Epithelial–mesenchymal transition (EMT) generates tumor cells with stem cell properties. The aim of our study was to investigate the effects of adipose tissue-derived stem cells (ASCs) on EMT of cancer cells and to further investigate the mechanisms involved. We demonstrate that conditioned medium from ASCs induces breast cancer cells (4T1) to express mesenchymal markers such as fibronectin, alpha smooth muscle actin and vimentin. Flow cytometry analyses show that ASC-conditioned medium promotes the expansion of CD44high/CD24low cancer stem cells. Soft agar assays using T47D, BT474 and MCF-7 breast cancer cells reveals that ASC conditioned medium promotes the anchorage-independent growth of cancer cells. These effects were inhibited by a neutralizing antibody against platelet-derived growth factor-D (PDGF-D). Furthermore, PDGF-D treated breast cancer cells grow faster in a mouse model, and this effect could be neutralized by a PDGF antibody. In conclusion, our data show that tissue-resident stem cells interact with the cancer microenvironment via PDGF-D, induce EMT in the cancer cells in a paracrine fashion, thereby increasing the number of cancer stem cells and increase tumor growth in a PDGF dependent manner. Our findings shed new light on mechanisms where local tissue-resident stem cells are able to promote the growth of breast cancer cells. Possibly this could open up a novel selective therapeutic strategy targeting EMT pathways and the specific communication between tissue-resident normal stem cell and cancer stem cells, assuming that the blockage of PDGF-D pathways is critical for tumor growth but would not affect normal tissue homeostasis.
Epithelial–mesenchymal transition (EMT) is a process by which epithelial cells acquire molecular alterations that result in the loss of apical polarity and the acquisition of a more spindle-shaped morphology.1–3 EMT involves the loss of epithelial cell–cell junctions; reorganization of the actin cytoskeleton; upregulation of mesenchymal molecular markers such as vimentin, fibronectin and N-cadherin and downregulation of epithelial markers such as E-cadherin and epithelial-specific antigen.4 These processes are ultimately thought to promote tumor progression through the generation of cancer stem cells.
EMT is triggered by an interplay of extracellular signaling originating from matrix such as collagen or from growth factors, including fibroblast growth factor, epidermal growth factor and platelet-derived growth factors (PDGFs)-A and −B.5–7 PDGF-D has recently been characterized, and overexpression of PDGF-D in PC3 prostate cancer cells has been shown to lead to the induction of EMT8. It is not known whether PDGF-D is present within the tumor microenvironment naturally or how it might contribute to EMT. Bone marrow-derived mesenchymal stem cells (MSCs) contribute to tumorigenesis by producing cytokines and growth factors.9 We recently showed that adipose tissue-derived stem cells (ASCs) contribute to the growth and progression of breast cancer cells.10–14 A recent study showed that bone marrow-derived MSC conditioned medium (CM) contains factors that promote EMT in breast cancer cell lines.15 The aim of our study was to investigate the effects of ASCs on EMT and the possible mechanisms involved.
Material and Methods
Adipose tissue samples were weighed and extensively washed with phosphate-buffered saline (PBS). The minced specimens were incubated in PBS containing Liberase Blendzyme 3 (Roche Diagnostics, Indianapolis, IN) at a final activity of 2 U/mL for 45 min at 37°C on a shaker (20 rpm). After trituration, the tissue was passed through a 100-μm filter, and adipocytes were separated from stromal-vascular fraction by centrifugation at 450g for 10 min. The pelleted cells were resuspended in growth medium consisting of alpha modification of Eagle's minimum essential medium, 20% fetal bovine serum, 2 mM L-glutamine, 100 U/mL and 100 μg/mL streptomycin. The cells were cultured on tissue culture plates in a 5% CO2 incubator at 37°C. The cultures were washed daily for the removal of red blood cells and nonattached cells. After 80% confluence of Passage 0, cells were seeded at a density of 3,000 cells/cm2 (passage 1), hASCs were subcultured every 3–5 days, and cells between Passages 1 and 3 were used for all experiments.
