Author contributions: A.F.V.: conception and design, collection and/or assembly of data, data analysis and interpretation, and manuscript writing; S.R.: collection and/or assembly of data and data analysis and interpretation; M.P.A., M.R.D., and N.M.: collection and/or assembly of data; A.A., G.F., R.G., R.S., and F.S.: data analysis and interpretation; J.F.C.-T.: provision of study material or patients; R.B.C.: conception and design, provision of study material or patients, and data analysis and interpretation; J.P.: conception and design, data analysis and interpretation, manuscript writing, and final approval of manuscript.
Disclosure of potential conflicts of interest is found at the end of this article.
First published online in STEM CELLSEXPRESS March 2, 2012.
Although the luminal progenitor cell of the normal mammary gland hierarchy has been proposed as the cell-of-origin for basal-like breast cancers, finding the cancer stem cell (CSC) phenotype for this malignancy has proven a difficult task, mostly due to the lack of specific markers. Recently, basal-like sporadic and familial cases of breast cancer have been linked to BRCA1 gene inactivation, which enables the upregulation of the target-repressed CDH3/P-cadherin gene, an important biomarker of basal-like breast carcinomas. Previously, we demonstrated that P-cadherin overexpression can mediate aggressive behavior in these tumors. Thus, our aim was to test whether P-cadherin mediates stem cell properties in basal-like breast carcinomas. Using a series of breast cancer cell lines and primary tumors, we showed that P-cadherin was directly associated with the expression of the breast stem markers CD44, CD49f, and aldehyde dehydrogenase 1 in the basal subtype. Moreover, cell population enriched for P-cadherin expression comprised increased in vitro mammosphere-forming efficiency and capacity to grow colonies in three-dimensional cultures as well as greater tumorigenicity. Importantly, an association was found with stem-/progenitor-like phenotypes of the breast, including the luminal progenitor population, CD49f+CD24+. Additionally, P-cadherin expression conferred resistance to x-ray-induced cell death, sustaining a role for this molecule in another stem cell property. In summary, we demonstrated, for the first time, that P-cadherin mediates stem cell properties, which could be explored in order to better define the CSC phenotype of basal-like breast tumors and the cell-of-origin of this malignancy. STEM CELLS2012;30:854–864
Several studies have shown that solid tumors may contain a small subpopulation of cancer cells that are tumorigenic and have the ability to self-renew and generate all the diverse cancer cells present within the tumor mass. This experimental evidence supports the cancer stem cell (CSC) hypothesis which proposes that a hierarchy exists in the solid tumors comparable to the one found in normal tissue differentiation . In breast, these cells are named breast CSCs. Breast CSCs share important properties with mammary stem cells, namely the ability to proliferate and resist to radiation- and chemotherapy-induced cell death, allowing them to survive and to cause tumor recurrence [2, 3]. However, the identification of breast CSCs has been a hard task due to the current technical constraints and the high intertumor and intratumor heterogeneity observed in breast cancer . Most authors make use of cell surface proteins, usually adhesion-related molecules, in an attempt to define a subpopulation of cells that represents the breast CSC population. In 2003, Michael Clarke's group isolated a subset of breast cancer cells with the phenotype ESA+/CD44+/CD24−/low, which were able to self-renew and were highly tumorigenic at a low cell inoculum . Since then, several other phenotypes/markers to isolate breast CSCs have been described.
For the basal-like breast cancer molecular subtype, in particular, which constitute 10% of all breast cancer cases, few descriptions exist concerning the isolation of their breast CSCs. Meyer et al. isolated CD44+/CD49fhigh/CD133-2high cancer cells from estrogen-receptor (ER)-negative patient tumors, which showed increased tumorigenic ability . Wright et al. described the phenotype CD133+ as able to isolate CSCs from tumors developed in BRCA1−/− mice . Hwang-Verslues et al. characterized the CSC phenotype PROCR+/ESA+ for the human basal MDA-MB-231 breast cancer cell line .
It was recently demonstrated that basal-like breast cancers have a molecular phenotype comparable to the luminal progenitor of the normal breast [9, 10], which raised the hypothesis that markers of luminal progenitors would be good CSC markers for basal-like lesions. Additionally, it has been shown that inactivation of BRCA1 gene in the luminal layer of the normal breast originates breast carcinomas in mice that resemble basal-like carcinomas in humans . In fact, BRCA1 is a major regulator of normal luminal maturation  and it is essential for the repression of a panel of genes that are typically expressed in basal-like carcinomas of the breast, such as the CDH3/P-cadherin gene . These evidences suggest that P-cadherin can be an important CSC marker for this type of tumor lesions.
