontaneous MMTV-Wnt-1 mouse mammary tumor), orthotopically transplanted into ovariectomized C57BL/6 mice to emulate claudin-low and basal-like breast cancer, respectively. We found that (i) DIO increases, whereas CR inhibits, claudin-low (M-Wnt) tumor progression; (ii) CR inhibits, whereas DIO does not affect, basal-like (E-Wnt) tumor progression; and (iii) DIO promotes, whereas CR suppresses, EMT as evidenced by altered protein expression of epithelial (e.g., E-cadherin) and mesenchymal (e.g., fibronectin, N-cadherin) markers in M-Wnt and E-Wnt tumors.

Mouse model studies of claudin-low mammary cancer and/or TICs typically involve xenotransplantation of sorted cell lines or ex vivo cells into immunodeficient mice (25). Xenograft models are limited for studying links between energy balance, TICs, and breast cancer because immunodeficient mice have aberrant mammary gland development, lack normal immune/inflammatory responses, and resist developing DIO and CR phenotypes. Syngeneic transplant models of claudin-low breast cancer derived from p53-null or IGF-1 receptor–overexpressing mouse mammary tumors (23, 26) appear poorly suited for modeling energy balance/breast cancer links given the established roles of p53 and IGF-1 pathways in the anti-cancer effects of many interventions (27-29). Our M-Wnt transplant model of claudin-low mammary cancer overcomes many existing limitations by using (i) cells derived from a spontaneous MMTV-Wnt-1 mouse mammary tumor that, like basal-like breast cancers in women, are responsive to CR and obesity (17, 30) and (ii) a wild-type, syngeneic host with normal immune function, mammary gland development, and metabolic responses to energy balance modulation. In addition, a matched transplant model of basal-like mammary cancer, also using a MMTV-Wnt-1 mammary tumor-derived cell line (E-Wnt), complements the claudin-low model to facilitate direct comparisons between these understudied intrinsic subtypes.

The M-Wnt cell line is among the few reported murine mammary cancer cell lines to closely mimic the pathology and molecular profile of human claudin-low breast tumors (4, 23, 26). M-Wnt cells have mesenchymal morphology, are stably enriched in putative TICs (>60% CD44high/CD24low, with high ALDH activity), display molecular signatures of EMT resembling human claudin-low breast tumors, readily form mammospheres in suspension culture, are highly invasive and migratory, and generate ER-negative, PR-negative claudin-low mammary tumors when as few as 50 cells are injected orthotopically. In contrast, E-Wnt cells have epithelial morphology, display molecular profiles similar to human and mouse basal-like breast tumors, and (versus M-Wnt cells) are TIC-sparse (<5% CD44high/CD24low, with low ALDH activity), poorly form mammospheres, have minimal invasion and migration capacity, and form basal-like mammary tumors (PR-negative, weakly ER-positive) when 5 × 105 cells are injected orthotopically. Transplanted MMTV-Wnt-1 tumor brei grows well in ovariectomized C57BL/6 female mice (17) and MMTV-Wnt-1 mice form spontaneous mammary tumors when crossed with ER-α knockout mice or treated with tamoxifen, suggesting that Wnt-driven mammary tumorigenesis is not estrogen-dependent (28, 31). With M-Wnt or E-Wnt cell lines, no presorting of cells is required before transplantation, and the morphologic and molecular features of the transplanted cells are recapitulated in resultant tumors.

Our findings of dietary effects on M-Wnt and E-Wnt tumors indicate a mechanistic link between energy balance, EMT, and TICs in breast cancer progression. We speculate that DIO prepares fertile soil (tumor microenvironment), including changes in EMT, intratumoral adipocytes, and local and systemic hormones, growth factors, and cytokines, for enhanced tumor progression and that determinants of growth in this fertile soil include the plant variety (intrinsic breast cancer subtype) and/or the seed density (extent of TIC enrichment). In contrast, CR may discourage tumor progression by acting on the soil antithetically to DIO, including promoting epithelial differentiation, discouraging EMT, preventing intratumoral adipocytes, and decreasing systemic hormones, growth factors, and cytokines. Future studies are warranted to determine whether the mammary tumor-enhancing effects of DIO depend on the extent of TIC-enrichment in different subtypes of breast cancer and whether CR targets different (and perhaps TIC-independent) pathways than DIO to impact breast cancer progression.

The present findings extend our previous reports (30, 32) and are the first (to our knowledge) to establish an effect of dietary energy balance interventions on EMT in mammary cancer. We report that in both M-Wnt and E-Wnt tumors, DIO deceases E-cadherin and increases N-cadherin and fibronectin protein expression, whereas CR increases E-cadherin expression. DIO in both tumor types also increases expression of several EMT- and TIC-associated genes, including TGFβ, Snail, FOXC2, and Oct4, each of which is modulated by obesity-related growth factors (9, 10, 11, 33, 34). Components of EMT may thus represent novel targets for preventing and/or controlling breast cancer and novel biomarkers of response to energy balance modulation or other interventions, particularly in obese women. In contrast, CR did not impact TIC-associated gene expression, which suggests that DIO and CR exert their effects on mammary tumor progression via some shared and some distinct pathways.

