Mitochondrial biogenesis is required for the anchorage-independent survival and propagation of stem-like cancer cells

1Tumor-initiating stem-like cells (TICs) are a small sub-population of tumor cells that are resistant to most anti-cancer therapies, including radio- and chemotherapy, and are able to expand and regenerate tumors, after conventional therapy is completed. As such, TICs are responsible for tumor recurrence, metastatic dissemination, and, ultimately, patient death. As a consequence, TICs have become very attractive targets for novel cancer therapies. TICs share some features with stem cells, and many studies have investigated the signaling pathways regulating their proliferation, asymmetric cell division and migrating properties, as well as their ability to undergo proliferation under anchorage-independent conditions. In fact, this latter property is being widely exploited to isolate TICs, which are able to survive and clonally expand as tumor-spheres, when placed in non-adherent settings. Tumor-spheres generated from breast cancer cells are known as mammospheres.

2The metabolic requirements of TICs remain a fairly unexplored area of investigation. We and others have recently shown that TICs rely mainly on mitochondrial metabolism, as compared to the more differentiated tumor cell population. Also, it was previously shown that the radio-resistance of TICs from gliomas and breast cancers correlates with higher mitochondrial respiration.

3We have demonstrated that >60 mitochondrial proteins are up-regulated in mammospheres derived from breast cancer cell lines (MCF7 and T47D), relative to cells grown in monolayers. Furthermore, pharmacological inhibition with oligomycin A, an inhibitor of the ATP synthase, greatly reduced mammosphere formation. More importantly, treatment with doxycycline, an FDA-approved antibiotic targeting mitochondrial ribosomes as a known ‘side effect’, inhibited the formation of tumor-spheres generated from cell lines across several different tumor types, as well as from primary tumor samples.

4These studies indicate that mitochondria in TICs are potentially druggable targets for the prevention of tumor recurrence and metastasis. The aim of our current study was to further validate mitochondrial function as a critical target for the development of new anti-cancer therapies. Here, we employed XCT790 as an investigational compound to inhibit mitochondrial oxidative phosphorylation. XCT790 is a well-established inverse agonist of Estrogen-Related Receptor α (ERRα), which functions as an essential cofactor of PGC1α. Importantly, PGC1α is required for the transcription of nuclear-encoded mitochondrial genes and drives mitochondrial biogenesis. XCT790 has been shown to inhibit colony formation in soft agar induced by PGC1α, and to induce cell death in chemo-resistant cancer cells. However, to our knowledge, no studies have previously investigated the effects of XCT790 on the propagation and survival of TICs.

5Here, our results show that XCT790 inhibits the proliferation of TICs from MCF7 breast cancer cells, as assessed using i) mammosphere formation and by ii) CD44/CD24 immuno-staining, coupled with FACS analysis. We also demonstrate that XCT790 suppresses the activation of well-established pathways governing TIC proliferation and survival, including Sonic hedgehog, TGFβ-SMAD, STAT3, and Wnt signaling. As expected, XCT790 profoundly reduced mitochondrial respiration. Importantly, XCT790-induced TIC inhibition could be rescued by treatment with Acetyl-L-Carnitine (ALCAR), a mitochondrial fuel and cofactor, which enhances mitochondrial oxidative metabolism and biogenesis. We also show that overexpression of ERRα is sufficient to augment MCF7 cell mammosphere formation, which can be blocked by treatment with mitochondrial inhibitors, such as XCT790 and oligomycin A, an inhibitor of ATP synthase. Finally, we show that mammosphere formation driven by FOXM1, a transcription factor downstream of Wnt/β-catening signaling that governs stemness, can also be blocked by treatment with three different classes of mitochondrial inhibitors (XCT790, oligomycin A, or doxycycline).

6Thus, our results firmly establish that mitochondrial biogenesis and metabolism are required for the survival of TICs, indicating that mitochondrial-targeted therapies are a promising novel strategy for targeting TICs for the treatment of relapsed and resistant cancers. Mitochondrial-targeted therapies should also be considered for cancer prevention.