P-cadherin, a classic type I adhesion molecule, is normally expressed in the myoepithelial/basal layer of the breast and is frequently overexpressed in basal-like breast carcinomas [13–15]. We have found that P-cadherin expression is linked to aggressive tumor behavior, increasing the production of metaloproteases (MMPs) by cancer cells to the extracellular matrix as well as inducing cancer cell invasion, migration, and motility, due to a mechanism involving alterations in the actin cytoskeleton and signaling through small guanosine triphosphatase (GTPase)-binding proteins [16, 17]. However, P-cadherin is also involved in homeostatic processes, such as cell differentiation, development, and embryogenesis, illustrating an indirect effect of this adhesion molecule in stem cell biology. P-cadherin deficient female mice present abnormal mammary gland morphology, showing premature differentiation of the breast and increased risk of developing preneoplastic lesions, such as alveolar hyperplasia and ductal dysplasia . In fact, P-cadherin seems to be important in the maintenance of an undifferentiated state in the malignant setting, as breast tumors with P-cadherin expression show loss of cell polarity . The role of P-cadherin in development and differentiation is also seen during embryonic histogenesis, since this cadherin is present in the extraembryonic ectoderm and visceral endoderm, structures originating the placenta . P-cadherin also has a direct effect in normal stem cells, since it was identified as a stem cell surface marker in human embryonic stem cells . Early hair progenitor cells were also isolated as P-cadherin+ (and K14+/α6-integrin [CD49f+] cells) . Furthermore, the stem cell related transcription factors β-catenin, p63, and C/EBP-β were shown to induce P-cadherin promoter activation [22–24]. Interestingly, in breast, P-cadherin is found in the cap cells, characteristic stem cells that are the precursors of myoepithelial cells [25, 26], and in the myoepithelial layer , eventually contributing to the suprabasal stem cell niche. Herein, we used human mammary cell lines (normal and malignant) as well as a series of invasive breast carcinomas to provide evidence that P-cadherin expression is important in the CSC context not only as a biomarker that better defines the basal-like breast CSC phenotype but also as a protein with direct relevance in stem cell activity in this specific molecular subtype.
MATERIALS AND METHODS
Human breast cell lines T47D, MDA-MB-468, BT-20, BT-549, and MCF-10A were obtained from ATCC (American Type Culture Collection, Manassas, VA, www.atcc.org). The human breast cancer cell line MCF-7/AZ was obtained from a collection developed in the laboratory of Prof. Marc Mareel (Ghent University Hospital, Ghent, Belgium), which was genetically manipulated to overexpress P-cadherin (MCF-7/AZ.Pcad). The control cell line (MCF-7/AZ.mock) shows low P-cadherin levels, identical to the parental cell line . MCF-10A cells were cultured in Dulbecco's modified Eagle's medium (DMEM): F12, supplemented with 5% heat inactivated horse serum, 10 μg/ml insulin, 0.5 μg/ml hydrocortisone, 20 ng/ml epidermal growth factor (EGF), and 100 ng/ml cholera toxin (Sigma-Aldrich, St Louis, MO, www.sigmaaldrich.com). All the other cell lines were grown in DMEM supplemented with 10% fetal bovine serum (FBS) and with 1% antibiotic solution (penicillin–streptomycin) (Invitrogen, Carlsbad, CA, www.invitrogen.com). All cell lines were routinely cultured in a humidified atmosphere with 5% CO2 and at 37°C. Cells were used in experiments when reached 70%–80% confluence.
Fluorescence-Activated Cell Sorting Analysis and Sorting for P-Cadherin and for the Breast Stem Markers CD44, CD24, and CD49f
Cells were harvested with versene/0.48 mM EDTA (Invitrogen). Detached cells were washed with phosphate buffered saline (PBS) supplemented with 0.5% FBS and resuspended in the stain buffer (2 mM EDTA + 0.5% bovine albumin in PBS). A single-cell suspension was labeled by fluorescence-conjugated antibodies at a concentration of 1–10 in stain buffer: fluorescein isothiocyanathe (FITC)-conjugated CD44, phycoerythrin (PE)-conjugated CD24, and FITC-conjugated CD49f. These antibodies were obtained from BD Biosciences (San Diego, CA, www.bdbiosciences.com). P-cadherin monoclonal antibody APC-conjugated was obtained from R&D (Minneapolis, MN, www.rndsystems.com) and used at the same concentration as above. A live-dead stain (Invitrogen) and the primary antibodies or the respective isotype controls (BD Biosciences) were incubated at 4°C, in the dark, for 15 minutes. The labeled cells were then washed in the stain buffer and analyzed on a LSR-II (BD Biosciences). In multicolor experiments, fluorescent minus one samples were used to determine the gating strategy.
For the sorting experiments, the normal cell line MCF-10A, two luminal cell lines, and two basal cell lines were selected and subpopulations were sorted according to P-cadherin expression (highest 20% expressing cells vs. lowest 20% expressing cells). Briefly, cells were stained for P-cadherin (APC) and a live-dead dye in stain buffer, as above. Cells were then passed through a 50 μm mesh to remove clumps and resuspended in stain buffer prior to sorting. Cells were sorted using BD Influx or FACS ARIA-II (BD Biosciences) and collected into 10% Hanks buffered solution (Invitrogen). The purity of sorted population was checked and the sorted population contained 80%–95% purified cells. In addition, a further sample was also collected from cells passed through the laser under pressure, but not sorted, to act as a control for the effect of the pressure on the cells. No differences in cell behavior were detected.
Aldehyde Dehydrogenase Activity
Aldehyde dehydrogenase (ALDH) activity was assessed using the ALDEFLUOR assay (Stem Cell Technologies, Grenoble, France, www.stemcell.com). In brief, cells were incubated in ALDEFLUOR assay buffer containing ALDH substrate (1.5 μmol/l per 106 cells). In each experiment, a sample of cells was also incubated with 15 μmol/l of diethylaminobenzaldehyde, a specific ALDH inhibitor, to use as a negative control. P-cadherin monoclonal antibody (APC-conjugated; R&D) was incubated with the sorted ALDEFLUOR positive and negative cells in assay buffer for 20 minutes on ice. Incubation with the isotype antibody (BD Biosciences) was used as a control. Cells were washed in assay buffer and analyzed in the FACS analyzer LSR-II (BD Biosciences).
