|Home | About | Journals | Submit | Contact Us | Français|
Y-box binding protein-1 (YB-1) is an oncogenic transcription/translation factor expressed in >40% of breast cancers, where it is associated with poor prognosis, disease recurrence, and drug resistance. We questioned whether this may be linked to YB-1’s ability to induce the expression of genes linked to cancer stem cells such as CD44 and CD49f. Herein, we report that YB-1 binds the CD44 and CD49f promoters to transcriptionally up-regulate their expressions. The introduction of wild-type YB-1 or activated P-YB-1S102 stimulated the production of CD44 and CD49f in MDA-MB-231 and SUM 149 breast cancer cell lines. YB-1-transfected cells also bound to the CD44 ligand hyaluronan more than the control cells. Similarly, YB-1 was induced in immortalized breast epithelial cells and up-regulated CD44. Conversely, silencing YB-1 decreased CD44 expression as well as reporter activity in SUM 149 cells. In mice, expression of YB-1 in the mammary gland induces CD44 and CD49f with associated hyperplasia. Further, activated mutant YB-1S102D enhances self-renewal, primary and secondary mammosphere growth, and soft agar colony growth, which were reversible via loss of CD44 or CD49f. We next addressed the consequence of this system on therapeutic responsiveness. Here we show that paclitaxel induces P-YB-1S102 expression, nuclear localization of activated YB-1, and CD44 expression. The over-expression of wild-type YB-1 promotes mammosphere growth in the presence of paclitaxel. Importantly, targeting YB-1 sensitized the CD44High/CD24Low cells to paclitaxel. In conclusion, YB-1 promotes cancer cell growth and drug resistance through its induction of CD44 and CD49f.
Breast cancer relapse confers a poor prognosis, decreasing survival rates from 80% to 60% when the recurrence is local, and from 80% to 10% for metastatic disease. Relapse is postulated to be mediated by the persistence of cancer stem cells that have survived an initial treatment regimen with radiation (1) and/or chemotherapy (2-4) possibly due to a selective resistance of the cancer stem cells to these agents. Human breast cancer stem cells have been identified as a phenotypically restricted subset of CD44+/CD24−/low cells that form tumours in immunocompromised mice in limiting dilution transplant assays (5). The same CD44+/CD24−/low subset has also been associated with breast tumour mammosphere generation in vitro (3). However, the nature of the essential molecular properties of breast cancer stem cells has remained poorly defined.
YB-1 (Y-box binding protein-1) is a transcription/translation factor that is commonly overexpressed in many cancers, including human breast cancer (40%) (6-8). This elevated expression of YB-1 in human breast cancer correlates with high rates of relapse (6). Targeted overexpression of YB-1 in the mouse mammary gland leads to the development of mammary tumours (9), confirming a role of YB-1 as an oncogene in that tissue. YB-1 is directly phosphorylated, and therefore activated, on its serine 102 site by Akt (10) and even more potently by ribosomal S6 kinase (RSK) (11), a major component of the MAP kinase pathway. Thus, YB-1 is positioned as a key player in both the PI3K/Akt and MAPK pathways. YB-1 regulates genes that promote breast cancer cell growth and survival including EGFR (12), Her-2 (erbB2) (7), PIK3CA (13), and the MET (c-Met) receptor (14), and biologically, is essential for breast cancer cell growth in vitro (7, 10-12) and in vivo (15). YB-1 expression has also been shown to be associated with drug resistance via the induction of genes such as MDR-1 (16, 17). Consistent with this, inhibition of YB-1 was found to sensitize breast cancer cells to paclitaxel (15), a chemotherapeutic agent commonly used in the clinic to treat advanced breast cancer. To elucidate the transcriptional programming activity of YB-1, we performed chromatin immunoprecipitation-on-chip (ChIP-on-chip) assays. This screen revealed a subset of genes known to be active in and important to a number of stem cell populations, including c-KIT, BMI-1, members of the WNT and NOTCH signaling pathways, as well as CD44 and CD49f (also known as α-6 integrin) (14). We also found that YB-1 transcripts were present in purified CD44+CD49f+ subpopulations of primitive human mammary progenitor cells populations isolated from normal reduction mammoplasties (14). Taken together, this led us to test the hypothesis that YB-1 plays a key role as an oncogene by transactivating genes associated with a cancer stem cell phenotype.