7We have previously shown that mitochondrial proteins are up-regulated in mammospheres derived from breast cancer cell lines (MCF7 and T47D), relative to cells grown in monolayers. Furthermore, pharmacological inhibition with oligomycin A, an inhibitor of ATP synthase, greatly reduced mammosphere formation. More importantly, treatment with doxycycline, an FDA-approved antibiotic that targets mitochondrial ribosomes as a known ‘side effect’, inhibited the formation of tumor-spheres derived from 12 cell lines across eight different tumor types, as well as from primary tumor samples.

8To further investigate the role of mitochondria in TICs, here we employed a well-established inverse agonist of ERRα, namely XCT790. ERRα is a cofactor of PGC1α/β, which is required for the transcription of nuclear mitochondrial genes and mitochondrial biogenesis. As such, XCT790 is believed to suppress mitochondrial function. Notably, treatment with XCT790 dose-dependently inhibits MCF7 mammosphere formation, with an IC-50 of 10 μM (Figure 1A). We also validated that XCT790 dose-dependently reduces the protein expression of PGC1α in MCF7 cell monolayers by immuno-blot analysis (data not shown), by up to 70-80%, over this concentration range.

9These results were independently confirmed by evaluating the expression of bona-fide TIC markers CD44/CD24 by FACS. Under these conditions, CD44(+)high/CD24(−)low cells are considered to represent the TIC sub-population. XCT790 treatment significantly reduced the number of CD44(+)high/CD24(−)low cells in a dose-dependent fashion, relative to vehicle alone controls (Figure 1B).

10We next set out to investigate if XCT790 can target TICs in the presence of the total cancer cell population. To this end, MCF7 cells were treated with XCT790 (at 5 or 10 μM) as monolayers for 2 days and then re-plated on low-attachment plates in the absence of XCT790, to generate mammospheres for 5 days. Under these conditions, XCT790 pre-treatment dose-dependently reduced MCF7 cell mammosphere formation, by up to ~70% (Figure 1C), indicating that XCT790 can target the TIC population also when present in a heterogeneous cell population.

11Also, we asked if decreased mammosphere formation induced by XCT790 could be rescued by treatment with the mitochondrial cofactor Acetyl-L-Carnitine (ALCAR). ALCAR plays a key role in mitochondrial oxidative metabolism, by enhancing fatty acid β-oxidation. ALCAR stimulates mitochondrial biogenesis and is also directly converted to acetyl-CoA, a mitochondrial fuel. To this end, mammosphere formation was assessed after treatment with XCT790 (at 20 μM) and increasing concentrations of ALCAR. Figure 1D shows that ALCAR rescues the decrease in mammosphere formation induced by XCT790, in a dose-dependent manner. Thus, mitochondrial function is required for the efficient clonal expansion and anchorage-independent growth of TICs.

12We then examined if XCT790 affects the viability of the total cancer cell population, or if it specifically inhibits the viability of MCF7 cells in mammospheres. To this end, MCF7 cells were treated with increasing concentrations of XCT790 as monolayers for 3 days (Figure 2A) or 5 days (Figure 2B). Cell viability was then assessed using the SRB assay. Note that 5-day treatment did not affect the viability of the MCF7 cell monolayers, as profoundly as MCF7 cell mammospheres (Figure 2B). For example, treatment with 10 μM XCT790 reduces mammosphere formation by 50% (Figure 1A), whereas the viability of monolayer cells is reduced by only 20%. Thus, XCT790 preferentially reduces the viability of MCF7 cell mammospheres, relative to bulk cancer cells.

13To further corroborate the idea that XCT790 inhibits cancer stem cell-like features, we next analyzed the effects of XCT790 on a series of well-established signaling pathways, which have been shown to promote stemness. For this purpose, we employed a panel of eight MCF7 cell lines carrying different luciferase reporters, to monitor the activation state of a variety of different signaling networks, including Sonic hedgehog, TGFβ-SMAD, STAT3, Wnt, Interferon (IFN)-α/β-STAT1/2, NRF2-dependent antioxidant responses, IFN-γ-STAT1 and Notch pathways. Notably, several pathways were significantly inhibited by XCT790 treatment, including stem cell signaling (Sonic hedgehog, TGFβ-SMAD, STAT3, Wnt) and IFN-α/β-STAT1/2 signaling (Figure 3A). However, no effects were observed on the NRF2-antioxidant response, IFN-γ-STAT1 and the Notch pathways (Figure 3B). Thus, XCT790 inhibits the activation of several signal transduction pathways related to cancer stem-like features.

