Metformin directly acts on mitochondria to alter cellular bioenergetics
© Andrzejewski et al.; licensee BioMed Central Ltd. 2014
Received: 13 June 2014
Accepted: 25 July 2014
Published: 28 August 2014
Skip to main content
© Andrzejewski et al.; licensee BioMed Central Ltd. 2014
Received: 13 June 2014
Accepted: 25 July 2014
Published: 28 August 2014
Metformin is widely used in the treatment of diabetes, and there is interest in ‘repurposing’ the drug for cancer prevention or treatment. However, the mechanism underlying the metabolic effects of metformin remains poorly understood.
We performed respirometry and stable isotope tracer analyses on cells and isolated mitochondria to investigate the impact of metformin on mitochondrial functions.
We show that metformin decreases mitochondrial respiration, causing an increase in the fraction of mitochondrial respiration devoted to uncoupling reactions. Thus, cells treated with metformin become energetically inefficient, and display increased aerobic glycolysis and reduced glucose metabolism through the citric acid cycle. Conflicting prior studies proposed mitochondrial complex I or various cytosolic targets for metformin action, but we show that the compound limits respiration and citric acid cycle activity in isolated mitochondria, indicating that at least for these effects, the mitochondrion is the primary target. Finally, we demonstrate that cancer cells exposed to metformin display a greater compensatory increase in aerobic glycolysis than nontransformed cells, highlighting their metabolic vulnerability. Prevention of this compensatory metabolic event in cancer cells significantly impairs survival.
Together, these results demonstrate that metformin directly acts on mitochondria to limit respiration and that the sensitivity of cells to metformin is dependent on their ability to cope with energetic stress.
The biguanide metformin is well established as an important drug in the treatment of type II diabetes [1–3]. Pharmaco-epidemiologic evidence [4, 5] and laboratory models [6, 7] have suggested that metformin may have antineoplastic actions, and this has led to renewed interest in the molecular actions of the drug . One popular view is that metformin acts as an inhibitor of complex I of the electron transport chain. However, the notion that metformin acts directly on mitochondria to inhibit complex I is controversial [9–15]. Recent work on the sensitivity of cancer cells to the direct actions of metformin further highlighted the controversy surrounding the mode of action of metformin. These studies demonstrate that cancer cells that are deficient in mitochondrial functions (rho0 cells) are sensitive to the action of metformin , and that cancer cells harboring complex I mutations are more sensitive to the action of metformin compared with cancer cells without these mutations .
While there is controversy regarding the molecular mechanisms underlying the action of metformin, there is a general agreement that the drug causes energetic stress, and that this results in a variety of cell lineage-specific secondary effects. The liver is an important target organ in the context of diabetes. This organ is exposed to a relatively high concentration of metformin via the portal circulation following oral ingestion, and hepatocytes express high levels of membrane transporters required for drug influx . Metformin-induced hepatocyte energetic stress leads to a reduction in gluconeogenesis [18–20], leading to improvements in hyperglycemia and hyperinsulinemia. These metabolic actions also represent a candidate mechanism relevant to the subset of cancers that are insulin-responsive . Recent work has indicated that metformin treatment alters the hepatocellular redox state by inhibiting mitochondrial glycerophosphate dehydrogenase .
Understanding the actions of metformin on energy metabolism, particularly on mitochondrial functions, is important in the context of interest in ‘repurposing’ the compound for possible applications in oncology. There is increasing evidence that mitochondrial metabolism plays an important role in supporting tumor growth, by providing ATP as well as metabolic intermediates that can be used for anabolic reactions . Also, functional mitochondrial complex I has been shown to be essential for the promotion of aerobic glycolysis and the Warburg effect . In support of these points, PGC-1α or ERRα, two known central regulators of mitochondrial metabolism have been shown to promote the growth of liver, colon, breast, prostate and melanoma cancers [25–29]. Here, we demonstrate the influence of metformin on mitochondrial bioenergetics in cells and in isolated mitochondria.
