Metformin inhibits thyroid cancer cell growth. We sought to determine if variable glucose concentrations in medium alter the anti-cancer efficacy of metformin. Thyroid cancer cells (FTC133 and BCPAP) were cultured in high-glucose (20 mM) and low-glucose (5 mM) medium before treatment with metformin. Cell viability and apoptosis assays were performed. Expression of glycolytic genes was examined by real-time PCR, western blot, and immunostaining. Metformin inhibited cellular proliferation in high-glucose medium and induced cell death in low-glucose medium. In low-, but not in high-glucose medium, metformin induced endoplasmic reticulum stress, autophagy, and oncosis. At micromolar concentrations, metformin induced phosphorylation of AMP-activated protein kinase and blocked p-pS6 in low-glucose medium. Metformin increased the rate of glucose consumption from the medium and prompted medium acidification. Medium supplementation with glucose reversed metformin-inducible morphological changes. Treatment with an inhibitor of glycolysis (2-deoxy-d-glucose (2-DG)) increased thyroid cancer cell sensitivity to metformin. The combination of 2-DG with metformin led to cell death. Thyroid cancer cell lines were characterized by over-expression of glycolytic genes, and metformin decreased the protein level of pyruvate kinase muscle 2 (PKM2). PKM2 expression was detected in recurrent thyroid cancer tissue samples. In conclusion, we have demonstrated that the glucose concentration in the cellular milieu is a factor modulating metformin's anti-cancer activity. These data suggest that the combination of metformin with inhibitors of glycolysis could represent a new strategy for the treatment of thyroid cancer.
Targeting cancer cell metabolism has emerged as a novel approach to prevent or treat cancers (Pollak 2012). Recent epidemiological and laboratory studies have demonstrated that the anti-diabetic drug metformin has anti-cancer activity (Pollak 2012, Margel et al. 2013, Yin et al. 2013, Thompson 2014). Currently, there are more than 200 ongoing clinical trials in patients with breast, ovarian, lung, colorectal, and pancreatic cancers using metformin alone or in combination with other therapies (ClinicalTrials.gov). Recent meta-analyses, however, have indicated that metformin-inducible reduction of cancer incidence and mortality does not affect all populations equally (Gandini et al. 2014). Clarification of the molecular mechanisms underlying cancer cell sensitivity to metformin could provide critical information for selective use of metformin in clinical trials.
Experimental studies have demonstrated that metformin may influence tumorigenesis, both indirectly, through the systemic reduction of insulin levels, and directly, via the induction of energetic stress (Pernicova & Korbonits 2014). Metformin has been shown to have a strong anti-proliferative effect in many cancer cell lines due to activation of AMP-activated protein kinase (AMPK), inhibition of mammalian target of rapamycin (mTOR) activity, and cell cycle arrest (Ben Sahra et al. 2010, Choi & Park 2013). Metformin can also inhibit mTOR signaling in an AMPK-independent manner through induction of REDD1, a known negative regulator of mTOR (Ben Sahra et al. 2011). Induction of autophagy was demonstrated in cancer cells after treatment with metformin, and inhibition of autophagy by Beclin1 knockdown was sufficient for prevention of the anti-proliferative effects of metformin on endometrial cancer cells (Takahashi et al. 2014). Metformin is considered as a mitotoxic drug and studies in human cancer cells have demonstrated its inhibitory effect on mitochondrial complex I activity (Wheaton et al. 2014).
In cell lines derived from follicular and papillary thyroid cancers, metformin inhibited cell growth, activated AMPK, down-regulated p70S6K/pS6 signaling, antagonized the growth-stimulatory effects of insulin, inhibited clonal cell growth, and reduced thyroid cancer sphere formation (Chen et al. 2012, Klubo-Gwiezdzinska et al. 2013). Metformin treatment also significantly delayed tumor growth in animal models of thyroid cancer (Cho et al. 2014). Diabetic patients treated with metformin who were subsequently diagnosed with thyroid cancer, and were treated with thyroidectomy and 131I, had an increased likelihood of complete response and higher remission rates as compared to matched thyroid cancer diabetic patients not receiving metformin (Klubo-Gwiezdzinska et al. 2013). Together these data suggested the potential utility of metformin as an adjunct treatment for non-diabetic patients with thyroid cancer.
