Signaling by urokinase-type plasminogen activator receptor (uPAR) can cause epithelial-mesenchymal transition (EMT) in cultured breast cancer cells. In this report, we show that uPAR signaling can also induce cancer stem cell (CSC)–like properties. Ectopic overexpression of uPAR in human MDA-MB-468 breast cancer cells promoted the emergence of a CD24−/CD44+ phenotype, characteristic of CSCs, while increasing the cell surface abundance of integrin subunits β1/CD29 and α6/CD49f that represent putative mammary gland stem cell biomarkers. uPAR overexpression increased mammosphere formation in vitro and tumor formation in an immunocompromized severe combined immunodeficient (SCID) mouse model of orthotopic breast cancer. Hypoxic conditions that are known to induce EMT in MDA-MB-468 cells also increased cell surface β1/CD29, mimicking the effects of uPAR overexpression. Antagonizing uPAR effector signaling pathways reversed the increase in cell surface integrin expression. Whereas uPAR overexpression did not induce EMT in MCF-7 breast cancer cells, CSC-like properties were nevertheless still induced along with an increase in tumor initiation and growth in the orthotopic setting in SCID mice. Notably, in MCF-7 cell mammospheres, which display a well-defined acinus-like structure with polarized expression of E-cadherin and β1-integrin, cell collapse into the central cavity was decreased by uPAR overexpression, suggesting that uPAR signaling may stabilize epithelial morphology. In summary, our findings show that uPAR signaling can induce CSC-like properties in breast cancer cells, either concomitantly with or separately from EMT. Cancer Res; 70(21); 8948–58. ©2010 AACR.
The epithelium of the normal mammary gland includes basal and luminal cells. Stem cells are present principally within the basal layer and responsible for normal mammary gland development and regeneration (1). Normal mammary gland stem cells (MGSC) express high levels of the integrins β1/CD29 and α6/CD49f (1–3), which serve not only as MGSC markers but may also play a functional role in stem cell behavior, reflecting their activity in cell adhesion and cell signaling (4, 5).
According to the cancer stem cell (CSC) hypothesis, a subset of tumor cells may be responsible for development and progression of leukemia, lymphoma, and solid malignancies (6–8). CSCs show the capacity for self-renewal and the ability to generate daughter cells, which differentiate into the various morphologies and phenotypes observed in the mature cancer. How to best recognize a CSC within a complex tumor cell population remains a topic of intense investigation; however, in breast cancer, CSCs are frequently identified by a CD44high/CD24low phenotype using flow cytometry (9). Integrin subunits, which are expressed in normal MGSCs, may also be important in breast CSCs (10, 11).
Considerable evidence has emerged, suggesting that the cell signaling and transcription regulatory pathways that induce epithelial-mesenchymal transition (EMT) in carcinoma cells overlap with those activated in CSCs (12, 13). The urokinase-type plasminogen activator (uPA) receptor (uPAR) induces EMT in cancer cells by activating diverse cell signaling pathways, including the Ras–extracellular signal-regulated kinase (ERK) pathway, the phosphoinositide 3-kinase (PI3K)–Akt pathway, and Rac1 (14, 15). Because uPAR expression is increased in hypoxia, uPAR-induced EMT may be important in large, poorly vascularized tumors. uPAR-induced EMT is reversible. Strategies that have been successfully used to reverse uPAR-induced cancer cell EMT include reoxygenation, blocking uPA binding to uPAR, and targeting cell signaling pathways downstream of uPAR (14).
uPAR is a glycosyl phosphatidylinositiol-anchored receptor (16). Its activity in cell signaling requires integrin coreceptors, such as α4β1 and α5β1 (17–19). Receptor tyrosine kinases, such as the epidermal growth factor receptor, may also be engaged in uPAR-dependent cell signaling (20, 21). It is well known that many cancers show an increased propensity for metastasis in humans when uPAR expression is high (22). This may be explained by the effects of uPAR on cell signaling, its ability to promote activation of proteases near the leading edge of cell migration, the function of uPAR as a vitronectin adhesion receptor, or its more recently described role in EMT (16, 23–25). Although the ability of uPAR to promote protease activation near the cell surface has been studied most exhaustively, at least in mice, uPAR promotes cancer progression by protease-independent mechanisms (26).
