Background
The mammalian target of rapamycin (mTOR) pathway, hyperactive in numerous cancer types including breast cancer, is an attractive therapeutic target. Disappointingly, mTOR inhibitors only show clinical benefit in selected settings and efficacy is limited. Moreover, toxicity, including fatigue and mucositis limit clinical use [
1]. mTOR signaling is central in the integration of cellular signals involved in growth and cellular energy status [
2]. Therefore, the metabolic context of mTOR inhibition in cancer cells is essential for understanding and improving its anti-tumor effects and toxicity profile.
The mTOR protein is the catalytic subunit of two structurally and functionally different protein complexes: mTORC1 and mTORC2. mTOR complex 1 (mTORC1) is sensitive to growth factor signaling, oxygen levels and nutrient availability. Downstream, mTORC1 inhibits the transcriptional repressor eukaryotic initiation factor 4B binding protein (4EBP1), and activates S6 ribosomal protein (S6), leading to expression of proteins essential for the regulation of cell growth. mTOR complex 2 (mTORC2) regulates AKT activity through phosphorylation and is involved in cell survival and proliferation. Moreover, mTORC2 induces expression of glycolytic enzymes, pentose phosphate pathway enzymes and glutaminase and increases cellular lipogenesis [
3]. Everolimus, the most commonly used mTOR inhibitor, directly inhibits mTORC1, but also (indirectly) inhibits mTORC2 [
4,
5]. This mTORC2 inhibition may underlie the induction of hyperglycemia in a large proportion of patients treated with everolimus [
6,
7]. High glucose levels can stimulate tumor growth in patients and are associated with resistance to breast cancer chemotherapy [
8,
9]. It is currently unknown whether hyperglycemia counteracts anti-proliferative effects of everolimus. Cancer patients on everolimus treatment are regularly treated with anti-diabetic drugs, especially metformin, to reduce glucose levels. Metformin is a widely prescribed, well-tolerated, effective treatment for type 2 diabetes mellitus. Moreover, epidemiological evidence and retrospective clinical data indicate, that metformin has intrinsic anti-cancer properties [
10,
11]. At the cellular level, metformin inhibits complex I of the mitochondrial respiratory chain leading to compensatory increases in glycolytic flux and activated AMP-activated kinase (AMPK) [
12]. This results in growth inhibition of tumor cells through inhibition of mTOR, cell cycle arrest, activation of autophagy and possibly apoptosis [
13]. Thus, everolimus and metformin both inhibit mTOR signaling and, moreover, differentially target tumor cell glucose metabolism.
We hypothesized that the combination of everolimus and metformin would synergistically inhibit cell growth in a glucose concentration dependent manner. To test this hypothesis and predict potential clinical value of the combination, culture conditions optimally reflecting in-vivo tumor metabolic circumstances are required. Strikingly, in most in vitro studies, media containing up to 25 mM glucose are used. This is 4–5-fold higher than the mean fasting blood serum glucose levels of healthy individuals. Additionally, poorly vascularized areas of tumors may have even lower glucose concentrations and hypoxia may be present. In the present study, we therefore investigated the growth inhibitory effects and underlying signal transduction and metabolic mechanisms of everolimus and metformin treatment alone, and in combination, at physiological glucose concentrations in hypoxic and normoxic conditions in breast cancer cell lines.
Methods
Reagents and cell culture
Everolimus (Sigma-Aldrich, Zwijndrecht, The Netherlands) was dissolved in dimethyl sulfoxide (DMSO) to a concentration of 20 mM and diluted in phosphate buffered saline (PBS, 0.14 M NaCl, 2.7 mM KCl, 6.4 mM Na
2HPO
4.2H
2O, 1.5 mM KH
2PO
4, pH 7.2–7.5) prior to use. Metformin (Sigma-Aldrich, Zwijndrecht, The Netherlands) was dissolved to a concentration of 1 M in PBS and stored at −20 °C until use. The human tumor cell lines used were purchased from the American Type Culture Collection (ATCC, Manassas, USA). The luminal A MCF-7 (catalog number HTB-22) and luminal A T47D (catalog number HTB-133) breast cancer cells were cultured in RPMI containing 11 mM glucose, supplemented with 10% FCS at 37 °C in 5% CO
2. Triple negative MDA-MB-231 breast cancer cells (catalog number HTB-26) [
14] were cultured in DMEM containing 11 mM glucose, supplemented with 10% fetal calf serum (FCS) and 1 mM glutamine at 37 °C in 5% CO
2. Cultures in 5.5 mM glucose were maintained by adding the appropriate amount of glucose-free RPMI/DMEM to standard RPMI (all Gibco Thermo Fisher Scientific, Bleiswijk, The Netherlands). Glucose concentrations in cell culture media were measured using the Accu-Chek Aviva glucose meter (Roche, Almere, The Netherlands). Accuracy of measurements of glucose concentrations in cell culture media was confirmed using a calibration curve constructed using fresh culture medium with known glucose concentrations. The detection limit of the Accu-Check is 0.6 mM glucose. Experiments using 2.75 mM glucose in the cell culture media were performed using cells that were cultured in 5.5 mM glucose and were prepared in 2.75 mM glucose containing medium 24 h before the start of the experiment. For hypoxia experiments, cells were placed in an incubator with 1% oxygen and 5% CO
2 after the addition of reagents.