Human breast cancer cell lines, MDA-MB-231, MDA-MB-468, and SUM 149, were purchased and maintained as previously described (13, 14). The cell lines was characterized for CD44 and CD24 expression by flow cytometry (Supplemental Figure 1A-E). Cell auto-fluorescence was taken into account.
ChIP using a polyclonal chicken antibody to precipitate endogenous YB-1 was performed as previously described (7). Three primer sets were designed to flank seven putative YB-1 binding sites (ATTG) in the first two kilobases of the CD44 promoter, and similarly, two primer sets flanking eight putative YB-1 binding sites in the first two kilobases of the CD49f promoter were also designed. Details in Supplemental Materials and Methods.
Immunofluorescence assays were performed as previously described (13, 14). Antibodies used were rat anti-human and anti-mouse fluorescein isothiocyanate (FITC)-conjugated anti-CD49f antibody (1:25, Clone GoH3, #555735, BD Pharmingen), rat anti-human phycoerythrin (PE)-conjugated anti-CD44 antibody (1:100, Clone 515, #550989, BD Pharmingen), unconjugated rabbit anti-human phospho-YB-1S102 antibody (P-YB-1S102, 1:100, Clone C34A2, #2900, Cell Signaling Technology), secondary anti-rabbit FITC antibody incubation (1:200, #711-096-152, Jackson ImmunoResearch Laboratories, Inc). Details in Supplemental Materials and Methods.
A single cell suspension was achieved by first scraping the adherent cells and digesting with a dispase solution (Stem Cell Technologies, Vancouver, BC) and DNase (Sigma Aldrich) at 37°C for 10 minutes. Cells were diluted into a 2% FBS PBS buffer and passed through a 40 μm cell strainer to achieve a single cell suspension. A single cell suspension was stained with anti-CD44 conjugated to phycoerythrin (PE) (BD Pharmingen). 7-aminoactinomycin D (7-AAD) viability dye (BD Pharmingen) was added to a 2% FBS PBS resuspension buffer. For cell sorting, the top and bottom 10% of cells with the highest and lowest CD44-PE fluorescence were collected into 50% FBS PBS buffer and seeded subsequently into a 96-well plate for immunofluorescence staining or into growth assays. Sorted cells were seeded at 1 × 104 cells per well in 96-well plates. Details in Supplemental Materials and Methods.
A single cell suspension of MDA-MB-231 or SUM 149 cells was seeded into non-adherent mammosphere culturing conditions and counted after 7 days. The single cells were seeded into secondary cultures under the same conditions and counted after 7 days. Cells were dissociated enzymatically into a single cell suspension with dispase digestion and mechanical dissociation by repeated pipetting. Cells were seeded at 5 × 103 (MDA-MB-231) and 2 × 104 (SUM 149) per well into Ultra-Low Attachment coated 6-well culture plates (Corning, Lowell, MA) in a 1:1 DMEM/F12 (Invitrogen) basal media freshly supplemented with 20 ng/mL human basic fibroblast growth factor (bFGF, Invitrogen), 20 ng/mL epidermal growth factor (EGF, Invitrogen), 10 μg/mL heparin (Sigma Aldrich), 1:50 B27 supplement without Vitamin A (Sigma Aldrich). Spheres containing approximately >15 cells were counted 7 days after seeding. In the self-renewal serial passaging experiments, all cells from primary mammosphere cultures were collected, centrifuged at 350 g for 5 minutes and dissociated into a single cell suspension with 0.25% trypsin for 5 minutes at 37°C and counted.
SiRNAs were transfected into cells using the Lipofectamine RNAiMAX transfection reagent (Invitrogen) according to the manufacturer’s protocol. Unless otherwise specified, in all experiments, 20 nM of siRNAs were incubated with the cells for 96 hours before assessment of biochemical/biological effects. Details in Supplemental Materials and Methods.
Immunoblotting analyses were performed as described previously (7). Antibodies used were as follows: anti-total YB-1 (T-YB-1; 1:2000, #ab12148, Abcam, Cambridge, MA; 1:1000, #2749, Cell Signaling Technology, Danvers, MA; 1:2000, a polyclonal antibody designed against the C-terminus of YB-1, produced by and a generous gift from Dr. Colleen Nelson, University of British Columbia, Vancouver, BC), anti-P-YB-1S102 (1:1000, Clone C34A2, #2900, Cell Signaling Technology), anti-CD44 (1:500, Clone EPR1013Y, #ab51037, Abcam), anti-CD49f (1:1000, #3750, Cell Signaling Technology), anti-Flag (1:2000, Clone M2, #F3165, Sigma Aldrich), anti-total Akt (T-Akt; 1:1000, #9272, Cell Signaling Technology), anti-pan-Actin (1:1000, #4968, Cell Signaling Technology), and anti-vinculin (1:1000, Clone hVIN-1, #V9131, Sigma Aldrich).