Tissue Microarray and Immunohistochemistry
A series of 466 primary invasive breast carcinomas were retrieved from the files of the Department of Pathology, Hospital Xeral-Cíes, Vigo, Spain, which were diagnosed between 1978 and 1992. Breast cancer patient follow-up information was available for 455 cases ranging from 1 to 120 months after the diagnosis. Representative tumor areas were carefully selected and at least two tissue cores (0.6 mm in diameter) were deposited into a tissue microarray. Immunohistochemistry was performed for P-cadherin (BD Biosciences) (1/50, 60 minutes, RT) and the stem cell markers CD44 (clone 156-3C11; Cell Signaling Technology, Danvers, MA, www.cellsignal.com) (1/100, 30 minutes, RT), CD24 (clone Ab2-SN3b; Neomarkers, Fremont, CA) (1/100, 60 minutes, RT), CD49f (HPA012696; Sigma-Aldrich, St. Louis, MO) (1/10, 120 minutes, RT), and ALDH1 (clone EP1933Y; Abcam, Cambridge, UK, www.abcam.com) (1/100, 60 minutes, RT).
High temperature (98°C) antigenic retrieval with Tris-EDTA (P-cadherin) or citrate buffer (CD44, CD24, CD49f, and ALDH1) was performed before primary antibody incubation. The primary antibodies were detected using a secondary antibody with horseradish peroxidase polymer (Cytomation Envision System HRP; DAKO, Carpinteria, CA, www.dako.com) using diaminobenzidine as chromogen, according to the manufacturer's instructions. This series of invasive breast carcinomas was previously characterized in terms of patient and tumor parameters (Supporting Information Table S1) .
P-cadherin, CD44, CD24, and CD49f staining were detected mainly at the membrane of tumor cells and the scoring was considered as follows: (0), 0%–10% of positive tumor cells; (1+), 10%–25% of positive tumor cells; (2+), 25%–50% of positive tumor cells; (3+), >50% of positive tumor cells. For P-cadherin, CD44, and CD49f, the cases which were classified as (0) were considered negative, whereas (1+), (2+), and (3+) were established as positive cases. For CD24, the cases were divided into negative/low (−/low), when considered (0) or (1+), or into positive cases when classified as (2+) or (3+). Immunohistochemical staining of ALDH1 was classified as positive when more than 1% of tumor cells showed clear cytoplasmic positivity [30, 31]. Since the immunohistochemical result was not interpretable for some of these markers, the statistical analyses were performed using only the breast tumor cases with available data. This study was conducted under the national regulative law for the usage of biological specimens from tumor banks, where the samples are exclusively available for research purposes in the case of retrospective studies.
P-Cadherin and Stem Cell Markers Knockdown and Immunoblot Analysis
P-cadherin (CDH3 gene) expression was silenced by a specific siRNA, target sequence: AAGCCTCTTACCTGCCGTAAA. Inhibition of P-cadherin was maintained for at least 72 hours after cell transfection, confirmed by Western blot. Inhibition of the expression of CD49f (ITGA6 gene), target sequence: CAGGGTAATAAACTTAGGTAA, and CD44 (CD44 gene), target sequence: AACTCCATCTGTGCAGCAAAC, was also performed.
All transfections were carried out using HiPerFect transfection reagent (Qiagen, Hilden, Germany) in a final concentration of 2 nM siRNA (Qiagen), according to manufacturer instructions. A siRNA scrambled sequence was included as a control (Qiagen).
Protein analysis was performed by Western blot using the following antibodies: anti-P-cadherin (BD Transduction, San Diego, CA, www.bdbiosciences.com), anti-CD49f (HPA012696; Sigma-Aldrich), anti-CD44 (Cell Signaling Technology, Danvers, MA), anti-CD44v6 (VFF-7; Abcam, Cambridge, U.K.), and anti-α-tubulin (DM1A; Sigma-Aldrich). Secondary antibodies were peroxidase conjugated, from Santa Cruz Biotechnology (Heidelberg, Germany, www.scbt.com) and detection was performed using the Amersham Hyperfilm and Amersham ECL Detection substrate (GE Healthcare, Chalfont St. Giles, U.K., www.gehealthcare.com).
Monolayer cells were enzymatically detached with 0.125% trypsin-EDTA (Sigma, St. Louis, MO), manually disaggregated with a 25-gauge needle to a single-cell suspension, and resuspended in cold PBS. Cells were plated at 500 cm−2 in nonadherent culture conditions, in flasks coated with 1.2% poly(2-hydroxyethylmethacrylate)/95% ethanol (Sigma). Cells were grown for 5 days, in DMEM/F12 containing B27 supplement, 500 ng/ml hydrochortisone, 40 ng/ml insulin, 20 ng/ml EGF, and maintained in a humidified incubator at 37°C and 5% (vol/vol) CO2. Mammosphere-forming efficiency (MFE) was calculated as the number of mammospheres (≥50 μm) formed, divided by the cell number plated, being expressed as a percentage.