MCF-7, BT-474, T-47D and 4T1 (American Type Culture Collection, Manassas, VA) cells were grown in Dulbecco's modified Eagle's medium F12 (Invitrogen, Carlsbad, CA) supplemented with 10% fetal bovine serum and penicillin–streptomycin at 37°C in 5% CO2.
Preparation of CM
For the collection of CM, hASCs were cultured in alpha modification of Eagle's minimum essential medium with 10% fetal bovine serum, 2 mM L-glutamine, 100 U/mL and 100 μg/mL streptomycin for 48-hr intervals. The CM was collected and centrifuged at 1,000g for 5 min. The CM was filtered using a 0.45-μm membrane and used for culturing cancer cells. The supernatant CM was transferred to breast cancer cell lines to study the paracrine effect. For all functional assays using hASCs, CM was produced using a similar method.
Total cellular protein extract was isolated from harvested cells using lysis buffer [50 mM Tris (pH 7.5), 150 mM NaCl and 0.5% NP-40] containing Protease Inhibitor Cocktail Tablets, from Roche (Cat#: 11836153001), and protein concentration was determined. From each sample, 50 μg of total protein was resolved on a 4–15% Bis-Tris Gel with running buffer and transferred to PVDF membranes. The membranes were then probed with various antibodies such as anti-β-actin (Abcam, Cambridge, MA), anti-E-cadherin (BD Biosciences, San Diego, CA), anti-fibronectin (BD Biosciences), anti-vimentin V9 (Neomarkers) and α-smooth muscle actin (Abcam).
About 2 × 104 cells were seeded on a 4-well Lab-Tek II chamber slide. After 24 hr, the cells were incubated with hASC-CM and cultured for 72 hr. The cells were washed with PBS twice and permeabilized and fixed in 2% paraformaldehyde and 0.1% Triton-X 100 in PBS buffer at 40°C for 30 min. The cells were then washed three times with PBS, incubated with anti-E-cadherin antibodies for 2 hr, washed three times with PBS for 15 min and incubated with anti-mouse Alexa Fluor secondary antibodies for 1 hr. Cells were then washed for 15 min, incubated with 4DAPI for 10 min, washed with PBS and mounted using the SlowFade Light Antifade Kit (Invitrogen). All samples were subjected to immunofluorescence microscopy and photographed at identical exposure times.
Soft agar assay
MCF-7, BT474 and T474D cells were cultured in hASC-CM for 3 days. Cells were harvested, and soft agar analysis was performed with 0.3% agar on top of 0.6% agar with 1 × 104 cells in each group. The cells were plated in six-well plates in triplicate and incubated for 2 weeks. Every 2 days, 10% alpha DMEM or hASC-CM was replenished on the top layer in the presence or absence of neutralizing antibodies against human PDGF-A, C and D. Colonies were counted in 10 random fields per well and photographed.
Total RNA was isolated using Trizol (Invitrogen). hASCs from two different donors (Passage 2) and breast cancer cells (MCF-7, BT474 and T47D) were grown with hASC-CM and 10% fetal bovine serum for 72 hr. A reverse transcription system (Invitrogen) was used to reverse transcribe 2 μg of RNA according to the manufacturer's instructions. PCR was performed using the specific primers according to Mani et al.16 for E-cadherin, ZO-1, alpha smooth muscle actin, Slug, Snail, fibronectin and vimentin. Human glyceraldehyde-3-phosphate dehydrogenase (GAPDH) was used as a control. The products were separated by 2% agarose gel electrophoresis.