Wild-type male C57BL/6J mice were purchased from The Jackson laboratory (Bar Harbour, ME, USA). NT2196 and NMuMG cells were kindly provided by Dr. William Muller (McGill University, Montréal, Canada) and have been described elsewhere . MCF7 and MCF10A cells were purchased from ATCC. All reagents were purchased from Sigma-Aldrich unless otherwise stated.
All cell culture material was purchased from Wisent Inc. unless otherwise specified. NT2196 and NMuMG cells were grown as previously published . MCF7 cells were grown in Dulbecco’s Modified Eagle Medium (DMEM) media with 10% fetal bovine serum, supplemented with penicillin and streptomycin. MCF10A cells were grown in DMEM/Ham’s F12 50/50 Mix Media supplemented with 5% horse serum, 20 ng/mL human epidermal growth factor (hEGF), 0.5 μg/mL hydrocortisone, 10 μg/mL insulin, penicillin and streptomycin. All cells were grown at 37°C, 5% CO2 (Thermo Forma, Series II Water Jacketed CO2 Incubator). For the experiments comparing the impact of growth in glucose or galactose media on respiration, MCF7 cells were cultured either in standard glucose DMEM or in galactose (25 mM) medium that has the same composition as DMEM except that the glucose has been replaced with galactose. Cells were cultured in glucose or galactose medium for a period of 20 to 25 days after being put into culture. Cells were then treated with either ddH20 (control) or metformin (0.5 mM) for a period of 24 hours, after which respiration was assessed as previously described .
A fixed number of cells were plated in 6-well plates (9.6 cm2/well). Every 24 hours, the media was removed and cells were treated with ddH20 (control) or metformin (0.5 mM and 5.0 mM). At the respective time points (24, 48, and 72 hours), the media was removed and stored into tubes (to collect floating cells); the adherent cells were washed with phosphate buffered saline (PBS), trypsinized and resuspended in the media collected, which was centrifuged at 2,500 rpm for 5 minutes. The media was removed (and used for lactate and glucose measurements; The media was removed (and used for lactate and glucose fold change measurements in the presence of metformin) and the cell pellet was resuspended), and the cell pellet was resuspended in a known volume of fresh media. Both total and live cell counts were obtained using Trypan Blue Stain (0.4%, Gibco) and a TC10 automated cell counter (Bio-Rad).
MCF10A, MCF7, NT2196 and NMuMG cells were grown in 6-well plates (9.6 cm2/well) to 60% confluency. The media in each well was removed and centrifuged at 13,000 rpm for 10 minutes to remove cellular debris, placed into new tubes and analyzed with a Nova BioProfile 400 analyzer. Wells that contained only media in the absence of cells were also analyzed in this manner to serve as blanks. To account for cell number, cells were counted as described above. To calculate lactate production and glucose consumption, the concentration of either lactate or glucose present in each condition was subtracted from that of blank wells and this value was then normalized for total cell count.
Respiration measurements with cultured cells or isolated mitochondria were performed using a Digital Model 10 Clark Electrode (Rank Brothers, Cambridge, UK). Respiration with cultured cells was carried out in their respective growth medium while respiration with isolated mitochondria was carried out in KHEB (120 mM KCl, 5 mM KH2PO4, 3 mM 4-(2-hydroxyethyl)-1-piperazineethanesulfonic acid (HEPES), 1 mM ethylene glycol tetraacetic acid (EGTA) and 0.3% bovine serum albumin (BSA) (w/v), pH 7.2) assay medium. Respiration traces for isolated mitochondria were digitized using DigitizeIt Software (Version 1.5). This software extracts values from traces using the background graph paper found on the trace as a reference. Simply, the respiration traces were imported, the axes were defined manually based on the corresponding values found on the graph paper of the trace, and data values were generated by the software and plotted using GraphPad Prism 5 Software.
Mice were sacrificed at approximately 6 months of age with approval from the McGill University Animal Care Committee. Mitochondria from skeletal muscle were isolated as previously described . The integrity of mitochondrial suspensions were evaluated by quantifying respiratory control ratio (RCR) values that are obtained by dividing the rate of oxygen consumption in the presence of ADP (state 3) by that in the presence of oligomycin (state 4). Only mitochondrial suspensions displaying RCR values greater than 3 in control conditions were used.
