Alpha-6 integrin is necessary for the tumourigenicity of a stem cell-like subpopulation within the MCF7 breast cancer cell line
Article first published online: 12 OCT 2007
Copyright © 2007 Wiley-Liss, Inc.
International Journal of Cancer
Volume 122, Issue 2, pages 298–304, 15 January 2008
How to Cite
Cariati, M., Naderi, A., Brown, J. P., Smalley, M. J., Pinder, S. E., Caldas, C. and Purushotham, A. D. (2008), Alpha-6 integrin is necessary for the tumourigenicity of a stem cell-like subpopulation within the MCF7 breast cancer cell line. Int. J. Cancer, 122: 298–304. doi: 10.1002/ijc.23103
- Issue published online: 16 NOV 2007
- Article first published online: 12 OCT 2007
- Manuscript Accepted: 18 JUL 2007
- Manuscript Received: 14 MAY 2007
- Fund for Addenbrooke's
- Cancer Research UK
- mammary epithelial stem cells;
- breast cancer stem cells;
- integrin alpha-6
The identification of mammary epithelial stem cells raises the hypothesis that these cells may be crucial in the pathogenesis of breast cancer. To further support this, a highly tumourigenic sub-population of cancer cells has recently been identified in primary and metastatic breast cancer samples. In this study, a sub-population of cells displaying features normally attributed to stem cells was identified within the breast cancer cell line MCF-7. This sub-population is capable of growth in anchorage-independent conditions as spherical organoids, displays resistance to proapoptotic agents and significantly greater tumourigenicity than its parental line, with as few as 1,000 cells able to form tumours in immunodeficient mice. Cells within this sub-population can be enriched by serial passages in anchorage-independence, and are characterized by over-expression of the adhesion molecule α6-integrin. Alpha-6 integrin proves to be required for the growth and survival of these cells, as the knockdown of ITGA6 causes mammosphere-derived cells to lose their ability to grow as mammospheres and abrogates their tumourigenicity in mice. These findings support the existence of a highly tumourigenic sub-population in breast cancer cells. Furthermore, it shows α6-integrin as a potential therapeutic target aimed at tumour-generating subsets of breast cancer cells. © 2007 Wiley-Liss, Inc.
Over the last decade, several reports support the existence of a subset of cells bearing stem cell features within tumours.1, 2, 3, 4, 5 In solid malignancies, it has been shown that only a minority of cells have the ability to form colonies in vitro or give rise to neoplastic growth when injected in immuno-deficient mice.6, 7 This could be explained by the fact that all tumour cells are similar, and extensive proliferation is merely a stochastic phenomenon. On the other hand, it could be that only a rare subset of phenotypically distinct cells has the ability to form new tumours, and cells within this subset, termed cancer stem cells, do so with greater efficiency. In 1997, Bonnet and Dick showed that a subset of cells could be identified as the initiator of Acute Myeloid Leukaemia, on the basis of surface markers characteristic of normal haematopoietic stem cells.1 This provided strong support for the cancer stem cell hypothesis. Similar results have since been reported for neural malignancies3 and breast cancer,2, 4 using a variety of experimental approaches including analysis of surface markers, and ability to grow in anchorage-independence, a feature previously described to characterize mammary epithelial stem cells.8 These studies show that tumour-initiating cells (or cancer “stem” cells) are responsible for tumour formation and progression and that they bear some of the phenotypic features of normal stem cells, such as self-renewal and the ability to generate a heterogeneous progeny. These findings clearly have potential implications in the treatment of solid malignancies.
A capacity attributed to bone marrow stem cells is their ability to efflux lipophilic dyes through the activity of the ATP-binding cassette transporter protein ABCG2/Bcrp1. Because of its characteristic appearance, this subset of cells has been termed “side population” (SP).9 It has been observed that a number of primary tumours and established cancer cell lines contain a SP.10, 11 Human breast SP cells have been reported as containing the normal human breast stem cell compartment12 although the mouse mammary SP is not enriched for mammary epithelial stem cells.13, 14
We decided to explore whether the commonly used breast cancer cell lines T47D, MDA-MB-231, SK-BR-3 and MCF-7 displayed any features of “stemness.” The identification of a subset of cells within any of these established lines would provide a highly useful and reproducible model of breast cancer stem cells.