Numerous factors may influence the anti-cancer effectiveness of treatment with metformin. Recent studies have indicated that the anti-proliferative effects of metformin in cancer cells are highly dependent on the glucose concentration in the extracellular milieu (Wahdan-Alaswad et al. 2013, Zordoky et al. 2014). It has been demonstrated that metformin inhibited cellular proliferation in the presence of glucose, but induced cell death upon glucose deprivation (Birsoy et al. 2012, Zordoky et al. 2014). Experimental in vitro and in vivo studies have demonstrated that use of the glycolysis inhibitor 2-deoxy-d-glucose (2-DG) in combination with metformin leads to significant cell death associated with a decrease in cellular ATP, prolonged activation of AMPK, and sustained autophagy (Cheong et al. 2011). In a xenograft model of human breast cancer, combined treatment with the mitotoxic drug Mito-CP and 2-DG led to significant tumor regression (Cheng et al. 2012).
Cancer cells exhibit diverse responses to glucose limitation. Over-expression of glycolytic genes (pyruvate kinase muscle (PKM), ENO1, and ALDOA) was demonstrated in cancer cells with high sensitivity to glucose deprivation (Birsoy et al. 2014). The glucose-utilization gene signature has been proposed as a biomarker for identifying tumors that are sensitive to treatment with biguanides.
There is limited information regarding the effects of micro-environmental factors on the efficacy of treatment with metformin in differentiated thyroid cancer. In this study, we examined the effects of glucose in the extracellular milieu on thyroid cancer cell response to treatment with metformin. We also determined how pharmacological blockade of glycolysis with 2-DG affected sensitivity of thyroid cancer cells to treatment with metformin.
Materials and methods
Human thyroid tissue samples
The protocol was approved by the Institutional Review Boards at the Medstar Washington Hospital Center and the Human Use Committee at the Uniformed Services University of the Health Sciences. Paraffin embedded thyroid tissue samples were obtained from 16 patients that developed progressive disease after initial treatment and underwent a second surgery for metastatic lesions. Informed consent from individual patients was not necessary because all data were rendered anonymous.
Thyroid cancer cell lines and reagents
Human thyroid cancer cell lines derived from follicular thyroid cancer 133 (FTC133) and papillary thyroid cancer (BCPAP) were obtained from Dr Motoyasu Saji (The Ohio State University) with permission from the researchers who originally established the cell lines. These cell lines had been tested and authenticated by DNA analysis to be of thyroid origin (Schweppe et al. 2008). In our laboratory, we confirmed the presence of BRAF mutation in BCPAP cells, and loss of PTEN expression in FTC133 cells.
Cancer cells were propagated in conventional DMEM supplemented with 10% of fetal bovine serum (FBS), 100 U/ml penicillin, and 100 μg/ml streptomycin in a humidified 5% CO2 incubator. The cells were sub-cultured with 0.5% trypsin and 0.02% EDTA (Sigma–Aldrich) when the cell confluency reached 80%. All experiments were performed using thyroid cancer cell lines which had been passaged <25 times. For glucose deprivation experiments, cells were washed with PBS twice, and then medium was replaced with non-glucose containing DMEM (Invitrogen). The glucose concentration was maintained at 0 mM or adjusted to either 5 mM (low-glucose) or 20 mM (high-glucose). Cells were incubated with either control medium or medium containing metformin (Sigma Chemical Co.). The pharmacological inhibitor of glycolysis, 2-DG was obtained from Sigma Chemical Co.
