In vitro breast cancer cell lines mirror most of the features found in breast tumors, they cluster into basal-like and luminal expression subsets, display heterogeneity, and carry the recurrent genomic and transcriptional abnormalities associated with clinical outcome (1, 2). However, individually, they exhibit increased lineage-restricted profiles and are enriched for cellular phenotypes that are normally a minor component in vivo, and thus they fail to represent the intratumoral heterogeneity of breast tumors (3).
Cells purified from breast tissues are categorized according to the cell surface expression of CD44 and CD24, which distinguishes CD44−/CD24+ cells (luminal epithelial cells) from CD44+/CD24− cells (basal cells) (2), however a considerable heterogeneity in CD44 and CD24 expression was seen both between and within breast tumors (4). It has been shown that a small subpopulation of CD44+/CD24− stem/progenitor breast cancer cells can give rise to the heterogeneity of differentiated cells that comprise the bulk of the tumor (5, 6). Al-Hajj et al. (5) demonstrated that solid tumors contain a distinct population of so-called cancer-initiating cells that have the unique ability to form tumors in mice. They identified cell surface markers that define a subset of cells that drives tumorigenesis and generates tumor cell heterogeneity. The CD44+/CD24− lineage cells they isolated had extensive proliferative potential, demonstrated by the ability of a few cells to give rise to tumors that could be serially transplanted in immunocompromised mice. This extensive proliferative potential contrasted with the bulk of CD44− and/or CD24+ cancer cells that lacked the ability to form detectable tumors. Not only did the CD44+/CD24− cells give rise to additional tumorigenic CD44+/CD24− cells but also they generated phenotypically diverse nontumorigenic cells that constituted the bulk of tumors. Subsequent in vitro studies revealed enrichment of the CD44+/CD24− and CD44−/CD24+ subsets in basal-like and luminal breast cancer cell lines, respectively (7, 8). These data were confirmed in primary breast carcinomas (4) with CD44 being positively associated with stem cell-like characteristics and CD24 expression related to differentiated epithelial features, according to tumor subtype and histologic stage (9). Consequently, these subsets can be considered predictors of survival in breast cancer patients. Mammary gland stem/progenitor cells can be propagated through mammosphere culture (10). The ability to form mammospheres is a characteristic of cancer stem cells. Tumors are broadly classified histopathologically by expression of either luminal cytokeratins (CK18/19) or basal cytokeratin (CK5), alpha smooth muscle actin (SMA) and vimentin (11).
We evaluated whether cell cultures maintain the cellular heterogeneity found in primary tissues and whether they may be used for in vitro modeling of subtypes of breast cancer. We compared breast cancer cell cultures with standard cell lines previously classified on the basis of gene and protein expression profiling (2, 3). We first determined CD44 and CD24 expression and generated mammospheres from cell cultures of breast tumors. To quantify inter- and intra-tumor heterogeneity we studied the expression profiles of a panel of surface markers of cell cultures of well-characterized human breast cell lines. We grouped the samples according to their cell surface expression and found markers homogeneously and heterogeneously expressed in cell lines and breast cancer cell cultures. Our data demonstrate that breast cancer cell cultures preserve the heterogeneity of the individual tumor and flow cytometry can recognize and quantitate markers to make “signatures” of these cells. In this context, heterogeneous myoepithelial cell phenotypes appear to be a constant feature.
Ethics and Study Design
Residual breast cancer specimens were collected, after informed consent, from patients undergoing surgery for early breast cancer at the Azienda Ospedaliera Universitaria Federico II (Naples, Italy). Patients did not receive neoadjuvant or adjuvant endocrine therapy. Samples were collected following a biobanking standard operating procedure as previously described (12). The samples were anonymously encoded to protect patient confidentiality and used under protocols approved by the Azienda Ospedaliera Universitaria Federico II Ethics Committee. 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.