14Next, we set out to investigate the mechanism(s) by which XCT790 inhibits mammosphere formation and CSC features. XCT790 is an inverse agonist of ERRα, which is a cofactor of PGC1α required for the transcription of mitochondrial genes and mitochondrial biogenesis. Thus, the metabolic profile of MCF7 cells treated with XCT790 was analyzed. To this end, MCF7 cells were treated with XCT790 for 48 hours, stained with various MitoTracker probes and analyzed by FACS. Figure 4A shows that XCT790 induces a decrease in mitochondrial membrane potential, as assessed with MitoTracker Orange, which accumulates in mitochondria with an active mitochondrial potential. Surprisingly, XCT790 induces also an increase in mitochondrial mass, as assessed with MitoTracker Green (Figure 4B), localizing to mitochondria regardless of mitochondrial membrane potential. Similar results were obtained with MitoTracker Deep Red, which also measures mitochondria mass (data not shown). Analysis of the ratio of mitochondrial membrane potential versus mitochondrial mass indicates that XCT790 induces a profound decrease in mitochondrial membrane potential per mitochondria (Figure 4C).

15Also, the metabolic profile of MCF7 cells treated with XCT790 was examined using the Seahorse XFe96 analyzer, by employing a mitochondrial stress test. Notably, the oxygen consumption rate (OCR) was greatly reduced by treatment with XCT790 (Figure 5A). Further quantification revealed significant reductions in basal and maximal respiration (Figure 5B-5C), as well as ATP levels (Figure 5D), upon XCT790 treatment. Thus, XCT790 significantly reduces the rates of oxidative mitochondrial metabolism. To further validate these results, a glycolytic stress test was performed using the Seahorse XFe96 analyzer, by measuring the extracellular acidification rate (ECAR) of MCF7 cells treated with XCT790. Notably, XCT790 also significantly reduces ECAR, a marker of glycolysis (Figure 6A). Finally, plotting the ratio of OCR to ECAR demonstrates that XCT790 shifts MCF7 cells from a highly energetic to a metabolically quiescent state (Figure 6B).

16XCT790 is an inverse agonist of ERRα. As XCT790 inhibits mammosphere formation, we then asked if ERRα expression is sufficient to promote 3D-spheroid formation in MCF7 cells. To this end, ERRα was over-expressed in MCF7 cells using a lentiviral approach. Empty Vector (EV) cells were generated in parallel. Figure 7A shows that MCF7 cells over-expressing ERRα show a 50% increase in mammosphere forming capacity. Notably, MCF7 cells over-expressing ERRα show increased mitochondrial membrane potential, as assessed by MitoTracker Orange, as well as increased mitochondrial mass, as assessed by MitoTracker Green and MitoTracker Deep Red, as expected (Figure 7B).

17We then used a pharmacological approach to inhibit ERRα-driven mammosphere formation, using oligomycin A, a mitochondrial inhibitor, as well as XCT790, a specific ERRα inverse agonist. Importantly, mammosphere formation induced by ERRα expression was inhibited by either treatment with oligomycin A (Figure 7C), or XCT790 (Figure 7D). These results indicate that ERRα activity and mitochondrial function are normally required for the clonal expansion of TICs.

18To examine if mitochondrial function is essential for TIC expansion driven by well-established stem-like signaling, we decided to overexpress FOXM1 in MCF7 cells. Empty Vector (EV) control cells were generated in parallel. The transcription factor FOXM1 was recently shown to induce the expansion of human normal stem cells, as well as of cancer stem-like cells. Mechanistically, it is believed that FOXM1 acts downstream of key signaling pathways essential for stem cell regulation and tumorigenesis, such as Wnt/β-catenin and 14-3-3ζ signaling cells.