Cell proliferation, viability, and caspase activity assays
Cell proliferation rate was determined by cell counting using Vi-CELL Cell Viability Analyzer from Beckman Coulter (Fullerton, CA, USA). Cell viability was assessed by alamarBlue Assay (Invitrogen) according to manufacturer instructions. Cell viability also was determined by evaluation of mitochondrial membrane potential with the fluorogenic lipophilic cation JC-1 (Cayman Chemical Company, Ann Arbor, MI, USA). Detection of mitochondrial membrane potential was performed according to the manufacturer's instructions, followed by fluorescent microscopy. Cytotoxicity was examined using the LIVE/DEAD Cell Imaging Kit (Invitrogen). FAM-FLICA Caspase 3/7 Assay for apoptosis detection via activated caspases 3 and 7 was obtained from ImmunoChemistry Technologies, LLC (Bloomington, MN, USA). Dead cells were detected using propidium iodine staining. All experiments were repeated at least three times and the average values of representative experiments are reported.
Evaluation of glucose concentration and pH in the medium
The glucose concentration was measured using the glucose oxidase method (GlucCell, CescoBioproducts, Atlanta, GA, USA).
Protein extraction and western blot analysis
Thyroid cancer cells were incubated with ice-cold cell lysis buffer; 25 μg of total protein lysates were subjected to SDS–PAGE (7%). The separated proteins were transferred to nitrocellulose membranes (Invitrogen) by electrophoretic blotting. Membranes were incubated overnight with the primary antibody against phospho-p70S6K (Thr389), total-p70S6K, phospho-pS6 (Ser235/236), phospho-AMPK (Thr172), total AMPK, phospho-AKT (Ser473), poly (ADP-ribose) polymerase (PARP), cleaved-caspase 3, LC3B, p62, BIP, PKM2 (Cell Signaling, Boston, MA, USA), cyclin D1, perforin, and b-actin (Santa Cruz Biotechnologies). Membranes were then incubated with the secondary antibody and, after washing protein bands on the blots were visualized using the Li-COR Odyssey Imaging System (LI-COR Biosciences, Lincoln, NE, USA). Image Studio Lite version 3.1 (LI-COR) was used for band densitometry measurement and quantification of protein levels.
RNA extraction and quantitative real-time PCR
Total RNA was isolated from thyroid cancer cells using RNeasy Mini Kit (Qiagen) according to the manufacturer's protocol. SYBR Green-based qPCR master mixes were obtained from Qiagen. Expression of genes with an established role in regulation of glycolysis was examined using ‘glycolysis’ real-time PCR arrays (Qiagen). Expression of PKM (with specific isoforms being PKM1 and PKM2), glucose-6-phosphate dehydrogenase (G6PD), and fructose-1,6-biphosphatase 1 (F1) were examined using specific primers.
Immunostaining was performed on paraffin-embedded tissue sections and on cancer cell lines that were cultured on eight-chambered SuperCell Culture Slides (Fisher Scientific, Pittsburgh, PA, USA). Endogenous peroxidase activity was quenched by incubation in 3% hydrogen peroxide. Sections and cancer cell lines were incubated overnight with primary anti-PKM2 and immunostaining was performed using the DAKO Universal LSAB Kit (DAKO, Carpinteria, CA, USA). Negative controls were performed as described earlier except that the primary antisera were not included. We adopted the semi-quantitative scoring system in considering the staining intensity and area extent. Every tumor was given a score according to the intensity of the staining (no staining, 0; weak staining, 1; moderate staining, 2; and strong staining, 3) and the extent of stained tumor areas (0%, 0; 1–10%, 1; 11–50%, 2; 51–80%, 3; and 81–100%, 4). The final immunoreactive score was determined by multiplying the intensity scores with the extent of positivity scores of stained cells, with the minimum score of 0 and a maximum score of 12.
Cells were grown on coverslips and fixed with 4% PFA in PBS followed by treatment with 0.2% Triton X-100 in PBS. Perforin primary antibody (1:100) was incubated for 1 h at room temperature and visualized by means of AlexaFluor-488 (1: 300, Invitrogen). Coverslips were mounted in antifade (SlowFade; Invitrogen) and examined under a confocal microscope (Zeiss 700).
Live cell imaging was performed using Leica AF6000 Time-Lapse Imaging System.
Metformin inhibits thyroid cancer cell viability in low-glucose medium
We first examined the effect of low (5 mM) and high (20 mM) glucose on growth of thyroid cancer cells. During the first 48 h, no difference was noted in the proliferation rate of cells growing in either high- or low-glucose conditions (Fig. 1A). Starting from 72 h, the number of cells was higher in cells grown in high-glucose medium, and the difference was more prominent in BCPAP cells. In glucose-free DMEM medium supplemented only with 10% FBS, the rate of growth was decreased in both cells lines.