Nonadherent 24-well ultra-low binding plates were used (Corning, NY, USA). Fetal bovine serum (FBS) was purchased from Gibco (Invitrogen, Milan, Italy). Standard medium consisted of: minimal essential Dulbecco/Ham F12 (1:1) (DMEM/F12) (Sigma-Aldrich), supplemented with 2 mM glutamine (Sigma-Aldrich), P+S, 15 mM HEPES (Sigma-Aldrich) and 5% FBS. MCF-7, MDA-MB231 (MDA), Hs578T, and MCF10A were from ATCC (American Type Culture Collection). Multicolor flow cytometry was performed with anti-human monoclonal antibodies (MoAbs) that were conjugated with phycoerythrin (PE), fluorescein isothiocyanate (FITC), phycoerythin-Cy7 (PE-Cy7) or Alexa Fluor 647. PE-conjugated MoAbs against CD10, CD29, CD49f , CD54, CD55, CD59, CD61, CD66c, CD151, CD166, and CD200, and FITC-conjugated MoAbs against CD9, CD26, CD47,CD49b, CD66b, CD81, CD90, CD164, CD165, CD227, CD324, and CD326 were from BD Biosciences and BD Pharmingen (San Jose, CA, USA); PE-conjugated MoAb against CD133 from Miltenyi Biotech (Auburn, CA, USA); AlexaFluor647-conjugated MoAb against CD24, and PE-Cy7-conjugated MoAb against CD44 were from Biolegend (San Diego, CA, USA); PE-conjugated MoAb against CD184 was from Immunotech (Marseille, France); PE-conjugated MoAb anti-CD105 was from Serotec (Kidlington, Oxford, UK). For western blot analysis the following MoAb CK5 (AE14) code: sc-80606, CK19 (BA17) code: sc-53258 and SMA (alpha SMA) (CGA7) code: sc-53015 were diluted 1:200. The secondary antimouse antisera (sc-2005) were diluted 1:3000. The polyclonal antibodies used were as follows: pNeu (Tyr-1248)-R code: sc-12352, diluted 1:100, CK18 (H-80) code: sc-28264 diluted 1:200, ER-alpha (HC-20) code: sc-543 diluted 1:400 and EGFR (1005) code: sc-03 diluted 1:1000. The secondary antirabbit antisera (sc-2004) were diluted 1:3000. All the antibodies were purchased from Santa Cruz Biotechnology (Santa Cruz, CA, USA). The monoclonal anti-Tubulin (clone DM 1A) code: T-9026 diluted 1:400 and vimentin (clone V9) code: V-6630 diluted 1:1000 were from Sigma-Aldrich (Milan, Italy). All dilutions were made in TBS-Tween 20 (Thermo Fisher Scientific, Milan, Italy) with 5% non fat milk (Applichem, Darmstadt, Germany).
Breast Tumor Specimens and Cell Culture Conditions
Tumor histotype, size, grading and markers including ERα were determined with standard procedures, and HER2 was determined with HercepTest™ (Dako, Carpintera, CA, USA). The following data were retrieved from the pathology reports of each patient: patient #1, invasive lobular carcinoma (ILC), pT1 N0, ERα-positive (Estrogen Receptor alpha) (80%) and without c-erbB2 overexpression; patient #2, invasive ductal carcinoma (IDC) with a mucinous component, pT2 N0, ERα-positive (80%) and c-erbB2-negative; patient #3, IDC, pT1 N1, ERα-negative (<5%) and c-erbB2 overexpression (+++); patient #4, IDC, pT1 N1, ERα-positive (80%), and c-erbB2 overexpression (+++). Patient #5, IDC, pT2 N0, and ERα-negative, without c-erbB2 overexpression; patient #6, IDC arising from a contralateral recurrence, pT1c Nx, ERα-positive (>90%) and c-erbB2-negative; patient #7, IDC, pT1, N0, ERα-negative (<5%) and c-erbB2-negative. All patients had a high grade of malignancy (G3).
Breast cancer tissue specimens were collected, processed, cultured and frozen as cell suspension via standardized operative procedures for banking (13) and cell population expansion (14). Control MCF-7, MDA-MB231, Hs578T, and MCF10A were cultured in DMEM plus 10% FBS. Frozen cells were thawed, allowed to adhere and harvested 15–20 days in standard medium.