SUM149 EV and shYB-1 cells were seeded in a 6-well plate (4 × 105 cells/well) and allowed to adhere. The cells were then transfected with pGL3b (1 μg) or pGL3b-CD44P (1 μg, Addgene, deposited by Dr. Robert Weinberg) (2) in triplicate using Lipofectamine 2000 (Invitrogen). Cells were lysed 24 hours post-transfection in passive lysis buffer (Promega) and luciferase activity was measured.
Matrigel basement membrane (BD Biosciences) were added at 40 μL per well in a 96-well plate and incubated briefly at 37°C for solidification. MDA-MB-231 and SUM 149 cells in single cell suspension were added in their respective media at 2 × 103 and 5 × 103 cells/well. Growth was assessed in photographed fields 7 days after seeding.
In a 6-well plate format, 2 or 4 μg of a Flag:EV, Flag:YB-1WT, or Flag:YB-1S102D plasmid were transfected with 5 or 10 μL of Lipofectamine 2000 (Invitrogen) respectively as specified by the recommended manufacturer’s protocol. Cells were harvested 96 hours post-transfection.
Stable EV, shYB-1, Flag:EV, Flag:YB-1WT, and Flag:YB-1S102D cell lines were established by transfecting 4 μg of DNA constructs as described. 400 μg/mL of G418 (Calbiochem, EMD Chemicals, San Diego, CA) was added to the culture media and replaced every 3 to 4 days. Cells were continually split at low density to allow for optimal selection of transfectants with acquired G418 resistance.
YB-1 transgenic (Tg 2 line) and wild-type (WT) mice were revived from cryo-preserved embryos generated and characterized previously (9), where expression of a human hemagglutinin (HA)–tagged YB-1 cDNA was controlled by the ovine β-lactoglobulin promoter (BLG). For our study, histological sections were obtained from the mammary glands of 6 to 8-month old female mice post-lactation after two cycles of mating and nursing. The mice were mated twice and nursed their pups for 3 weeks each time. The YB-1 transgenic and control mice were euthanized at the end of their second lactation (day 20 to 22 after the birth of the pups) and mammary gland tissues were collected. The mammary glands were dipped into Tissue-Tek O.C.T. Compound (Sakura Finetek USA, Torrance, CA), frozen, and kept at −80°C. OCT-embedded tissues were sectioned (6 μM) using a Cryostat and placed on Fisherbrand Superfrost Plus glass slides (Thermo Fisher Scientific, Waltham, MA) and kept at −80°C. The sections were defrosted at room temperature and incubated with phosphate-buffered saline (PBS) for 5 minutes, fixed with 2% formaldehyde for 20 minutes, rinsed twice with PBS. Subsequent immunofluorescence staining and imaging protocol was performed as described above. The only exception is a primary purified rat anti-mouse CD44 antibody (1:100, Clone IM7, #550538, BD Pharmingen) and a secondary anti-rat Alexa 546 antibody (Molecular Probes, Invitrogen) were used. A hemagglutinin-specific antibody (Santa Cruz Biotechnology) was used with immunohistochemistry methods. Details in Supplemental Materials and Methods.
SUM 149 cells were treated with commonly employed chemotherapeutic agents at 10 μM and cell viability was assessed as previously described (18). The cells were also stained with anti-CD44-PE (1:5, Clone G44-26, #555479, BD Pharmingen) and assessed as previously described (18). SUM 149 cells were also treated with paclitaxel (Sigma Aldrich) dissolved in DMSO at 10 nM for 48 or 72 hours before performing immunoblotting and ChIP as described. MDA-MB-231 Flag:EV and Flag:YB-1WT cells were treated in monolayer or in the mammosphere assay in the presence of 0 nM or 10 nM of paclitaxel with DMSO as vehicle control for 7 days. MDA-MB-231 cells grown in monolayer had the media changed at Day 4 and cells were harvested for immunoblotting on Day 7.