The three-dimensional (3D) on-top method was used. Briefly, single-cell suspensions were seeded at a density of 250 or 1,000 cells per well into eight-well glass chamber slides containing 50 μl of 100% growth factor-reduced matrigel per well (a biologically active matrix material resembling the mammalian cellular basement membrane [BD Biosciences]). Cells were plated in growth medium containing H14 medium (DMEM/F12 with insulin 250 ng/ml, transferrin 10 μg/ml, sodium selenite 2.6 ng/ml, estradiol 10−10 M, hydrocortisone 1.4 × 10−6 M, prolactin 5 μg/ml, and EGF 10 ng/ml, according to Kenny et al.  with 5% growth factor-reduced matrigel and 1% FBS).
Cells were incubated at 37°C, with replacement of the growth medium containing 2% growth factor-reduced matrigel every 2–3 days, to allow 3D structures to form. The size and number of the structures formed were assessed microscopically after 21 days.
Cell X-Ray Irradiation
Normal and cancer cells were plated in mammosphere culture conditions and immediately irradiated with 2 Gy or 4 Gy, respectively. Irradiations were performed using a 320 kV X-ray system (Gulmay Medical Ltd., Camberley, U.K.). The machine was operated at 300 kV, 10 mA, with filtration fitted in the X-ray beam to give a radiation quality of 2.3 mm Cu half-value layer. Samples were positioned at a distance of 500 mm from the X-ray focus, where the dose rate was determined to be 1.37 Gy/minute.
In Vivo Assessment of P-Cadherin Tumorigenic Capacity
The basal-like cell line MDA-MB-468 was sorted according to P-cadherin expression into two subpopulations: P-cadhigh and P-cadlow fractions, as described above. The sorted cells were xenotransplanted at varying dilutions (106, 105, or 5 × 104 cells in 100 μl cell suspension) into the subcutaneous region, under the mammary fat pad of 4–5-week-old female N:NIH(s)II:nu/nu nude mice, using a 25-gauge needle. Mice were maintained and housed at IPATIMUP Animal House, sited at the Medical Faculty of the University of Porto, in a pathogen-free environment, under controlled conditions of light and humidity. Animal experiments were carried out in accordance with the European Guidelines for the Care and Use of Laboratory Animals, directive 2010/63/UE. Mice (four per group) were weighted, and tumor width and length were measured with calipers every week. Tumor volume was estimated by using the equation, V = 0.5 × a × b2, where V is volume, a is the length of the major axis of the tumor, and b is the length of its minor axis. Mice were euthanized 3 months after tumor cell inoculation.
Mammosphere-forming ability and growth in 3D cultures (size and number) were compared using two-tailed unpaired t test. Immunohistochemical associations between the molecular markers were assessed by Pearson's correlation and Chi-squared tests. Survival curves were estimated by the Kaplan-Meier method and compared using the log-rank test to assess significant differences for overall survival. A maximum cutoff value of 120 months (10 years) was considered. For multivariate survival analysis, Cox regression models were fitted to estimate hazard ratios (HRs) and the corresponding 95% confidence interval. Statistical analyses were carried out using SPSS statistics 17.0 software (SPSS Inc., Chicago, IL), and a significant level of 5% was considered. Flow cytometry data were analyzed with the Flowjo software package (Treestar, Ashland, OR).
Normal and Tumorigenic Breast Cell Lines, with a Basal-Like Phenotype, Are Enriched for the Expression of P-Cadherin as well as for the Stem Cell Markers CD44, CD49f and Show an Increased ALDEFLUOR-Positive Subpopulation
P-cadherin is normally expressed in the basal layer of the mammary epithelium  and is frequently upregulated in basal-like breast carcinomas [13–15, 27, 33]. In order to study whether this protein was associated with the expression of CSC markers in vitro, a panel of breast cancer cell lines was used. This series of cell lines comprised two main subgroups: the ER-positive luminal cell lines (MCF-7/AZ and -T47D) and the ER-negative basal-like cell lines (MDA-MB-468, BT-20, and BT-549) [34, 35]. In order to demonstrate that this putative association was not restricted to cancer cells, a normal mammary cell line was also included in the study (MCF-10A), which was previously described as harboring a basal gene expression signature .
In accordance to what is seen in human breast carcinomas [13, 27, 33], high/moderate cell surface expression of P-cadherin (>50% of positive cells by fluorescence-activated cell sorting [FACS]) was preferentially observed in basal-like cell lines, in contrast to luminal cell lines, which show lower levels of this protein (Fig. 1A). The same association was observed for CD44 and CD49f, with increased cell surface expression of these markers in the basal-like phenotype (P-cadherin high/moderate cells), in contrast to the luminal phenotype (P-cadherin low cells) (Fig. 1B). Other studies are in accordance with our data indicating that, in breast cell lines, luminal cells usually express lower levels of CD44 [36–38] and CD49f [35, 39] in comparison with basal cells. Importantly, the later molecules are already well established as CSC markers of the basal phenotype [6, 7] as well as putative mammary gland stem cell biomarkers [9, 40, 41]. All the cell lines expressed high levels of CD24, independently of the level of P-cadherin expression. In addition to the cell surface markers, it was previously shown that normal and cancer human mammary epithelial cells with high ALDH enzyme activity have stem/progenitor properties . The analysis of ALDH activity in this panel of cell lines revealed that the P-cadherinmod/high basal cells showed a significant subpopulation with this putative stem cell profile (>15%), in contrast to P-cadherinlow luminal cells (Fig. 1B). Again, these results are in accordance with other studies that have shown that basal cells have an increased ALDEFLUOR+ subpopulation, compared with the luminal ones [42, 43].