Material and methods
Mammosphere culture and dissociation
MCF-7, SK-BR-3, T-47-D, MDA-MB-231 and BT-474 cells grown as adherent cultures were trypsinized with 0.05% trypsin/0.53 mM EDTA-4Na (Invitrogen). Cell were then washed twice with PBS and subsequently counted and plated in 90-mm plates (Corning) at a density between 20,000 and 50,000 viable cells/ml in primary culture.
Cells were grown in a serum-free mammary epithelial basal medium (MEBM, Cambrex) supplemented with 1× B27 (Invitrogen), 20 ng/ml EGF (Sigma) and 20 ng/ml b-FGF (Invitrogen). Plates were then sealed with Parafilm® (American Can Company) and grown in a Sanyo CO2 incubator MCO-17AIC at 37°C with 5% CO2.
Mammospheres were collected by gravity or gentle centrifugation (800g, 10 sec) after 7–10 days, and dissociated enzymatically (10–15 min in 0.05% trypsin, 0.53 mM EDTA-4Na; Invitrogen) and mechanically by pipetting.
These cells were sieved through a 40-μm nylon mesh (Falcon), analysed microscopically for single cellularity and counted.
Successive generations were plated in the same media but at 1,000 cells/ml density.
Mammosphere-forming efficiency was assessed performing a limiting dilution assay. Cells were grown either in attachment or anchorage independence and trypsinized once they reached 60–70% confluence. Cells were strained through a 40-μm nylon mesh to ensure single cellularity and 1 cell per well was seeded into five 96-well plates. Cells were then allowed to grow in anchorage independence for 10–14 days, at which point the number of wells containing spherical organoids was counted.
Cells were trypsinized and filtered through a 40-μm nylon mesh to ensure single cellularity. Subsequently, they were resuspended at 106 cells/ml and preincubated in DMEM supplemented with 2% Foetal Bovine Serum, for 30 min at 37°C.
Cells were then incubated with 2.5μg/ml Hoechst33342 (Sigma) in DMEM + 2% FBS, at 37°C for 90 min in a shaking water bath. All SP studies included a control sample supplemented with 50 μM verapamil (Sigma). At the end of incubation cells were immediately transferred on ice and each subsequent step, inclusive of analysis and sorting, was performed at 4°C. After staining, cells were washed with ice-cold staining media and resuspended in a suitable volume.
Analysis and sorting were performed with Mo-Flo high-speed cytometer (Dako), using a 60 mW multiline UV from a coherent I-90 laser. Hoechst blue was detected at 424/24 nm, and Hoechst red at 590/30. The channels were separated by a 555LP dichromic beam-splitter. Sorting was carried out using a 100-μm nozzle and at 35-Ψ sheath pressure, using the purity-1 mode.
All procedures carried out on animals were subject to ethical review, approved and licensed under the Animals (Scientific Procedures) Act 1984 by the Home Office.
Principles of reduction, refinement and replacement were applied throughout the whole duration of the project, and where animals could be substituted by in-vitro alternatives, the latter were preferred.
All tumourigenicity assays were carried out according to the guidelines set by the United Kingdom Co-ordinating Committee on Cancer Research (UKCCCR).
Animals were monitored daily and if signs of pain, suffering or discomfort appeared, they were dealt with in consultation with the Named Veterinary Surgeon. A week was allowed for the animals to recover, after which, were the signs still present, animals were humanely killed according to the methods indicated by the Home Office.
MCF-7 and MSS-derived cells were dissociated as previously described and counted. Cells were then resuspended in a mixture of RPMI and Matrigel™ (reconstituted basement membrane, BD Biosciences) 1:1, at variable cell densities, ranging from 107 to 104 cells/ml.