19MCF7 cells over-expressing FOXM1 show a 3.3-fold increase in mammosphere formation, relative to empty-vector controls, as expected (Figure 8A). In order to better dissect the molecular changes driven by FOXM1, MCF7 cell monolayers over-expressing FOXM1 and EV control cells were subjected to proteomics analysis. Proteomic datasets of up-regulated proteins in the MCF7-FOXM1 cells were then analyzed with respect to their metabolic protein profiles.

20Table 1 shows that several key molecules related to mitochondria, glycolysis and the EMT are up-regulated in FOXM1-over-expressing MCF7 cells, indicating that FOXM1-induced stemness is associated with increased metabolic flexibility. Notably, keratin-19 (KRT19), a well-established marker of circulating tumor cells and CSCs, is increased by >20-fold in MCF7-FOXM1 cells. Also, four key mitochondrial proteins (HSPD1, ACADVL, ATP5B, COX4I1) are increased by >10-fold each in MCF7-FOXM1 cells; SLC25A4 (the mitochondrial ATP/ADP exchanger) is increased by nearly 6-fold. Similarly, twenty-two other mitochondrial proteins are increased by >1.5 fold.

21Moreover, Table 2 shows that several key molecules related to protein synthesis are also up-regulated in FOXM1-overexpressing MCF7 cells. We have previously shown that protein synthesis is a mechanism for enhancing the proliferation of TICs, and that known inhibitors of protein synthesis, such as puromycin and rapamycin, are very effective at reducing mammosphere formation. Similarly, the EMT is also regarded as a key feature of TICs. Consistent with the notion that FOXM1 drives stemness, ribosome-related proteins and EMT markers are increased by FOXM1 expression (Tables 1 and 2).

22To test if mitochondrial inhibitors could block FOXM1-driven mammosphere formation, we then used several independent pharmacological approaches. Interestingly, formation of FOXM1-driven mammospheres was inhibited by treatment with oligomycin A (Figure 8B), XCT790 (Figure 8C), and doxycycline (Figure 8D). Doxycycline is an FDA-approved antibiotic that we have recently shown to inhibit proliferation of TICs, by targeting their mitochondrial ribosomes. Thus, mitochondrial protein translation and function are required for FOXM1-driven mammosphere formation.

23Finally, we asked if mitochondrial function is required for 3D-spheroid formation of other cell lines, such as H295R adrenocortical carcinoma cells. Figure 9 shows that XCT790 dose-dependently inhibits 3D-spheroid formation of the H295R adrenocortical carcinoma cell line, with an IC-50 of 10 μM. These results indicate that mitochondrial function is required for the 3D-spheroid formation of cell lines other than MCF7 breast cancer cells.

24Quantitatively similar results were obtained when XCT790 was tested on several other epithelial cancer cell lines, indicating that the ability of XCT790 to inhibit 3D-spherioid formation reflects a general property of CSCs (data not shown).

25To assess the possible clinical relevance of our results, we also determined whether the proteomic targets that we identified in MCF7-FOXM1 cells were transcriptionally over-expressed in human breast cancer cells in vivo. Towards this end, we exploited a clinical data set of tumor samples from 28 breast cancer patients. These tumor samples were subjected to laser-capture micro-dissection, to separate epithelial cancer cells from adjacent tumor stroma.

26Tables 3 and 4 present a summary of these findings. Overall, greater than fifty FOXM1 targets (related to mitochondria, glycolysis, the EMT, and protein synthesis) that we identified in MCF7-FOXM1 cells were also transcriptionally elevated in human breast cancer cells in vivo. As such, the new FOXM1 protein targets that we identified in MCF7-FOXM1 cells may be especially relevant for improving human breast cancer diagnosis and therapy.