To determine the effects of glucose on thyroid cancer cell response to metformin, we performed treatment with metformin in thyroid cancer cells growing in high- and in low-glucose medium. The AlamarBlue Assay showed that in high-glucose medium, treatment with metformin at concentration 5 mM for 48 h was required for inhibition of cell viability (Fig. 1B). In low-glucose medium, thyroid cancer cell viability was significantly reduced after exposure to metformin at concentrations ranging from 0.1 to 0.5 mM. ViCell analysis demonstrated that metformin had no significant effect on the number of dead cells in high-glucose medium, but induced cell death in low-glucose medium.
Western blot analysis demonstrated that metformin decreased expression of cyclin A, and its effects were more prominent in low-glucose medium (Fig. 1C). Treatment with metformin was not associated with induction of caspase-3 and PARP cleavage in cancer cells growing in high- or low-glucose medium. Caspase 3/7 activity assay showed no caspase activation in thyroid cancer cells treated with metformin for 48 h in low-glucose medium (Supplementary Figure 1, see section on supplementary data given at the end of this article).
Analysis of autophagy-related molecules showed that treatment with metformin resulted in cell type specific responses. Loss of p62 was observed in FTC133 cells treated with metformin in low-glucose condition. In BCPAP cells, switch from high- to low-glucose medium was associated with loss of p62 (Fig. 1D). The lipidated form of LC3B (LC3B-II) was detected after exposure to metformin in FTC133 cells cultured in low-glucose but not in high-glucose medium. In BCPAP cells, LC3B expression was detected in cells cultured in high- and low-medium conditions, and was not significantly affected by metformin.
These results showed that thyroid cancer cells were more sensitive to treatment with metformin in low-glucose compared to high-glucose medium metformin.
Metformin induces oncosis in thyroid cancer cells in low-glucose medium
To clarify the effects of metformin on thyroid cancer cells, we performed real-time imaging of thyroid cancer cells treated with metformin. In high-glucose medium, metformin (5 mM) had no significant effect on cellular morphology. In low-glucose medium, time-dependent morphological changes suggesting oncosis were observed in metformin treated thyroid cancer cells (Supplementary Videos 1 and 2). These morphological changes occurred gradually, with time points corresponding to the following stages: non-reactive stage (24 h), cell swelling (24–36 h), cell shrinkage (36–40 h), ‘anchored’ detachment (40–48 h), and membranous blebbing with cytoplasmic ballooning (48–72 h). Representative images of morphological changes in thyroid cancer cells under treatment with metformin are presented in Fig. 2A.
Analysis of mitochondrial membrane potential by JC-1 staining showed cytoplasmic localization of JC-1 aggregates in non-treated thyroid cancer cells (Fig. 2B). The perinuclear accumulation of JC-1 aggregates was noted in cells that were treated with metformin for 48 h, suggesting that metformin affected localization of mitochondria in thyroid cancer cells. Loss of mitochondrial membrane potential in cells that detached from the plates after treatment with metformin for 72 h was associated with propidium iodine nuclear staining, indicating cell death.
Since increased permeability of the endoplasmic reticulum (ER) membrane is considered an early event of oncosis, we examined expression of BiP, a well-characterized regulator of transport across the ER membrane. Treatment with metformin for 48 h was associated with overexpression of BiP in cells cultured in low-, but not in high-glucose medium (Fig. 2C). Expression of a molecular marker of oncosis (perforin) was also induced in thyroid cancer cells treated by metformin (Fig. 2C). Confocal microscopy demonstrated membranous localization of perforin in metformin treated thyroid cancer cells (Fig. 2D).
Together these results suggest that oncosis may represent a mechanism of metformin-inducible thyroid cancer cell death.
AMPK is activated in thyroid cancer cells cultured in low-glucose medium
Oncotic cell death is triggered by ATP depletion, and AMPK is a known sensor of decreased ATP/AMP ratio. Therefore, we examined the effects of metformin on phosphorylation of AMPK and activation of pS6 in thyroid cancer cells growing in high- and low-glucose medium.