All quantitative data are represented as mean ± standard error of the mean (s.e.m.) of 3 independent experiments. P-values were generated using the paired Student’s T-test where * denotes p <0.05, ** denotes p <0.01, and *** denotes p <0.001.
Several putative YB-1 binding sites were located 2 kilobases upstream of the transcriptional start site of the CD44 and CD49f genes, initially identified in a ChIP-on-Chip screen (14) (Figure 1A, upper panel). To validate these results, we performed conventional ChIP analyses and provide direct evidence of the ability of YB-1 to bind to the promoters of both CD44 and CD49f genes in two human breast cancer cell lines (MDA-MB-468 and SUM 149, Figure 1A, lower panel). Immunohistochemical analysis of SUM 149 cells grown either in monolayer cultures or as xenografts illustrated coincident staining of P-YB-1S102 and CD44 protein in individual cells (Figure 1B). Therefore, we first asked whether the expression of CD44 would fluctuate concordantly with changes in activated YB-1, P-YB-1S102. Quantification of this association of CD44 expression with the presence of P-YB-1S102 was demonstrated by isolating the highest and lowest 10% of CD44-expressing viable SUM 149 cells by FACS sorting and co-staining with CD44 and P-YB-1S102 antibody. There, we observed a >2-fold difference in P-YB-1S102 levels between the subsets expressing the highest and lowest levels of CD44 (Figure 1C, upper panels), corresponding to enhanced growth in mammosphere cultures and soft agar (Figure 1C, lower panels).
We next asked whether YB-1 is a determining regulator of CD44 and CD49f expression in breast cancer cells by knocking down YB-1 expression in MDA-MB-231 and SUM 149 cells using three siRNA oligonucleotides designed to target YB-1 transcripts specifically. The result was a decrease in both transcript and protein levels of CD44 and CD49f (Figure 2A and 2B). To obtain stable inhibition of YB-1 expression, SUM 149 cells were transfected with anti YB-1 shRNAs. This produced sustained decreases in CD44 and CD49f transcript and protein levels as compared to control cells transfected with an empty vector (EV) (Figure 2C). Stable knockdown of YB-1 transcripts also resulted in a decreased expression of CD104 (β4 integrin), a beta chain partner of CD49f (Figures 2B and 2C). Further, YB-1 shRNA-expressing cells showed a 50% decrease in CD44 promoter activity (Figure 2D), suggesting YB-1 as a significant regulator of CD44.
Conversely, when we forced a transient elevation in the expression of either Flag:YB-1WT or a constitutively active Flag:YB-1S102D, increases in both CD44 and CD49f transcript levels were obtained in both cases, with the largest increases resulting from the constitutively active YB-1 construct (Figure 3A). A similar result was seen when these constructs were stably expressed in the same cells (Figure 3B). Importantly, we confirmed that a heightened expression of CD44 and CD49f was concomitantly induced in the same transfected YB-1-overexpressing cells (Figure 3C). Likewise, expression of YB-1 in immortalized breast epithelial cells which typically have very little of this protein caused a measurable increase in CD44 mRNA and protein (Supplemental Figure 2). To functionalize the effect in breast cancer cells, we determined that MDA-MB-231 cells that stably express wild-type YB-1 or YB-1D102 have a greater affinity for the CD44 ligand hyaluronan (Supplemental Figure 3).
To determine whether elevated YB-1 expression would have a similar effect on the expression of CD44 and CD49f in primary mammary epithelial cells in vivo, we evaluated the mammary glands of 6 to 8-month old transgenic mice expressing human hemagglutinin (HA)–tagged YB-1 cDNA under the control of the ovine β-lactoglobulin promoter (BLG) (9). To confirm transgene expression we stained the tissues for HA and showed an increase in expression with an associated elevation in P-YB-1S102 (Figure 3D i-ii). These mice had higher levels of CD49f and CD44 protein (Figure 3D iii-iv and Supplemental Figure 4 A-B) as compared to mammary cells from wild-type animals (n=4/group). The increased expression of CD49f in the YB-1 transgenic mice was further validated with a second antibody (Supplemental Figure 4B). The specificity of the CD44 antibody staining was confirmed by staining B6/129 CD44−/− mouse kidney cells (Supplemental Figure 4C). We also noted that the cells in the mammary glands from the YB-1 transgenic mice were hyperplastic and displayed atypical nuclei in comparison to those in the wild-type controls when stained with H&E (Supplemental Figure 4D). Taken together, these findings indicate that YB-1 regulates the expression of CD44 and CD49f both in human transformed mammary cells in vitro and in primary mouse mammary cells generated in vivo.