27Here, we specifically tested the hypothesis that new mitochondrial biogenesis is required for the survival and propagation of stem-like cancer cells. For this purpose, we used an investigational inhibitor of the ERRα-PGC1α/β signaling pathway, namely XCT790, to interrogate the role of mitochondrial biogenesis and function in the anchorage-independent growth of CSCs. Importantly, treatment with XCT790 blocked mammosphere formation and prevented anoikis-resistance in CD44(+)high/CD24(−)low MCF7 cells. Mechanistically, XCT790 inhibited mitochondrial function (OCR) and effectively reduced signaling along a number of classical stem cell transduction pathways, such as the Hedgehog/GLI, TGFβ/SMAD, and Wnt/β-catenin signaling. Overexpression of either ERRα or FOXM1 (a down-stream target of Wnt-signaling) in MCF7 cells significantly augmented mammosphere formation, which could be prevented by a number of mitochondrial inhibitors. In this regard, doxycycline treatment was sufficient to overcome the effects of FOXM1 on mammosphere formation. Interestingly, doxycycline is an FDA-approved antibiotic that has been used for nearly 50 years to treat a wide variety of bacterial and parasitic infections, without significant side effects. Finally, unbiased proteomics analysis of MCF7-FOXM1 cells revealed the up-regulation of specific mitochondrial proteins, glycolytic enzymes, EMT markers and components of the protein synthesis machinery. Importantly, these FOXM1 target proteins were also transcriptionally up-regulated in patient samples in vivo, in human breast cancer epithelial cells isolated by laser-capture micro-dissection, highlighting their clinical relevance.

28Many other recent studies have also directly implicated augmented mitochondrial function in resistance to chemotherapy and radiation, in the survival of treatment-resistant cancer stem-like cells, in the propagation and motility of circulating tumor cells, as well as in tumor growth and cancer cell metastasis in pre-clinical animal models in vivo.

29Consistent with our current findings, we have previously shown that a PGC1/NRF1 gene signature predicts tumor recurrence, metastasis and poor overall survival in ER(+)/Luminal-A breast cancer patients. This PGC1/NRF1 gene signature was also elevated in >2,000 human tumors excised from breast cancer patients, including both ER(+) and ER(−) cases. Moreover, recombinant over-expression of PGC1α/β, or other genes that promote mitochondrial biogenesis (MitoNEET/POLRMT), in MDA-MB-231 cells, was indeed sufficient to functionally increase tumor growth by up to ~3-fold.

30Asymmetric cell division is an essential characteristic of stem cells. Interestingly, Sabatini, Weinberg and colleagues examined the segregation of “newly-synthesized” and “older” mitochondria in immortalized mammary epithelial cells, during asymmetric cell division, a property that is also a characteristic of CSCs. They observed that “newly-synthesized” mitochondria were preferentially enriched in stem cells during asymmetric cell division, while “old” mitochondria were segregated into the daughter cells. These results imply that asymmetric cell division in stem cells somehow requires new mitochondrial biogenesis, for the propagation and maintenance of the stem cell phenotype.

31As such, these findings may help to mechanistically explain our current results that new mitochondrial biogenesis is required for the efficient propagation of and survival of CSCs (Figure 10), as we have seen using the mammosphere and anoikis-resistance assays, to measure stem cell activity. Taken together, these data implicate mitochondrial biogenesis and mitochondrial function as critical targets for new drug discovery, to overcome tumor recurrence, distant metastasis and drug-resistance, especially in cancer patients with clinically advanced disease.

32MCF7 breast cancer cells were purchased from the ATCC. H295R adrenocortical carcinoma cells were a generous gift of Dr. Antonio Stigliano (University of Rome). Gibco-brand cell culture media (DMEM/F12) was purchased from Life Technologies. XCT790 was purchased from Tocris, oligomycin A and doxycycline were purchased from Sigma-Aldrich. Lentiviral vectors for the expression of ERRα (#Z1441) and FOXM1 (#U1376) were obtained commercially from Genecopoeia, along with the appropriate empty vector controls, in the Lv-105 (PuroR) vector system.

33A single cell suspension was prepared using enzymatic (1x Trypsin-EDTA, Sigma Aldrich, #T3924), and manual disaggregation (25 gauge needle) to create a single cell suspension. Cells were plated at a density of 500 cells/cm2 in mammosphere medium (DMEM-F12/B27/EGF(20ng/ml)/Pen-Strep) in non-adherent conditions, in culture dishes coated with (2-hydroxyethylmethacrylate) (poly-HEMA, Sigma, #P3932). Cells were grown for 5 days and maintained in a humidified incubator at 37°C at an atmospheric pressure in 5% (v/v) carbon dioxide/air. After 5 days for culture, spheres >50 μm were counted using an eye piece graticule, and the percentage of cells plated which formed spheres was calculated and is referred to as percentage mammosphere formation, and was normalized to one (1 = 100 %MSE, mammosphere forming efficiency). For pharmacological inhibition of mammosphere formation, cells were directly seeded on low-attachment plates in the presence of XCT790 or oligomycin or doxycycline at the indicated concentrations.