Time-course experiments demonstrated that culturing of thyroid cancer cells in high-glucose medium for up to 48 h had no significant effects on p-AMPK (Fig. 3A). In contrast, in low-glucose medium a progressive increase in p-AMPK over time was documented, especially in BCPAP cells. At 48 h, the level of p-AMPK was increased by more than twofold when compared to control cells.
In high-glucose medium, treatment with metformin for 48 h was associated with a dose-dependent induction of p-AMPK and inhibition of p-pS6 (Fig. 3B). In low-glucose medium, metformin increased the level of p-AMPK with subsequent down-regulation of p-pS6 at 24 and 36 h (data not shown). At 48 h of incubation in low-glucose medium, however, the level of p-AMPK was not significantly different between metformin-treated and non-treated cells (Fig. 3C). Interestingly, despite the negligible effects of metformin on p-AMPK at 48 h, there was significant down-regulation of p-pS6.
These data show that thyroid cancer cell response to metformin includes AMPK-dependent as well as by AMPK-independent mechanisms.
Glucose depletion increases cytotoxic effects of metformin on thyroid cancer cells
Since growth inhibitory effects of metformin were prominent in thyroid cancer cells cultured in low-glucose conditions, we performed experiments using glucose-free medium supplemented with FBS. As demonstrated in Fig. 4A, metformin significantly decreased the number of viable cells within 24 h. BCPAP cells were more sensitive to metformin treatment in these conditions.
Measurement of glucose in the medium during treatment with metformin showed that in both high- and low-glucose media, metformin accelerated the decline in glucose concentration. Data from measurements in high-glucose medium are presented in Fig. 4B, with an identical trend noted in low-glucose medium. The increased consumption of glucose from the medium with metformin treatment was prominent in FTC133 cells but not in BCPAP cells. Treatment with metformin also prompted acidification of medium consistent with induction of glycolysis. In FTC133 and BCPAP cells treated with metformin (1 mM for 48 h), the levels of pH decreased from 7.35±0.02 to 6.8±0.12 (P=0.01) and from 7.25±0.1 to 6.8±0.2 (P=0.01) respectively.
To determine if glucose concentration in the medium is a factor influencing metformin action, we adjusted the level of glucose to 20 mM after cell incubation with metformin in low-glucose medium for 48 h. Supplementation of glucose reversed metformin-inducible morphological changes (Fig. 4C), and prevented metformin-inducible death in thyroid cancer cells. In contrast, supplementation of medium with FBS was not sufficient to rescue thyroid cancer cell from metformin-inducible cell death.
These data demonstrate that glucose concentration in the medium is a factor influencing metformin activity against thyroid cancer cells.
Treatment with 2-DG potentiates the effects of metformin on thyroid cancer cells
To confirm that activation of glycolysis is a critical mechanism underlying thyroid cancer cell survival when treated with metformin, we used the anti-glycolytic agent 2-DG. In high-glucose medium, treatment with 2-DG (5 mM) for 48 h had no significant effects on cell viability. In FTC133 cells, treatment with 2-DG increased p-AMPK in high- as well as in low-glucose medium, and down-regulated p-pS6, especially in low-glucose medium. In BCPAP cells, the level of p-AMPK was increased in low-glucose medium and was not significantly affected by 2-DG. However, treatment with 2-DG (5 mM) down-regulated p-pS6 in BCPAP cells in low-glucose medium (Fig. 4D). Treatment with 2-DG (5 mM) also resulted in induction of BiP expression in FTC133 and BCPAP cells. Analysis of apoptotic markers showed that as a single agent, 2-DG had no significant effect on caspase-3 and PARP cleavage at 24 h. Treatment with 2-DG (5 mM) for 36 and 48 h was associated with slight induction of caspase-3 cleavage.