Viability assay and colony survival assay
For the viability assays MCF7, T47D and MDA-MB-231 cells were plated at a density of 2000, 2500 or 3000 cells per well, respectively, in 96 wells plates (4 wells/condition) and subsequently incubated with metformin and everolimus at the desired concentrations for 4 days in the same culture medium, that was also used for cell culture. For MCF7 and T47D RPMI-media containing 11 or 2.75 mM glucose was used. For MDA-MB-231 DMEM containing 11 or 2.75 mM glucose was used. After 4 days 20 μl 3-(4,5-dimethylthiazol-2-yl)-2,5-diphenyltetrazolium bromide solution (5 mg/ml in PBS) was added to each well. After 4 h of incubation formazan crystals were dissolved in 200 μl DMSO and absorption at 520 nm wavelength was determined with a plate reader (iMark, BioRad, Veenendaal, The Netherlands). No major effects of metformin on the relationship between cell numbers and MTT conversion were observed. For each experiment MTT results were visually checked by light microscopy. For the colony survival assay cells were plated in 6-wells plates. 250 cells/well were plated and allowed to adhere for at least one hour before treatment. When glucose was replenished, 2.75 mM glucose was added every other day for in total 3 times to achieve a total amount of usable glucose of 11 mM during the course of the experiment. Pilot data demonstrated that this procedure ensured the presence of relatively stable glucose levels during the course of the drug treatment. After 8 days of treatment, cells were fixed and stained with Coomassie blue. Colonies consisting of at least 50 cells were counted.
Western blotting analysis
MCF-7 and MDA-MB-231 cells were lyzed in MPER (Thermo Scientific, Bleiswijk, The Netherlands) and diluted 1:1 with SDS sample buffer (4% SDS, 20% glycerol, 0.5 mol/l Tris-HCl (pH 6.8), 0.002% bromophenol blue). Lysates were resolved by SDS-PAGE and transferred to PVDF membranes. Membranes were incubated overnight at 4 °C and probed with the following antibodies: rabbit-anti-AKT, rabbit-anti-pAKT (Thr308), rabbit-anti-S6, rabbit-anti-pS6, rabbit-anti-4EBP1 (all Cell Signaling Technologies, Leiden, The Netherlands) in a 1:1000 dilution or anti-HIF1α (BD Biosciences, Breda, The Netherlands) and mouse-anti-actin (MP Biomedicals, Santa Ana, USA) in a 1:10,000 dilution. Primary antibodies were stained using HRP-coupled goat anti-rabbit or rabbit anti-mouse IgG and developed with Lumi-Light (Roche, Almere, The Netherlands). Images were captured with the ChemiDoc MP imaging system (Bio-Rad, Veenendaal, The Netherlands) and Image Lab Software.
Quantification of autophagy, reactive oxygen species (ROS), and cell death
MCF-7 and MDA-MB-231 cells were transfected with a GFP-LC3 containing retrovirus (kindly provided and developed by H Folkerts, Department of Experimental Hematology, University Medical Centre Groningen, the Netherlands). Upon upregulation of autophagy the LC3-GFP protein forms aggregates that can be visualized using fluorescence microscopy. Bafilomycin A1, a known inhibitor of the late phase of autophagy, efficiently blocks turnover of autophagic vesicles, thereby increasing LC3-GFP foci. GFP-LC3 expressing MCF-7 and MDA-MB-231 were grown on cover slips and treated with metformin, everolimus and 20 nM bafilomycin (Sigma-Aldrich, Zwijndrecht, The Netherlands) for the indicated duration. Cells were washed with cold PBS and fixed with 3.7% paraformaldehyde. Cover slips were mounted on glass plates using Kaiser’s mounting medium. Fluorescent GFP-LC3 foci per individual cell were counted. Moreover, cleavage of the LC3 protein was determined using Western Blotting with an anti-LC3 antibody (Cell Signaling Technology, Leiden, The Netherlands). ROS measurement was performed using H2DCF (Sigma-Aldrich, Zwijndrecht, The Netherlands). Hydrogen peroxide treated cells were used as a positive control. After harvesting by trypsinization, cells were washed once with PBS and subsequently incubated with 10 μM H2DCF for 30 min at 37 °C. Samples were washed with cold PBS and analyzed using a FACSCalibur (Becton Dickinson, Breda, The Netherlands). Analysis was performed using Flowing software 2.5 (Informer Technologies, Inc).