Mammosphere Formation and Diameter
Cells were dissociated and seeded, 1000 cells/well, in ultra low attachment surface 24-well plates (Corning) in DMEM/F12 plus 0.5% FCS medium, according to Dontu et al. (10). The medium was renewed twice weekly. Mammospheres were cultured for 15 days and measured under an inverted microscope Axiovert 40 C (Zeiss, Milan, Italy) equipped with a Canon powershot A640 camera (Zeiss). Digital images were analyzed with AxioVision software (Zeiss).
Flow Cytometry (FCM) Analysis
Cell suspensions were analyzed for the expression of 28 surface antigens, namely, CD9 (p24) (15), CD10 (CALLA) (16), CD24 (HSA) (5), CD26 (DPPIV) (17), CD29 (β1 integrin) (18), CD44 (H-CAM) (5), CD47 (Rh-associated protein) (19), CD49b (α2 integrin) (20), CD49f (α6 integrin) (21), CD54 (ICAM) (22), CD55 (DAF) (23), CD59 (MIRL) (24), CD61 (β3 integrin) (18), CD66b (CEACAM8) (25), CD66c (CEACAM6) (26), CD81 (TAPA1) (27), CD90 (Thy1) (28), CD105 (Endoglin) (29), CD133 (prominin1) (30), CD151 (PETA-3) (31), CD164 (MUC-24) (32), CD165 (AD2) (33), CD166 (ALCAM) (34), CD184 (CXCR4) (35), CD200 (OX2) (36), CD227 (MUC1) (37), CD324 (E-Cadherin) (38), CD326 (EpCAM) (37) (Table 1). We used a four color flow cytometric strategy to measure the expression of these surface markers based on an FCM panel in which cells were stained with MoAbs anti-CD24 and anti-CD44 and with antibodies against all the other analyzed surface antigens (Table 2). Briefly, each experiment consisted of 13 tubes, each containing MoAb anti-CD24-AlexaFluor647, MoAb anti-CD44-PE-Cy7, one PE-conjugated and one FITC-conjugated antibody against two of the other CD antigens to be analyzed. For all experiments, after enzymatic detachment from non saturated cultures, cells were counted, resuspended at 1–3 × 106 in PBS and stained by incubation at 4°C for 20 min with the appropriate amount of above the MoAbs in PBS. After staining, all samples were washed twice with PBS, pelletted and suspended in 0.5 ml of FACS buffer (FACS Flow Sheat Fluid, BD Biosciences) for FACS analysis. A few minutes before FACS acquisition, cells were incubated at room temperature in the dark with a vital dye (SytoxBlue, Invitrogen) to exclude dead cells from the analysis. In all experiments, for each cell type, a negative control was treated with the same procedure but without antibody staining. This negative control represents the background caused by the cellular autofluorescence. The samples were analyzed with a FACSAria flow cytometer and the FACSDiva software (Becton Dickinson, Franklin Lakes, NJ, USA). For each sample run, 10,000 to 20,000 events were recorded and analyzed. We used a three gating strategy: 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, we gated Sytox Blue-negative cells. The levels of expression of surface markers were reported as percentage of positive cells in Count versus FITC- or PE-CD histograms.
Table 1. Molecular identity and function of the cell surface markers analyzed by flow cytometry
Table 2. Four-color flow cytometry panel for expression analysis of surface markers
Each experiment consisted of 13 tubes. Each tube contained the MoAb anti-CD24-AlexaFluor647, the MoAb anti-CD44-PE-Cy7, one PE-conjugated and one FITC-conjugated antibody against two of all the other CD antigens to be analyzed.