CD44+/CD24−/low cells were sorted using a fluorescence-activated cell-sorting scan (BD Biosciences). Breast WT and ASC-CM-treated cancer cells were fractionated into CD44high/CD24low and CD44low/CD24high subpopulations by FACS. We then plated a single-cell in a 96-well plate by introducing the CD44high/CD24low and CD44low/CD24high cells at a dilution yielding an average of 1 cell per well. We examined these wells under a microscope to verify that in any of the wells only one cell was present. After the cells were cultured for 3 weeks, the mammospheres were counted as populations derived from single-cell clones arising from CD44high/CD24low cells and CD44low/CD24high cells. The cells were cultured in serum-free medium containing Dulbecco's modified Eagle's medium F12 supplemented with 2% B-27 (Invitrogen), 40 ng/ml recombinant human basic fibroblast growth factor (Chemicon, Temecula, CA), 100 ng/ml thrombin (R&D Systems, Minneapolis, MN), 20 ng/ml epidermal growth factor, 1 mM 2-mercaptoethanol, 1 ng/ml leukemia-inhibiting factor and 1% ITS (Sigma–Aldrich, St. Louis, MO).
Flow cytometry analysis
BT474 and T47D cells were cultured with and without hASC-CM and then trypsinized. Cells were washed in PBS and stained with phycoerythrin mouse anti-human CD24 (BD Pharmingen, San Jose, CA), allophycocyanin anti-human CD44 (BD Pharmingen) or mouse immunoglobulin (BD Pharmingen) as the isotype control according to the manufacturer's instructions and incubated at 4°C for 15 min. After incubation, cells were washed once with PBS and then separated (no fewer than 30,000 cells) on either an Aria cell sorter (BD Biosciences) or a MoFlo high-performance cell sorter (Dako Cytomation, Carpinteria, CA). Flow cytometry gates were determined by the analysis of unstained cells, isotype-specific stains.
Murine breast cancer 4T1 cells were either treated with 100 ng/ml PDGF-D (R&D Systems) for 72 hr or untreated before injection. In separate experiments, 4T1 cells were grown in ASC-CM for 72 hr in the presence of either a neutralizing antibody against PDGF-D (1 μg/ml) (R&D Systems) or isotype matched normal mouse IgG (1 μg/ml) (SouthernBiotech) for 72 hr before injection. Untreated 4T1 cells served as control.
Eight-week-old female Balb/c mice were each injected with the same number of 10 × 103 of the respective 4T1 cells subcutaneously into the inguinal mammary fat pad with Matrigel (BD Bioscience)17 at 1:1 ratio. Tumor sizes were measured weekly with a caliper, and tumor volumes were calculated with the equation V = (L × W2) × 0.5, where L is length and W is width of a tumor.
ASC-CM induces EMT in breast cancer cells in a paracrine fashion
hASCs are spindle shaped and feature plastic adherent characteristics (Fig. 1I). Human breast cancer cell lines T47D, MCF-7 and BT474 grew in clusters (Fig 1IIa); however, when cultured in the presence of ASC-CM, some cancer cells became dissociated, and the morphology of these cells switched to a mesenchymal type (Fig. 1IIb). An important feature of epithelial cells is the expression of E-Cadherin. As shown by immunostaining with antibodies against E-Cadherin, E-cadherin expression decreased when epithelial cells were treated with ASC-CM (Fig. 1III). The reduced expression of E-Cadherin was accompanied by an increased expression of mesenchymal markers: RT-PCR data revealed that the expression of several important EMT markers such as Slug, Snail, alpha smooth muscle actin, fibronectin and vimentin increased when cancer cells were grown in ASC-CM (Fig. 2I). The protein expression of fibronectin, alpha smooth muscle actin and vimentin was confirmed by Western blotting (Fig. 2II).
ASC-CM promotes mammosphere formation in human breast cancer cells
To determine whether cells that underwent EMT express stem cell properties, we isolated CD44+/CD24−/low cells from T47D and BT474 cell populations by flow cytometry. The sorted cells were grown in either regular culture medium or ASC-CM before the mammosphere assay. The CD44+/CD24−/low cells formed an increased number of spheres when treated with ASC-CM (Fig. 3).
ASC-derived PDGF-D promotes anchorage-independent growth of cancer cells and the enrichment of cancer stem cells
Since overexpression of PDGF-D in prostate cancer cells was shown to induce EMT,8 we tested whether PDGF-D was present in ASCs and whether PDGF-D played a role in ASC-CM-induced EMT. Our data from RT-PCR and Western blot analysis show that ASCs expressed PDGF-A, −C and −D but not −B (Fig. 4). Soft agar assay revealed that ASC-CM promoted cancer cell colony formation, and this effect was inhibited in the presence of a neutralizing antibody against PDGF-D but not by antibodies against PDGF-A and −C (Fig. 5). Representative images of the respective soft agar assays are shown in Supporting Information Figure 1. We also analyzed the total volume of the colonies and our data demonstrate that the volume of the colonies in the presence of ASC conditioned medium was significantly higher than that of the control. This effect was inhibited by the presence of a neutralizing antibody against PDGF-D (Supporting Information Figs. 2a and 2b). To further confirm the involvement of PDGF-D in EMT, we analyzed the percentage of CD44high/CD24low subpopulations in two different breast cancer cell lines: BT474 and T47D. We found that treatment with ASC-CM significantly increased the percentage of CD44high/CD24low cells, and this effect was inhibited as well by a PDGF-D-neutralizing antibody (Fig. 6).
PDGF-D treated breast cancer cells grow faster in vivo
To evaluate the effect of PDGF-D treated cancer cells on tumor growth in vivo, we performed animal experiments using a murine 4T1 breast cancer model. 4T1 cells were exposed to PDGF-D for 72 hr before injection. Our data show that the tumor volume is significantly higher in the PDGF-D treated group compared to the untreated group (Fig. 7a). In a separate experiment, 4T1 cells were grown in ASC-CM in the presence of either a neutralizing antibody against PDGF-D or an isotype-matched normal mouse IgG. Untreated 4T1 cells served as control. Our data show that ASC-CM treated 4T1 cells formed larger tumors compared to untreated 4T1 cells and this effect was abolished in the presence of a neutralizing antibody against PDGF-D (Fig. 7b).
The important finding of this study is that stem cells resident in the tissue representing the local tumor microenvironment are capable of interacting with tumor cells and play a pivotal role in tumor progression and growth. We demonstrate that one pathway of this interaction is mediated through PDGF-D, which is critical for the anchorage-independent growth of cancer cells and the enrichment of CD44high/CD24low cancer stem cells.
MSCs have been reported to facilitate the entry of breast cancer cells into bone marrow18 and are actively recruited to the tumor microenvironment,19 and may play a role in cancer growth and metastasis.20 Recent literature suggests that EMT plays a critical role in the metastatic cascade and that MSCs are involved in the initiation of EMT. Martin et al.21 elucidated the effects of bone marrow-derived MSCs on breast cancer cells in a direct coculture system. They observed a significant upregulation of EMT-specific markers (N-cadherin, vimentin, Twist and Snail) following the coculture of breast cancer cells with bone marrow-derived MSCs. These changes were predominantly mediated through cell contact and appeared to be specific to bone marrow-derived MSCs.
Vimentin and N-cadherin are associated with EMT.16, 22 Vimentin upregulation is commonly observed in more invasive basal cancer subtypes and has been positively correlated with poor prognosis in breast cancer patients.23 In our study, we found that vimentin was upregulated in MCF-7, BT-474 and T47D breast cancer cells treated with ASC-CM. Expression of fibronectin and alpha smooth muscle actin was increased as well while expression of E-cadherin was decreased.