In this study, we show that breast cancer cells acquire robust CSC-like properties when uPAR is overexpressed and uPAR-dependent cell signaling is activated. In MDA-MB-468 cells, acquisition of CSC-like properties is associated with EMT; however, in MCF-7 cells, CSC-like properties develop independently of EMT. The difference may reflect availability of uPA, which is important to activate some cell signaling pathways downstream of uPAR, and other receptor systems, which control the physiology of these cells.
Materials and Methods
Anti-CD24-FITC (ML5), anti-CD44-phycoerythrin (PE; G44-26), and isotype-matched IgGs were from BD Biosciences. Anti-CD29-FITC (HMβ1-1) and anti-CD49f-PE (GoH3) were from Biolegend. Rat monoclonal antibody that recognizes β1 integrin (AIIB2) was from the Developmental Studies Hybridoma Bank (University of Iowa). Monoclonal antibody that detects activated β1 integrin (HUTS4) was from Millipore. Monoclonal human uPAR-specific antibody (ATN658) and polyclonal human uPAR-specific antisera were provided by Dr. Andrew Mazar. E-cadherin–specific monoclonal antibody HECD-1 and polyclonal α6 integrin-specific antibody were from Abcam. Monoclonal antibody that recognizes vimentin was from Sigma-Aldrich. Secondary antibodies conjugated with Alexa Fluor 488, Alexa Fluor 569, and Alexa Fluor 647 were from Invitrogen. PI3K inhibitor (LY294002) and mitogen-activated protein kinase (MAPK)/ERK kinase 1 (MEK1) inhibitor (PD098059) were from EMD Biosciences. B27 serum-free supplement was from Invitrogen. Basic fibroblast growth factor (bFGF) and epidermal growth factor (EGF) were from R&D Systems. Quantitative PCR (qPCR) reagents, including primers and probes for human uPAR and hypoxanthine phosphoribosyltransferase-1 (HPRT-1), were from Applied Biosystems. Primer sets for β1 integrin were synthesized by Integrated DNA Technologies.
Human uPAR overexpressing MDA-MB-468 cells (468/uPAR) and empty vector–transfected MDA-MB-468 cells (468/EV) are previously described (15). MCF-7 cells that overexpress human uPAR (MCF-7/uPAR) were prepared by transfecting cells with pcDNA-uPAR using Lipofectamine 2000. After selection for 14 days with 500 μg/mL hygromycin, single-cell clones were established and screened for uPAR expression by immunoblot analysis. The cells were maintained in DMEM (HyClone) supplemented with 10% fetal bovine serum, 100 units/mL penicillin, and 100 μg/mL streptomycin.
Cells (3 × 106) were plated in 10-cm dishes and, in some studies, treated with LY294002 (10 μmol/L) or PD098059 (50 μmol/L) in serum-free medium for 18 hours. Cells were resuspended in fluorescence-activated cell sorting buffer [2% bovine serum albumin (BSA) in PBS] and incubated with anti-CD29-FITC (0.5 μg/106 cells), anti-CD49f-PE (10 μL/test), anti-CD24-FITC (10 μL/test), anti-CD44-PE (10 μL/test), or isotype-matched IgG for 30 minutes on ice. To detect uPAR or activated β1 integrin, cells were incubated with primary antibodies (1 μg/106 cells) for 1 hour. Primary antibodies were detected using secondary antibodies conjugated with Alexa Fluor 488 or Alexa Fluor 647. Results were analyzed using a FACSCanto II (BD Biosciences) and FlowJo software.
Biotinylation of cell surface proteins
Cells in monolayer culture (1.5 × 106) were washed three times with ice-cold PBS and then treated with EZ-link sulfo-NHS-SS-biotin (1 mg/mL, Pierce) for 15 minutes on ice. Biotinylation reactions were terminated with 100 mmol/L glycine in PBS. After washing with PBS, cell extracts were prepared in radioimmunoprecipitation assay (RIPA) buffer [20 mmol/L sodium phosphate, 150 mmol/L NaCl (pH 7.4), 1% NP40, 0.1% SDS, and 0.5% deoxycholic acid] with protease inhibitor cocktail (Roche). Biotinylated membrane proteins were precipitated with streptavidin-sepharose (GE Healthcare). Proteins were eluted with SDS sample buffer, resolved by SDS-PAGE, electrotransferred to polyvinylidene difluoride (PVDF) membranes, and probed with primary antibodies.