Four previously prepared female SCID (by intra-peritoneal injection of VP-16 at 30 mg/kg mouse 5 days before tumour injection; this is required to induce further immuno-suppression in these animals to achieve greater “take-rates” as reported by Al Hajj et al.2) mice were anaesthetized by inhalation and injected subcutaneously in the flank region with, respectively, 106, 105, 104 or 103 MSS-derived cells, resuspended in 100 μl of the above mixture. Contra-laterally, animals were injected with equivalent numbers of MCF-7 cells. Mice were then allowed to recover for 24 hr, after which they were monitored daily for signs of tumour growth.
DNA content analysis
MCF-7 cells and MSS-derived cells were stained for DNA content according to the following protocol. They were trypsinized and then resuspended in 200 μl of ice-cold PBS to which 4 ml of cold 70% ethanol were added. Cells were then incubated at 4°C for a minimum of 45 min to a maximum of overnight incubation. Subsequently, they were centrifuged at 1,500g for 10 min at 4°C and resuspended in PI master mix (Propidium Iodide 40 μg/ml, RNAse 100 μg/ml; Sigma) at 37°C for 30 min. Cells were then analysed with LSR II cytometer, and PI staining intensity was determined by fluorescence at 670 nm from the 488 nm laser.
Adherent cells were trypsinized, washed twice in PBS and pelleted by centrifugation. Mammospheres where washed and allowed to settle by gravity, then pelleted. The resulting cell pellets were resuspended in 10% un-buffered formalin and allowed to fix overnight. Cells were then pelleted at high speed (10 min at 3,000g) and processed to a paraffin wax block using a Shandon tissue processor (Thermo Electron Co.). Four-micrometer sections were cut from each block and mounted on polyanionic slides. The slides were dried at 56°C overnight.
The following day, the sections were deparaffinised and rehydrated to distilled water. Sections were placed in 0.5% hydrogen peroxide/methanol for 10 min to block endogenous peroxidase. Following antigen-retrieval, sections were washed in Tris Buffered Saline (TBS) for 5 min and placed in serum for 10 min. Subsequently, sections were incubated with anti-Ki-67 monoclonal antibody (Mib-1 clone, Dako, 1:300), followed by 2 washing steps in TBS and then incubated with appropriate biotinylated secondary antibody. This step was again followed by 2 washes. The slides were then incubated with ABC reagent (Avidin-Biotin Complex) and incubated with suitable peroxidase substrate according to Dako Chemate™ ABC protocol (Dako).
The slides were then washed thoroughly under running tap water and counterstained with Mayer's hematoxylin solution; then finally dehydrated and mounted.
Cells were grown in 6-well plates in either standard or anchorage-independent conditions. At 50–60% confluence, culture media was replaced with serum-free D-MEM containing 20 μM C2 or 10 μM tamoxifen (Sigma), and cells were incubated at 37°C for 24 hr. Cells were then washed and trypsinized to achieve single cellularity. Cells were next washed and fixed in 4% paraformaldehyde (Sigma) at 4°C for 30 min. Cells were subsequently spun onto cytology slides and stained with 1 μg/ml Hoechst 33258 (Sigma) for 10 min in the dark at room temperature. Slides were then mounted with Glycergel (Dako) and apoptotic cells were counted using a fluorescence microscope. A minimum of 500 cells was examined for each condition and the results were expressed as a percentage of apoptotic cells observed of the total number of counted cells. Experiments were performed in triplicates for each treatment condition.
Flow cytometric analysis of α 6-integrin expression
Following trypsinization, MCF-7 and MSS-derived cells were strained through a 40-μm nylon mesh to ensure single cellularity. Cells were then resuspended in ice-cold PBS (containing 2% heat inactivated foetal bovine serum and 0.09% sodium azide, BD Biosciences) to a density of 106 cells/100μl. PE-Cy5-conjugated anti-CD49f (α6-integrin) and PE-conjugated anti-CD29 (β1-integrin) monoclonal antibodies (both from BD Biosciences) were added to the cell suspension at concentrations suggested by the manufacturer and cells were incubated at 4°C in the dark for 30–45 min. Samples were then washed twice and finally resuspended in PBS (volume adjusted to cell number). Cells were analyzed and sorted using a Mo-Flo high-speed sorter (Dako, Denmark). PE was detected using a 200 mW 488-nm laser line with channels defined as 580/30 nm (yellow, FL2). PE-Cy5 was detected using a 200mW 488-nm laser line with channels defined as 670/40 nm (orange, FL3).