Decreased expression of E-cadherin and the resultant cellular dissociation are other markers consistent with EMT.24 Previous studies25, 26 have shown that when MCF-7 cells were cocultured with MSCs, the morphology of MCF-7 cells changed from a cluster to a dispersed pattern. MCF-7 cells cocultured with MSCs also expressed low levels of the intercellular adhesion molecules E-cadherin and epithelial-specific antigen. Our data concur with these previous reports, and in addition, we show that ASCs promote EMT in a paracrine fashion. Consistent with our findings, Sasser et al.27 showed that MSC-derived soluble factors enhance the growth rate of various breast cancer lines, such as MCF-7, T47D, BT474 and ZR-75-1 in a three-dimensional tumor environment. ZEB family members are among the candidate transcriptional repressors of the epithelial cell–cell adhesion molecule E-cadherin, the downregulation of which is a hallmark of EMT. Kumar et al.28 showed that aberrant expression of tissue transglutaminase (TG2) promotes EMT in mammary epithelial cell lines. At the molecular level, TG2 expression resulted in a loss of E-cadherin and increased expression of various transcriptional repressors (Snail 1, ZEB1, ZEB2 and Twist1). Sanchez-Tillo et al. showed that ZEB1 represses E-cadherin and induces EMT by recruiting the SWI/SNF chromatin-remodeling protein BRG1.29 These reports are indeed relevant, and we plan to investigate the role of ZEB1 and ZEB2 in our system in the future.
A relationship between EMT and breast cancer stem cells has emerged with evidence that the induction of EMT in immortalized human mammary epithelial cells results in the acquisition of mesenchymal traits and the expression of stem-cell markers.16 Similarly, cell lines derived from Brca1-deficient mouse mammary tumors have shown features of EMT with enrichment in cells that were CD44high/CD24low.30 Indeed, the CD44high/CD24low phenotype has been used to define human breast cancer stem cells.31–33 Similarly, we observed that treatment of breast cancer cells with ASC-CM led to increased expression of CD44 and decreased expression of CD24. We found that within the BT474 and T47D lines, the percentage of CD44high/CD24low cells was significantly reduced in the presence of a PDGF-D-neutralizing antibody. Both CD44+/CD24−/low markers and mammosphere assays have been widely used for the characterization of cancer stem cells. However, neither of these two assays or any other assays alone are sufficient to characterize cancer stem cells. For example, CD44+/CD24−/low markers are enriching for stem cells but the cell samples also contain non-stem cell populations. The same holds true for cells that have the capability to form spheres. Double selection using the combination of these two assays effects an increase in the percentage of cancer stem cells.
Previous studies have shown that PDGF-D is a potent transforming growth factor for NIH/3T3 cells, and the transformed cells displayed stress fiber reorganization, increased proliferation rate, anchorage-independent growth in soft agar, ability to induce tumors in nude mice and upregulation of vascular endothelial growth factor.34
We recently reported that tumor cell-derived PDGF-B promotes the migration of ASCs.11 We further found that ASCs, once incorporated into the tumor microenvironment, differentiate into myofibroblasts under the influence of transforming growth factor beta.14 The myofibroblasts produced stromal cell-derived factor-1 which acts on CXC chemokine receptor Type 4 in tumor cells and promotes tumor growth and invasion.10, 13 The results reported in this studies now indicate that ASC-CM contains additional paracrine signaling that promote EMT in several breast cancer lines and that PDGF-D derived from ASC plays an important role. Previous studies had shown that PDGF-D expression is associated with accelerated tumor growth in prostate and pancreatic cancers.35 Inhibition of PDGF-D signaling prevented glioma formation.35 Our in vivo data demonstrating that PDGF-D treated 4T1 cells grow faster are in line with these previous reports. As well demonstrate our animal data, that the tumor promoting effect of culturing the cancer cells in PDGF D containing cultured media is neutralized by addition of an antibody against PDGF D. While the PDGF media has a highly stimulating effect on tumor size, after 17 days the size of tumors without PDGF pre-treatment and with pre-treatment plus the addition of an PDGF D antibody is comparable and significantly lower compared to the ASC-CM treated group.
Our findings shed new light on mechanisms where local tissue-resident stem cells are able to promote the growth of breast cancer cells. Possibly this could open up a novel selective therapeutic strategy targeting EMT pathways and the specific communication between tissue-resident normal stem cells and cancer stem cells, assuming that the blockage of PDGF D pathways is critical for tumor growth but would not effect normal tissue homeostasis.
This research was supported in part by the Department of Defense Breast Cancer Research Program (to YHS) and by the Alliance of Cardiovascular Researchers (to EA).