Cell extracts were prepared in RIPA buffer containing complete protease inhibitor cocktail. Protein concentrations were determined by bicinchoninic acid assay (Sigma-Aldrich). Equal amounts of cell extracts were resolved by SDS-PAGE, electrotransferred to PVDF membranes, and probed with primary antibodies.
Cells were trypsinized and mechanically disrupted to obtain single-cell suspensions. The cells were then cultured in mammosphere medium (DMEM/F12, B27, 20 ng/mL bFGF, 20 ng/mL EGF, 100 units/mL penicillin, and 100 μg/mL streptomycin) in ultralow attachment 24-well plates (Corning) at a density of 2,000, 500, or 50 cells/500 μL for 7 to 10 days. Mammospheres were imaged and counted under phase-contrast microscopy. Only mammospheres exceeding 100 μm in diameter were counted.
As a second assay to assess mammosphere formation, cultures were established in ultralow attachment 96-well plates at a density of one cell per well. Mammosphere formation was assessed after culturing for 7 to 10 days. This second method assured that mammospheres were formed by a single cell.
Tumor formation assay
Animal experimentation was done in accordance with protocols approved by the University of California San Diego Animal Care Program. Anesthetized 8-week-old C.B-17/lcrCrl-scid-BR mice (Charles River Laboratories) were inoculated in the fourth and contralateral fourth mammary fat pads with 50, 100, or 1,000 468/EV or 468/uPAR cells suspended in 50 μL of Matrigel (Sigma). MCF-7/EV or MCF-7/uPAR cells were also injected into mammary fat pads (1 × 106 cells per injection). Primary tumor growth was monitored every 2 to 3 days. One month after injection of MDA-MB-468 cells or 10 weeks after injection of MCF-7 cells, mice were euthanized and the mammary fat pads were visually inspected for tumor. Tumor formation was confirmed by histologic analysis. Data processing and statistical analysis were performed using GraphPad Prism (GraphPad Software, Inc.) and Microsoft Excel (Microsoft Corporation).
Histology and immunohistochemistry
Formalin-fixed tissue was paraffin-embedded. Serial 4-μm sections were stained with H&E. Immunohistochemistry was performed using the Vantana Discovery XT System (Vantana). Sections were pretreated with citric acid buffer and then incubated with polyclonal antibody specific for human uPAR (1:200) followed by peroxidase-conjugated secondary antibody. Peroxidase activity was imaged using 3,3′-diaminobenzidine. Slides were examined using a Leica DM2500 light microscope. Images were acquired using a Leica DFC420 digital camera and Leica Application Suite software.
Total RNA was isolated using the RNeasy kit (Qiagen). cDNA was synthesized using the iScript kit (Bio-Rad). qPCR was performed using an Applied Biosystems instrument and a one-step program: 95°C, 10 minutes; 95°C, 30 seconds; and 60°C, 1 minutes for 40 cycles. HPRT-1 gene expression was measured as a normalizer. Results were analyzed by the relative quantity (ΔΔCt) method. All experiments were performed in triplicate with internal duplicate determinations.
Cell adhesion assays
Medium-binding 96-well plates (Corning) were coated with fibronectin (5 μg/mL) or type I collagen (5 μg/mL) for 18 hours at 4°C and then blocked with 1% (w/v) BSA for 2 hours. Cells were washed and resuspended in serum-free DMEM at a concentration of 0.5 × 106 cells/mL and allowed to adhere for 15 minutes at 37°C. Adherent cells were fixed with 4% formaldehyde and stained with crystal violet. The dye was eluted with 10% acidic acid, and absorbance at 570 nm was determined. Each value represents the mean of 18 replicates in three separate experiments.