FACS analysis of the ALDEFLUOR+ and ALDEFLUOR− cell compartments of cell lines from the two main groups showed that there was an enrichment of P-cadherin expression in the ALDEFLUOR+ stem cell compartment, pointing for a direct association between these two markers. Interestingly, this association was only found in the basal-like group of cell lines, as shown in MDA-MB-468 (Fig. 1C). In order to more clearly demonstrate the link between P-cadherin and the ALDEFLUOR+ subpopulation, we decided to perform transient inhibition of P-cadherin in the basal cells MDA-MB-468, BT-20, BT-549, and MCF-10A. Measurement of ALDH1 activity showed that the inhibition of P-cadherin leads to a decrease in the number of ALDEFLUOR+ cells in all cell lines, being more significant in MCF-10A and BT-549. In the cell line BT-549, the ALDEFLUOR+ subpopulation significantly decreased from 50.9% to 33.2% (Fig. 1D). These results clearly show an association between P-cadherin and the stem cell pool.
The inhibition of P-cadherin also clearly impacted in the expression of the stem cell marker CD49f. In all basal-like cell lines analyzed, P-cadherin knockdown was accompanied by a reduction of CD49f expression (Fig. 1E). Expression of CD44 was also affected, namely the CD44v6 isoform, with a reduction found in the cell line BT-549 after P-cadherin inhibition (Fig. 1E). We also performed transient knockdown of CD44 or CD49f and studied the impact of these stem cell markers on P-cadherin expression. Inhibition of CD44 caused a slight reduction of P-cadherin expression in the BT-549 cell line and inhibition of CD49f did not affect P-cadherin expression in the three basal cell lines (Fig. 1E).
P-Cadherin Expression Is Associated with the Phenotype of the Luminal Progenitor from the Normal Breast Differentiation Hierarchy: CD49f+CD24+
Combinations of the markers CD44, CD24, and CD49f have been used in the literature in order to define subpopulations within cell lines or tissues that have stem or CSC properties [5, 41, 44–49]. Specifically, in the CSC field, the phenotype CD44+CD24−/low is believed to exhibit CSC properties  and some authors also suggest that the CD44+CD24+ phenotype similarly harbors stem-like properties [50, 51]. Importantly, CD44 expression in the normal human hierarchy is reported in both luminal and basal lineages by Visvader  and Raouf et al. , with a lower expression in the luminal compartment as seen in breast cell lines. Studies performed by Shipitsin et al. with cancerous and normal breast tissue indicate that cells enriched for CD44 expression represent a more basal ER-negative phenotype, with mammary epithelial progenitor-like properties . Additionally, in the normal human breast, the luminal progenitor is reported to be CD49f+CD24+ [9, 10, 41, 47] (also EpCAM+MUC1+CD133+Thy1−CD10−) [46, 49, 54, 55].
In this work, cells were stained with a combination of target proteins, which include P-cadherin, CD44, CD24, and CD49f. Stringent cell sorting was applied to separate and analyze the cell subpopulations with stem-like properties described above within each of the breast cell lines, after gating for P-cadherin expression. Separation of the highest 20% P-cadherin (P-cadhigh)-expressing cell subpopulation from the lowest 20% P-cadherin (P-cadlow)-expressing cell subpopulation revealed that, in all the cell lines studied, the P-cadhigh cell subpopulation was enriched for the stem cell-associated markers CD44, CD49f, and CD24, in comparison with the P-cadlow subpopulation, as shown in Figure 2A, 2B. Conversely, when parental cells were separated into the four possible subpopulations by CD44/CD24 and CD49f/CD24 expression, the CD44+CD24+ and CD49f+CD24+ cell subpopulations were the ones enriched for P-cadherin expression (Fig. 2C and Supporting Information Fig. S1).
The population described by Al Hajj et al. as the CSC phenotype, CD44+CD24−/low, is decreased in the P-cadhigh fraction (0.20% vs. 24.2% in the P-cadlow fraction, in the example shown in Fig. 2A). However, the enrichment of the P-cadhigh fraction for the phenotype CD44+/CD24+ (from 50.9% in the P-cadlow to 99.6% in the P-cadhigh) indicates that this adhesion molecule is associated with stem-like properties [50, 51]. Furthermore, the P-cadhigh cells are also enriched in CD49f+/CD24+ cells (from 57.4% in the P-cadlow fraction to 99.9% in the P-cadhigh), which are evocative of an association with the luminal progenitor profile of the breast [9, 10, 47].