34After incubation with XCT790 (5 or 10 μM for 48 hours), MCF7 cell monolayers were trypsinized and seeded for mammosphere cultures for 5 days, without XCT790.

35Lentiviral plasmids, packaging cells and reagents were from Genecopoeia. 48 hours after seeding, 293Ta packaging cells were transfected with lentiviral vectors encoding ERRα, FOXM1 or empty vector (EX-NEG-Lv105), using Lenti-PacTM HIV Expression Packaging Kit according to the manufacturer's instructions. Two days post-transfection, lentivirus-containing culture medium was passed through a 0.45 μm filter and added to the target cells (MCF7 cells) in the presence of 5μg/ml Polybrene. Infected cells were selected with a concentration of 1.5 μg/ml of puromycin.

36Following XCT790 treatment for 48 hours, the TIC population was enriched by seeding on low-attachment plates. Under these conditions, the non-TIC population undergoes anoikis (a form of apoptosis induced by a lack of cell-substrate attachment) and TICs are believed to survive. The surviving TIC fraction was analyzed by FACS analysis. Briefly, 1 × 104 MCF7 monolayer cells were treated with XCT790 (5μM and 10 μM) for 48h in 6-well plates. Then, cells were trypsinized and seeded in low-attachment plates in mammosphere media. After 10h, MCF7 cells were spun down and incubated with CD24 (IOTest CD24-PE, Beckman Coulter) and CD44 (APC mouse Anti-Human CD44, BD Pharmingen cat.559942) antibodies for 15 minutes on ice. Cells were rinsed twice and incubated with LIVE/DEAD dye (Fixable Dead Violet reactive dye; Invitrogen) for 10 minutes. Samples were then analyzed by FACS (Fortessa, BD Bioscence). Only the live population, as identified by the LIVE/DEAD dye staining, was analyzed for CD24/CD44 expression. Data were analyzed using FlowJo software.

37Cell viability was assessed by sulphorhodamine (SRB) assay, based on the measurement of cellular protein content. After treatment with XCT790 for 3 or 5 days in 96 well plates, cells were fixed with 10% trichloroacetic acid (TCA) for 1h in cold room, and dried overnight at room temperature. Then, cells were incubated with SRB for 15 min, washed twice with 1% acetic acid, and air dried for at least 1h. Finally, the protein-bound dye was dissolved in 10 mM Tris pH 8.8 solution and read using the plate reader at 540 nm.

38The Cignal Lenti reporter assay (luc) system (Qiagen) was chosen for monitoring the activity of several signaling pathways in MCF7 cells. The responsive luciferase constructs encode the firefly luciferase reporter gene under the control of a minimal (m)CMV promoter and tandem repeats of response elements for each pathway. The following constructs were used: TCF/LEF(luc) for Wnt signaling (CLS-018L); STAT3(luc) for transcriptional activity of STAT3 (CLS-6028L); RBP-Jk(luc) for Notch-induced signaling (CLS-014L); ARE(luc) for Nrf2- and Nrf1-mediated antioxidant response (CLS-2020L); GAS(luc) for IFNγ-induced Stat1-signaling (CLS-009L); ISRE(luc) for (IFN)-α/β-STAT1/2 signaling (CLS-008L); SMAD(luc) for TGFβ-induced signaling (CLS-017L); GLI(luc) for Sonic hedgehog signaling (CCS-6030L). Briefly, 1 × 105 MCF7 cells were seeded in 12-well plates. Once cells were attached, the viral particles were diluted 1:10 in complete culture media containing polybrene (sc-134220, Santa Cruz), and added to the cells. Puromycin treatment (#P9620, Sigma) was started 48 hours later in order to select stably infected cells.