Cells were plated on glass coverslips, fixed in 4% formaldehyde, permeabilized in 0.2% Triton X-100, and incubated with antibodies specific for E-cadherin (1:100) or vimentin (1:400) followed by secondary antibodies conjugated with Alexa Fluor 488 or Alexa Fluor 594. Preparations were mounted on slides using ProLong Gold with 4′,6-diamidino-2-phenylindole (DAPI; Invitrogen) and examined using a Leica DMIRE2 fluorescence microscope. Images were obtained using a 63× oil immersion objective and a Hamamatsu digital camera with SimplePCI software.
Mammospheres were allowed to form in ultralow attachment T25 flasks. The mammospheres were collected by centrifugation at 200 × g, suspended, fixed in 4% formaldehyde, permeabilized with 0.2% Triton X-100, and incubated with antibodies specific for β1 integrin (AIIB2, 1 μg/100 μL) and E-cadherin. Secondary antibodies were conjugated with Alexa Fluor 488 or Alexa Fluor 594. Control preparations were treated equivalently, except for the omission of primary antibody. Stained mammospheres were mounted on chamber slides using ProLong Gold (Invitrogen) and examined using a spectral confocal microscope (C1-si, Nikon).
uPAR overexpressing MDA-MB-468 cells acquire CSC-like properties
Hypoxia increases uPAR expression, activates uPAR-dependent cell signaling, and induces EMT in MDA-MB-468 breast cancer cells (15). Equivalent changes in cell phenotype are observed when uPAR is overexpressed. Because others have suggested that the cell signaling pathways responsible for EMT in cancer cells overlap with those activated in CSCs (13), we tested whether uPAR overexpression in MDA-MB-468 cells induces CSC-like properties. Figure 1A shows that the abundance of cell surface uPAR was substantially increased in uPAR-overexpressing (468/uPAR) cells. Although the level of CD44 was already high in control (468/EV) cells, a subpopulation of the cells shifted to a higher level of CD44 expression when uPAR was overexpressed. CD24 was significantly decreased in the majority of the 468/uPAR cells (Fig. 1B).
Next, we examined whether uPAR overexpression alters the abundance of cell surface integrins that have been described as MGSC biomarkers. Total β1/CD29 was substantially increased at the cell surface in 468/uPAR cells, as was activated β1, as determined with antibody HUTS4. α6/CD49f was also increased, although the increase observed by flow cytometry was less pronounced than that observed for β1/CD29 (Fig. 1C). To confirm that cell surface α6/CD49f is regulated in 468/uPAR cells, we biotinylated cell surface proteins for isolation by affinity precipitation. Figure 1D shows that cell surface α6/CD49f was increased in 468/uPAR cells, whereas the total level of α6/CD49f was unchanged.
468/uPAR and 468/EV cells formed mammospheres in suspension culture, providing evidence that cells with self-renewal capacity were present (Fig. 1E). The mammospheres did not show a well-organized structure compared with those formed by MCF-7 cells (see below). Nevertheless, the number of mammospheres formed by 468/uPAR cells was significantly increased (P < 0.01) compared with the number formed by 468/EV cells (Fig. 1F). The ability of uPAR overexpression to promote mammosphere formation was shown in different studies in which suspension cultures were established at varied density. To confirm that mammospheres formed from a single cell, single-cell cultures of 468/uPAR cells and 468/EV cells were established in ultralow attachment 96-well plates. Figure 1G shows that the percentage of cultures that generated a mammosphere was significantly increased when uPAR was overexpressed (n = 4, P < 0.01).
To test whether uPAR overexpression promotes tumor initiation by MDA-MB-468 cells in vivo, we performed limiting dilution assays, injecting 1,000, 100, or 50 cells into mammary fat pads in severe combined immunodeficient (SCID) mice. As shown in Table 1, when 1,000 or 100 cells were injected, 468/uPAR and 468/EV cells formed tumors equivalently. However, when 50 cells were injected, 468/uPAR cells formed tumors at a significantly increased rate (P < 0.05). The results of our flow cytometry experiments, mammosphere assays, and limiting dilution studies support a model in which uPAR overexpression engenders CSC-like properties in MDA-MB-468 cells.