NT2196, NMuMG, MFC10A and MCF7 cells were grown in the presence of ddH20 (control) or specific doses of metformin for 24 hours. 1 × 106 cells were used for respiration measurements. Calculations of coupled and uncoupled respiration were performed according to . Briefly, coupled respiration is calculated by subtracting total respiration from oligomycin-insensitive (2.5 μg/mL/1 × 106 cells) respiration. Uncoupled respiration represents oligomycin-insensitive respiration. Nonmitochondrial respiration represents respiration that is insensitive to myxothiazol (10 μM). Cells displayed no detectable nonmitochondrial respiration.
For the metformin incubation experiments, mitochondria (0.6 mg/mL) were incubated in KHEB media at 37°C in a temperature-controlled water bath (Fisher Scientific, Isotemp 3006S) in the presence of either complex I (equimolar 30 mM malate and pyruvate) or complex II (25 mM succinate and 50 μM rotenone) substrates, either in the presence of ddH20 (control) or 10 mM metformin for 30 minutes. Samples were resuspended every 10 minutes. After 30 minutes, the 100 μL reaction was diluted in 400 μL KHEB media (final equimolar concentration of 6 mM malate and pyruvate or 5 mM succinate and 10 μM rotenone, in the absence or presence of 2 mM metformin). Respiration was recorded immediately, followed by the addition of ADP (500 μM, state 3), oligomycin (2.5 μg oligomycin/mg mitochondrial protein, state 4) and FCCP (1.5 μM).
MCF10A and MCF7 cells were cultured in 6-well plates (9.6 cm2/well) to 80% confluency, after which ddH20 (control) or metformin (0.5 mM, 5.0 mM) was added to the media for 24 hours. The media was then exchanged for [U-13C]glucose (Cambridge Isotope Laboratories, Tewksbury, MA, USA, CLM-1396, 99% atom 13C)-labeled media for a period of 1 hour. Cells were then rinsed once with 4°C saline solution (9 g/L NaCl) and quenched with 80% methanol (<20°C). Isolated mitochondria from murine skeletal muscle were resuspended in KHEB media at a concentration of 1.5 mg/mL. Samples were incubated in a temperature-controlled water bath (Fisher Scientific, Isotemp 3006S) at 37°C in the presence of 1 mM malate and 1 mM [U-13C]pyruvate for 30 minutes, either in the presence of ddH20 (control) or 5 mM metformin. Samples were then quenched in 80% methanol (<20°C). The remaining procedure is identical for cellular and mitochondrial extracts. Metabolite extraction was carried by sonication at 4°C (10 minutes, 30 sec on, 30 sec off, high setting, Diagenode Bioruptor). Extracts were cleared by centrifugation (14,000 rpm, 4°C) and supernatants were dried in a cold trap (Labconco) overnight at -4°C. Pellets were solubilized in pyridine containing methoxyamine-HCl (10 mg/mL) by sonication and vortex, centrifuged and pellets were discarded. Samples were incubated for 30 minutes at 70°C (methoximation), and then were derivetized with MTBSTFA at 70°C for 1 h. Next, 1 μL was injected into an Agilent 5975C GC/MS configured for single ion monitoring (SIM) according to . Data analyses were performed using the Chemstation software (Agilent, Santa Clara, USA). Mass isotopomer distribution analyses were performed according to [34, 35].
As metformin had a significant impact on mitochondrial metabolism in breast cancer cells (Figure 2), we then compared the effect of this drug between cancer cells and nontransformed controls given that they display differences in mitochondrial metabolism (Figure 1). Metformin caused a decrease in mitochondrial respiration in both breast cancer cells and nontransformed controls (Figures 3A,B). However, the decrease in respiration was larger in nontransformed cells compared with breast cancer cells (Figure 3A,B). Metformin also caused a decrease in respiration upon acute treatment (15 minute incubation), [see Additional file 1, Additional file 2: Figure S1] in the murine control cells (NMuMG), while no change was observed in the murine breast cancer cells (NT2196). Furthermore, metformin caused a shift in the mitochondrial coupling status in favor of uncoupled respiration, which was greater in magnitude in nontransformed cells compared with cancer cells (Figure 3C,D).