Western Blot Analysis
Whole cell extracts were obtained by lysing samples in 50 mmol/L of TRIS (pH = 7.5), 100 mmol/L of NaCl, 1% NP40, 0.1% Triton, 2 mmol/L of EDTA, 10 μg/mL of aprotinin, and 100 μg/mL of phenylmethylsulfonyl-fluoride. Protein concentration was measured using the Bio-Rad protein assay (Bio-Rad Laboratories, Milan, Italy). Polyacrylamide gels (8–15%) were prepared as reported (13). Prestained molecular weight standards BenchMark were from Invitrogen (Milan). Proteins separated on the polyacrylamide gels were blotted on a nitrocellulose membrane (Protran, Whatman; Germany). The membrane was stained with Ponceau S (Sigma-Aldrich) to evaluate the success of transfer, and to locate the molecular weight markers. Free protein binding sites on the nitrocellulose were blocked with nonfat dry milk and a Tween 20/TBS solution. The membranes were washed and stained with specific primary antibodies and with secondary antisera conjugated with horseradish peroxidase diluted 1:3000. The luminescent signal was visualized with the ECL Western blotting detection reagent kit (Amersham) and quantified by scanning with a Discover Pharmacia scanner equipped with a Sun Spark Classic Workstation. Scion Image version beta 4.0.3 software (Scion, Frederick, Maryland) was used to quantify signal intensity.
Each experiment was carried out two to four times and found to be reproducible. Human tissue samples were not pooled, therefore each sample served as its own control. Error bars are presented as standard error of the mean (SEM). We used analysis of variance to identify statistically significant differences among means. The non parametric Spearman correlation test was used to compare breast cancer cells and control cell lines.
CD44/CD24 Expression Profile in Breast Cancer Cell Cultures
Cells purified from breast tissues are categorized according to the cell surface expression of CD44 and CD24, which distinguishes CD44−/CD24+ cells (luminal epithelial cells) from CD44+/24− cells (basal cells) (2); however, a considerable heterogeneity in CD44 and CD24 expression was reported both between and within breast tumors (4). To determine whether these markers could be used to classify luminal and basal breast cancer cells in cell cultures, we used a panel of breast cancer cell cultures (n = 7) and cell lines (n = 4), and compared the results with results obtained in standard cell lines previously classified on the basis of gene and protein expression profiling [1-3]. Although all cell cultures displayed a CD44+CD24-/low phenotype regardless of tumor subtype, biparametric dot plots (Fig. 1A) of CD44-PE-Cy7 versus CD24-AlexaFluor647 show that the prevalence of CD24−/low cells varied among samples. MCF7, MCF10A, cultures #4, and #7 show a low prevalence of this phenotype (Fig. 1B). Although it seems to emerge a clear-cut different CD44/CD24 expression pattern for the MCF7 cell line and all the other cell lines and samples, quantitative analysis of positive CD44+/CD24− versus positive CD44+/CD24+ of cultures #4 and #7 displayed a trend similar to that of MCF7 and MCF10A cells, with a high percentage of CD24+ cells (79.0% and 89.9 of CD24+ cells, respectively, as reported in Table 1) and a low percentage of CD44+/CD24− cells. Conversely, in culture #5, all the cells were CD24− (0% of CD24+ cells) (Fig. 1, panel B), which is in agreement with the signal intensity for CD44 (Fig. 1, panel C) and CD24 (Fig. 1, panel D).
Intensity of CD44 Expression
We used mean fluorescence intensity (MFI) analyses to estimate the number of CD44 molecules per cell (Fig. 1C). CD44 expression was very low in MCF7 cells (MFI = 15, linear scale range: 0–10,000), and high (>1000) in all the breast cancer cell cultures, as well as in MCF10A, Hs578T, and MDAMB231 cells.
Intensity of CD24 Expression
As shown in Figure 2D, CD24 expression was low in all cell lines and cultures (1 ≤ MFI ≤ 109; MFI linear scale range: 0–10,000). Within this low range of intensity, CD24 expression was low in all cultures, and very low in the MCF10A and MDAMB231 cell line (1 ≤ MFI ≤ 28), whereas it was higher in Hs578T and MCF7 cell lines and in culture #4 (MFI > 50).
Morphology and Mammosphere Formation
One of the associations in cell line collections is the relationship between the transcriptional phenotype and distinctive biological characteristics such as morphology. Adherent cultures of breast cancer cells had a mixed morphology with few polygonal cells and a prevalence of spindle-like cells (Fig. 2 left side of each pair of panels). Mammary gland progenitor cells can be propagated through mammosphere culture. We evaluated whether the cell cultures possessed the ability to form spheres. To this aim, we passed the cells (1000/well) to low attachment plates with low serum (0.5% FBS); within 2–3 days the cells formed aggregates that had the characteristic morphology of floating mammospheres (Fig. 2, right side of each pair of panels). If, with a fixed number of seeded cells, the number of mammospheres formed does not change 15 days after seeding in nonadherent conditions, we presume that the diameter of spheres is inversely related to the number of dead cells and thus represents a surrogate index of stem cell survival in nonadherent conditions. We measured the diameter and the number of spheres per well. Although all the cell lines (not shown) and breast cancer cell cultures formed spheres, the average diameter of aggregates (Fig. 2, histogram) is heterogeneous, ranging from 50 to 180 μm. Breast cancer cell cultures #4 (mean = 50 μm) and #1 (mean = 180 μm) contained the smallest and largest size mammospheres, respectively. The mean diameters of the remaining cultures were: culture #2 = 116 μm, #3 = 75 μm, #5 = 100 μm, #6 = 120 μm, and #7 = 94 μm.
Immunophenotypic Characterization of Breast Cancer Cell Cultures
The 28 surface markers were selected based on their potential to distinguish stem cell populations (CD24, CD44, CD49f, CD133, CD200, and CD326), to mediate cell adhesion and migration (CD9, CD29, CD49b, CD49f, CD61, and CD184), metastasis (CD44 and CD184) and signaling (CD66c) (Table 1). We used a four color flow cytometric strategy to measure the expression of these surface markers based on an FCM panel in which cells were stained with MoAbs anti-CD24 and anti-CD44 and with antibodies against all the other analyzed surface antigens (Table 2). We used a three gating strategy to define the target cell population which should be analyzed for the expression of surface markers (Fig. 3).
Table 3 shows the percentage of cells expressing surface markers in all cells and cultures. All markers showed a unimodal profile in all cell lines and cell cultures, which is indicative of homogeneity of a cellular population within the sample. An exception to the general unimodal pattern was the expression of CD105 (endoglin) in sample #1 (Fig. 3). Ten of 28 surface markers were homogeneously expressed, at high or low level, in all cell lines and cell cultures (Table 3). These were CD29 (β1 integrin), CD55 (DAF), CD59 (MIRL), CD66b (CEACAM8), CD66c (CEACAM6), CD81 (TAPA1), CD151 (PETA-3), CD165 (AD2), CD166 (ALCAM), and CD324 (E-cadherin). However, 18/28 surface markers were heterogeneously expressed (Table 4). CD90 (Thy-1), which is expressed on normal basal cells but not on luminal cells (39), was highly expressed in all cell cultures and in the Hs578T cell line (100% of positive cells), not expressed in the luminal MCF7 cell line, and intermediately expressed (60.4%) in the basal MCF10A cells. CD10 (CALLA), a marker of myoepithelial cells, was highly expressed in all cell cultures (100% of positive cells) and poorly expressed in the MCF7 and MDAMB231 cell line. CD326 (EpCAM) and CD227 (MUC-1), two key markers of luminal epithelial cells, were highly expressed in the luminal MCF7 cell line, 100% and 85.3% of positive cells, respectively, their expression was low or absent in the Hs578T and MDAMB231 cell lines and intermediate (77.8% and 33.85%, respectively) in the MCF10A cell line. In the cell cultures, whereas CD326 was always expressed at very low levels, the expression of CD227 was 42.4% and 40.0% for samples #2 and #7, respectively. CD54 (ICAM-1), a molecule whose function is required for invasion of metastatic breast cancer cells, is highly expressed either in the cell cultures (from 85.1 to 99.7 of positive cells) and in the cell lines (MCF7 = 94.7%, MDA = 83%, HS578T = 100%, MCF10A = 64.0%). However, because the differences exceed 5%, CD54 is considered to be heterogeneously expressed.
Molecules with expression differences ≤5% were considered homogeneously expressed, whereas those with differences >5% were considered heterogeneously expressed.
Negative or expressed at very low level: ≤7%; highly expressed: ≥99.1%.
Percentage of cells within each cell line or primary culture that were antibody-positive and therefore expressing the indicated antigens. The percentages of antibody-positive cells were calculated by comparison with the appropriate negative control.
Surface markers expression profiles of human breast cancer cell lines and cell cultures.
Molecules with expression differences ≤5% were considered homogeneously expressed, whereas those with differences >5% were considered heterogeneously expressed.
Percentage of cells within each cell line or cell culture that were antibody-positive and therefore expressing the indicated antigens. The percentages of antibody-positive cells were calculated by comparison with the appropriate negative control.
Cell Culture Phenotype Correlates with the Myoepithelial/Mesenchymal Phenotype
We used the non parametric Spearman correlation test to compare the phenotypic heterogeneity of breast cancer cell cultures with that of control cell lines (Table 4). Although displaying heterogeneous expression of surface markers (Table 5), cell cultures were strongly correlated with Spearman correlation coefficients (rs), which ranged from 0.7 to 0.9 (0.7 < rs > 0.9). All cell cultures were strongly correlated with the cell line Hs578T (0.6 < rs > 0.9) and, albeit to a lesser extent, to the MDAMB231 and to the MCF10A (0.1 < rs > 0.6) cell line. They were not correlated with the MCF7 cell line (−0.3 < rs > −0.04, p > 0.05).
Table 5. Correlations between human breast cancer cell lines and cell cultures based on heterogeneously expressed surface markers
To determine whether cultured cells expressed the epithelial or mesenchymal phenotype we immunoblotted for cytokeratins (CK), vimentin and SMA (Fig. 4). CK18 and CK19 are epithelial markers, whereas CK5, SMA and vimentin are myoepithelial/mesenchymal markers (40). Densitometric analysis, reported as percentage increase versus tubulin, showed that CK18 and CK19 were highly expressed (increased by more than 0.5%) in MCF7 cells (2.08% and 1.9%) and in culture #4 (1.05% and 1.3%) (Fig. 5). CK5, SMA and vimentin were not expressed in MCF7 epithelial cells, whereas they were expressed in the Hs578T and MDAMB231, MCF10A cells, and in all the breast cancer culture cells. Cultures #1, #4, #6 expressed high levels of SMA (an increase >0.5%).
We have immunophenotyped breast cancer cell cultures and cell lines for 28 surface markers to analyze the cellular phenotype, and show that a panel of CD antigens can be used to determine inter-tumor heterogeneity.
In vitro culture of human breast cancer cells usually leads to selection for basal/myoepithelial cells and most human breast cancer cell lines display a partial loss of the myoepithelial differentiation program and a partial gain of the luminal differentiation program (1). This culture profile is thought to reflect the existence of rare stem cells in the basal compartment (41). As yet, there is no in vivo equivalent of these basal/myoepithelial cells.
According to the cancer stem cell hypothesis, tumors are driven by cells that display the properties of self-renewal and differentiation. Self-renewal is demonstrated by the ability of the enriched stem cell populations to form mammospheres and grow in an anchorage-independent environment. Putative breast cancer stem cells can be identified by their surface expression of CD24 and CD44 (42). The CD24 and CD44 subsets of tumor cell populations are genetically highly heterogeneous and differ in their tumor-initiating potential and response to drugs (43). The heterogeneity of cells within individual tumors may be indicative of a cancer stem cell subpopulation that gives rise to more differentiated cell populations, and the features they display in vitro may represent the intratumoral heterogeneity of human breast tumors.
To determine the cellular heterogeneity of tumors, we used a four-color flow cytometric strategy in which CD24 and CD44 stained cells were also stained with MoAb against two other antigens of the 28 to be analyzed (Table 2). The 28 surface markers were selected based on the potential to distinguish between cell types of epithelial or mesenchymal lineage or to identify putative cancer stem/progenitor cells. Every marker showed a unimodal profile in all samples, which indicates the lack of cellular subpopulations. The only exception to this general pattern is the expression of CD105 (endoglin) in sample #1, the significance of which requires further investigation.
The analysis showed that 10 surface markers were homogeneously expressed in all cell lines and breast cancer cell cultures. A wide range of expression of the surface markers analyzed was found, with some markers, such as CD324 (E-cadherin) exhibiting very low values, and others, such as CD166 (ALCAM), displaying high values.