Mammosphere inhibition assay
MCF-7 cells and mammospheres were trypsinized as described above and single cells were plated at a density of 105 cells/ml in 35-mm culture dishes.
Cells were grown in serum-free mammary epithelial basal medium (Cambrex) supplemented with 1× B27 (Invitrogen), 20 ng/ml EGF (Sigma) and 20 ng/ml b-FGF (Invitrogen). Anti-α6-integrin monoclonal (GoH3) antibody (BD Pharmingen) was added to the media mixture at the following concentrations: 0, 0.5 and 1.2 μg/ml.
Experiments were performed in triplicate. Plates were then sealed with Parafilm® (American Can Company) and grown in a Sanyo CO2 incubator MCO-17AIC at 37°C with 5% CO2. Sphere formation was assessed in each treatment group as the ability or lack of ability to form mammospheres.
Inhibition of α 6-integrin expression by SiRNA
Alpha-6 integrin activity was also inhibited by RNA interference (siRNA) of the ITGA6 gene. ITGA6-KD in MCF-7 and MSS-derived cells was carried out using SMARTpool® siRNA reagents (Dharmacon, IL) following the manufacturer's protocol. About 24 hr after transfection cells were harvested by trypsinization and resuspended in MCF-7 or MSS growth media to assess their ability to grow in anchorage-independent conditions. To confirm the silencing specificity of our SMARTpool® siRNA reagents, and to distinguish sequence specific silencing from nonspecific effects, negative controls were performed by using siCONTROL nontargeting siRNA Pools (Dharmacon, IL). Furthermore, the use of pooled siRNAs is an accepted strategy for reducing off-target effects.
About 48 hr after the transfection, cells were harvested for total RNA extraction and subsequent ITGA6 expression analysis by QT-PCR, or for injection into immuno-deficient mice. Cells were harvested in TriReagent® (Sigma) and total RNA was extracted following manufacturer's protocol. QT-PCR was performed using Gene specific Taqman® assays (Applied Biosystems, CA). Three housekeeping genes (GAPDH, HPRT1 and RPLP0) were used. Reactions were performed in triplicates with each of the housekeeping genes (9 replicates in total) and MDA-MB-231 total RNA was used as calibrator. Experimental procedures were performed following the manufacturer's instruction and RT-PCR was carried out with an ABI 7900HT system (Applied Biosystems). Relative gene expressions were calculated as follows. First,all CT values were adjusted bysubtracting the highest CT (40) among all samples, then relative expression = 1/2▵▵CT [(CT,Target − CT Avg.HK in Targets) − (CT,Cal. − CT Avg.HK in Cal.)] as described before,15, 16, 17 (HK, housekeeping genes; Avg, Average; Cal, Calibrator).
A stem cell-like subpopulation exists within the breast cancer line MCF-7
We first attempted to explore the existence of a side-population within these cancer cell lines. Following trypsinization and staining with Hoechst 33342, cells were analyzed by flow cytometry, separating the red and blue channels with a 555LP dichromic beam-splitter. MCF-7 cells showed a distinct SP, accounting for ∼2.4% (±0.4%, n = 3) of the whole population and confirming the hypothesis that this established cell line contains at least 2 phenotypically distinct sub-populations, one of which bears stem cell features (Fig. 1a). On the other hand, T47D, MDA-MB-231 and SK-BR-3 did not show any SP fraction. We then decided to grow MCF-7 cells in anchorage independence. Whereas the majority of cells underwent anoikisis as expected, some cells displayed the ability to grow as spherical organoids (mammospheres, MSS), as shown in Figure 1b. To confirm that these organoids represented the progeny of an individual cell, rather than the aggregation of quiescent cells, we performed a limiting dilution assay with individual cells plated in each well of 96-well plates. Approximately 2.2% (±0.4%, n = 5) of observed wells contained MSS identical to the ones we had previously observed. We explored whether these cells had the ability to self-renew, a key feature typical of stem cells. To show this, we trypsinized primary MSS and replated single cells in 96-well plates. Primary MSS-derived cells were not only able to give rise to new MSS, but they were able to do so more efficiently than the parental MCF-7 cell line (Table I), with spheres observed in 8% (±1.22%, n = 5) of wells. Significant progressive enrichment in MSS-generating cells was observed up to fifth-generation spheres (p values = 0.007, 0.009, 0.009, 0.008 and 0.027, respectively), when the ability of single cells to form new organoids plateaued at 31.6% (±3%, n = 5) of cells (Table I).