Four days prior to cell death measurements, cells were plated at the desired density, treated with metformin and everolimus and supplemented with 2.75 mM glucose (1 M stock solution) each day. On the day of analysis, cells were harvested by trypsinization and washed once in calcium-buffer. Cells were subsequently incubated in a 1:12 dilution of annexin V-FITC antibody (IQ products, Groningen, The Netherlands) in calcium buffer for 20 min on ice. Samples were washed with calcium-buffer and resuspended in calcium-buffer containing 0.5 μg/ml propidium iodide (PI). Cells were analyzed immediately using a FACSCalibur (Becton Dickinson, Breda, The Netherlands). Analysis was performed using Flowing Software 2.
Quantification analyses of mitochondrial respiration and glycolysis
Mitochondrial and glycolytic function of MCF7 and MDA-MB-231 cells was determined using a Seahorse XF24 Extracellular Flux Analyzer (Seahorse Bioscience, North Billerica, USA). Cells were seeded with an appropriate density in specialized V7 Seahorse tissue culture plates (3 wells/condition). After 2 days cells were treated with indicated concentrations of metformin, everolimus or a combination and incubated for another 2 days. On the day of the measurements, cells were washed once with PBS and once with unbuffered 1 mM sodium pyruvate containing XF assay medium (pH 7.4) and 11 mM or 2.75 mM glucose, respectively. The assay commenced after cells had been incubated in 500 μl unbuffered XF assay medium (pH 7.4) for 1 h. Baseline oxygen consumption rate (OCR) and extracellular acidification rate (ECAR) were determined. To gather detailed information about the mitochondrial and glycolytic function of the cell lines MCF-7 and MDA-MB-231 in response to treatment with metformin and everolimus a mitochondrial stress test was performed. Using the ATP-synthase inhibitor oligomycin, the mitochondrial uncoupler carbonyl cyanide 4-(trifluoromethoxy)phenylhydrazone (FCCP) the complex I inhibitor rotenone and the cytochrome C reductase inhibitor antimycin A (all Sigma-Aldrich, Zwijndrecht, The Netherlands) a detailed profile of basal respiration, maximal respiration and induction of glycolysis can be gathered. Three technical replicates were performed per sample. OCR and ECAR were normalized for the amount of cellular protein in each well using the seahorse XF24 software. Protein amount was determined using the Bradford assay. The three measurements of each step of this mitochondrial stress test were combined for analysis.
Statistical analyses
Data are presented as mean ± standard deviation (SD). Different experimental conditions were compared using unpaired Student’s t-tests. Statistical analyses were performed using Prism v.5 (GraphPad). A P-value of <0.05 (two-tailed) was considered significant.
Discussion
In the present study, we show that everolimus and metformin both inhibit mTOR activity and have additive inhibitory effects on glucose metabolism, tumor cell growth and colony formation. These effects are evident in high and low glucose conditions and not reduced in the presence of hypoxia. These results support further in vivo investigation of everolimus combined with metformin as a putative anti-cancer therapy.
We found that the inhibitory effects of metformin on growth and colony formation of breast cancer cells were additive to the effects of everolimus in high and low glucose conditions, even when relatively low concentrations of both drugs were used. A previous study with different mutant p53 breast cancer cell lines cultured in high glucose media, demonstrated efficacy of metformin even at lower concentrations in both MTT and mammosphere assays, while higher concentrations of everolimus were required compared to our study [
19]. Wang et al. also showed in vivo efficacy of the combination in xenograft bearing mice. Metformin sensitivity has been related to the presence of mutant p53 [
20] and everolimus sensitivity to the presence of wild type p53 [
21]. In our cell line panel, everolimus was indeed effective in wild-type p53 cells (MCF7) and less effective in mutant p53 cells (MDA-MB231 and T47D), but the preferential sensitivity of metformin in mutant p53 cells was not observed. Thus, more studies are required to investigate metformin and everolimus sensitivity in relation to the p53 status in breast cancer models. Interestingly, the combination of everolimus and metformin effectively inhibited colony and mammosphere forming capacity of wild type and mutant p53 breast cancer cells [Fig.
6c], [
19]. These results suggest that tumor initiating cells are also sensitive to this combination in addition to bulk tumor cells as measured in the MTT assay, making this combination even more attractive to be further explored in breast cancer.