To test whether the ability of YB-1 to regulate CD44 and/or CD49f expression was important for YB-1-dependent tumour cell growth(15), we next examined the effect of treating MDA-MB-231 cells with anti-CD44 and anti-CD49f siRNAs on mammosphere formation (Figure 4A). SiRNAs directed against transcripts for YB-1 as well as CD44 and CD49f all inhibited primary mammosphere formation by >50%. Conversely, forced increased expression of Flag:YB-1WT or Flag:YB-1S102D increased the yield of mammospheres by up to 8-folds (Figure 4B). Moreover, this effect was inhibited if the Flag:YB-1S102D expressing cells were treated with either CD44 or CD49f siRNAs (Figure 4B). Further, this effect is perpetuated over time through passaging as SUM 149 cells transduced to obtain stable high expression of either Flag:YB-1WT or Flag:YB-1S102D also produced a higher number of cells in mammospheres in secondary non-adherent cultures than control cells transduced with an empty vector plasmid (Flag:EV) (Figure 4B), an indicator of cell self-renewal(19, 20). Consistent with the mammosphere assays, analogous experiments with siRNA-treated MDA-MB-231 and SUM 149 cells, showed both CD44 and CD49f to be required for the clonal anchorage-independent growth of both of these cell lines in soft agar (Figure 4C). CD49f was needed for SUM 149 cells but not MDA-MB-231 cells to form colonies (Figure 4C). Conversely, forced expression of Flag:YB-1S102D increased the ability of transfected cells to grow in soft agar and this was again blocked when the expression of either CD44 or CD49f was silenced (Figure 4C). The ability of MDA-MB-231 cells to proliferate and form extensive structures in Matrigel cultures was also inhibited when the cells were pre-treated with CD44 or CD49f or YB-1 siRNAs (Figure 4D). These findings demonstrate a critical role of YB-1 in enabling breast cancer cells to grow in 3-dimensional assays that is dependent on the downstream up-regulation of CD44 and CD49f. Notably, the expressions of CD44 and CD49f appear to be inter-dependent in these cell lines, suggesting a functional association.
Preceding findings intimate that the increased expression of YB-1 associated with breast cancer relapse in patients might be due to a selection of cells showing an increased expression of CD44. To investigate this possibility, we screened a panel of commonly used anti-cancer drugs (all at 10 μM) for their ability to kill SUM 149 cells in vitro in parallel with an assessment of the proportion of CD44High cells present 72 hours later. All compounds killed >80% of the input cells and the percentage of CD44High cells amongst the survivors was consistently higher than in the control cells (Supplemental Table 1). The most marked effect in this regard was obtained with paclitaxel treatment which produced an almost 10-fold increase in the proportion of viable CD44High cells (Supplemental Table 1). Next, we determined that paclitaxel at a much lower and more clinically relevant concentration of 10 nM was able to promote nuclear expression of P-YB-1S102 (Supplemental Figure 5). Likewise, the chemotherapeutic agent cisplatin had the same effect (Supplemental Figure 5). Related to these findings, paclitaxel increased YB-1 binding to the CD44 promoter after 48 hours (Figure 5A), which also induced P-YB-1S102, correlating with activation of the RSK pathway based on higher levels of P-RSKS380 and a second substrate P-GSKβS9 (Figure 5A). The induction of this pathway then led to increased CD44 after 72 hours (Figure 5B). In MDA-MB-231 Flag:YB-1WT stably expressed cells, after 10 nM paclitaxel treatment, there was a further increase in CD44 expression and a greater amount of Flag:YB-1WT detected per cell on the whole (Figure 5C). Further, the Flag:YB-1WT-transfected MDA-MB-231 cells formed more mammospheres in the presence of paclitaxel (10 nM) as compared to control transduced cells (Figure 5D). Finally, we sorted the SUM 149 cells for CD44High/CD24Low by FACS (Figure 6A) and confirmed enrichment by immunofluorescence, comparing CD44High sorted cells to unsorted cells for CD44 expression (Figure 6B). The CD44High/CD24Low were then transfected with YB-1 siRNAs for 48 hours and treated with paclitaxel (10 nM) for 24 hours. Notably the CD44High/CD24Low treated with the scrambled control were resistant to paclitaxel however if YB-1 was silenced growth of this TIC population was significantly attenuated based on crystal violet staining (Figure 6C) and quantification of cell viability using Hoechst33342 staining (Figure 6D).