In Human Breast Carcinomas, P-Cadherin Expression is Associated with Poor Patient Outcome as well as with the Expression of the Stem Cell Markers CD44, CD49f, and ALDH1
The expression of P-cadherin and the breast stem cell markers CD44, CD24, CD49f, and the isoform ALDH1 was analyzed by immunohistochemistry in a large series of 466 invasive human breast carcinomas. This analysis was performed in order to validate, in primary breast carcinomas, the association of P-cadherin expression with the stem cell markers previously observed in vitro. P-cadherin membrane expression was found in 24.5% of the cases (114/466), 63.2% of them were basal-like carcinomas (Supporting Information Table S2). CD44 membrane staining was present in 51.2% (237/463) of the cases. CD49f membrane staining was positive in only 11.5% (49/427) of the invasive breast carcinomas. Regarding the membranous stain of CD24, the majority of the cases (88.6%—410/463) were classified as negative/low, and only 11.4% (53/463) of the tumors had clear membrane staining. Concerning ALDH1 expression, a minority of cases (7.1%—33/463) were classified as positive, showing a clear cytoplasmic expression in tumor cells. Importantly, P-cadherin-positive cases were significantly enriched for the stem cell markers CD44 (p = .003), CD49f (p < .001), and ALDH1 (p < .0001), and no significant correlation was found between P-cadherin and CD24 expression (p = .747) (Table 1). Furthermore, the statistical analysis demonstrated that CD44, CD49f, and ALDH1 expressions were also significantly enriched in the basal-like carcinomas, in contrast to CD24 (Supporting Information Table S2).
Table 1. Association between P-cadherin expression and the expression of the stem cell markers CD44, CD24, CD49f, and ALDH1 analyzed by immunohistochemistry in a series of 466 primary invasive breast carcinomas
Kaplan-Meier survival curves showed that P-cadherin-positive tumors were significantly associated with poor overall survival (log rank, p = .023) (Fig. 3A). Furthermore, when taken into account the classic prognostic markers tumor size, histological grade, and node involvement, the expression of P-cadherin appeared as an independent factor, demonstrating the importance of this protein as a poor prognostic marker in breast cancer (HR = 1.486, p = .037, by multivariate Cox analysis) (Supporting Information Table S3). In contrast, the expression of the stem cell markers CD44, CD24, CD49f, or ALDH1 alone was not significantly associated with the clinical outcome (Supporting Information Fig. S2). By multivariate Cox analysis, the expression of CD24 was the only significant stem cell marker influencing survival, in this case related to a better prognosis (HR = 0.439, p = .014) (Supporting Information Table S3). Interestingly, however, when P-cadherin and CD24 were combined, the positive expression of both markers was highly associated with the worst patient overall survival (log rank, p = .014) (Fig. 3C), whereas tumors with a P-cadherin−CD24+ phenotype were associated with a good prognosis. The tumors with combined expression of P-cadherin and CD44 were also associated with a worst patient overall survival (Fig. 3B) (log rank, p = .047). Although a strong association between P-cadherin and CD49f expression was found, the combined expression of both these markers was not significantly associated with survival (log rank, p = .271) (Fig. 3D).
P-Cadherin Expression Confers Increased Self-Renewal Ability, Improved Cell Growth in 3D Cultures, and Radioresistance in Breast Cancer Cell Lines
The previous in vitro and in vivo indications, showing that P-cadherin is associated with the expression of stem cell makers in breast cancer, led us to study the cellular effects mediated by the expression of P-cadherin, namely in stem cell activity.
Stem cell activity can be measured by means of the mammosphere-forming ability of a specific cell line/subpopulation. In different cell lines, the heterogeneous parental cell population were separated by sorting according to P-cadherin levels of expression (high 20% vs. low 20%) and they were studied for stem cell activity. We found that P-cadhigh cell fractions from basal-like cell lines showed a significant increased MFE when compared with the P-cadlow cell fraction (Fig. 4A). The MFE from the basal P-cadhigh fractions reached up to two times the levels of the negative fractions. This result was seen both in the basal tumorigenic cells as well as in the normal MCF-10A breast cell line, suggesting that P-cadherin contributes to the stem cell activity in both normal and malignant contexts.
We also studied the clonogenic capacity of P-cadherin positivity in a 3D proliferation permissive environment containing matrigel, a matrix resembling the basal lamina of the normal breast. Using the same sorting procedure, in the luminal MCF-7/AZ and the basal BT-549 cells, we found that there was an increase in the number of 3D structures formed by the P-cadhigh cell subpopulation, compared with the P-cadlow subpopulation in the basal-like cell line (Fig. 4B). Furthermore, we found that the grape-like structures formed by P-cadhigh cells were bigger than the 3D structures formed by the P-cadlow subpopulation (95.8 μm vs. 50.1 μm, p = .003, data not shown). The same does not hold true for luminal cells for both parameters analyzed.
To verify the specific role of P-cadherin in stem cell activity, we used genetically manipulated cell lines, by in vitro transduction or silencing of P-cadherin. We showed that stable transduction of P-cadherin in MCF-7/AZ cells led to an increase of mammosphere formation (>25% of MFE) when compared with the mock cell line (Fig. 5A, 5B). This difference in mammosphere formation persists for at least two passages (data not shown), suggesting a role of P-cadherin in mediating self-renewal or survival. Likewise, transient inhibition of P-cadherin expression by siRNA assays showed that mammosphere-forming ability was negatively affected, although not completely abrogated in most of the cell lines studied (Fig. 5A, 5C). These results indicate that P-cadherin contributes to the stem cell activity of these cells.
One of the features attributed to breast CSCs is the increased resistance to irradiation, which allows them to survive and persist in tissues after treatment, contributing to disease relapse. We investigated the effect of x-ray irradiation alone and in combination with P-cadherin inhibition on the stem cell activity measured by MFE. We found that inhibition of P-cadherin in the tumorigenic basal-like cell lines potentiated the effect of irradiation-induced cell death, decreasing the number of mammospheres formed. This same effect was not observed in the normal breast cell line MCF-10A, where irradiation treatment or P-cadherin silencing alone had a negative effect in MFE; however, when the two treatments were combined, no additive or synergistic effect was observed concerning stem cell activity (Fig. 5D).