|Cell type||MSS forming efficiency|
|MCF-7||0.022 ± 0.004|
|MSS I||0.08 ± 0.012|
|MSS II||0.18 ± 0.02|
|MSS III||0.23 ± 0.01|
|MSS IV||0.27 ± 0.02|
|MSS V||0.316 ± 0.03|
Furthermore, a significant (p value <0.05) enrichment in SP content was observed when comparing MCF-7 cells with first-generation MSS-derived cells as the latter contained ∼6.1% SP (±0.05%, n = 3) (Fig. 1c).
When MSS-derived cells and MCF-7 cells were compared immunohistochemically using antibodies against the oestrogen and progesterone receptors, no significant difference was identified, therefore suggesting that MSS-forming cells might be oestrogen receptor positive or anyway capable of generating a population vastly positive for the oestrogen receptor (data not shown).
Mammosphere-derived cells are more tumourigenic than MCF-7 parental line
We then tested whether in addition to these phenotypic differences MSS-derived cells also differed functionally from the parental MCF-7 line. To do so we investigated the ability of either cell type to generate tumours after injection in the subcutaneous layer of the flank region of adult female SCID mice. About 4–6 weeks after the injection, MCF-7 cells were able to produce tumours when at least 106 cells were injected, but failed to do so at lower cell doses (105, 104, 103) (Fig. 2a). Remarkably, primary MSS-derived cells were able to generate neoplasms in 4/4, 4/4, 4/4 and 3/4 animals when 106, 105, 104 and 103 cells were injected, respectively (Fig. 2a). The frequency of tumour initiating cells within the MCF-7 and primary MSS populations was calculated using the statistical software program for limiting dilution analysis l-calc (Stemsoft, Vancouver). The frequency of tumour-initiating cells in the MCF-7 population was calculated at 1/434,130 (95%CI 1/1,282,169 to 1/146,992). The frequency of primary MSS tumour-initiating cells was 1/721 (95%CI 1/2,454 to 1/212).
MSS-derived cells have proliferative and antiapoptotic features
In an attempt to explain the higher tumourigenicity observed with MSS-derived cells, we studied the cell cycle activity and response to proapoptotic agents in both MCF-7 and MSS-derived cells. When DNA content was analyzed by flow cytometry after incubation with a propidium iodide solution, G0/G1, S, and G2/M fractions in MSS-derived cells (n = 5) were, respectively, 65.4% (±0.97%), 14.2% (±0.52%) and 16.4% (±0.21%) of the whole population. On the other hand, MCF-7 cells (n = 5) displayed 78.5% (±0.51%), 5.74% (±0.12%) and 11.7% (±0.23%) of the whole population in G0/G1, S and G2/M phase, respectively. The differences between the percentages of cells in each cell cycle phase were statistically significant (p = 0.009) (Fig. 2b). The greater proliferative activity of MSS-derived cells was confirmed by immunohistochemistry using Mib-1 staining, with MSS-derived cells and MCF-7 cells showing intense staining in 95.4% (n = 3, ±0.35%) and 73.3% (n = 3, ±0.93%) of cells, respectively (p = 0.046) (Fig. 2c).