uPAR-initiated cell signaling regulates the subcellular distribution of β1 integrin
Because MDA-MB-468 cells express uPA endogenously, overexpression of uPAR results in activation of PI3K and ERK/MAPK (15). uPAR also physically associates with integrins in the plasma membrane (17). Either of these nonmutually exclusive mechanisms could be responsible for the increase in cell surface β1 integrin observed in 468/uPAR cells. To test whether cell signaling is involved, 468/uPAR cells were treated with the MEK inhibitor, PD098059 (50 μmol/L), or with the PI3K inhibitor, LY294002 (10 μmol/L). In three separate experiments, both reagents significantly decreased the level of cell surface β1/CD29, α6/CD49f, and activated β1/CD29, as determined with antibody HUTS4 (Fig. 2).
As a second model system to study cells in which increased uPAR expression drives EMT, we exposed MDA-MB-468 cells to 1.0% O2 for 48 hours. Figure 3A shows that uPAR mRNA expression increased 3-fold, confirming our previous results (15). β1 integrin subunit mRNA was unchanged. The cell surface abundance of β1 integrin was examined by biotinylation and affinity precipitation. Figure 3B shows that cell surface β1 integrin was substantially increased in MDA-MB-468 cells that were exposed to hypoxia. β1 integrin in whole-cell extracts distributed into two bands: the mature form found at the cell surface, which is more slowly migrating, and the more rapidly migrating immature form. Although the total level of β1 integrin (sum of two bands) did not change significantly in hypoxia, the distribution of β1 integrin between the two bands shifted so that more mature β1 integrin was present. This result confirms our affinity precipitation data, showing that hypoxia increases the abundance of cell surface β1 integrin.
uPAR overexpression induces CSC-like properties independently of EMT in MCF-7 breast cancer cells
uPAR overexpression does not induce EMT uniformly in all cancer cells (15). The degree to which the cancer cells express uPA endogenously and the function of other receptors that regulate cancer cell physiology may be involved. MCF-7 cells are estrogen receptor–positive breast cancer cells that express low levels of uPAR and undetectable uPA (24). We overexpressed uPAR in MCF-7 cells and isolated two single-cell clones (C4 and C5). uPAR overexpression was confirmed by flow cytometry (Fig. 4A). Neither of the uPAR-overexpressing MCF-7 cell clones showed signs of EMT. Figure 4B shows the results of experiments performed with C5 cells, which expressed the highest level of cell surface uPAR. Compared with cells that were transfected with empty vector (MCF-7/EV), C5 MCF-7/uPAR cells showed an unchanged level of E-cadherin and no vimentin. By contrast, 468/uPAR cells showed a substantial decrease in E-cadherin and high levels of vimentin, as previously shown (15). These results were confirmed by immunoflorescence microscopy (Fig. 4C). Note the robust vimentin immunopositivity in uPAR-overexpressing MDA-MB-468 cells. uPAR-overexpressing MCF-7 cells retained E-cadherin at cell-cell junctions, whereas vimentin was not detected.
To determine whether uPAR overexpression induces CSC-like properties in MCF-7 cells, C4 and C5 MCF-7/uPAR cells were analyzed by flow cytometry (Fig. 5A). CD24 expression was decreased in the majority of the C4 and C5 MCF-7/uPAR cells compared with MCF-7/EV cells. Again, as is frequently the case in cancer cell lines, CD44 was already expressed at high levels in the control cells; uPAR overexpression slightly decreased CD44 in the C5 clone. However, both β1/CD29 and α6/CD49f were increased in the C4 and C5 MCF-7/uPAR cells.
Because the increase in cell surface integrin expression induced by uPAR overexpression in MCF-7 cells was less robust than that observed in 468/uPAR cells, we performed cell adhesion assays to a test whether integrin function was regulated. Adhesion of C5 MCF-7/uPAR cells and control MCF-7/EV cells to the β1 integrin ligands, type 1 collagen and fibronectin, was compared. Figure 5B shows that adhesion of C5 cells to both substrata was significantly increased (P < 0.01).