P-Cadherin Expression Confers Increased Tumorigenic Ability to Breast Cancer Cells
The indications given by in vitro functional assays, showing that P-cadherin expression has a role in the maintenance of stem cell properties and poor overall survival of breast cancer patients, led us to test whether cancer cell population enriched for P-cadherin have a higher capacity to promote tumor growth in nude mice. The tumorigenicity of the P-cadherin high and low subpopulations was addressed using the basal-like MDA-MB-468 breast cancer cells, which were inoculated in different dilutions, and the results obtained are shown in Table 2. Interestingly, we could observe that cancer cells enriched for P-cadherin expression have a higher capacity to promote tumor growth, since the number of tumors formed was higher in the P-cadhigh group, compared with the control group, in which the same number of parental cells was injected. Furthermore, tumors derived from the P-cadhigh cells were larger in size than the tumors derived from the P-cadlow fraction.
Table 2. P-cadherin capacity to promote tumors was evaluated by xenografting P-cadhigh and P-cadlow cells sorted from the basal-like MDA-MB-468 breast cancer cell line in the nude mouse model
Increasing evidence supports that cancers are propagated by a small cell population, the CSCs, which originate both tumorigenic and nontumorigenic cells and are responsible for tumor heterogeneity, therapy resistance, and disease recurrence. The identification and analysis of CSCs are mandatory in carcinomas with high patient mortality rate, early relapses, and lack of a targeted therapy. One of such poor-prognosis cancer is the basal-like subtype of breast cancer.
Basal-like breast carcinomas include tumors that are characterized by an expression signature similar to that of the basal cells of the breast . However, features of the luminal gene expression profile are often also attributed to these cancers ; therefore, basal-like breast carcinomas are often considered to have a mixed phenotype. More recently, the cell-of-origin for basal-like breast cancers was defined as the luminal progenitor cell from the normal breast hierarchy [9, 10], but the phenotype for the CSC has proven a tough task.
Thus, within this work, we established a positive association between P-cadherin, a basal marker, and the stem cell markers CD44, CD24, and CD49f in human breast cell lines. Indeed, we found that all the stem cell markers analyzed segregated with P-cadherin in the same subset of cells, suggesting that this adhesion molecule could be contributing to a stem-like aggressive phenotype in the breast. Accordingly, the expression of all these markers has been previously linked to the acquisition of a malignant phenotype or to a decreased breast cancer patient survival [44, 58, 59]. Significantly, our study shows that breast cancer cases that express P-cadherin and CD44, or P-cadherin and CD24, have the worst patient prognosis. Additionally, P-cadherin could be an important regulator of these stem cell markers in tumor cells, since when this adhesion molecule was knocked down, a reduction was found in the expression of CD49f and CD44v6.
Based on these results, we believe that we got experimental evidence that P-cadherin can be classified as a stem cell marker in breast cancer. Although the simultaneous expression of all the three stem markers (CD44, CD24, and CD49f) in the same cell was never clearly described, studies with CD49f and CD24 have shown that these markers combined in the same cell are able to identify cells with progenitor-like properties in the mouse [10, 41] and human [9, 47] mammary epithelial tissue. Specifically, Keller et al. showed that the normal human breast tissue contains an epithelial subtype with the phenotype EpCAMhighCD49f+CD24+, consistent with a definition of luminal progenitor cells, according to the characterization of lineage markers . Additionally, CD44 has also been used in combination with CD24 to show that the phenotype CD44+CD24+ identifies cells with stem-like properties, including tumorigenicity, in breast cell lines [50, 51] and in ER-negative human breast tumors . In this last study, Meyer et al. also showed that CD44posCD49fhigh cells are enriched for tumor-initiating ability in ER-negative breast cancers .
The association between the expression of P-cadherin and the luminal progenitor phenotype CD49f+CD24+ in several cancer and normal cell lines supports the notion that P-cadherin could potentially be used together with other markers in the identification of the luminal progenitor of the normal breast. In fact, luminal progenitor cells have been described as the cell-of-origin for basal-like cancers [9, 10], in which BRCA1 inactivation is a common feature and is considered to halt the maturation toward the luminal phenotype . This would explain the upregulation of P-cadherin found in basal-like cancers, since BRCA1 is a P-cadherin transcriptional repressor . Notably, Sarrio et al. recently reported that, in a normal nontumorigenic human breast cell line with a basal-like profile, the subpopulation defined as EpCAM+CD49f+ exhibits progenitor-like properties and, interestingly, this subpopulation of cells specifically expressed P-cadherin .
An enrichment of P-cadherin expression in cells that have the stem/progenitor phenotype CD44+CD24+ was also found in this study. Importantly, cells with the phenotype CD44+CD24+ have been described as having tumorigenic ability and represent a dynamic population that can originate the CSC phenotype CD44+CD24−/low [50, 51, 60]. In fact, Pece et al. showed that cells with CSC activity are CD24+  and that the phenotype CD44+CD24+, isolated from several human breast cancer cell lines, shows tumorigenic activity in murine xenograft models . This indicates that targeting cells with the phenotype CD44+CD24+, which have increased P-cadherin expression, could help to eradicate CSCs.