Four markers (CD66b, CD66c, CD165, and CD324) displayed negative/low expression in all the cell lines and breast cancer cells tested. Every marker is involved in cell adhesion. CD66b and CD66c mAbs react with CEACAM-8 and CEACAM-6, respectively. Members of the CEACAM family play a role in invasion and metastasis, are involved with cellular differentiation (44), and are expressed in breast cancer carcinomas (45). CD324 reacts with E-cadherin. E-cadherin functions as a tumor suppressor gene. In breast cancer, loss of E-cadherin expression is a necessary component for invasion and metastasis (46), and is a marker of epithelial-mesenchymal transition (40). Reduction of CD324 (E-cadherin) expression is associated with breast carcinomas of basal-like and triple negative phenotype (47) and with the acquisition of cancer stem cell associated characteristics like increased CD44+/CD24− ratio (48).
Six markers displayed homogenous high expression in all the cells, namely CD29 (beta1-integrin), CD55 (DAF), CD59 (MIRL), CD81 (TAPA1), CD151 (PETA-3), CD166 (ALCAM). CD55 (DAF), and CD59 (MIRL) are two complement regulatory proteins expressed on cells to avoid autologous complement attack. In breast cancer, there is a high variability in complement regulatory protein expression, with some tumors expressing only one inhibitor, and others expressing various combinations of two or three inhibitors (49). CD81 (TAPA1) and CD151 (PETA-3) belong to the four-transmembrane domain (tetraspanin) family of proteins involved in regulating tumor cell motility and invasiveness, mainly through their effect on the adhesive and signaling function of integrins (27). Integrin–tetraspanin complexes modulate tumor cell–cell and cell–substrate interactions (50, 51). CD166 (ALCAM) is overexpressed in breast cancer metastases (52). High values of CD166 are associated with high values of CD44 and either high or low values of CD24 (53).
The homogeneous low or high expression of surface markers suggests that cell lines and breast cancer cell cultures share many of the molecular features of their acquired ability to grow in vitro, and acquired adaptation to culture conditions. This is in agreement with previous reports (1, 3, 41).
Although the markers in the homogenously expressed group could be considered molecules ubiquitously expressed for the biological adaptation of cells to in vitro conditions, the heterogeneously expressed group is likely to consist of molecules that identify cells of different origin. In agreement with others (3), we found that heterogeneity within samples, cell lines and breast cancer cell cultures was remarkably restricted (unimodal profile), which indicates that in vitro culturing enriches for cellular phenotypes that may represent the progeny of different stem/progenitor cells. If the different distribution of marker expression is influenced by the culture conditions, then the expression profile patterns in the cell lines and breast cancer cell cultures would appear not to be related. We performed correlations using Spearman's rank to avoid assumptions regarding distribution. We found no correlation between breast cancer cells and MCF7, a weak to moderate correlation with MDA-MB231 and with MCF10A, statistically significant correlations (p< 0.001) with Hs578T and a strong statistically significant correlation within the breast cancer cell cultures (Table 5). Since heterogeneity within breast cancer cell cultures does not depend on culture conditions because all the primary cultures undergo the same protocol of isolation, the flow cytometry analysis shows the potential to identify individual features of a breast tumor.
Studies designed to define the cell of origin in experimental model systems report that mesenchymal stem cells, isolated from various normal and pathological tissues, show phenotypic heterogeneity (reviewed in (54)) and it has been suggested that they reside in virtually all postnatal tissues. The resulting cultures are morphologically heterogeneous; in fact, they contain cells ranging from spindle-shaped cells to polygonal cells and some cuboidal cells.
Mesenchymal stem cells express a number of markers, none of which are specific to mesenchymal stem cells. Although many of these markers display variable expression due to differences in tissue source, the method of isolation and culture, and species differences, it is generally agreed that adult human mesenchymal stem cells can express CD44 (H-CAM), CD54 (ICAM), CD90 (Thy-1), CD105 (Endoglin), CD184 (CXCR4), various integrin molecules, such as CD49b (α2integrin), CD49f (α6integrin), CD29 (β1integrin), CD61 (β3integrin), and other adhesion molecules including CD54 (ICAM) and CD166 (ALCAM) (54). Within our breast cancer cell cultures, which were all CD44+, concomitant high expression of surface markers associated with the mesenchymal stem cell phenotype included only three antigens (CD10-CALLA, CD54-ICAM, CD90-Thy-1) and was paralleled by low expression CD326-EpCAM.