An important implication of these data is that a constant supply of glucose to cells is critical to attenuate the energetic stress caused by metformin by fuelling aerobic glycolysis. Therefore, we tested whether cells that are forced to rely exclusively on mitochondrial metabolism for ATP production are more sensitive to metformin. We cultured human breast cancer cells (MCF7) in media where the glucose had been replaced by galactose . MCF7 cells grown in galactose media displayed an approximate two-fold increase in mitochondrial respiration compared with MCF7 cells grown in glucose media (Figure 3K). Importantly, MCF7 cells grown in galactose media devoted a larger proportion of their respiration for ATP production than those grown in glucose (Figure 3L). These results validate the experimental design by showing that cancer cells grown in the presence of galactose increase mitochondrial respiration, and elevate the proportion of their mitochondrial respiration devoted to support ATP production compared to cells grown in glucose (Figure 3K,L). Metformin caused an approximate 20% decrease in respiration for MCF7 cells grown in glucose media (Figure 3K). However, when MCF7 cells were grown in galactose media, metformin had a more profound impact on mitochondrial respiration, which decreased by more than two-fold upon metformin treatment (Figure 3K). Metformin caused a significant increase in the proportion of uncoupled respiration for MCF7 cells grown in either glucose or galactose (Figure 3L). However, the impact of metformin on the proportion of uncoupled respiration was much greater for MCF7 cells grown in galactose than glucose, given that at baseline, these cells were more coupled than those grown in glucose (Figure 3L). Importantly, MCF7 cells grown in galactose media and exposed to 5 mM metformin for 48 hours exhibited strikingly more cell death than MCF7 cells grown in glucose media (Figure 3J,M). Together, these results demonstrate that cells that cannot engage aerobic glycolysis due to limiting glucose levels, are entirely dependent on mitochondria for ATP production, and are thus more susceptible to the action of metformin.
To probe the impact of metformin on mitochondria, we used mitochondria that were incubated with either complex I or II substrates. Comparison of the effect of metformin on the respiration rate of mitochondria that were incubated with complex I or II substrates allows one to pinpoint whether metformin is acting on complex I or II, given that complexes III to V are involved in both complex I- and II-dependent respiration. Metformin reduced state 3 and state 4 respiration, as well as the maximal respiratory capacity of mitochondria respiring on complex I substrates (Figure 5C,E), but had no significant effect on these parameters when mitochondria were respiring on complex II substrates (Figure 5D,F). Finally, metformin also acutely decreased complex I-dependent respiration in isolated mitochondria from cultured MCF7 and MCF10A cells [see Additional file 1, Additional file 2: Figures S2 and S3]. Together, these results demonstrate that metformin can directly act on mitochondria and limit complex I-dependent respiration.
Our results confirm that mitochondria are key targets of metformin despite reports suggesting cytoplasmic actions [11, 13]. This is in keeping with prior evidence for an inhibitory effect on complex I together with a membrane potential-driven accumulation of positively charged drug within the mitochondrial matrix . Our data argue against an indirect action of metformin on mitochondria . While this manuscript was in preparation, a study by the Chandel group has shown that the ability of metformin to limit tumour growth in vivo is dependent on mitochondrial complex I . Also, a study by the Hirst group has demonstrated that metformin can limit the activity of purified complex I . These papers support our data showing a direct effect of metformin on mitochondrial respiration.