Our data still show that P-cadherin is not a simple bystander present in cells harboring a stem or progenitor phenotype, but it can also be a regulator of the stem cell function. First, we showed that high P-cadherin expression is associated with an increased capacity of cancer cells to promote tumor growth in nude mice. Second, we found that P-cadhigh cell lines have increased ALDH activity and that P-cadherin expression is increased in the ALDEFLUOR+ subpopulation. Previous studies have shown that ALDH activity and expression is linked to increased stem cell activity, tumorigenicity, and poor prognosis in breast carcinomas [31, 61]. Furthermore, P-cadherin gene expression manipulation (either by silencing with siRNA or by P-cadherin/CDH3 overexpression), as well as separation of cell population by sorting, showed that this adhesion molecule mediates stem cell activity and the self-renewal of mammospheres in basal breast cell lines. The growth in 3D matrigel matrix was also increased in the P-cadhigh subpopulation relative to the P-cadlow subpopulation of basal-like breast cancer cells. It is still not clear whether the increase in clonogenic capacity in 3D cultures mediated by P-cadherin, and the increase in stem cell activity measured by MFE, is the result of increased survival or alterations in the growth rate that affects proliferation/quiescence of the prospective P-cadhigh stem cells. Analysis of the cell cycle revealed that P-cadherin does not affect the subpopulations S and G2M (Supporting Information Fig. S4). Importantly, the fact that cells with lower expression of P-cadherin still show some tumorigenic ability, as well as a certain level of growth in anchorage-independent conditions and in matrigel 3D cultures, indicates that stem cell properties are still present in the P-cadherin-depleted fraction. Although P-cadherin does not seem to be an exclusive marker of CSCs, our results show that a further enrichment of the stem cell population could potentially be achieved by a combination of P-cadherin with other biomarkers. For basal-like carcinomas, several studies have used combinations of the following biomarkers to define the CSC subpopulation: ESA, CD44, CD24, CD49f, CD133-2, PROCR, and ALDEFLUOR [5–8, 61]. Not surprisingly, expression of P-cadherin alone in invasive breast cancer confers bad prognosis, and we observed that this marker is associated with the stem cell markers CD44, CD49f, and ALDH1, and all these markers with the basal-like molecular subtype. The isolation of P-cadherin-positive subpopulations within the CD44+ or within CD24+ fractions is of potential interest, since in our series of human breast carcinomas, the tumors that exhibit P-cadherin+CD44+ or P-cadherin+CD24+ phenotype seem to be particularly aggressive. Interestingly, the phenotype P-cadherin+CD44+CD49f+CD24+ could also support the isolation of progenitor cells from the normal breast, as mentioned above.
Additionally, it is known that basal-like breast cancers are particularly aggressive because they resist to current therapeutic strategies, usually recurring in a short time frame. In particular, resistance to radiotherapy has been reported for breast cancer, and relapses often occur. Although poorly understood, this type of resistance could be mediated by several mechanisms controlling cell survival and DNA repair mechanisms, allowing CSCs to survive and to give rise to a new tumor . In this study, we show that, when given x-ray irradiation, the silencing of P-cadherin contributes to decreased survival of the stem cell population in the tumorigenic cells but not in the normal. Hence, inhibition of P-cadherin could be an approach to increase sensitization of tumorigenic cells to radiotherapy, potentially allowing the reduction of the current doses of radiation administered to the malignant tissue and causing less damage to the normal cells.
We found, for the first time, that P-cadherin confers stem cell features to breast tumorigenic cells that could be linked to the aggressive behavior of basal-like breast cancers. We show that this molecule is associated with increased stem cell activity (tumorigenicity in athymic nude mice, mammosphere formation, and growth in 3D matrix) in basal-like cell lines but not in luminal cells. P-cadherin is associated with already described stem cell markers that define the luminal progenitor phenotype and which, in our series of primary invasive breast cancers, is associated with the shortest overall patient survival. We also found that inhibition of P-cadherin sensitizes cancer cells to x-ray-induced cell death.
In the future, CSC therapies for the aggressive basal-like breast carcinomas can eventually involve the targeting of P-cadherin cell surface protein. In fact, phase-I clinical trials are underway, which will help to define whether P-cadherin constitutes a good therapeutic target in breast cancer . Importantly, our results show that anti-P-cadherin treatment could improve radiation therapy in patients.
This work was supported by a grant (Project 96633-2010) from Fundação Gulbenkian, Portugal. We acknowledge Fundação Luso Britânica and Fundação para a Ciência e Tecnologia (FCT), Portugal, for fellowship support (A.V. was supported by SFRH/BD/41363/2007) and for the Programa Ciência 2007 (J.P.) (POPH—QREN—Tipology 4.2). We thank Dina Leitão and Regina Pinto for preparing the TMAs. The Programme for Advanced Medical Education is sponsored by Fundação Gulbenkian, Fundação Champalimaud, Portuguese Ministry of Science, Technology and Higher Education and FCT, Portugal. IPATIMUP is an Associate Laboratory of the Portuguese Ministry of Science, Technology and Higher Education and is partially supported by FCT.
DISCLOSURE OF POTENTIAL CONFLICTS OF INTEREST
The authors indicate no potential conflicts of interest.