The inhibitory effect on mTOR has been described for each drug individually [
22]. Here, we demonstrate that mTOR signaling is additively inhibited by the combination treatment. For everolimus it was expected that reduced mTOR activation would lead to reduced transcription of glycolytic enzymes and therefore a shift to mitochondrial respiration [
23]. In contrast, we observed that everolimus inhibits mitochondrial respiration, which was not compensated by an enhanced glycolysis rate, as previously reported for hepatocellular carcinoma cells as well [
24]. In human pancreatic cancer cell lines, however, everolimus treatment reduced the rate of glycolysis. Unfortunately, the effect on mitochondrial respiration was not reported [
25]. At a mechanistic level, it has been shown that mTOR stimulates translation of mitochondrial mRNAs by inhibiting 4EBPs [
26]. Consequently, mTORC1 inhibition leads to less mitochondrial mRNA translation and less mitochondrial respiration, which is in agreement with our results that everolimus treatment resulted in reduced p-S6 levels and mitochondrial respiration in both MCF7 and MDA-MB231 cells. As expected treatment with the mitochondrial complex I inhibitor metformin inhibited mitochondrial respiration and induced glycolysis [
27‐
29]). Metformin treated cells had also a reduced maximal respiratory capacity. This might be an indirect effect of the reduced mTORC1 activity caused by metformin treatment. The additive metabolic effect of everolimus combined with metformin has not been described before. Our metabolic measurements show that a combination of both drugs, even at relatively low concentrations, completely reduced p-S6 and disrupted mitochondrial respiration. Moreover, similar results were obtained in high and low glucose containing media.
We demonstrate that metformin can induce a metabolic shift to increased glycolysis. The increased glucose utilization in the presence of metformin, especially under low glucose conditions, results in an earlier onset of glucose starvation and more cell death in MCF7 and MDA-MB231 breast cancer cells. This finding is in accordance with previous reports using glucose-free culture conditions [
28,
30‐
32]. Here, we show that under stable low glucose conditions, thus preventing glucose starvation by replenishment of glucose, metformin treatment still results in growth inhibition, but cell death does not occur. Since glucose concentrations are relatively stable in vivo, our model under stable low glucose conditions in this respect is likely to reflect the situation in tumors in vivo [
33,
34]. This is indirectly supported by in vivo observations showing that tumors from metformin treated patients did not have increased numbers of apoptotic cells compared to placebo treated patients [
35]. Previous in vitro studies in breast cancer cell line models identified apoptosis as a mechanism of metformin’s anti-cancer effects. Unfortunately, glucose concentrations were not measured [
22,
36,
37]. Our results strongly suggest that these findings were likely to be caused by in vitro glucose starvation due to metformin-induced increases in glucose utilization.
Other tumor-microenvironmental factors such as intratumoral pH, glutamine concentration, and oxygen tension are likely to influence the in vivo efficacy of metformin [
38,
39]. Like metformin treatment, hypoxia shifts cellular metabolism towards glycolysis. This effect could either lead to synergy or reduce metformin effects, as shown in a sarcoma cell line model [
39]. However, the sarcoma cell line study CoCl
2 was used as a hypoxia mimetic in a short-term assay. We demonstrate that although in in vitro short-term experiments the effects of metformin in hypoxic conditions appeared to be reduced, longer-term assays showed that metformin activity is maintained. In the short-term setting, hypoxia alone already decreased cell proliferation, so that possible effects of metformin on proliferation were harder to detect. Indeed, we found that p-S6 had almost completely disappeared under hypoxic, low glucose conditions suggesting that these cells were not actively proliferating. During the long-term clonogenic assay, cells sufficiently adapted to hypoxia and formed colonies. Metformin was effective against these colonies under hypoxic conditions, suggesting that metformin will retain its efficacy in the hypoxic areas of an in vivo tumor.
Blood plasma levels of metformin in patients (around 0.05 mM) are at least twenty-fold lower than those used in most in vitro cell line studies (1–40 mM) [
40]. Although there is evidence that the acidophilic properties of metformin may cause accumulation in the mitochondrial matrix, thereby decreasing the systemic levels required, the low concentrations which potentiate everolimus in our model are promising [
41].
Treatment of cancer cells with inhibitors of both glycolysis and mitochondrial respiration, leading to synthetic metabolic lethality is a promising therapeutic strategy to kill cancer cells. Combination of metformin with the glycolytic inhibitor dichloroacetate effectively induced cell death in breast and ovarian cancer cell lines [
27]. Also in vivo combination treatment with the mitochondrial inhibitor phenformin and the lactate dehydrogenase inhibitor oxamate has been successful [
42]. Combining metformin with everolimus is a novel example of dual metabolic targeting.
Acknowledgements
Not applicable.