Mammosphere assays were performed to compare the self-renewal activity of C5 MCF-7/uPAR and MCF-7/EV cells. Mammospheres formed by MCF-7 cells showed substantial differentiation into spheroid-like structures (Fig. 5C). Regularly contoured external borders and partially hollow central cavities were observed. As shown in Fig. 5D, uPAR overexpression in MCF-7 cells increased the frequency of mammosphere formation 4-fold to 5-fold irrespective of the number of cells that were introduced into each well (P < 0.05 at each density). In experiments with single-cell cultures, uPAR overexpression increased mammosphere formation 3-fold (P < 0.05), confirming the results shown in Fig. 5D (results not shown).
Nonmalignant mammary epithelial cells, such as MCF10A cells, form polarized, acinus-like spheroids in three-dimensional culture, recapitulating the grandular structure of normal mammary glands (27). In the acinus-like structures, β1 integrin is localized basally and E-cadherin to cell junctions. Figure 6A shows that, in mammospheres formed by MCF-7 cells, β1 integrin polarized principally to the inner surface of the central cavity. Some β1 integrin was also detected at cell junctions, reminiscent of normal mammary gland structure in vivo (3, 28). E-cadherin localized to cell-cell junctions and also, to some extent, to the inner surface of the central cavities, colocalizing with β1 integrin. In the control MCF-7/EV cells, the inner cavities of most mammospheres were partially collapsed by cells growing inward. By contrast, in mammospheres formed by uPAR-overexpressing MCF-7 cells, collapse of cells into the central cavity was much less frequent. As a result, these mammospheres more rigorously approximated a normal mammary gland acinus-like structure. The distribution of β1 integrin and E-cadherin in mammospheres formed by C5 MCF-7/uPAR cells was similar to that observed in MCF-7/EV cells. These studies confirm that, in MCF-7 cells, CSC-like properties are induced by uPAR overexpression independently of signs of EMT.
uPAR overexpression in MCF-7 cells promotes tumor initiation in vivo
In the absence of estrogen supplementation, orthotopic tumor formation by MCF-7 cells in SCID mice is limited even when a large number of cells is injected (29). As shown in Table 2 (see Supplementary Data), uPAR overexpression significantly increased the frequency of tumor formation from equivalently sized inoculums (P < 0.05). Tumors formed by uPAR-overexpressing cells grew much larger in the time frame of the study (Fig. 6B); the mean volume of the tumor was increased >20-fold. The malignant cells within the tumors remained robustly uPAR-immunopositive in vivo, as determined by immunohistochemistry (Fig. 6C). Nonmalignant mouse cells that infiltrated the tumor were immunonegative for uPAR, as determined with our human uPAR-specific antibody.
In diverse human malignancies, high levels of uPAR expression are associated with an increased propensity for cancer progression and metastasis (22, 30). Studies analyzing uPAR at the cellular and molecular level have revealed multiple candidate mechanisms. The ability of uPAR to accelerate a number of extracellular reactions, including conversion of the zymogen form of uPA into the active two-chain form (tcuPA; ref. 31) and plasminogen activation by tcuPA (32), promotes proteolysis and remodeling of the extracellular matrix (ECM). Because uPAR polarizes to the leading edge of cellular migration (33), increased ECM remodeling may facilitate cancer cell migration and invasion through tissue boundaries.
uPAR may also promote cancer progression by its effects on cell adhesion and cell signaling. The importance of this mechanism is supported by recent mouse model experiments (26). In addition to uPA, uPAR functions as a receptor for the ECM protein vitronectin (25). Binding of uPA or vitronectin to uPAR activates distinct cell signaling pathways, which collectively support cell migration and survival (23, 24, 26, 34–36). Understanding how uPAR-dependent cell signaling regulates cancer cell physiology is an important goal. In some cancer cell lines, increased uPAR expression and activation of uPAR-dependent cell signaling in hypoxia induce EMT (15). Our evidence suggests that the full continuum of uPAR-initiated cell signaling pathways, including those activated downstream of uPA and vitronectin, may be necessary for EMT (14, 15). However, the factors that control whether uPAR induces EMT remain incompletely understood.
In this study, we showed that uPAR overexpression does not induce EMT in MCF-7 cells. In fact, the polarized epithelial morphology of these cells may be strengthened by uPAR, because collapse of cells into central cavities of mammospheres was decreased. Whether EMT occurs in human cancer in vivo and thus represents a relevant pathway affecting cancer progression remains unsettled (37). Hypoxia-induced EMT, under the control of uPAR, is reversible (14). Thus, the fact that metastases in organs such as the lungs frequently show well-defined epithelial morphology does not preclude that EMT occurred as a step in the metastasis cascade.