Although obesity, inflammation, and breast cancer are individually associated with upregulation of TGFβ and other EMT pathway components (9, 10), their combined interactions are poorly characterized. In our study, DIO increased EMT marker expression (including TGFβ) and intratumoral adipocyte accumulation, whereas CR decreased EMT and prevented intratumoral adipocyte accumulation, in M-Wnt and E-Wnt tumors. Furthermore, intratumoral adipocytes were present in M-Wnt, but not E-Wnt, tumors from control diet–fed mice. Thus, connections between EMT, TIC enrichment, and the presence of intratumoral adipocytes are plausible and potentially important in mammary tumor development and progression (35, 36).

Study limitations include inadequate assessments of (i) the impact of energy balance on metastases and (ii) the direct effect of high-fat intake independent of obesity. Our previous studies showing that Wnt-1 tumors grow faster in genetically obese db/db mice (that overconsume control diet) than wild-type control mice strongly support an obesity (independent of dietary fat) effect (30). We did not elucidate the relative roles of individual obesity-related growth factors/cytokines in EMT and breast cancer, although additional studies on leptin and IGF-1 effects in obesity-driven EMT and metastases are underway.

In summary, in novel murine models of 2 highly aggressive forms of breast cancer, dietary energy balance interventions impact progression of claudin-low tumors (affected by both DIO and CR) and basal-like tumors (affected by CR only) and also modulate EMT in both mammary tumor subtypes (each affected by both DIO and CR). To our knowledge, this is the first study to show that dietary energy balance interventions differentially affect tumor progression and EMT in these breast cancer subtypes. Taken together, our findings suggest that components of the EMT and TIC pathways represent possible targets for breaking the obesity–breast cancer link, particularly for preventing and/or controlling TIC-enriched subtypes that confer poor prognosis and are often therapy-resistant, such as claudin-low breast cancer.

No potential conflicts of interest were disclosed.

Conception and design: S.M. Dunlap, L.J. Chiao, L. Varticovski, S.D. Hursting

Development of methodology: S.M. Dunlap, L. Varticovski

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): S.M. Dunlap, L.J. Chiao, L. Nogueira, J. Usary, C.M. Perou, S.D. Hursting

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): S.M. Dunlap, L.J. Chiao, L. Nogueira, J. Usary, C.M. Perou, L. Varticovski, S.D. Hursting

Writing, review, and/or revision of the manuscript: S.M. Dunlap, C.M. Perou, L. Varticovski, S.D. Hursting

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): L. Nogueira, S.D. Hursting

The authors thank Crystal Salcido and Mollie Wright for their technical contributions.

The study was supported by Breast Cancer Research Foundation (UTA09-001068) and NIEHS (P30ES007784) to S.D. Hursting; USAMRMC BCRP Fellowship (W81XWH-09-1-0720) to S.M. Dunlap; and the intramural program at the National Cancer Institute, NIH (L. Varticovski).

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

1.
Reeves
GK
,
Pirie
K
,
Beral
V
,
Green
J
,
Spencer
E
,
Bull
D
. 
Cancer incidence and mortality in relation to body mass index in the Million Women Study: cohort study
.
BMJ
2007
;
335
:
1134
.
2.
Sjostrom
L
,
Gummesson
A
,
Sjostrom
CD
,
Narbro
K
,
Peltonen
M
,
Wedel
H
, et al
Effects of bariatric surgery on cancer incidence in obese patients in Sweden (Swedish Obese Subjects Study): a prospective, controlled intervention trial
.
Lancet Oncol
2009
;
10
:
653
62
.
3.
Michels
KB
,
Ekbom
A
. 
Caloric restriction and incidence of breast cancer
.
JAMA
2004
;
291
:
1226
30
.
4.
Prat
A
,
Parker
JS
,
Karginova
O
,
Fan
C
,
Livasy
C
,
Herschkowitz
JI
, et al
Phenotypic and molecular characterization of the claudin-low intrinsic subtype of breast cancer
.
Breast Cancer Res
2010
;
12
:
R68
.
5.
Herschkowitz
JI
,
Simin
K
,
Weigman
VJ
,
Mikaelian
I
,
Usary
J
,
Hu
Z
, et al
Identification of conserved gene expression features between murine mammary carcinoma models and human breast tumors
.
Genome Biol
2007
;
8
:
R76
.
6.
Yang
XR
,
Chang-Claude
J
,
Goode
EL
,
Couch
FJ
,
Nevanlinna
H
,
Milne
RL
, et al
Associations of breast cancer risk factors with tumor subtypes: a pooled analysis from the Breast Cancer Association Consortium studies
.
J Natl Cancer Inst
2011
;
103
:
250
63
.
7.
Stark
A
,
Schultz
D
,
Kapke
A
,
Nadkarni
P
,
Burke
M
,
Linden
M
, et al
Obesity and risk of the less commonly diagnosed subtypes of breast cancer
.
Eur J Surg Oncol
2009
;
35
:
928
35
.
8.
Khramtsov
AI
,
Khramtsova
GF
,
Tretiakova
M
,
Huo
D
,
Olopade
OI
,
Goss
KH
. 
Wnt/beta-catenin pathway activation is enriched in basal-like breast cancers and predicts poor outcome
.
Am J Pathol
2010
;
176
:
2911
20
.
9.
Alessi
MC
,
Bastelica
D
,
Morange
P
,
Berthet
B
,
Leduc
I
,
Verdier
M
, et al
Plasminogen activator inhibitor 1, transforming growth factor-beta1, and BMI are closely associated in human adipose tissue during morbid obesity
.
Diabetes
2000
;
49
:
1374
80
.
10.
Samad
F
,
Yamamoto
K
,
Pandey
M
,
Loskutoff
DJ
. 
Elevated expression of transforming growth factor-beta in adipose tissue from obese mice
.
Mol Med
1997
;
3
:
37
48
.
11.
Kushiro
K
,
Nunez
NP
. 
Ob/ob serum promotes a mesenchymal cell phenotype in B16BL6 melanoma cells
.
Clin Exp Metastasis
2011
;
28
:
877
86
.
12.
Creighton
CJ
,
Chang
JC
,
Rosen
JM
. 
Epithelial-mesenchymal transition (EMT) in tumor-initiating cells and its clinical implications in breast cancer
.
J Mammary Gland Biol Neoplasia
2010
;
15
:
253
60
.
13.
Svirshchevskaya
EV
,
Mariotti
J
,
Wright
MH
,
Viskova
NY
,
Telford
W
,
Fowler
DH
, et al
Rapamycin delays growth of Wnt-1 tumors in spite of suppression of host immunity
.
BMC Cancer
2008
;
8
:
176
.
14.
Dontu
G
,
Abdallah
WM
,
Foley
JM
,
Jackson
KW
,
Clarke
MF
,
Kawamura
MJ
, et al
In vitro propagation and transcriptional profiling of human mammary stem/progenitor cells
.
Genes Dev
2003
;
17
:
1253
70
.
15.
Wright
MH
,
Calcagno
AM
,
Salcido
CD
,
Carlson
MD
,
Ambudkar
SV
,
Varticovski
L
. 
Brca1 breast tumors contain distinct CD44+/CD24- and CD133+ cells with cancer stem cell characteristics
.
Breast Cancer Res
2008
;
10
:
R10
.
16.
Charafe-Jauffret
E
,
Ginestier
C
,
Iovino
F
,
Tarpin
C
,
Diebel
M
,
Esterni
B
, et al
Aldehyde dehydrogenase 1-positive cancer stem cells mediate metastasis and poor clinical outcome in inflammatory breast cancer
.
Clin Cancer Res
2010
;
16
:
45
55
.
17.
Nunez
NP
,
Perkins
SN
,
Smith
NC
,
Berrigan
D
,
Berendes
DM
,
Varticovski
L
, et al
Obesity accelerates mouse mammary tumor growth in the absence of ovarian hormones
.
Nutr Cancer
2008
;
60
:
534
41
.
18.
Lashinger
LM
,
Malone
LM
,
McArthur
MJ
,
Goldberg
JA
,
Daniels
EA
,
Pavone
A
, et al
Genetic reduction of insulin-like growth factor-1 mimics the anticancer effects of calorie restriction on cyclooxygenase-2-driven pancreatic neoplasia
.
Cancer Prev Res
2011
;
4
:
1030
40
.
19.
Herzig
M
,
Savarese
F
,
Novatchkova
M
,
Semb
H
,
Christofori
G
. 
Tumor progression induced by the loss of E-cadherin independent of beta-catenin/Tcf-mediated Wnt signaling
.
Oncogene
2007
;
26
:
2290
8
.
20.
Mani
SA
,
Guo
W
,
Liao
MJ
,
Eaton
EN
,
Ayyanan
A
,
Zhou
AY
, et al
The epithelial-mesenchymal transition generates cells with properties of stem cells
.
Cell
2008
;
133
:
704
15
.
21.
Kalluri
R
,
Weinberg
RA
. 
The basics of epithelial-mesenchymal transition
.
J Clin Invest
2009
;
119
:
1420
8
.
22.
Li
Y
,
Hively
WP
,
Varmus
HE
. 
Use of MMTV-Wnt-1 transgenic mice for studying the genetic basis of breast cancer
.
Oncogene
2000
;
19
:
1002
9
.
23.
Campbell
CI
,
Thompson
DE
,
Siwicky
MD
,
Moorehead
RA
. 
Murine mammary tumor cells with a claudin-low genotype
.
Cancer Cell Int
2011
;
11
:
28
.
24.
Zhu
M
,
Yi
M
,
Kim
CH
,
Deng
C
,
Li
Y
,
Medina
D
, et al
Integrated miRNA and mRNA expression profiling of mouse mammary tumor models identifies miRNA signatures associated with mammary tumor lineage
.
Genome Biol
2011
;
12
:
R77
.
25.
Marotta
LL
,
Almendro
V
,
Marusyk
A
,
Shipitsin
M
,
Schemme
J
,
Walker
SR
, et al
The JAK2/STAT3 signaling pathway is required for growth of CD44CD24 stem cell-like breast cancer cells in human tumors
.
J Clin Invest
2011
;
121
:
2723
35
.
26.
Herschkowitz
JI
,
Zhao
W
,
Zhang
M
,
Usary
J
,
Murrow
G
,
Edwards
D
, et al
Comparative oncogenomics identifies breast tumors enriched in functional tumor-initiating cells
.
Proc Natl Acad Sci U S A
2012
;
109
(
8
):
2778
83
.
27.
Colbert
LH
,
Westerlind
KC
,
Perkins
SN
,
Haines
DC
,
Berrigan
D
,
Donehower
LA
, et al
Exercise effects on tumorigenesis in a p53-deficient mouse model of breast cancer
.
Med Sci Sports Exerc
2009
;
41
:
1597
605
.
28.
Fuchs-Young
R
,
Shirley
SH
,
Lambertz
I
,
Colby
JK
,
Tian
J
,
Johnston
D
, et al
P53 genotype as a determinant of ER expression and tamoxifen response in the MMTV-Wnt-1 model of mammary carcinogenesis
.
Breast Cancer Res Treat
2011
;
130
:
399
408
.
29.
Hursting
SD
,
Smith
SM
,
Lashinger
LM
,
Harvey
AE
,
Perkins
SN
. 
Calories and carcinogenesis: lessons learned from 30 years of calorie restriction research
.
Carcinogenesis
2010
;
31
:
83
9
.
30.
Zheng
Q
,
Dunlap
SM
,
Zhu
J
,
Downs-Kelly
E
,
Rich
J
,
Hursting
SD
, et al
Leptin deficiency suppresses MMTV-Wnt-1 mammary tumor growth in obese mice and abrogates tumor initiating cell survival
.
Endocr Relat Cancer
2011
;
18
:
491
503
.
31.
Bocchinfuso
WP
,
Hively
WP
,
Couse
JF
,
Varmus
HE
,
Korach
KS
. 
A mouse mammary tumor virus-Wnt-1 transgene induces mammary gland hyperplasia and tumorigenesis in mice lacking estrogen receptor-alpha
.
Cancer Res
1999
;
59
:
1869
76
.
32.
Padovani
M
,
Lavigne
JA
,
Chandramouli
GV
,
Perkins
SN
,
Barrett
JC
,
Hursting
SD
, et al
Distinct effects of calorie restriction and exercise on mammary gland gene expression in C57BL/6 mice
.
Cancer Prev Res
2009
;
2
:
1076
87
.
33.
Yang
Z
,
Norwood
KA
,
Smith
JE
,
Kerl
JG
,
Wood
JR
. 
Genes involved in the immediate early response and epithelial-mesenchymal transition are regulated by adipocytokines in the female reproductive tract
.
Mol Reprod Dev
2012
;
79
:
128
37
.
34.
Lindley
LE
,
Briegel
KJ
. 
Molecular characterization of TGFbeta-induced epithelial-mesenchymal transition in normal finite lifespan human mammary epithelial cells
.
Biochem Biophys Res Commun
2010
;
399
:
659
64
.
35.
Dirat
B
,
Bochet
L
,
Dabek
M
,
Daviaud
D
,
Dauvillier
S
,
Majed
B
, et al
Cancer-associated adipocytes exhibit an activated phenotype and contribute to breast cancer invasion
.
Cancer Res
2011
;
71
:
2455
65
.
36.
Subbaramaiah
K
,
Howe
LR
,
Bhardwaj
P
,
Du
B
,
Gravaghi
C
,
Yantiss
RK
, et al
Obesity is associated with inflammation and elevated aromatase expression in the mouse mammary gland
.
Cancer Prev Res
2011
;
4
:
329
46
.