39Luciferase Assay System (E1501, Promega) was performed in all luciferase reporter MCF7 cells treated with XCT790. Briefly, 6 × 103 MCF7 cells were seeded in black-walled 96-well plates and then were treated with XCT790 (10 μg/ml). As control, vehicle-treated cells were run in parallel. Four replicates were used for each condition. After 48 hours of treatment, luciferase assay was performed according to the manufacturer's instructions. Light signal was acquired for 2 minutes in photons/second in the Xenogen VivoVision IVIS Lumina (Caliper Life Sciences), and the results were analysed using the Living Image 3.2 sofware (Caliper Life Sciences). Luminescence was normalized using total proteins, as assessed with the Bradford protein assay.

40To measure mitochondrial activity, cells were stained with MitoTracker Orange (#M7510, Invitrogen), whose accumulation in mitochondria is dependent upon membrane potential. To measure mitochondrial mass, cells were stained with MitoTracker Green (#M7514 Invitrogen), or MitoTracker Deep Red (#M22426, Invitrogen), both localizing to mitochondria regardless of mitochondrial membrane potential. Briefly, MCF7 cells were treated with XCT790 for 48 hours. Cells were then incubated with pre-warmed MitoTracker staining solution (diluted in PBS/CM to a final concentration of 10 nM) for 30-60 min at 37 °C. All subsequent steps were performed in the dark. Cells were washed in PBS, harvested, and re-suspended in 300 μL of PBS. Cells were then analyzed by flow cytometry. Data analysis was performed using FlowJo software.

41Extracellular acidification rates (ECAR) and real-time oxygen consumption rates (OCR) for MCF7 cells treated with XCT790 or vehicle alone control were determined using the Seahorse Extracellular Flux (XF96) analyzer (Seahorse Bioscience, MA, USA). MCF7 cells were maintained in DMEM supplemented with 10% FBS (fetal bovine serum), 2 mM GlutaMAX, and 1% Pen-Strep. 7,000 cells were seeded per well into XF96-well cell culture plates, and incubated overnight at 37°C in a 5% CO2 humidified atmosphere. After 24h, cells were treated with XCT790 (5μM or 10 μM) for 48h. After 48h of treatment, cells were washed in pre-warmed XF assay media (for OCR measurement, XF assay media was supplemented with 10mM glucose, 1mM Pyruvate, 2mM L-glutamine and adjusted at pH 7.4). Cells were then maintained in 175 μL/well of XF assay media at 37°C, in a non-CO2 incubator for 1h. During the cell incubation time, we loaded 25 μL each of 80mM glucose, 9μM oligomycin, 1M 2-deoxyglucose (for ECAR measurement) and 25 μL each of 10μM oligomycin, 9μM FCCP, 10μM rotenone, 10μM antimycin A (for OCR measurement), in XF assay media into the injection ports in the XFe-96 sensor cartridge. ECAR and OCR measurements were normalized by protein content. Data set was analyzed by XFe-96 software and GraphPad Prism software, using one-way ANOVA and Student's t-test calculations. All experiments were performed in quintuplicate, three times independently, such that each data point represents the average of 15 replicates.

42Cell lysates were prepared for trypsin digestion by sequential reduction of disulphide bonds with TCEP and alkylation with MMTS. Then, the peptides were extracted and prepared for LC-MS/MS. All LC-MS/MS analyses were performed on an LTQ Orbitrap XL mass spectrometer (Thermo Scientific, San Jose, CA) coupled to an Ultimate 3000 RSLCnano system (Thermo Scientific, formerly Dionex, The Netherlands). Xcalibur raw data files acquired on the LTQ-Orbitrap XL were directly imported into Progenesis LCMS software (Waters Corp., Milford, MA, formerly Non-linear dynamics, Newcastle upon Tyne, UK) for peak detection and alignment. Data were analyzed using the Mascot search engine. Five replicates were analyzed for each sample type (N = 5). Statistical analyses were performed using ANOVA and only fold-changes in proteins with a p-value less than 0.05 were considered significant.

43To firmly establish the clinical relevance of our results from the quantitative proteomics analysis of mammosheres, we re-analyzed the transcriptional profiles of epithelial breast cancer cells and adjacent tumor stromal cells that were physically separated by laser-capture microdissection (from N = 28 human breast cancer patients).