Here we present YB-1 as the first oncogene identified which induces both CD44 and CD49f. YB-1 utilizes CD44 and CD49f to promote self-renewal, mammosphere growth, soft agar colony formation, and drug resistance in breast cancer cells. The YB-1/CD44/CD49f relationship was also evident in the mammary glands of YB-1 transgenic mice as well as in immortalized normal breast epithelial cells. These findings have broad implications given the important role for CD44 in mediating tumour initiation and cancer relapse. It also gives us some insight into why YB-1 may be so highly associated with breast cancer recurrence as previously reported (6). Consistent with this idea, we showed that CD44 cells were resistant to chemotherapy whereby the agents stimulate YB-1 to bind to the CD44 promoter. Furthermore, we report that eliminating YB-1 in the CD44High/CD24Low population can improve cellular response to paclitaxel.
We question whether YB-1 somehow hijacks normal breast stem cells during neoplastic progression. This extends our initial finding that YB-1 is detectable in primary mammary progenitor cells from women who have undergone reduction mammoplasties (14). In the study presented herein, we show that inducing YB-1 in the immortalized breast epithelial cells, HTRY, caused an increase in CD44 mRNA and protein. Consistent with this, both CD44 and CD49f are selectively expressed on normal stem cells in the breast and other tissues (21-24). We therefore propose that early activation of YB-1 during mammary tumourigenesis involves a primitive cell population.
CD44 is a widely recognized marker of many cells with cancer-initiating (5, 25-27) and, in some cases, metastatic activity (28-30). As such, understanding its regulation has implications in tumour cell invasion, escape from apoptosis and drug resistance. CD44 couples to a number of receptor tyrosine kinases such as EGFR, the c-Met receptor and CD49f to sti mulate signal transduction (14, 31). It also partners with the oncogene RHAMM to promote signalling through the ERK1/2 pathway whereby CD44 stimulates tumour cell invasion (32, 33). Importantly, CD44 expression is thought to confer drug resistance by directly inducing downstream stem cell associated and/or drug resistant genes including Nanog and MDR-1 through its interaction with hyaluronan (34-36). This is attributable to activation of the PKC and STAT3 signaling pathways (34-36). Notably, as CD44 expression was increased, cells were less sensitive to paclitaxel due to STAT3 activation and consequent up-regulation of MDR-1. YB-1 expression has already been reported to be associated with chemoresistance perhaps through the regulation of MDR-1 (16, 17). Bridging these studies is the evidence for YB-1’s role in invasion (13) that may also be attributed to CD44 and CD49f. It is foreseeable that it induces other genes associated with cancer stem cells as well. We reported that YB-1 also regulates the stem cell-associated Met receptor (c-Met), EGFR, and Her-2 (erbB2) (7, 14), signaling partners of CD44 and CD49f (37-40), suggesting a cooperative complex on the surface of TICs that activates downstream pathways such as the PI3K-Akt (40-42) and MAPK pathways (33, 43) (Supplemental Figure 6). The crosstalk between these receptors and their downstream effectors provide interesting opportunities for future therapeutic possibilities. We conclude that YB-1 inhibition may be a new approach to eliminating TICs and by extension possibly cancer recurrence.
We thank Lisa Xu of the Child and Family Research Institute (CFRI) Core FACS Facility, as well as Lucy Yang Zhao and Peter Eirew (UBC, Vancouver, BC, Canada) for technical assistance. We would also like to acknowledge Michel Roberge (UBC, Vancouver, Canada) for his assistance in accessing the Chemical Biology Network. This research was supported by grants from the Canadian Breast Cancer Research Alliance (S.E.D. /C.J.E.), the National Institutes of Health (R01-CA114017, S.E.D./I.M.B.), the Canadian Stem Cell Network (A.R./C.J.E.) and the Canadian Breast Cancer Foundation (CBCF) British Columbia and Yukon Division (C.J.E.). K.T. was a recipient of a CIHR Canada Graduate Scholarship Master’s Award and A.A., a CIHR MD/PhD Award. K.T./A.A./A.H.D. were recipients of Michael Smith Foundation for Health Research Graduate Studentships. A.L.S. was awarded the CFRI and CBCF Post-doctoral Fellowships. A.R. held a CBCF British Columbia and Yukon Division Fellowship and a CIHR Fellowship.