In a cell type that undergoes uPAR-induced EMT (MDA-MB-468) and in a cell type that does not (MCF-7), uPAR overexpression engendered cells with biomarkers and properties of CSCs. In MDA-MB-468 cells, uPAR overexpression significantly increased the likelihood of tumor initiation by a small number of cancer cells in vivo. uPAR overexpression was associated with a CD44high/CD24low phenotype, increased cell surface β1/CD29 and α6/CD49f, and a significant increase in mammosphere formation frequency. Thus, in this cell type, the correlation between EMT and CSC-like properties was upheld. In MCF-7 cells, CSC-like properties associated with uPAR overexpression included decreased CD24 expression, an increase in the cell surface abundance of β1/CD29 and α6/CD49f and an increased frequency of mammosphere formation. Although the requirement for estrogen supplementation in vivo precluded typical serial dilution studies with MCF-7 cells, we did show that uPAR overexpression significantly increases the frequency of tumor initiation and tumor growth in SCID mice. We previously showed that uPAR-induced EMT is reversible (14). Because the signaling pathways downstream of uPAR that are responsible for EMT and CSC-like properties may be at least partially overlapping, it is possible that uPAR-induced CSC-like properties may also be dynamic and reversible.
MDA-MB-468 cell mammospheres showed poorly defined structure consistent with the loss of epithelial morphology and EMT. By contrast, mammospheres formed by MCF-7 cells showed a highly ordered and polarized structure with β1 integrin localized principally to a single surface and E-cadherin at cell-cell junctions. Interestingly, the β1 integrin subunit localized mainly to the internal surface of the mammosphere, opposite the location in a normal mammary acinus or duct. MCF-7 cells are frequently thought to have arisen from luminal epithelium (38). The increase in expression of cell surface β1 integrin, which accompanies uPAR overexpression, may indicate a transformation to a more basal cell phenotype.
In response to uPAR overexpression or hypoxia, β1 integrin in MDA-MB-468 cells relocated to the cell surface from intracellular pools. It is well established that uPAR associates with integrins in the plasma membrane (17, 18, 39–41) and that integrins function as coreceptors in uPAR-initiated cell signaling (17, 23, 41, 42). Receptors other than uPAR, which activate similar signal transduction pathways, may also cause relocation of β1 integrin to the cell surface (43). This process may provide a positive feedback loop by which uPAR-initiated cell signaling is strengthened.
When the cell surface abundance of uPAR was increased, the level of activated β1 integrin was increased, as determined with antibody HUTS4. Integrins exist in conformations that favor ligand binding or not; fluctuation between these conformational states may be controlled by cell signaling pathways (44), many of which are controlled downstream of uPAR (23, 45). Thus, it is not surprising that HUTS4 reactivity in uPAR-overexpressing MDA-MB-468 cells was decreased by inhibitors of PI3K and ERK/MAPK activation. However, these results do not rule out the possibility that uPAR also regulates integrin conformation by cis-interactions within the plasma membrane. Regulation of integrin activation by uPAR represents another possible positive feedback pathway by which uPAR and integrins may cooperate to control cancer cell physiology.
For the most part, uPAR has been studied as a gene product that controls invasion and metastasis of existing cancers. The function of uPAR as a gene product involved in generating CSC-like properties refocuses attention on this receptor to earlier steps in cancer development. In normal human adults, uPAR is sparsely expressed in tissues and organs (22). Thus, uPAR could serve as a target for cancer diagnostics development or therapeutics aimed at early stages of cancer.
Disclosure of Potential Conflicts of Interest
No potential conflicts of interest were disclosed.
Grant Support: NIH grant R01 CA94900.
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
Note: Supplementary data for this article are available at Cancer Research Online (http://cancerres.aacrjournals.org/).
- Received May 28, 2010.
- Revision received August 25, 2010.
- Accepted September 1, 2010.
- ©2010 American Association for Cancer Research.