There is clinical  and experimental  evidence that metformin use is associated with modest weight loss, in contrast to many antidiabetic medications. This is consistent with our observation that metformin causes inefficient mitochondrial metabolism, as demonstrated by the increase in the fraction of uncoupled respiration. Classic uncouplers also cause inefficient mitochondrial metabolism and have been shown to cause substantial weight loss, but are too toxic for clinical use . Interestingly, recent preclinical work suggests that targeting the uncoupling agent DNP to the liver, the organ most impacted by metformin due to its pharmacokinetics following oral administration, reduces toxicity . However, it is important to recognize that although metformin causes inefficient mitochondrial metabolism, it should not be considered as a classic uncoupler.
Recently, it has been shown that cancer cells that are more sensitive to low glucose are defective in oxidative phosphorylation (OXPHOS) regulation and more sensitive to biguanides . The low glucose condition is a setting that is advantageous for cells displaying robust mitochondrial capacities, due to the fact that cells need to rely on alternate fuel sources that are metabolized by mitochondria [38, 47]. Furthermore, because they inhibit mitochondrial metabolism, biguanides exacerbate the OXPHOS defects of cells sensitive to low glucose, explaining their greater sensitivity to metformin under low glucose conditions . We found that cells cultured in the absence of glucose and in the presence of galactose displayed increased mitochondrial metabolism and were drastically more sensitive to the effects of metformin than cells grown in the presence of glucose. It has also been shown that cancer cells grown in the absence of glucose and presence of glutamine were more affected by metformin treatment than cells grown in the presence of glucose . Together, these data support the notion that metformin inhibits OXPHOS, and thus cells that are forced to rely on OXPHOS are more affected by the actions of metformin. Furthermore, these data show that in the setting of inhibition of OXPHOS, cancer cells compensate by increasing glycolysis. We demonstrate that when metformin inhibits OXPHOS, either in isolated mitochondria or in intact cells, the citric acid cycle is inhibited, and accepts less glucose carbon, thus favoring lactic acid production. Importantly, if this compensation is restricted by a lack of glucose, or by inhibition of oncogenes that drive glycolysis [29, 49], even in the presence of other nutrients that require mitochondrial function for generation of ATP, cell viability is threatened.
While the concept of inducing energetic stress in cancers by using metformin is appealing, pharmacokinetic issues must be considered. It is by no means clear that conventional anti-diabetic doses of metformin achieve active concentrations in neoplastic tissue. Many cancers express cell surface transport molecules such as OCT1, which are required for cellular uptake at low ambient drug concentrations, at far lower levels than in the liver, where the drug is active. However, once inside cells, the greater membrane potential of mitochondria from cancer cells [50, 51] should facilitate metformin uptake compared with mitochondria from nontransformed cells. Thus, although metformin at high doses has some in vivo antineoplastic activity , it may be considered a ‘lead compound’ for pharmacokinetic optimization for possible applications in oncology.
We demonstrate that metformin directly acts on mitochondria to limit citric acid cycle activity and OXPHOS, as demonstrated in isolated mitochondria as well as in intact cells. The metformin-mediated decrease in mitochondrial function was accompanied by a compensatory increase in glycolysis. Hence, the sensitivity of cells to metformin is dependent on their capacity to engage aerobic glycolysis. Biguanides could thus potentially be used in oncology to exploit the metabolic vulnerability of cancer cells.
Bovine serum albumin
Citric acid cycle
Dulbecco’s Modified Eagle Medium
Ethylene glycol tetraacetic acid
Gas chromatography/mass spectrometry
human epidermal growth factor
Mass isotopomer distribution
Organic cation transporter
Phosphate buffered saline
Respiratory control ratio
Single ion monitoring.
This work was supported by grants from the Canadian Institutes of Health Research (grant number MOP-106603 to JSP) and Terry Fox Foundation (grant number TFF-116128 to JSP and MP). We acknowledge salary support from the McGill Integrated Cancer Research Training Program (MICRTP) (to SA, M219196C0G), Maysie MacSporran Studentship (to SA, F201699C10), Canderel (to SPG), and Fonds de Recherche du Québec-Santé (FRQS) (to JSP). We would also like to thank Erzsebet Nagy Kovacs for technical help with the mice.
This article is published under license to BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/4.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly credited. The Creative Commons Public Domain Dedication waiver (http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated.