The surface glycoprotein THY is a marker of myoepithelial precursor cells, which are basal cells with epithelial–mesenchymal intermediate phenotype originating from the ectoderm. Myoepithelial precursor cells are lost during progression from in situ to invasive carcinoma. To define the functional role of Thy1-positive cells within the myoepithelial population, we tracked Thy1 expression in human breast cancer samples, isolated THY1-positive myoepithelial progenitor cells (CD44+/CD24low/CD90+), and established long-term cultures (parental cells). Parental cells were used to generate a xenograft model to examine Thy1 expression during tumor formation. Post-transplantation cell cultures lost THY1 expression through methylation at the THY1 locus and this is associated with an increase in EGFR and NOTCH1 transcript levels. Thy1-low cells are sensitive to the EGFR/HER2 dual inhibitor lapatinib. High THY1 expression is associated with poorer relapse-free survival in patients with breast cancer. THY1 methylation may track the shift of bipotent progenitors into differentiated cells. Thy1 is a good candidate biomarker in basal-like breast cancer.

Implications:

Our findings provide evidence that THY1 expression is lost in xenografts due to promoter methylation. Thy1-low cells with increased EGFR and Notch1 expression are responsive to target therapy. Because DNA methylation is often altered in early cancer development, candidate methylation markers may be exploited as biomarkers for basal-like breast cancer.

The human breast gland is a ductal tree covered with a monolayer of polarized epithelial cells whose basal surface lies on contractile myoepithelial cells that are confined by the basement membrane and surrounded by an interstitial stroma. Myoepithelial cells originate from the ectoderm and are basal cells, namely, cells in the basal position adjacent to the basement membrane (1). Interest in basal cells was stimulated after molecular gene profiling divided breast cancer into five intrinsic subtypes, one of which displays basal-like gene expression (2). Basal-like breast cancers are generally aggressive (3) and most are triple-negative, that is, they test negative for estrogen receptor (ER), progesterone receptor (PgR), and HER2 (HER; ref. 4). Treatment of patients with basal-like triple-negative breast cancer (TNBC) is challenging because of the heterogeneity of the disease and the absence of well-defined druggable targets (5).

In breast cancer, myoepithelial cells are considered tumor suppressors because they inhibit epithelial cell growth and invasion (6), and because their oncosuppressive function disappears during progression from in situ to invasive carcinoma (7). Disappearance of the basement membrane and of the myoepithelial cell layer distinguishes invasive from in situ carcinomast (8), and the gene expression profiles of myoepithelial cells associated with in situ cancer are distinct from those in normal breast (9). The signals that initiate these changes are unknown, although it is recognized that tumor-associated fibroblasts and myofibroblasts counteract the tumor suppressor function of myoepithelial cells by promoting tumorigenesis (10, 11) and cancer progression (11).

THY1 (also known as “CD90”) is a surface glycoprotein of 25–28 kDa (12) that is expressed on the cytoplasmic membrane of diverse cell types (13). The structural gene for human THY1 lies on the long arm of chromosome 11 (11q23.3; ref. 14). Thy1 triggers a variety of cellular functions, namely, proliferation, differentiation, wound repair, and apoptosis. In lung cancer, Thy1 differentiates between malignant pleural mesothelioma and lung carcinoma (15). In melanoma, it contributes to metastasis seeding by mediating the adhesion of melanoma cells to endothelial cells (16). Thy1 has been associated with tumor suppression in human ovarian cancer (17). It is also a cancer stem cell marker in esophageal cancer (18), high-grade gliomas (19), and hepatocarcinoma (20). In breast cancer, Thy1-expressing cells are undifferentiated cancer progenitor/stem cells (21). Thy1 is upregulated in the epithelial-to-mesenchymal transition (EMT) core signature (22).

Moreover, it mediates the interactions of breast cancer stem cells with tumor-associated macrophages to maintain and reinforce the cancer stem cell state (23).

We previously demonstrated that sorted CD44+/CD24low cells display differential expression of surface markers that identify heterogeneous myoepithelial phenotypes (24). Among these surface markers, THY1/CD90 was commonly found highly expressed. In this study, we have investigated the relevance of Thy1 as a tracer biomarker of myoepithelial precursor cells also in relation to receptor profile, in our model of sorted breast cancer stem/progenitor cells. We show that xenotransplantation of CD44+/CD24low/CD90+ myoepithelial cells in mouse reduces THY1 expression through methylation of THY1 in conjunction with the acquisition of the Notch1–EGFR signaling.

Cell lines and materials

Cell lines MCF-7, MDA-MB-231 (MDA), BT474, and MCF10A were from ATCC. HER2-18 cells (25) were kindly provided by Dr. R. Schiff. Cells were maintained in standard medium consisting of minimal essential Dulbecco/Ham F12 (1:1) (DMEM/F12; Sigma-Aldrich) supplemented with 2 mmol/L glutamine (Sigma-Aldrich), 1% penicillin/streptomycin (Life Technologies), 15 mmol/L HEPES (Sigma-Aldrich), and 5% FBS and kept at 37°C in a humidified atmosphere of 5% CO2. Cell cultures were routinely checked for Mycoplasma with Hoechst 33258 (Sigma-Aldrich) staining; Mycoplasma-negative cell lines were used for experiments. Adherent and nonadherent 24-well ultra-low–binding plates were used (Corning). FBS was purchased from Gibco (Invitrogen). The monoclonal anti-pancytokeratin and anti-vimentin antibodies were purchased from Sigma-Aldrich. Multicolor flow cytometry was performed with antihuman mAbs that were conjugated with phycoerythrin (PE), fluorescein isothiocyanate (FITC), PE-Cy7 (PE-Cy7), or Alexa Fluor 647. Phycoerythrin-conjugated mAbs against CD10, CD29, CD49f, CD61, and FITC-conjugated mAbs against CD49b, CD90, CD227, CD324, and CD326 were from BD Biosciences and BD Pharmingen; PE-conjugated mAbs against CD133 were from Miltenyi Biotec; Alexa Fluor 647–conjugated mAbs against CD24 and PE-Cy7–conjugated mAbs against CD44 were from BioLegend.

Ethics and study design

Residual breast cancer and paired normal specimens were collected, after informed consent, from patients undergoing surgery for breast cancer at the Azienda Ospedaliera Universitaria Federico II (Naples, Italy). Nine patients with breast cancer were recruited. Pathologic diagnosis was made based on the histology of tumor specimens that had been examined by experienced pathologists. Tumor histotype, size, grading, and markers including ERα were determined with standard procedures, and HER2 was determined with the Hercep Test TM (Dako). The receptor profile of human breast tumors, namely, ER status, PgR status, and HER2, are summarized in Supplementary Table S1. The breast cancer–intrinsic subtype was determined using surrogate IHC definitions according to Goldhirsch and colleagues (26).

Sample collection

Breast cancer tissue (S#) and paired normal (N#) specimens were collected using a biobanking standard operating procedure as reported previously (27). The samples were anonymously encoded to protect patient confidentiality and used according to protocols approved by the Azienda Ospedaliera Universitaria Federico II Ethics Committee (Ethical Committee Approval Protocol # 107/05). The primary objective of the approved protocol was to expand human breast cancer cells to characterize the protein expression profile of in vitro cultured cells.

Primary cultures and breast cancer stem/progenitor cells

Within 2 hours after surgery, fragmented aliquots of fresh specimens were processed as reported previously (28). Briefly, the samples were extensively rinsed with PBS and suspended in standard culture media supplemented with 10% FBS. After three cycles of differential centrifugation, cells were seeded overnight in minimal DMEM/F12 medium (1:1; Sigma-Aldrich), supplemented with 2 mmol/L glutamine (Sigma-Aldrich), penicillin/streptomycin (100 μg/mL streptomycin, 100 U/mL penicillin), 15 mmol/L HEPES (Sigma-Aldrich), and 5% FBS. After exposure to trypsin (0.25% in 1 mmol/L EDTA; trypsin–EDTA solution, Invitrogen) for 2 minutes at 37°C, the floating aggregates were transferred to 24-well plates and cultured in standard medium (DMEM/F12 + 0.5% FBS) for 21–30 days at 37°C in a humidified atmosphere of 5% CO2. Cells were continuously passaged with trypsin–EDTA until only the tumor epithelial cell population remained. The epithelial origin of the cells was confirmed by Western blot analysis with monoclonal anti-pancytokeratin antibody (Supplementary Fig. S1). Multiple vials of cells were cryopreserved. Frozen cells were thawed, allowed to adhere, and harvested within 15–20 days in standard medium before sorting.

Flow cytometry and sorting

Flow cytometry experiments were performed as reported previously (24). Briefly, samples and control cells, harvested at subconfluence in 100-mm dishes, were dissociated by trypsin–EDTA, counted in a hemocytometer chamber, and 2 × 106 cells/sample were incubated for 5 minutes at room temperature with 50 μL of FBS. Cells were washed twice with PBS and stained at 4°C for 20 minutes with the appropriate amount of the fluorescence-labeled mAb in PBS. After staining, all samples were washed twice with PBS, centrifuged, and suspended in 0.5 mL of FACS buffer (FACS Flow Sheat Fluid, BD Biosciences) for FACS analysis. To exclude dead cells, immediately before FACS acquisition, cells were incubated at room temperature in the dark with a vital dye (SytoxBlue, Invitrogen). We used a four-color flow cytometric method to measure the expression of the markers based on a flow cytometry panel in which cells were stained with anti-CD24-Alexa Fluor 647 and anti-CD44-PE-Cy7 mAb, and with FITC-conjugated antibodies against CD90 (Thy1) and PE-conjugated anti CD133 (prominin 1; ref. 24). Fluorescence Minus One (FMO) control, was used as a negative control. After washing twice with PBS/0.5% BSA, cells were pelleted, suspended in 300 μL of PBS/0.5% BSA, and filtered through 50-μm filters. Cell analysis and sorting were performed with a FACSAria flow cytometer and with the FACS Diva software (Becton Dickinson). A total of 10,000 to 20,000 events were recorded and analyzed in each sample run. A three gating strategy was adopted: first, to exclude dead cells and debris, cells were gated on a two physical parameters dot plot measuring forward scatter (FSC) versus side scatter (SSC). Then, doublets were excluded by gating cells on FSC-Height versus FSC-Area dot plots, and, finally, SytoxBlue-negative cells were gated. The levels of expression of surface markers were reported as percentage of positive cells in Count versus FITC- or PE-CD histograms. For cell sorting, the cells were suspended in PBS with 2% FBS and 0.5 mmol/L EDTA, sequentially labeled with a cocktail of mAb anti-CD24-Alexa Fluor 647and anti-CD44-PE-Cy7, mixed with magnetic microbeads and separated using a magnet. The purity of sorted cells was evaluated by flow cytometry. Sorted CD44+/CD24low cells were cultured with standard medium for at least four/five passages before experiments. The steps of isolation of CD44+/CD24low/CD90high myoepithelial precursor cells (K#) are summarized in Supplementary Fig. S2. Cultures at the fourth to fifth passage were used for the experiments.

Immunofluorescence

For immunofluorescence analysis, cells were plated on glass coverslips, fixed, immunostained at 4°C overnight with rabbit polyclonal anti-Thy1 antibodies (H-110; SC-9163; Lot # B2514; Santa Cruz Biotechnology, Inc.) and treated for 30 minutes with goat anti-rabbit IgG secondary antibody tagged with Alexa Fluor 594 (1/300; A-11012; Lot # 1420898; Life Technologies). Nuclei were stained for 30 minutes at room temperature with 1:1 (vol/vol) Hoechst 33258 (94403, Sigma-Aldrich)/DRAQ5 (ab108410; Abcam). Immunofluorescence was visualized using a Zeiss LSM510 Meta argon/krypton laser-scanning confocal microscope.

Cumulative population doubling frequency

Experiments were performed in triplicate in 24-well plates using 2.5 × 103 cells/well. The cells, routinely cultured in 100-mm dishes, were enzymatically detached, counted, and 2.5 × 103 cells/well were seeded with standard medium. Cells were maintained in a sterile environment, and at the times indicated in the figures, they were trypsinized, counted in a hemocytometer chamber, and replated 1:2 in new wells. Cells viability was assessed by Trypan blue with paired triplicates. Lineage continuity for Thy1 in 3D culture was assessed with immunofluorescence (representative at Supplementary Fig. S1). For 2D experiments, after trypsinization, the cells were plated overnight in 5% FBS, allowed to adhere, and then switched to 0.5% FBS. The proliferation rate of the cells was measured by calculating the cumulative population doubling frequency (cpdf) in continuous culture from a known number of cells using the formula Ln(No/Nn)/Ln2, where Ln is the natural log and No and Nn are, respectively, initial and final cell numbers at each subcultivation. The sum of the cpdf of the subcultivation periods provides the cumulative final number of total counts.

Semiquantitative multiplex RT-PCR analysis and real-time PCR analysis

Total RNA was isolated from sample and control cells, and from breast tumor tissues using TRIzol Reagent (Invitrogen) according to the producer's instructions. Purity of RNA was checked by measuring the absorbance ratio at 260/280 nm in a Beckman Coulter spectrophotometer (Beckman Coulter) with appropriate purity values between 1.8 and 2.0. RNA was stored at −80°C in aliquots of 50 ng/L. The integrity of RNA was assessed on a standard 1% agarose/formaldehyde gel. The reverse transcription of 1.5 μg of total RNA was performed with the Super Script III Reverse Transcriptase Kit (Invitrogen) according to the manufacturer's instructions.

Multiplex PCR was performed in 50 μL reactions using the PTC-200 Peltier Thermal Cycler (Bio-Rad) and gene-specific sets of primers, including those for the internal standard β-actin. Agarose gel electrophoresis and staining with 0.3 mg/mL of ethidium bromide (Sigma) were carried out to assess template products.

Real-time PCR amplifications were carried out on a Step One Real-Time Thermocycler (Applied Biosystems) using the iTaq Universal SYBR Green Supermix (Bio-Rad). Experiments were performed in triplicate for each data point, and the expression of housekeeping β2-microglobulin gene (B2M, forward: 5′-GCA GAA TTT GGA ATT CAT CCA AT-3′; reverse: 5′- CCG AGT GAA GAT CCC CTT TTT-3′) was used for normalization.

Primer sequences are listed in Supplementary Table S2.

Nude mice cancer xenografts

The tumorigenicity of sorted cells was assessed by injecting the harvested cells into immunodeficient mice. Five-week-old female BALB/c athymic (null/null) mice (Charles River Laboratories) were maintained in accordance with the institutional guidelines of the University of Naples Animal Care Committee welfare policy (European Commission 86/609/EEC). All the animal experiments were approved by the Ethics Committee of the University of Naples Federico II Animal Care (ethical approval protocol # 83). Adherent harvested K197 cells were enzymatically dissociated, counted, diluted in PBS, mixed 1:1 with 200 μL Matrigel (CBP), and injected orthotopically in the fourth mammary fat pad of triplicate mice, as reported previously (29). We injected 1 × 102, 1 × 103, 1 × 104, 1 × 105, 1 × 106 of the K197 cells into the fourth mammary fat pad of immunodeficient mice. The experiment was performed twice. Tumor volume (cm3) was measured with calipers and calculated with the formula π/6 × largest diameter × (smallest diameter)2. Within 8 weeks, the tumors (size: 1–3 cm3) were excised, digested with a trypsin/collagenase mixture, and plated for in vitro growth. The steps of isolation of CD44+/CD24low/CD90-Thy1-low cells (Topo9) are summarized in Supplementary Fig. S3. The homogeneity of the cultures was confirmed with flow cytometry analysis of ten surface markers (Table 1). Cultures at their fourth/fifth passage were used for the experiments.

Table 1.

Surface marker percentage expression (%) and mean fluorescence intensity (MFI) in myoepithelial progenitors before (K197) and after transplantation (Topo9) and in cell lines

K197aTopo9aMCF7bMCF10Aa
%MFI%MFI%MFI%MFI
CD227-MUC1 0.0 49 0.6 41 90.2 605 42.6 227 
CD324-E-cadherin 0.5 45 0.8 52 6.0 90 3.3 148 
CD326-EpCAM 2.0 32 1.9 51 100.0 1377 72.1 425 
CD10-CALLA 8.4 65 15.0 20 18.7 82.4 388 
CD29-β1 Integrin 100.0 1474 100.0 1373 100.0 1300 100.0 3526 
CD49b-α2 Integrin 100.0 597 100.0 960 99.9 560 99.9 1996 
CD49f-α6 Integrin 100.0 519 100.0 554 23.8 66 100.0 6044 
CD61-β3 Integrin 60.0 11 33.0 18 0.4 25.2 121 
CD90-THY1 100.0 1065 11.0 76 0.2 45 58.6 208 
CD133-Prominin1 16.0 35 18.0 25 68.0 29 28.0 70 
Percentage (%) Pearson R score P-value Significance at P < 0.001 Correlation 
K197 vs Topo9 0.9818 1.5E-05 Yes Strong positive 
K197 vs MCF7 0.3412 0.3688 No Weak 
K197 vs MCF10A 0.7603 0.0174 No Positive 
MFI Pearson R score P Significance at P < 0.001 Correlation 
K197 vs. Topo9 0.9659 9.7E-05 Yes Strong positive 
K197 vs. MCF7 0.3263 0.3914 No Weak 
K197 vs. MCF10A 0.5006 0.1698 No Moderate Positive 
K197aTopo9aMCF7bMCF10Aa
%MFI%MFI%MFI%MFI
CD227-MUC1 0.0 49 0.6 41 90.2 605 42.6 227 
CD324-E-cadherin 0.5 45 0.8 52 6.0 90 3.3 148 
CD326-EpCAM 2.0 32 1.9 51 100.0 1377 72.1 425 
CD10-CALLA 8.4 65 15.0 20 18.7 82.4 388 
CD29-β1 Integrin 100.0 1474 100.0 1373 100.0 1300 100.0 3526 
CD49b-α2 Integrin 100.0 597 100.0 960 99.9 560 99.9 1996 
CD49f-α6 Integrin 100.0 519 100.0 554 23.8 66 100.0 6044 
CD61-β3 Integrin 60.0 11 33.0 18 0.4 25.2 121 
CD90-THY1 100.0 1065 11.0 76 0.2 45 58.6 208 
CD133-Prominin1 16.0 35 18.0 25 68.0 29 28.0 70 
Percentage (%) Pearson R score P-value Significance at P < 0.001 Correlation 
K197 vs Topo9 0.9818 1.5E-05 Yes Strong positive 
K197 vs MCF7 0.3412 0.3688 No Weak 
K197 vs MCF10A 0.7603 0.0174 No Positive 
MFI Pearson R score P Significance at P < 0.001 Correlation 
K197 vs. Topo9 0.9659 9.7E-05 Yes Strong positive 
K197 vs. MCF7 0.3263 0.3914 No Weak 
K197 vs. MCF10A 0.5006 0.1698 No Moderate Positive 

NOTE: Numbers of positive cells are mean percentage of average of triplicates of three (a) or two (b) experiments; SD, not reported, was < 10%.

5-Aza-2′-Deoxycytidine treatment

THY1-positive (K197) and THY1-low (Topo9) cells, 5 × 105cells/cell type, were seeded in 100-mm cell culture dishes (Corning) with standard medium. After an initial 24 hours of incubation, the cells were exposed to 5 μmol/L 5-aza-2′-deoxycytidine (5-AZA-dC; Sigma) for 12, 24, 36, 48, 72, and 96 hours. The medium was renewed every 24 hours. Control cultures lacking 5-AZA-dC treatment were incubated in the identical culture condition. At the time indicated, the cells were harvested for total RNA extraction.

Methylation-sensitive amplified polymorphism

Genomic DNA was extracted with phenol/chloroform technique (30). For measurement of Thy promoter methylation status, 1 μg of DNA was digested overnight at 37 C with HpaII or MspI (50 U/1μg DNA, Fermentas) restriction enzymes. DNA was recovered by phenol/chloroform extraction and ethanol precipitation, and resuspended in DNase/RNase free water. HpaII sensitivity was evaluated by amplifying a 276-bp fragment containing 4 CG upstream of the Thy promoter (position: −2328, −2297, −2259, and −2190 with respect to the first ATG). Methylation status was assessed by analyzing the efficiency of fragment amplification exposed to digestion (HpaII or MspI) on 2% agarose gel and quantified by real-time PCR. Amplification was performed on a PTC-200 Peltier Thermal Cycler (Bio-Rad) through 40 PCR cycles using the following temperature profile: 95°C for 40 seconds, 61°C for 40 seconds, 72°C for 1 minute, and one final elongation step at 72°C for 10 minutes. All reactions were preceded by a primary denaturation step at 95°C for 5 minutes. Real-time PCR amplifications were carried out on a Step One Real-Time thermocycler (Applied Biosystems) using the iTaq Universal SYBR Green Supermix (Bio-Rad). Cycling conditions were: one cycle at 95°C for 5 minutes, followed by 40 cycles of 95°C for 15 seconds, 61°C for 20 seconds, and 72°C for 30 seconds. Experiments were performed in triplicate for each data point. The following primers were used for Thy-1 DNA amplifications: forward 5′- CCAATGCGGGACCGCCTTCTCTTCC-3; reverse 5′-GTCTTGCATGGGCGCCTGACGGCG-3′.

Western blot analysis

Protein preparations were obtained by lysing samples in 50 mmol/L Tris (pH 7.5), 150 mmol/L NaCl, 1% Nonidet P40, 0.1% Triton, 1 mmol/L EDTA, 10 μg/mL aprotinin, and 100 μg/mL phenylmethylsulfonylfluoride. Protein concentration was measured by the Bio-Rad Protein Assay (Bio-Rad). Twenty-five–microgram aliquots were electrophoresed through 8%–15% SDS polyacrylamide gels. After transfer onto nitrocellulose membranes (Hybond-C pure; Amersham Italia), the membrane was stained with Ponceau S (Sigma) to evaluate the success of transfer, and to locate the molecular weight markers. Free protein-binding sites were blocked with nonfat dry milk and Tween-20/TBS solution. The membranes were washed, stained with specific primary antibodies and then with secondary antisera, conjugated with horseradish peroxidase (1:3,000; Santa Cruz Biotechnology). Antibodies were: Ab anti-E-Cadherin (1:1,000, Cell Signaling Technology); Ab anti-Notch1 (C-20): sc-6014-R (1:200, Santa Cruz Biotechnology); Ab anti-Fibronectin (P1H11): sc-18825 (1:100, Santa Cruz Biotechnology); Ab anti-Fibronectin (EP5): sc-8422 (1:200, Santa Cruz Biotechnology); Ab anti-ERalfa (F-10):sc-8002 (1:200, Santa Cruz Biotechnology). The luminescent signal was visualized with the ECL Western blotting Detection Reagent Kit (Amersham Italia) and quantified by scanning with a Discover Pharmacia scanner equipped with a Sun Spark Classic Workstation. Expression levels were calculated as the relative expression ratio compared with β-actin or tubulin using Image J (ImageJ.nih.gov).

Mammosphere formation assay and growth in soft agar

Cells were dissociated and seeded, 1,000 cells/well, in ultra-low attachment 24-well plates (Corning) in DMEM/F12 plus 0.5% FCS medium, as reported previously (31). The medium was renewed twice weekly. Mammospheres were cultured for 15 days and their diameter measured under an Axiovert 40 C inverted microscope (Zeiss) equipped with a Canon Powershot A640 camera (Zeiss). Digital images were analyzed with AxioVision software (Zeiss). For colony growth in soft agar, cells were trypsinized, counted, and 104 cells/dish were plated in 60-mm triplicate dishes with 0.3% agar on a 0.5% agar (Type I, Sigma) bottom layer with DMEM/F12 containing 0.5% FBS. Colonies, cultured for 60 days, were counted in 10 fields per dish. The fields to be counted were ID-numbered fields on a 7 × 7 horizontal–vertical transparency grid 60 mm in diameter. The same ID fields were counted for all dishes. Results are reported as mean ± SEM of three different experiments performed in duplicate.

Cell viability assay

Cell viability experiments were performed as described previously (32). Cells were seeded at 2,500 cells/well, in 24-well plates, treated with the reported concentrations of lapatinib for 7 days, and analyzed with the 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazoliumbromide (MTT) assay according to the manufacturer's instructions (Sigma-Aldrich). The percentage of absorbance of treated samples versus untreated is reported as a percentage of viable cells/controls. Experiments were performed three times; values represent means ± SD from triplicate samples for each treatment.

Kaplan–Meier curves

Kaplan–Meier curves were generated using the Kaplan–Meier plot software and a public database of microarray datasets (probes: 213869_x_at (Thy1-CD90);211551_at (EGFR, ERBB, ERBB1; http://kmplot.com/analysis; ref. 33). Kaplan–Meier plots were generated after averaging the probes. For the analysis, eligible patients were divided according to the median expression value, and ERα-negative/HER2-negative cases were included. P value was determined by log-rank test.

Statistical analysis

Flow cytometry, cell counting, sphere formation assay, and RT-PCR experiments were carried out 2–3 times and found to be reproducible. Human tissue samples were not pooled; each sample served as its own control. Values are presented as mean ± SEM of multiple experiments, each experiment was performed at least in triplicate, or as mean ± SD of triplicates when a representative experiment is shown. The statistical significance between two groups was determined with the Fisher exact test and multiple group comparisons were made with ANOVA, as reported. Pearson's correlation coefficient was used to calculate r. GraphPad software was used for all statistical analyses.

Ethical approval

All procedures performed in studies involving human participants were in accordance with the ethical standards of the institutional and/or national research committee and with the 1964 Helsinki declaration and its later amendments or comparable ethical standards.

Availability of data and materials

The datasets used and/or analyzed during this study are available from the corresponding author on reasonable request. The datasets generated and/or analyzed during the Kaplan–Meier curves are available in the [kmplot.com] repository (http://kmplot.com/analysis/).

CD44+/CD24low/THY1+ myoepithelial progenitor cells

In our previous studies we profiled heterogeneous THY1-expressing myoepithelial phenotype in breast carcinomas. To investigate the functional role of THY1-expressing cells in experimental models, we isolated breast cancer cells that displayed features of stem/progenitor cells from fresh surgical breast tumor tissues. Nine tumor specimens (S#) were chosen from our stored breast cancer collection of fresh frozen tissues belonging to various molecular subclasses of breast tumors: S40, S43, S79, S88, S193 and S197 were luminal breast cancers; S66 was HER2-positive; and S77 and S90 were triple-negative (Supplementary Table S1). Breast cancer stem cells are identified by one or more of the following features: a CD44+/CD24low phenotype, mammosphere formation in vitro, and ability to form new tumors when xenografted into immunodeficient mice (34, 35). We established nine primary cultures of breast cancer; to avoid fibroblasts, cells were continuously passaged until only the tumor epithelial cell population remained (Supplementary Fig. S1). Immunofluorescence experiments with antibodies against Thy1/CD90 showed a mixed population of THY-positive and THY1-negative cells in the primary cultures with a percentage ranging from 27% to 58% of Thy1/CD90–stained cells versus nonstained cells (Fig. 1A). From each culture, we sorted the CD44+/CD24low cells. THY1/CD90 expression assessed by immunofluorescence on cells harvested for 10 days after sorting, showed expression of THY1 on 90%–100% of the sorted CD44+/CD24low cells and on 98%–100% of cells maintained in culture for three months (Fig. 1A and B). To assess the ability of the sorted cells to form spheres, we seeded 100 cells/well per each cell culture under nonadherent conditions and found that all the nine cultures formed mammospheres in low-attachment plates (Supplementary Fig. S2). CD44/CD24 expression on cultures of cells stabilized for 1–3 months under adherent conditions (dot plots at Supplementary Fig. S2) was measured by calculating the mean fluorescence intensity (MFI). The analysis confirmed high expression of CD44 molecules per cell (1,153 ≤ MFI ≤ 7,967) and a low signal for CD24 (1 ≤ MFI ≤ 72) in all the breast cancer cell cultures. The MFI for CD44/CD24 in BT474 (16/4,078), HER2-18 (123/250), MCF7 (15/79), MCF10A (914/27), and MDA-MB-231 (4,906/10) cell lines served as control (Fig. 1C). To measure the doubling frequency of cells dissociated from mammospheres in adherent (2D) and nonadherent conditions (3D), we enzymatically detached cells to obtain single-cell suspensions, seeded in ultra-low attachment or in adherent plates and counted the cells each month thereafter. After two months of culture, growth and propagation was arrested in breast cancer cells in low adhesion conditions, whereas the paired adherent culture grew with a doubling time ranging between 2 and 3 days (lowest) and 7 and 10 days (highest) across cell types. The cumulative population doubling frequency (cpdf) at 120 days was 4.7 × 105 in cell cultures with a high growth rate (K197), and 2.8 × 105 in cell cultures with a low growth rate (K77). The median cpdf of the nine CD44+/CD24low/Thy1+ cell cultures (K40, K43, K66, K77, K88, K79, K90, K193, and K197) at 120 days was 3.26 × 105 for 2D cultures and 1.12 × 105 for 3D cultures (Fig. 1D). In all cases, long-term culturing of Thy1-positive cells under nonadherent conditions delayed the proliferation. In fact, the mean doubling time was 15–18 days during the initial two months, and was arrested thereafter.

Figure 1.

Breast tissue specimens contain THY1-expressing cells. A, Percentage of THY1/CD90-positive cells as assessed by immunofluorescence in primary cultures, in CD44+/CD24low cells cultured for 10 days after sorting and in long-term (3 months) cultures. Numbers of positive cells were counted from 7–10 representative fields of three slides per cell culture (n = 9 cultures). B, Immunofluorescence of THY1/CD90 on two representative cell cultures (K43 and K197) at three months after sorting; scale bar, 50 μm. C, Mean fluorescence intensity of CD24 (light gray) and CD44 (dark gray) markers expressed on the cell surface of breast cancer stem/progenitor cells (K40, K43, K66, K77, K88, K79, K90, K193, K197) and cell lines (BT474, HER2–18, MCF7, MCF10A, and MDA-MB231; log scale range: 1–10,000); the SE, not reported on graph, was ≤ 10%. For the analysis, cells were cultured for 15–20 days in standard medium, then detached, counted, and 1–3 × 106 cells were used. For each tube, 20,000 events were recorded and analyzed. D, Cumulative population doubling frequency (cpdf) of Thy1-positive cells. The median cpdf of the nine CD44+/CD24low/THY1+ cell cultures (K40, K43, K66, K77, K88, K79, K90, K193, and K197 is plotted. Triplicate dishes of each cell culture, plated at 2.5 × 103 cells/well, were cultured on adherent dishes (light gray) and as floating mammospheres on nonadherent wells (dark gray) and counted in a hemocytometer chamber for the time indicated. After counting, the cells, maintained in a sterile environment, were replated 1:2 in new wells. The cpdf was calculated with the formula Ln(No/Nn)/Ln2. Statistical analysis was done by χ2 squares (P = 0.069). Error bars indicate the SEM of triplicates.

Figure 1.

Breast tissue specimens contain THY1-expressing cells. A, Percentage of THY1/CD90-positive cells as assessed by immunofluorescence in primary cultures, in CD44+/CD24low cells cultured for 10 days after sorting and in long-term (3 months) cultures. Numbers of positive cells were counted from 7–10 representative fields of three slides per cell culture (n = 9 cultures). B, Immunofluorescence of THY1/CD90 on two representative cell cultures (K43 and K197) at three months after sorting; scale bar, 50 μm. C, Mean fluorescence intensity of CD24 (light gray) and CD44 (dark gray) markers expressed on the cell surface of breast cancer stem/progenitor cells (K40, K43, K66, K77, K88, K79, K90, K193, K197) and cell lines (BT474, HER2–18, MCF7, MCF10A, and MDA-MB231; log scale range: 1–10,000); the SE, not reported on graph, was ≤ 10%. For the analysis, cells were cultured for 15–20 days in standard medium, then detached, counted, and 1–3 × 106 cells were used. For each tube, 20,000 events were recorded and analyzed. D, Cumulative population doubling frequency (cpdf) of Thy1-positive cells. The median cpdf of the nine CD44+/CD24low/THY1+ cell cultures (K40, K43, K66, K77, K88, K79, K90, K193, and K197 is plotted. Triplicate dishes of each cell culture, plated at 2.5 × 103 cells/well, were cultured on adherent dishes (light gray) and as floating mammospheres on nonadherent wells (dark gray) and counted in a hemocytometer chamber for the time indicated. After counting, the cells, maintained in a sterile environment, were replated 1:2 in new wells. The cpdf was calculated with the formula Ln(No/Nn)/Ln2. Statistical analysis was done by χ2 squares (P = 0.069). Error bars indicate the SEM of triplicates.

Close modal

CD44+/CD24low/THY1+ cells are basal cells with intermediate epithelial–mesenchymal phenotype

To evaluate whether Thy1 expression was stable in cells sorted for stem features, we measured THY1 messenger RNA (mRNA) in stabilized cultures. As shown in Fig. 2A and B, THY1 was highly expressed in all nine THY1-positive cell cultures established from the CD44+/CD24low population of primary culture, and low in MCF7, BT474, HER2/18, MCF10A and MDA-MB231 cells. Messenger RNA levels were consistent with the high expression of the protein; indeed, the percentage of THY1/CD90 expression, measured by immunophenotyping, ranged from 87.9% to 100% (Fig. 2C). To estimate the number of Thy1 molecules per cell, we calculated the MFI of Thy1/CD90 expression, and found that it was low in all the cell lines BT474 (211), HER2-18 (189), MCF7 (45), MCF10A (208), and MDA-MB-231 (81) (45 ≤ MFI ≤ 211) but high in all the nine cell cultures (1,065 ≤ MFI ≤ 8,117; Fig. 2D). THY1 expression per cell was significantly higher versus the control cell lines (P = 0.0078).

Figure 2.

Thy1 expression in CD44+/CD24low breast cancer cells. A, RT-PCR of THY1 mRNA (235 bp) expression of the K40, K43, K66, K77, K88, K79, K90, K193, and K197 myoepithelial progenitor cells. Cell lines MCF7, MCF10a, BT474, HER2/18, and MDA-MB-231 served as references. Standardization was on the basis of β-actin cDNA levels. B, Densitometry RT-PCR for Thy1 in MCF7, MCF10a, BT474, HER2/18, and MDA-MB231, and in the K40, K43, K66, K77, K88, K79, K90, K193, and K197 cells. C, Percentage of surface protein expression of THY1-CD90 in MCF7, MCF10a, BT474, HER2/18, and MDA-MB231, and in the myoepithelial progenitor cells K40, K43, K66, K77, K88, K79, K90, K193, and K197. The levels of expression of surface markers were reported as percentage of positive cells in Count versus FITC- or PE-CD histograms. D, Protein expression of THY1-CD90 calculated as mean fluorescence intensity (MFI) on breast cancer stem/progenitor cells (K40, K43, K66, K77, K88, K79, K90, K193, and K197) and cell lines (MCF7, BT474, HER2-18, MCF10A, and MDA-MB-231); (log scale range: 1–10,000) is plotted. The SEM of triplicates, not reported on the graph, was ≤ 10%. Statistical analysis of differences of breast cancer stem/progenitor cells versus control cells was done by one-way ANOVA (P = 0.0078).

Figure 2.

Thy1 expression in CD44+/CD24low breast cancer cells. A, RT-PCR of THY1 mRNA (235 bp) expression of the K40, K43, K66, K77, K88, K79, K90, K193, and K197 myoepithelial progenitor cells. Cell lines MCF7, MCF10a, BT474, HER2/18, and MDA-MB-231 served as references. Standardization was on the basis of β-actin cDNA levels. B, Densitometry RT-PCR for Thy1 in MCF7, MCF10a, BT474, HER2/18, and MDA-MB231, and in the K40, K43, K66, K77, K88, K79, K90, K193, and K197 cells. C, Percentage of surface protein expression of THY1-CD90 in MCF7, MCF10a, BT474, HER2/18, and MDA-MB231, and in the myoepithelial progenitor cells K40, K43, K66, K77, K88, K79, K90, K193, and K197. The levels of expression of surface markers were reported as percentage of positive cells in Count versus FITC- or PE-CD histograms. D, Protein expression of THY1-CD90 calculated as mean fluorescence intensity (MFI) on breast cancer stem/progenitor cells (K40, K43, K66, K77, K88, K79, K90, K193, and K197) and cell lines (MCF7, BT474, HER2-18, MCF10A, and MDA-MB-231); (log scale range: 1–10,000) is plotted. The SEM of triplicates, not reported on the graph, was ≤ 10%. Statistical analysis of differences of breast cancer stem/progenitor cells versus control cells was done by one-way ANOVA (P = 0.0078).

Close modal

To investigate whether the CD44+/CD24low/THY1+ populations of sorted cells were tumorigenic, we injected the K197 cells into the fourth mammary fat pad of immunodeficient mice. As few as 1 × 104 CD44+/CD24low/THY1+ cells generated tumors with a calculated frequency of tumorigenic cells that resulted in 1/4,326 cells. Each of the dilutions that generated tumors showed a latency and a size when excised, at day 60, that correlated with the number of cells injected (R = 0.949, P < 0.05). Indeed, the mean tumor volume of triplicates was 950 ± 190 mm3 per 1 × 104 cells injected, 1,500 ± 250 mm3 per 1 × 105 cells, and 2,800 ± 260 mm3 per 1 × 106 cells (Fig. 3A). The excised tumors (measuring 1–3 cm3) were minced and dissociated by enzymatic digestion, and the cells derived were maintained in long-term culture. A signal for THY1 mRNA was detectable in carcinoma specimens S197 (Fig. 3B, left) and absent in either normal N197 tissue (Fig. 3B, middle) and xenotransplanted specimen Topo9 (Fig. 3B, right). THY1expression levels were confirmed by real-time PCR (Fig. 3C). We next performed flow cytometry to investigate whether the cell cultures derived from the transplanted tumors preserved the stem cell characteristics of the implanted cells. This analysis showed that, consistent with the implanted CD44+/CD24low population, tumors generated by this population recapitulated the CD44+/CD24low profile (Fig. 3D) and the MFI (Fig. 3E). We measured the mRNA expression to compare the receptor profiles of the implanted cells (K197 cells) with those of the cells isolated and cultured from mouse tumor tissue (Topo9 cells). Like the parental cells, transplanted cells expressed low levels of ERα, PgR, and HER2 (Fig. 3F), which indicates subtype relationship with the implanted cells. Immunoblot for ER and HER2 (Supplementary Fig. S4) confirmed that both K197 and Topo9 cells do not express potentially functional levels of these markers.

Figure 3.

Thy1 expression in xenotransplanted cells. A, CD44+/CD24low THY1-positive K197 cells were assayed for the ability to form tumors after injection into mice. Five-week-old female BALB/c athymic (null/null) mice were injected orthotopically in the fourth mammary fat pad, with K197 cells resuspended in Matrigel/PBS. Tumor volume (mm3) was measured by calculating π/6 × largest diameter × (smallest diameter)2, and correlated with the number of cells injected (R = 0.949, P < 0.05). Within 60 days, tumors were excised (250 mm3), digested with trypsin/collagenase mixture and mouse-derived tumor cells were harvested for in vitro growth (Topo9 cells). B, RT-PCR of THY1 mRNA (235 bp) expression in carcinoma specimens S197 (left), paired normal N197 (middle), and xenotransplanted specimen Topo9 (right) tissue. Standardization with β-actin cDNA levels. C, qRT-PCR of THY1 expression in carcinoma specimens S197, paired normal N197 and xenotransplanted specimen Topo9 tissue (S-Topo9). Experiments were performed in triplicate for each data point, and the expression of housekeeping β2-microglobulin gene (B2M) was used for normalization. D, Cells derived from xenotransplanted tumors recapitulate the CD44+/CD24low profile of the grafted stem/progenitor cells. Dot plots of flow cytometry analysis of parental K197 and Topo9 cells. Cells were cultured for 15 days in standard medium, and 1 × 106 cells were stained for the flow cytometric analysis of CD44PE-Cy7A and CD24-Alexa Fluor 647. The expression of each antigen is represented on a frequency distribution histogram (count vs. FITC or PE signal). The expression of the two markers is presented on a biparametric dot plot CD44-PE-Cy7 vs. CD24-AlexaFluor647 for each cell type. Vertical and horizontal markers delineate the quadrants used to identify the CD44/CD24 subsets and were set with the appropriate FMO control. E, The histogram reports the mean fluorescence intensity (MFI) of CD24 (light gray) and CD44 (dark gray) expressed on the cell surface of the indicated cells (log scale range: 1–10,000); SE, not reported on the graph, was ≤ 10%. F, Topo9 cells, cultured from xenotransplanted tumors, are triple-negative, as the grafted K197 THY1-positive cells. RT-PCR for mRNA of ERα (441 bp), PgR (121 bp) and HER2 (420 bp) of K197 and Topo9 cells. Standardization was based on β-actin cDNA levels. Protein expression determined with Western blot analysis at Supplementary Fig. S4. G, Percentage of THY1/CD90 surface protein as determined by flow cytometry with CD90-FITC. H, RT-PCR of THY1 mRNA (235 bp) expression in K197 (left), and xenotransplanted Topo9 (right) cells. Standardization with β-actin cDNA levels. I, qRT-PCR of THY1 expression in MDA-MB231 cell line, in K197 and xenotransplanted Topo9 cells. Experiments were performed in triplicate for each data point, and the expression of housekeeping β2-microglobulin gene (B2M) was used for normalization.

Figure 3.

Thy1 expression in xenotransplanted cells. A, CD44+/CD24low THY1-positive K197 cells were assayed for the ability to form tumors after injection into mice. Five-week-old female BALB/c athymic (null/null) mice were injected orthotopically in the fourth mammary fat pad, with K197 cells resuspended in Matrigel/PBS. Tumor volume (mm3) was measured by calculating π/6 × largest diameter × (smallest diameter)2, and correlated with the number of cells injected (R = 0.949, P < 0.05). Within 60 days, tumors were excised (250 mm3), digested with trypsin/collagenase mixture and mouse-derived tumor cells were harvested for in vitro growth (Topo9 cells). B, RT-PCR of THY1 mRNA (235 bp) expression in carcinoma specimens S197 (left), paired normal N197 (middle), and xenotransplanted specimen Topo9 (right) tissue. Standardization with β-actin cDNA levels. C, qRT-PCR of THY1 expression in carcinoma specimens S197, paired normal N197 and xenotransplanted specimen Topo9 tissue (S-Topo9). Experiments were performed in triplicate for each data point, and the expression of housekeeping β2-microglobulin gene (B2M) was used for normalization. D, Cells derived from xenotransplanted tumors recapitulate the CD44+/CD24low profile of the grafted stem/progenitor cells. Dot plots of flow cytometry analysis of parental K197 and Topo9 cells. Cells were cultured for 15 days in standard medium, and 1 × 106 cells were stained for the flow cytometric analysis of CD44PE-Cy7A and CD24-Alexa Fluor 647. The expression of each antigen is represented on a frequency distribution histogram (count vs. FITC or PE signal). The expression of the two markers is presented on a biparametric dot plot CD44-PE-Cy7 vs. CD24-AlexaFluor647 for each cell type. Vertical and horizontal markers delineate the quadrants used to identify the CD44/CD24 subsets and were set with the appropriate FMO control. E, The histogram reports the mean fluorescence intensity (MFI) of CD24 (light gray) and CD44 (dark gray) expressed on the cell surface of the indicated cells (log scale range: 1–10,000); SE, not reported on the graph, was ≤ 10%. F, Topo9 cells, cultured from xenotransplanted tumors, are triple-negative, as the grafted K197 THY1-positive cells. RT-PCR for mRNA of ERα (441 bp), PgR (121 bp) and HER2 (420 bp) of K197 and Topo9 cells. Standardization was based on β-actin cDNA levels. Protein expression determined with Western blot analysis at Supplementary Fig. S4. G, Percentage of THY1/CD90 surface protein as determined by flow cytometry with CD90-FITC. H, RT-PCR of THY1 mRNA (235 bp) expression in K197 (left), and xenotransplanted Topo9 (right) cells. Standardization with β-actin cDNA levels. I, qRT-PCR of THY1 expression in MDA-MB231 cell line, in K197 and xenotransplanted Topo9 cells. Experiments were performed in triplicate for each data point, and the expression of housekeeping β2-microglobulin gene (B2M) was used for normalization.

Close modal

To further investigate the relationship between the preimplantation population (K197) and the population obtained from the cultivation in vitro of the CD44+/CD24low cells from implanted tumors (Topo9), we profiled the phenotype by measuring the percentage of expressing cells and the MFI of three epithelial markers (MUC1, E-cadherin, and EpCAM), and seven stem/mesenchymal markers (CALLA, β1integrin, α2 integrin, α6 integrin, β3 integrin, Thy1, and prominin 1). As reported in Table 1, the percentage of expressing cells and the MFI of the markers of the transplanted cells, except Thy1, overlapped that of the parental cells. There was a strong positive correlation in percentage terms between the transplanted and the parental cells (P < 0.001), a weak correlation between the transplanted and the MCF7 cell line, and a possible, albeit not significant correlation between the transplanted and the MCF10A cells. MFI data were in agreement with the percentage data, and statistical analysis confirmed the high correlation between the transplanted and the parental cells (P < 0.001). Like the parental cells, the transplanted cells preserved an α2β1 Integrin-CD44high/MUC1-EpCAM-CD24low phenotype, but THY1 expression was lost (Fig. 3G). To verify this finding, we measured THY1 mRNA expression in the cells obtained after transplantation, and found a consistent decrease of THY1 mRNA in the cultures derived from the transplanted tumors versus the preimplantation cells (Fig. 3H). qRT-PCR normalized to B2M confirmed these differences (P < 0.05; Fig. 3I).

Xenotransplantation silences Thy1 via promoter methylation

The cellular plasticity between the epithelial and mesenchymal states is ascribed to epigenetic changes of cancer cells that result in cellular heterogeneity (36). In metastatic breast cancer, the THY1 gene is silenced by methylation in those tumors that are hormone receptor–negative and basal-like (37). To determine whether Thy1 expression is relevant in breast cancer, we interrogated public databases. Data mining of TCGA 450K DNA methylation, at the human pan-cancer methylation database, MethHC (http://methhc.mbc.nctu.edu.tw/php/index.php), confirmed the enrichment of THY1 hypermethylation in human breast carcinoma compared with normal breast tissue (P ≤ 0.005; Supplementary Fig. S5). To determine whether the expression of Thy1 in myoepithelial precursors is subject to epigenetic regulation, we explored the possibility that methylation might determine the disappearance of Thy1 when progenitors are implanted in mice. Thus, we treated the Topo9 cells with methylation inhibitors and found that THY1 expression was restored in a time-dependent fashion. Parental K197 cells and transplanted Topo9 cells were cultured, for various times (Fig. 4A), with the methylation inhibitor 5-aza-20-deoxycytidine (5-AZA-dC) and subjected to RT-PCR. THY1 was not methylated in parental K197 cells (Fig. 4A, left blot) and highly methylated in untreated Topo9 cells (Fig. 4A, right blot, lane 0). Treatment with the demethylation agent 5-AZA-dC time dependently restored THY1 gene expression in Topo9 cells. The latter effect progressively increased from 0 to 96 hours (Fig. 4A, right blot). These data support the hypothesis that methylation contributes to THY1 silencing. To test this hypothesis, we analyzed the THY1 promoter using the methylation-sensitive amplified polymorphism (MSAP) technique. HpaII (CpG methylation insensitive) digestion resulted in a 276-bp fragment in THY1-low cells (Topo9), but not in THY1-positive (K197) or control cells (C; Fig. 4B). At quantitative real-time PCR analysis of HpaII/MspI sensitivity, amplification rates (64% in THY1-low cells vs. 0.54% and 0.18% in THY1-positive and control cells, respectively) were significantly correlated with methylation status (Fig. 4C). These results indicate that methylation occurs on the THY1 loci in THY1-low Topo9 cells.

Figure 4.

Methylation on the THY1 loci in transplanted cells. A,THY1-positive (K197) and THY1-low (Topo9) cells were cultured, for the time indicated, with the methylation inhibitor 5-aza-20-deoxycytidine (5-AZA-dC) and qRT-PCR for THY1 (235 bp) was performed. Treatment with the demethylation agent 5-AZA-dC did not affect the methylation status of Thy1 in parental K197 THY1-positive cells (left blot) whereas it restored the expression of Thy1 mRNA in Topo9 THY1-low cells (right blot). This effect progressively increased from 0 to 96 hours. Standardization with β-actin. B, Methylation analysis of the THY1 promoter in K197 and Topo9 cells and control cells. After HpaII/MspI digestion (blocked/unblocked by CpG methylation, respectively), only methylated (undigested) genomic DNA produced a fragment. PCR-based assay of HpaII/MspI sensitivity showed a 276-bp fragment only in low expressing/hypermethylated samples (Thy1-low Topo9 cells) and not in high expressing/hypomethylated controls (Thy-positive K197 cells and control cells); no product is seen in MspI digestions. C, Real-time analysis of HpaII/MspI sensitivity. The differences in amplification rates (64% in THY1-negative cells vs. 0.54% and 0.18% in THY1-positive cells and control cells, respectively) relates to the methylation (mC) status. * indicates the difference between THY-low cells versus THY1-positive cells and control cells with P < 0.05; ** indicates the difference between control cells versus THY1-positive cells with P < 0.01.

Figure 4.

Methylation on the THY1 loci in transplanted cells. A,THY1-positive (K197) and THY1-low (Topo9) cells were cultured, for the time indicated, with the methylation inhibitor 5-aza-20-deoxycytidine (5-AZA-dC) and qRT-PCR for THY1 (235 bp) was performed. Treatment with the demethylation agent 5-AZA-dC did not affect the methylation status of Thy1 in parental K197 THY1-positive cells (left blot) whereas it restored the expression of Thy1 mRNA in Topo9 THY1-low cells (right blot). This effect progressively increased from 0 to 96 hours. Standardization with β-actin. B, Methylation analysis of the THY1 promoter in K197 and Topo9 cells and control cells. After HpaII/MspI digestion (blocked/unblocked by CpG methylation, respectively), only methylated (undigested) genomic DNA produced a fragment. PCR-based assay of HpaII/MspI sensitivity showed a 276-bp fragment only in low expressing/hypermethylated samples (Thy1-low Topo9 cells) and not in high expressing/hypomethylated controls (Thy-positive K197 cells and control cells); no product is seen in MspI digestions. C, Real-time analysis of HpaII/MspI sensitivity. The differences in amplification rates (64% in THY1-negative cells vs. 0.54% and 0.18% in THY1-positive cells and control cells, respectively) relates to the methylation (mC) status. * indicates the difference between THY-low cells versus THY1-positive cells and control cells with P < 0.05; ** indicates the difference between control cells versus THY1-positive cells with P < 0.01.

Close modal

THY1-low cells activate the Notch1–EGFR program

We examined THY1-low cells for cellular functions (EMT-marker expression, growth in 3D) that are critical when cells are allowed to adhere to a substrate (38, 39). The protein expression profile of the EMT markers CK18, CK19, CK5, vimentin, αSMA, and fibronectin (Fig. 5A) showed that THY1-low cells acquired a partial epithelial phenotype. In fact, these cells expressed CK18, CK19, and vimentin, reduced fibronectin, but they did not express CK5 or αSMA.

Figure 5.

THY1-methylated cells are Notch1-EGFR–expressing cells. A, Western blot analysis for EMT markers CK18, CK19, CK5, vimentin, and αSMA; THY1-low cells (Topo9) acquired CK18, CK19, and vimentin, and lost CK5 and αSMA. Standardization with tubulin. B, Representative phase-contrast microphotographs of cultures in 3D of mammospheres of THY1-positive (K197, left) and aggregates of THY1-low (Topo9, right) cells. Scale bar, 100 μm. C, Colony growth in soft agar of THY1-positive (K197) and THY1-low (Topo9) cells. Colonies were counted in 10 fields per dish on a horizontal/vertical grid. The mean results of two experiments of triplicates dishes are plotted. D, RT-PCR for EGFR (348 bp) and NOTCH1 (520 bp) in K197 and Topo9 cells. Standardization with β-actin. E, Western blot analysis for EGFR and cleaved NOTCH1 (NICD). Standardization with tubulin. F, RT-PCR for THY1 mRNA (235 bp) and EGFR (348 bp); G, and immunoblots for THY1 and EGFR protein expression in K197 and Topo9 cells treated, 24 hours after seeding, without (−) and with (+) 5 μmol/L 5-AZA-dC for 48 hours. Standardization with β-actin and tubulin, respectively. H, Lapatinib inhibited the proliferation of EGFR-expressing THY1-low (Topo9) cells. The effect of treatment on the survival of control cells (BT474, MCF7, and MDA-MB-231), parental THY1-positive (K197) and xenotransplanted THY1-low (Topo9) cells is expressed as percentage of viable cells over control. All cells were seeded at 2,500 cells/well, treated with lapatinib 0.1 μmol/L (gray) and 1 μmol/L (black) for seven days and analyzed by the MTT assay. The percentage of absorbance of treated samples versus untreated samples is reported as a percentage of viable cells/controls. Values represent means ± SD from triplicate samples for each treatment. Error bars indicate SD values. Asterisks indicate statistical significance, as determined by two-tailed Fisher exact test (**, two-sided P = 0.0011). I, Real-time RT-PCR quantification of THY1, NOTCH1, and EGFR expression in six carcinoma tissues paired with tissue used to obtain primary cultures of Thy1-positive cells; all samples were run in triplicate and normalized to B2M housekeeping.

Figure 5.

THY1-methylated cells are Notch1-EGFR–expressing cells. A, Western blot analysis for EMT markers CK18, CK19, CK5, vimentin, and αSMA; THY1-low cells (Topo9) acquired CK18, CK19, and vimentin, and lost CK5 and αSMA. Standardization with tubulin. B, Representative phase-contrast microphotographs of cultures in 3D of mammospheres of THY1-positive (K197, left) and aggregates of THY1-low (Topo9, right) cells. Scale bar, 100 μm. C, Colony growth in soft agar of THY1-positive (K197) and THY1-low (Topo9) cells. Colonies were counted in 10 fields per dish on a horizontal/vertical grid. The mean results of two experiments of triplicates dishes are plotted. D, RT-PCR for EGFR (348 bp) and NOTCH1 (520 bp) in K197 and Topo9 cells. Standardization with β-actin. E, Western blot analysis for EGFR and cleaved NOTCH1 (NICD). Standardization with tubulin. F, RT-PCR for THY1 mRNA (235 bp) and EGFR (348 bp); G, and immunoblots for THY1 and EGFR protein expression in K197 and Topo9 cells treated, 24 hours after seeding, without (−) and with (+) 5 μmol/L 5-AZA-dC for 48 hours. Standardization with β-actin and tubulin, respectively. H, Lapatinib inhibited the proliferation of EGFR-expressing THY1-low (Topo9) cells. The effect of treatment on the survival of control cells (BT474, MCF7, and MDA-MB-231), parental THY1-positive (K197) and xenotransplanted THY1-low (Topo9) cells is expressed as percentage of viable cells over control. All cells were seeded at 2,500 cells/well, treated with lapatinib 0.1 μmol/L (gray) and 1 μmol/L (black) for seven days and analyzed by the MTT assay. The percentage of absorbance of treated samples versus untreated samples is reported as a percentage of viable cells/controls. Values represent means ± SD from triplicate samples for each treatment. Error bars indicate SD values. Asterisks indicate statistical significance, as determined by two-tailed Fisher exact test (**, two-sided P = 0.0011). I, Real-time RT-PCR quantification of THY1, NOTCH1, and EGFR expression in six carcinoma tissues paired with tissue used to obtain primary cultures of Thy1-positive cells; all samples were run in triplicate and normalized to B2M housekeeping.

Close modal

To shed light on the growth features of THY1-low cells and to evaluate whether the cells exerted sphere-forming activity, we seeded the Topo9 cells under adherent and nonadherent conditions. In 2D cultures, THY1-low cells were smaller than THY1-positive cells and had and an epithelial-like morphology (Supplementary Fig. S3), the median cpdf was 6.3 × 105 at 120 days, that was double that of THY1-positive cells (reported in Fig. 1D). The 3D-harvested THY1-low cells in low adhesion formed aggregates in suspension instead of mammospheres (Fig. 5B). To understand whether the loss of Thy1 affects growth on substrate, we performed soft-agar experiments. After seeding in semisolid medium for 3 weeks, THY1-low Topo9 cell frequency was 4-fold higher than Thy1-positive K197 cell (Fig. 5C), which confirms the greater propensity of THY1-low cells to proliferate on adhesion to a semisolid substrate. These observations prompted us to investigate on pathways involved in EMT.

To evaluate whether THY1-silencing in the grafted cells signals a transition phenotype, we searched for signaling pathways through which cells interact during development and tissue homeostasis. Molecular analysis had demonstrated that both parental K197, THY1-positive, and transplanted Topo9, THY1-low cells are triple-negative (Fig. 3F). Triple-negative breast cancers generally express EGFR (40), which cross talks with the Notch pathway in this setting (41). We measured the levels of EGFR and NOTCH1 mRNA in THY1-positive and THY1-low cells (Fig. 5D). At densitometry, EGFR mRNA was three times higher in Topo9 cells than in parental K197cells (grayscale: 0.62 vs. 1.93, respectively). Moreover, Topo9 cells expressed NOTCH1 mRNA, whereas K197 parental cells did not (Fig. 5D). Immunoblots for EGFR and Notch1 (NICD) confirmed that Topo9 cells express potentially functional levels of these proteins (Fig. 5E). Further mRNA and protein analysis of Topo9 cells rescued after 48 hours with 5-AZA-dC showed a concomitant reduction of EGFR levels together with Thy1 induction (Fig. 5F-G).

Having identified activation of signaling pathways susceptible to targeting, we investigated the ability of THY1-low cells to respond to tyrosine kinase inhibition. In gastric cancers, Thy1 is a cancer stem cell marker and trastuzumab (humanized anti-HER2 antibody) treatment of high tumorigenic gastric primary tumor models reduces the Thy1 population in the tumor mass thereby suppressing tumor growth when combined with chemotherapy (42). As HER2 has no ligand, antibodies against this receptor inhibit its activation by preventing heterodimerization (43) with other members of the HER2 family (44). Heterodimerization results in intrinsic kinase activation. We treated K197 and Topo9 cells, and BT474, MCF7 and MDA-MB231 cells with increasing concentrations (0.7–2–5–10 μg/mL) of trastuzumab and measured cell viability with the MTT assay 4, 5, and 7 days later. Trastuzumab did not alter the viability of either the K197 or Topo9 cells (% of inhibition = 0). Because THY1-low Topo9 cells express the EGFR (HER1), we tested the effect of lapatinib, a reversible tyrosine kinase inhibitor that binds intracellularly and inhibits both the EGFR and HER2 activity. We found that lapatinib, in a concentration-dependent manner, significantly reduced the viability of THY1-low cells (Fig. 5H). Indeed, at concentrations of 0.1 μmol/L and 1 μmol/L, lapatinib reduced cell viability by 75% and 85%, respectively (P = 0.0011). These results suggest that THY-low cell proliferation is sustained by tyrosine kinase activation.

To determine whether Thy1 overepression might have clinical relevance in conjunction with EGFR and Notch1 expression status, we analyzed our series of breast carcinoma and performed real-time PCR experiments on the carcinoma tissue samples paired with those used to establish primary cultures of THY1-positive cells. At qRT-PCR, we examined six tissue specimens and found that THY1 was highly expressed in breast carcinoma tissues (range 0.82–1.9), while transcript levels for EGFR and NOTCH1 were very low (range 0.003–0.1; Fig. 5I).

We interrogated the Pancancer Analysis of Whole Genomes (PCAWG; https://www.ebi.ac.uk/gxa/) for the TCGA RNA-sequencing data, to compare combined expression of THY1, EGFR, and Notch1 in breast adenocarcinoma, normal-adjacent to breast adenocarcinoma, invasive lobular carcinoma, and normal breast. THY1 was more highly expressed in invasive lobular (24FPKM) and invasive adenocarcinoma (18FKM) than in tissue adjacent to breast carcinoma (11FPKM) or normal breast (7FPKM). Interestingly, the levels of EGFR and NOTCH1 appear to be directly related each other, but inversely related to THY1. Indeed, EGFR in normal tissues (13 FPKM) was more than 3-fold higher than in breast tumors (3/4 FPKM), as well NOTCH1 had higher levels (11 FPKM) in normal than in tumors (6/6 FPKM; Fig. 6A).

Figure 6.

Low EGFR-high THY1 expression in human breast correlates with poor prognosis. A, Distribution of various phenotypes with different ratio of Notch1-EGFR/Thy1 expression in a large collection of tumors derived from cancer databases. TCGA RNA-sequencing data were extracted from the PCAWG website (https://www.ebi.ac.uk/gxa/) to compare combined expression of THY1, EGFR and NOTCH1 in breast adenocarcinoma (IDC), normal-adjacent to breast adenocarcinoma, invasive lobular carcinoma (ILC), and normal breast. B and C, Kaplan–Meier analysis of relapse-free survival on subjects with ERα-negative HER2-negative breast cancer on the basis of THY1 (B) and EGFR (C). P value was determined by log-rank test. The group with the lowest THY1 expression were more likely to have a relapse-free survival. The group with highest EGFR expression were more likely to have a relapse-free survival. Multivariate analysis with ER (ESR1) and HER2 (ERBB2) was performed. The plots were generated using http://kmplot.com (probes: 213869_x_at (Thy1-CD90); 211551_at (EGFR, ERBB, ERBB1; ref. 33).

Figure 6.

Low EGFR-high THY1 expression in human breast correlates with poor prognosis. A, Distribution of various phenotypes with different ratio of Notch1-EGFR/Thy1 expression in a large collection of tumors derived from cancer databases. TCGA RNA-sequencing data were extracted from the PCAWG website (https://www.ebi.ac.uk/gxa/) to compare combined expression of THY1, EGFR and NOTCH1 in breast adenocarcinoma (IDC), normal-adjacent to breast adenocarcinoma, invasive lobular carcinoma (ILC), and normal breast. B and C, Kaplan–Meier analysis of relapse-free survival on subjects with ERα-negative HER2-negative breast cancer on the basis of THY1 (B) and EGFR (C). P value was determined by log-rank test. The group with the lowest THY1 expression were more likely to have a relapse-free survival. The group with highest EGFR expression were more likely to have a relapse-free survival. Multivariate analysis with ER (ESR1) and HER2 (ERBB2) was performed. The plots were generated using http://kmplot.com (probes: 213869_x_at (Thy1-CD90); 211551_at (EGFR, ERBB, ERBB1; ref. 33).

Close modal

Furthermore, to situate our model within the context of a broad spectrum of “in vitro” cell lines, we interrogated Oncomine (https://www.oncomine.org) for Neve collection of cultured cells (Supplementary Fig. S6). Also in this case, the results support the notion of an inverse relationship between THY1 and EGFR; as for Topo9 cells, the pattern with low THY1 and high EGFR is largely represented in most of the cell lines broadly used to model breast cancer.

We then performed a Kaplan–Meier analysis to measure relapse-free survival in subjects categorized ER- and HER2-negative based on high/low Thy1 and EGFR expression. At the time of analysis, the median follow-up was 200 months (range, 0.1–200 months). In this setting, high THY1 expression (THY1-high vs THY1-low P = 0.062) (Fig. 6B) as well as low EGFR expression (EGFR-high vs. EGFR-low P = 0.061; Fig. 6C) identifies those patients with a worst relapse-free survival. Multivariate analysis of ESR1 and HER2 expression (stages I and II vs. stage III; HR, 2.815; 95% confidence interval, 1.022–7.751; P = 0.045) excluded interaction of Thy1 and EGFR with ER and HER2, respectively. The analysis suggests that combined detection of Thy1 and/or EGFR expression might help to better identify subclasses of patients with breast cancer of basal origin.

Tumor cell heterogeneity, induced by a combination of genetic and epigenetic events that lead to cancer cell plasticity, is one of the cancer features responsible for drug resistance and treatment failure (45). Gene expression profiling has revealed intertumor heterogeneity and identified five intrinsic breast cancer subtypes: luminal A and B (ERα- and/or PgR-positive and HER2-negative), HER2-enriched (HER2-positive), basal-like (which includes triple negative) and normal-like, that differ in biologic, prognostic and predictive features (2, 3, 46). In routine clinical practice, IHC is used to identify the molecular subclasses (26). Within the subclasses, triple-negative (ER- and PgR- and HER2-negative), which accounts for 15%–20% of all invasive breast cancers, is the most heterogeneous (47), has a higher risk of relapse and responds poorly to targeted therapy (48). Eighty percent (80%) of triple-negative breast cancers harbor a signature that coincides with a high proportion of cells with the basal/myoepithelial phenotype (4). The “phenotypic” plasticity of cancer cells favors intratumor heterogeneity that, through the EMT, promotes the invasion and dissemination of cells, while the MET counterpart confers fitness advantage which enables cells to return to a highly proliferative state and mediates tumor relapse at metastatic sites (45). In tumor xenografts, the expansion of subclones with fitness advantages is ascribed to cosegregating genomic factors, such as methylation, that, as determinants of fitness, lead to reproducible clonal dynamics (49). DNA methylation is a dynamic process that contributes to tumor heterogeneity (50). Because DNA methylation is often altered in early cancer development, candidate methylation markers may serve as prognostic or predictive factors (37). In this context, we investigated the relevance of Thy1 as a biomarker of myoepithelial progenitor to gain insight into the role of basal myoepithelial cells in breast cancer heterogeneity.

Here we isolated and harvested, from human breast cancer tissues, THY1-expressing cells with phenotypic and functional stem cell characteristics. When we transplanted the (α2β1integrin-CD44)high (MUC1-EpCAM-CD24)low THY1-positive myoepithelial progenitors in nude mice, Thy1 expression was lost. Recent studies on invasive breast cancer report that THY1 is highly methylated in the group of HR-Basal-like-p53mutant (37). To evaluate whether epigenetic changes modified Thy1 expression, as occurs in patients with metastatic basal-like tumors, we treated the cells with methylation inhibitors and found that they time dependently restored THY1 mRNA expression. This result suggested the emergence of epigenetic-induced transiting phenotypes. In agreement with data obtained with different approaches (7, 34, 51), in our model of human myoepithelial progenitors, THY1-silenced cells displayed the alternative MET phenotype, which resulted in a better propensity to proliferate and differentiate. This behavior, together with the finding that THY1-low cells, activated the Notch1-EGFR program may explain the rarity of tumors displaying myoepithelial features notwithstanding the ubiquity of myoepithelial cells in breast tissue. The Notch signaling network is an evolutionarily conserved intercellular signaling pathway that regulates interactions between physically adjacent cells. In mouse mammary cells, Notch activation increases the proliferation potential of both bipotent and myoepithelial progenitor cells (52, 53). In normal breast, NOTCH1 mRNA is expressed in luminal cells and its effect on lineage commitment is irreversible (52). In breast cancer, Notch signaling pathways crosstalk with EGFR. Forced overexpression of Notch1 by transfection increases EGFR expression (53), although inhibition of EGFR or Notch signaling alone is not sufficient to suppress human breast cancer cell survival and proliferation (54). Our data suggest that THY1 methylation signals the acquisition of a cycling epithelial phenotype (CK18-CK19-positive/CK5-αSMA-negative) that activates the Notch1–EGFR pathway. Thy1 suppression, through methylation, at metastatic sites might be required for implantation and growth of clones with fitness advantages. Intriguingly, examination on disease progression, of the Sorlie Breast 2 dataset at Oncomine (https://www.oncomine.org) comparing expression status of 107 breast carcinomas at primary site versus 5 metastasis showed a decreased THY1 expression at metastatic site (Supplementary Fig. S7). This observation might have clinical implications because high Thy1 expression might identify, within the heterogeneous basal-like (and TNBC) subclass of breast cancer, a subset of primary tumors originating from precursors/bipotent myoepithelial cells that have worst prognosis. If THY1is methylated and the loss of THY1 coincides with the expression of EGFR, patients might have a better relapse-free survival. The tumors that we obtained after xenotransplantation have low levels of Thy1and high levels of EGFR and resemble those EGFR-overexpressing tumors with better prognosis. Finally, because lapatinib inhibits the viability of THY1-low cells, but not that of THY1-positive cells, the THY1-methylated/EGFR-expressing (THY/EGFR+) phenotype might define a subtype, within the basal-like, that may benefit from tyrosine kinase inhibition. Further studies are needed to determine the relative impact of these processes during cancer evolution.

In our in vitro/in vivo model of stable basal breast cancer progenitor cells in culture, Thy1 expression tracks the myoepithelial lineage. Quiescent myoepithelial progenitor cells are components of luminal and basal subtypes of breast cancer tissues, and can be identified on the basis of their expression of Thy1. THY1-expressing myoepithelial progenitors are quiescent in 3D culture and display a stable phenotypic profile of (α2β1integrin-CD44)high (MUC1-EpCAM-CD24)low that, in vivo, might be responsible for attachment of cells to the extracellular matrix. Upon transplantation, THY1 expression is silenced by methylation parallel to activation of Notch–EGFR signaling. This process marks the emergence of differentiated clones, namely, CK18/CK19/vimentin–positive/CK5/αSMA–negative clones. The latter proliferate extensively, and, notably, might be arrested by tyrosine kinase inhibition. Collectively, our results suggest that THY1 methylation may track the shift of THY1-positive bipotent progenitors into THY1-low differentiated cells. This behavior may explain the rarity of tumors with myoepithelial features, and considered to be of "basal" origin, such as some metaplastic carcinomas, notwithstanding the ubiquity of myoepithelial cells in breast tissue. An understanding of the dynamics of cellular states in breast cancer evolution can lead to a more accurate definition of the subtypes of breast cancer, and opens the way to new therapeutic strategies.

No potential conflicts of interest were disclosed.

Conception and design: G. Arpino, R. Lauria, S. De Placido, B.M. Veneziani

Development of methodology: M. Montanari, R. Lauria, B.M. Veneziani

Acquisition of data (provided animals, acquired and managed patients, provided facilities, etc.): M. Montanari, M.R. Carbone, L. Coppola, G. Arpino, R. Lauria, A. Nardone, F. Leccia, C. Garbi, R. Bianco, B.M. Veneziani

Analysis and interpretation of data (e.g., statistical analysis, biostatistics, computational analysis): M. Montanari, G. Arpino, F. Leccia, E.V. Avvedimento, B.M. Veneziani

Writing, review, and/or revision of the manuscript: M. Giuliano, G. Arpino, R. Lauria, M.V. Trivedi, R. Bianco, E.V. Avvedimento, S. De Placido, B.M. Veneziani

Administrative, technical, or material support (i.e., reporting or organizing data, constructing databases): M. Montanari, M. Giuliano, B.M. Veneziani

Study supervision: R. Lauria, B.M. Veneziani

Breast cancer tissue specimens were obtained from the Breast Cancer Tissue Bank, developed under the auspices of the BIONCAM (Biobanca Oncologica Campania) Project and maintained by the CRPO (Centro Regionale Prevenzione Oncologica), University of Naples “Federico II”. We thank Jean Ann Gilder (Scientific Communication srl) for editing the text. This work was supported by Ministero dell’Universitaria Ricerca, PRIN Grant 2015B7M39T (to S. DePlacido and B.M. Veneziani), Grant MOVIE of the Rete delle Biotecnologie in Campania (to B.M. Veneziani), PON 03PE_00146_1 BIOBIOFAR (to R. Bianco, S. De Placido, and B.M. Veneziani) M. Montanari is supported by a postdoctoral fellowship from POR CREME, M.R. Carbone is supported by a fellowship from Dottorato di Ricerca (PhD) in Medicina Molecolare e Biotecnologie Mediche, University Federico II of Naples, Italy.

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.
Nagle
RB
,
Böcker
W
,
Davis
JR
,
Heid
HW
,
Kaufmann
M
,
Lucas
DO
, et al
Characterization of breast carcinomas by two monoclonal antibodies distinguishing myoepithelial from luminal epithelial cells.
J Histochem Cytochem
1986
;
34
:
869
81
.
2.
Perou
CM
,
Sørlie
T
,
Eisen
MB
,
van de Rijn
M
,
Jeffrey
SS
,
Rees
C
, et al
Molecular portraits of human breast tumours.
Nature
2000
;
406
:
747
52
.
3.
Sorlie
T
,
Perou
CM
,
Tibshirani
R
,
Aas
T
,
Geisler
S
,
Johnsen
H
, et al
Gene expression patterns of breast carcinomas distinguish tumor subclasses with clinical implications.
Proc Natl Acad Sci U S A
2001
;
98
:
10869
74
.
4.
Foulkes
WD
,
Smith
IE
,
Reis-Filho
JS
. 
Triple-negative breast cancer.
N Engl J Med
2010
;
363
:
1938
48
.
5.
Pegram
MD
,
Lipton
A
,
Hayes
DF
,
Weber
BL
,
Baselga
JM
,
Tripathy
D
, et al
Phase II study of receptor-enhanced chemosensitivity using recombinant humanized anti-p185(HER2/neu) monoclonal antibody plus cisplatin in patients with HER2/neu-overexpressing metastatic breast cancer refractory to chemotherapy treatment.
J Clin Oncol
1998
;
16
:
2659
71
.
6.
Barsky
SH
,
Karlin
NJ
. 
Myoepithelial cells: autocrine and paracrine suppressors of breast cancer progression.
J Mammary Gland Biol Neoplasia
2005
;
10
:
249
60
.
7.
Polyak
K
,
Hu
M
. 
Do myoepithelial cells hold the key for breast tumor progression?
J Mammary Gland Biol Neoplasia
2005
;
10
:
231
47
.
8.
Lerwill
MF
. 
Current practical applications of diagnostic immunohistochemistry in breast pathology.
Am J Surg Pathol
2004
;
28
:
1076
91
.
9.
Allinen
M
,
Beroukhim
R
,
Cai
L
,
Brennan
C
,
Lahti-Domenici
J
,
Huang
H
, et al
Molecular characterization of the tumor microenvironment in breast cancer.
Cancer Cell
2004
;
6
:
17
32
.
10.
Bissell
MJ
,
LaBarge
MA
. 
Context, tissue plasticity, and cancer.
Cancer Cell
2005
;
7
:
17
23
.
11.
Hu
M
,
Yao
J
,
Carroll
DK
,
Weremowicz
S
,
Chen
H
,
Carrasco
D
, et al
Regulation of in situ to invasive breast carcinoma transition.
Cancer Cell
2008
;
13
:
394
406
.
12.
Bonewald
L
,
Wang
AC
,
Wang
IY
,
Balch
CM
,
Ades
EW
. 
Antigenic determinant(s) shared by the human Thy 1 and human IgG.
Immunobiology
1984
;
166
:
428
38
.
13.
Kroczek
RA
,
Gunter
KC
,
Germain
RN
,
Shevach
EM
. 
Thy-1 functions as a signal transduction molecule in T lymphocytes and transfected B lymphocytes.
Nature
1986
;
322
:
181
4
.
14.
Seki
T
,
Spurr
N
,
Obata
F
,
Goyert
S
,
Goodfellow
P
,
Silver
J
. 
The human Thy-1 gene: structure and chromosomal location.
Proc Natl Acad Sci U S A
1985
;
82
:
6657
61
.
15.
Kawamura
K
,
Hiroshima
K
,
Suzuki
T
,
Chai
K
,
Yamaguchi
N
,
Shingyoji
M
, et al
CD90 is a diagnostic marker to differentiate between malignant pleural mesothelioma and lung carcinoma with immunohistochemistry.
Am J Clin Pathol
2013
;
140
:
544
9
.
16.
Schubert
K
,
Gutknecht
D
,
Köberle
M
,
Anderegg
U
,
Saalbach
A
. 
Melanoma cells use thy-1 (CD90) on endothelial cells for metastasis formation.
Am J Pathol
2013
;
182
:
266
76
.
17.
Abeysinghe
HR
,
Pollock
SJ
,
Guckert
NL
,
Veyberman
Y
,
Keng
P
,
Halterman
M
, et al
The role of the THY1 gene in human ovarian cancer suppression based on transfection studies.
Cancer Genet Cytogenet
2004
;
149
:
1
10
.
18.
Tang
KH
,
Dai
YD
,
Tong
M
,
Chan
YP
,
Kwan
PS
,
Fu
L
, et al
A CD90+ tumor-initiating cell population with an aggressive signature and metastatic capacity in esophageal cancer.
Cancer Res
2013
;
73
:
2322
32
.
19.
He
J
,
Liu
Y
,
Zhu
T
,
Zhu
J
,
DiMeco
F
,
Vescovi
aL
, et al
CD90 is identified as a candidate marker for cancer stem cells in primary high-grade gliomas using tissue microarrays.
Mol Cell Proteomics
2012
;
11
:
M111.010744
.
20.
Yang
ZF
,
Ho
DW
,
Ng
MN
,
Lau
CK
,
Yu
WC
,
Ngai
P
, et al
Significance of CD90+ cancer stem cells in human liver cancer.
Cancer Cell
2008
;
13
:
153
66
.
21.
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
.
22.
Taube
JH
,
Herschkowitz
JI
,
Komurov
K
,
Zhou
AY
,
Gupta
S
,
Yang
J
, et al
Core epithelial-to-mesenchymal transition interactome gene-expression signature is associated with claudin-low and metaplastic breast cancer subtypes.
Proc Natl Acad Sci U S A
2010
;
107
:
15449
54
.
23.
Lu
H
,
Clauser
KR
,
Tam
WL
,
Fröse
J
,
Ye
X
,
Eaton
EN
, et al
A breast cancer stem cell niche supported by juxtacrine signalling from monocytes and macrophages.
Nat Cell Biol
2014
;
16
:
1105
17
.
24.
Leccia
F
,
Nardone
A
,
Corvigno
S
,
Del Vecchio
L
,
De Placido
S
,
Salvatore
F
, et al
Cytometric and biochemical characterization of human breast cancer cells reveals heterogeneous myoepithelial phenotypes.
Cytometry A
2012
;
81
:
960
72
.
25.
Benz
CC
,
Scott
GK
,
Sarup
JC
,
Johnson
RM
,
Tripathy
D
,
Coronado
E
, et al
Estrogen-dependent, tamoxifen-resistant tumorigenic growth of MCF-7 cells transfected with HER2/neu.
Breast Cancer Res Treat
1992
;
24
:
85
95
.
26.
Goldhirsch
A
,
Winer
EP
,
Coates
AS
,
Gelber
RD
,
Piccart-Gebhart
M
,
Thurlimann
B
, et al
Personalizing the treatment of women with early breast cancer: Highlights of the st gallen international expert consensus on the primary therapy of early breast Cancer 2013.
Ann Oncol
2013
;
24
:
2206
23
.
27.
Nardone
A
,
Cavaliere
C
,
Corvigno
S
,
Limite
G
,
Placido
S
,
Veneziani
BM
. 
A banking strategy toward customized therapy in breast cancer.
Cell Tissue Bank
2009
;
10
:
301
8
.
28.
Veneziani
BM
,
Criniti
V
,
Cavaliere
C
,
Corvigno
S
,
Nardone
A
,
Picarelli
S
, et al
In vitro expansion of human breast cancer epithelial and mesenchymal stromal cells: optimization of a coculture model for personalized therapy approaches.
Mol Cancer Ther
2007
;
6
:
3091
100
.
29.
Formisano
L
,
Nappi
L
,
Rosa
R
,
Marciano
R
,
D Amato
C
,
D Amato
V
, et al
Epidermal growth factor receptor activation modulates Src-dependent resistance to lapatinib in breast cancer models.
Breast Cancer Res
2014
;
16
:
R45
.
30.
Gupta
RC
. 
Nonrandom binding of the carcinogen N-hydroxy-2-acetylaminofluorene to repetitive sequences of rat liver DNA in vivo.
Proc Natl Acad Sci U S A
1984
;
81
:
6943
7
.
31.
Nardone
A
,
Corvigno
S
,
Brescia
A
,
D’Andrea
D
,
Limite
G
,
Veneziani
BM
. 
Long-term cultures of stem/progenitor cells from lobular and ductal breast carcinomas under non-adherent conditions.
Cytotechnology
2011
;
63
:
67
80
.
32.
Cavaliere
C
,
Corvigno
S
,
Galgani
M
,
Limite
G
,
Nardone
A
,
Veneziani
BM
. 
Combined inhibitory effect of formestane and herceptin on a subpopulation of CD44+/CD24low breast cancer cells.
Cancer Sci
2010
;
101
:
1661
9
.
33.
Gyorffy
B
,
Lanczky
A
,
Eklund
AC
,
Denkert
C
,
Budczies
J
,
Li
Q
, et al
An online survival analysis tool to rapidly assess the effect of 22,277 genes on breast cancer prognosis using microarray data of 1,809 patients.
Breast Cancer Res Treat
2010
;
123
:
725
31
.
34.
Al-Hajj
M
,
Wicha
MS
,
Benito-Hernandez
A
,
Morrison
SJ
,
Clarke
MF
. 
Prospective identification of tumorigenic breast cancer cells.
Proc Natl Acad Sci U S A
2003
;
100
:
3983
8
.
35.
Stingl
J
. 
Detection and analysis of mammary gland stem cells.
J Pathol
2009
;
217
:
229
41
.
36.
Pal
B
,
Chen
Y
,
Bert
A
,
Hu
Y
,
Sheridan
JM
,
Beck
T
, et al
Integration of microRNA signatures of distinct mammary epithelial cell types with their gene expression and epigenetic portraits.
Breast Cancer Res
2015
;
17
:
85
.
37.
Conway
K
,
Edmiston
SN
,
May
R
,
Kuan
PF
,
Chu
H
,
Bryant
C
, et al
DNA methylation profiling in the Carolina Breast Cancer Study defines cancer subclasses differing in clinicopathologic characteristics and survival.
Breast Cancer Res
2014
;
16
:
450
.
38.
Thiery
JP
,
Acloque
H
,
Huang
RY
,
Nieto
MA
. 
Epithelial-mesenchymal transitions in development and disease.
Cell
2009
;
139
:
871
90
.
39.
Valastyan
S
,
Weinberg
RA
. 
Tumor metastasis: molecular insights and evolving paradigms.
Cell
2011
;
147
:
275
92
.
40.
Nielsen
TO
,
Hsu
FD
,
Jensen
K
,
Cheang
M
,
Karaca
G
,
Hu
Z
, et al
Immunohistochemical and clinical characterization of the basal- like subtype of invasive breast carcinoma.
Clin Cancer Res
2004
;
10
:
5367
74
.
41.
Dong
Y
,
Li
A
,
Wang
J
,
Weber
JD
,
Michel
LS
. 
Synthetic lethality through combined notch-epidermal growth factor receptor pathway inhibition in basal-like breast cancer.
Cancer Res
2010
;
70
:
5465
74
.
42.
Jiang
J
,
Zhang
Y
,
Chuai
S
,
Wang
Z
,
Zheng
D
,
Xu
F
, et al
Trastuzumab (herceptin) targets gastric cancer stem cells characterized by CD90 phenotype.
Oncogene
2012
;
31
:
671
82
.
43.
Schmitz
KR
,
Ferguson
KM
. 
Interaction of antibodies with ErbB receptor extracellular regions.
Exp Cell Res
2009
;
315
:
659
70
.
44.
Alvarado
D
,
Klein
DE
,
Lemmon
Ma
. 
ErbB2 resembles an autoinhibited invertebrate epidermal growth factor receptor.
Nature
2009
;
461
:
287
91
.
45.
Brooks
MD
,
Burness
ML
,
Wicha
MS
. 
Therapeutic implications of cellular heterogeneity and plasticity in breast cancer.
Cell Stem Cell
2015
;
17
:
260
71
.
46.
van 't Veer
LJ
,
Dai
H
,
van de Vijver
MJ
,
He
YD
,
Hart
AAM
,
Mao
M
, et al
Gene expression profiling predicts clinical outcome of breast cancer.
Nature
2002
;
415
:
530
6
.
47.
Lehmann
BD
,
Bauer
JA
,
Chen
X
,
Sanders
ME
,
Chakravarthy
AB
,
Shyr
Y
, et al
Identification of human triple-negative breast cancer subtypes and preclinical models for selection of targeted therapies.
J Clin Invest
2011
;
121
:
2750
67
.
48.
Carey
LA
,
Perou
CM
,
Livasy
CA
,
Dressler
LG
,
Cowan
D
,
Conway
K
, et al
Race, breast cancer subtypes, and survival in the Carolina Breast Cancer Study.
JAMA
2006
;
295
:
2492
502
.
49.
Eirew
P
,
Steif
A
,
Khattra
J
,
Ha
G
,
Yap
D
,
Farahani
H
, et al
Dynamics of genomic clones in breast cancer patient xenografts at single-cell resolution.
Nature
2014
;
518
:
422
6
.
50.
Mazor
T
,
Pankov
A
,
Johnson
BE
,
Hong
C
,
Hamilton
EG
,
Bell
RJA
, et al
DNA methylation and somatic mutations converge on the cell cycle and define similar evolutionary histories in brain tumors.
Cancer Cell
2015
;
28
:
307
17
.
51.
Raouf
A
,
Zhao
Y
,
To
K
,
Stingl
J
,
Delaney
A
,
Barbara
M
, et al
Transcriptome analysis of the normal human mammary cell commitment and differentiation process.
Cell Stem Cell
2008
;
3
:
109
18
.
52.
Bouras
T
,
Pal
B
,
Vaillant
F
,
Harburg
G
,
Asselin-Labat
ML
,
Oakes
SR
, et al
Notch signaling regulates mammary stem cell function and luminal cell-fate commitment.
Cell Stem Cell
2008
;
3
:
429
41
.
53.
Dontu
G
,
Jackson
KW
,
McNicholas
E
,
Kawamura
MJ
,
Abdallah
WM
,
Wicha
MS
. 
Role of Notch signaling in cell-fate determination of human mammary stem/progenitor cells.
Breast Cancer Res
2004
;
6
:
R605
15
.
54.
Dai
J
,
Ma
D
,
Zang
S
,
Guo
D
,
Qu
X
,
Ye
J
, et al
Cross-talk between Notch and EGFR signaling in human breast cancer cells.
Cancer Invest
2009
;
27
:
533
40
.