Breast cancers are broadly classified histopathologically by the expression of either luminal cytokeratins (CK18/19) or basal cytokeratins (CK5), SMA and vimentin (11); the categorization in subtypes is critical for the prediction of clinical outcome and treatment response. To refine the portrait of breast cancer cell cultures, we examined whether the markers associated with myoepithelial/basal-like breast cancers (CK5, SMA, and vimentin) and those associated with luminal tumors (CK18 and CK19) were differentially expressed based on culture conditions. In the mouse mammary gland, studies based on CK expression show that CD24−, CD24low, and CD24hi populations correspond to nonepithelial, basal/myoepithelial, and luminal epithelial cells, respectively (55). Although not all basal-like tumors contain CD44+/CD24− cells, an association between basal-like phenotype and CD44+/CD24− cells has been recently reported (4).
Carcinomas belonging to the basal-like subtype of breast cancers are thought to be stem cell-derived or to have acquired stem cells features during transformation (11). Recent studies identified myoepithelial cells as early progenitors and as candidates for precursor cells of basal-like breast cancer (56–59). Myoepithelial cells are cells from the basal layer, defined as cells expressing both epithelial markers and contractile proteins (60). These cells are distinguished from basal cells in multilayered squamous epithelium because they express proteins characteristics of mesenchymal cells, such as vimentin (61), alpha SMA (62), and cytokeratins 5 (CK5, or basal cytokeratin) (63). However, studies aimed at defining the cell of origin in breast tumors show that the expression of CK5 is not restricted to myoepithelial cells in tissue cultures (64). We used western blot to evaluate the expression of these markers in all cell lines and breast cancer cell cultures. Although the overall levels of marker expression were different, all of the breast cancer cell cultures analyzed in this study expressed CK5, SMA, and vimentin, all markers of basal/myoepithelial cells. Although the majority of human breast cancers cell lines, including MCF7 and MDA-MB231, are cytokeratin 18 (CK18)- and cytokeratin 19 (CK19)-positive, all established breast epithelial cell lines from reduction mammoplasties are completely cytokeratin CK19-negative (65), and CK19 staining of luminal cells in the fetal breast are positive (60). Excluding culture #4, as well as MCF7 and MDA-MB231, all the cultures were weakly CK19-positive (<0.5), thus reinforcing the concept that these cells are likely to be derived from the basal layer.
In this study, we used a much larger vocabulary of surface marker present on mammary tumor-deriving cells than those used in previous studies to characterize clinical-pathological features of these cells. Our results, although confirming that in vitro culturing of breast cancer cells reduces luminal lineage-type of cells, it indicates the increased propensity for the selection of myoepithelial/basal breast cancer cells. Our data supports that spindle-cell morphology, ability to form mammospheres and their size, surface marker expression, and intracellular marker expression (CKs, SMA, vimentin) are associated with the extent of CD44/CD24 cells. Collectively, the heterogeneous expression of markers is representative of the diversity of breast tumors and the flow cytometry approach is a useful tool to determine the quantitative expression of markers in breast cancer cells. It remains to be established whether the targeting of surface markers in breast cancer cells will have an impact on the survival of tumor cells and also if surface markers could represent a tool to identify and test patients' customized therapy.
LL and AN equally contributed and assisted in provision of study material, conception and design, data analysis and interpretation. SC assisted in collection and assembly of data, and provision of study material. SDP and FS assisted in conception and design, collection and assembly of data, data analysis and interpretation, and financial support. LDV and BMV assisted in conception and design, collection and assembly of data, data analysis and interpretation, and manuscript writing. All authors read and approved the final manuscript.
Breast cancer tissue specimens were obtained from the Breast Cancer Tissue Bank developed under the auspices of the BIONCAM Project and maintained by the CRPO, University of Naples “Federico II”. The authors thank Jean Ann Gilder (Scientific Communication srl) for editing the text.