MSS-derived cells also show greater resistance to proapoptotic agents than their MCF-7 counterparts. When treated with 20 μM ceramide for 24 hr in serum-free media, 79.7% (n = 3, ±1.8%) of MCF-7 cells showed apoptotic changes in their nuclei following staining with Hoechst33258. Apoptotic changes were also observed in 75.2% (n = 3, ±1.8%) of MCF-7 cells following treatment with 10 μM tamoxifen. On the contrary, MSS-derived cells showed resistance to the above treatments, with only 17.8% (n = 3, ±1.1%) and 14.2% (n = 3, ±0.6%) of cells showing apoptotic changes in their nuclei following ceramide and tamoxifen treatment, respectively. The differences in apoptosis between these 2 cell types were statistically significant (p = 0.009) (Figs. 2d and 2e). The higher proliferative activity and resistance to apoptotic agents could be accounted for the greater tumourigenicity displayed by MSS-derived cells.
Alpha-6 integrin is highly expressed in MSS-derived cells
To investigate the formation of mammospheres, we decided to study the cell surface expression of α6-integrin, a molecule that has recently been used to identify mammary epithelial stem cells.14, 18 Using staining with a fluorochrome-conjugated monoclonal antibody against CD49f, MSS-derived cells were highly positive, with 72.65% (n = 6, ±6.4%) of cells CD49fhigh, while 23.28% (n = 4, ± 6.1%) of MCF-7 cells were CD49fhigh (Fig. 3a). The difference between the 2 cell types was significant (p = 0.01). Moreover, using quantitative real-time PCR (QT-PCR), MSS-derived cells, ITGA6 expression was 3.2-folds higher than in MCF-7 cells (Fig. 3b).
To explore the functional role of the α6-integrin we decided to use 2 approaches. First, we incubated cells with a blocking monoclonal antibody against CD49f (clone GoH3) previously described to inhibit α6-integrin function in HT1080 cells.19 Alternatively, we knocked down the ITGA6 gene with siRNA technology. The effectiveness of the knockdown strategy was confirmed by QT-PCR (Fig. 3b) (70% knockdown efficiency, ΔΔCT = 3.178). Both approaches induced the complete loss of MSS-forming ability by MCF-7 and MSS-derived cells (Figs. 3c and 3d).
Alpha-6 integrin is necessary for the tumourigenicity of MSS-derived cells
We next explored whether inhibiting α6-integrin affected the tumourigenicity of MCF-7 and MSS-derived cells. We observed that when 106 cells were injected, both MSS-derived cells and MCF-7 cells gave rise to new tumours as expected. However, ITGA6 knockdown cells of either type (MCF-7 and MSS-derived) failed to generate neoplasms (Fig. 3e). These findings led us to the conclusion that α6-integrin is necessary for both MCF-7 and MSS-derived cells to survive and proliferate in anchorage-independence and in vivo.
In summary, there is growing consensus that solid tumours are driven by cancer stem cells20, 21 and several reports have identified tumour-inducing cells or cancer stem cells in breast and neurological malignancies.2, 3, 4 In this study, we have identified a sub-population within the established breast cancer cell line MCF-7 bearing features conventionally attributed to stem cells. This sub-population has a higher proliferative index, shows a greater resistance to proapoptotic agents and has greater ability to produce tumours in immuno-deficient mice than its parental counterpart. These cells constitute ∼2% of the whole MCF-7 population and represent a useful, widely available model to study breast cancer stem cell biology. Moreover, we demonstrate that α6-integrin is over-expressed in MSS-derived cells. Alpha-6 integrin has been recently identified as a prospective marker of MESC and shown in its partnership with the β4-integrin to affect survival of SUM-149 cells in vivo.22 We show that its inhibition prevents both MCF-7 and MSS-derived cells from forming spherical organoids in vitro, and producing tumours in vivo. Taken together, these data identifythe α6-integrin as a molecule crucial for the survival of MCF-7 stem cells and therefore a potential therapeutic target aimed at tumour-generating subsets of breast cancer cells.
We thank Dr. John Stingl and Dr. Eleanor Bolton for helpful discussion and comments about the manuscript. We also thank Simon McCallum for technical assistance with flow cytometry analysis and sorting.