In a low-glucose environment, the combination of 2-DG with metformin resulted in robust activation of AMPK and down-regulation of p-pS6 at 6 h of treatment. The combination of 2-DG (5 mM) with metformin (1 mM) for 24 h resulted in activation of apoptosis as demonstrated by western blot with anti-caspase-3 and anti-cleaved PARP (Fig. 4E). Dose–response experiments demonstrated that pretreatment of thyroid cancer cells with 2-DG for 24 h sensitized them to clinically achievable concentrations of metformin. Under these conditions, treatment with metformin at micromolar concentrations (25 μM) induced caspase-3 and PARP cleavage at 24 h (Fig. 4F) and led to massive cell death at 48 h in the examined cell lines.
These data indicate that the inhibition of glycolytic activity by 2-DG increases thyroid cancer cells sensitivity to metformin, and suggest that anti-cancer effects of metformin can be mediated via regulation of genes controlling glycolysis.
Metformin inhibits expression of PKM2 in thyroid cancer cells cultured in low-glucose medium
We performed a RT-PCR screening of 84 glycolytic genes with well-established role in regulation of glycolysis using commercially available RT-PCR arrays. The mRNA extracted from normal thyroid tissue was used as control. The set of genes that were commonly up- or down-regulated in thyroid cancer cell lines comparing to the normal thyroid is presented in Table 1. Both FTC133 and BCPAP cell lines were characterized by over-expression of genes that were described in ‘glucose-deprivation sensitive’ cancer cells (Birsoy et al. 2014). We also performed qRT-PCR analysis of PKM2, G6PD, and F1 using RNA from 15 human thyroid papillary cancers and normal thyroid tissue. Compared to a normal thyroid, the mRNA levels of PKM2 and G6PD were higher in 14/15 PTCs (mean 2.59±1.3, P=0.003) and 13/15 PTCs (mean 3.6±2.0; P=0.007) respectively. The mRNA level of F1 was decreased in 10/15 PTCs (mean 0.72±0.4), however the difference was not statistically significant (P=0.08).
Glycolytic genes identified by real-time PCR screening as differentially expressed in FTC133 and BCPAP cells as compared to normal thyroid
|Target||Description||FTC133 vs normal||BCPAP vs NT|
|Fold change||P value||Fold change||P value|
|PKM2||Pyruvate kinase muscle 2||5.5404||0.0000||4.2029||0.0001|
|G6PC3||Glucose-6-phosphatase, catalytic, 3||4.7404||0.0000||5.2162||0.0001|
|ENO2||Enolase 2 (gamma, neuronal)||3.9041||0.0031||5.5906||0.0001|
|PYGM||Phosphorylase, glycogen, muscle||−5.6765||0.0000||−35.1877||0.0000|
|SUCLG2||Succinate-CoA ligase, GDP-forming, beta subunit||−5.7358||0.0000||−9.2663||0.0001|
|ALDOC||Aldolase C, fructose-bisphosphate||−7.2350||0.0000||−26.3002||0.0000|
|GALM||Galactose mutarotase (aldose 1-epimerase)||−64.8934||0.0000||−7.8462||0.0001|
Since PKM expression was significantly up-regulated in human PTCs, we focused our study on this regulator of glycolysis. The switch in a splice isoforms of the PKM is sufficient for the shift in cellular metabolism to aerobic glycolysis (Christofk et al. 2008). Therefore, we next examined the effects of metformin on expression of PKM1 and PKM2 isoforms in FTC133 and BCPAP cells. RT-PCR analysis showed that a switch from high- to low-glucose medium did not affect expression of PKM isoforms. Treatment with metformin was not associated with significant changes in PKM isoforms mRNA levels in FTC133 cells and in BCPAP cells (Fig. 5A). Western blot analysis demonstrated that metformin had no significant effects on protein expression of PKM2 in high-glucose medium, but significantly inhibited PKM2 in thyroid cancer cells cultured in low-glucose medium (Fig. 5B). The inhibitory effects of metformin on PKM2 protein expression were confirmed by immunostaining (Fig. 5C).
These data show that metformin decreases the level PKM2 protein in thyroid cancer cells, and suggest that this process occurs via regulation of PKM2 protein degradation.
Expression of PKM2 is increased in human thyroid cancers
Our in vitro experiments suggested that detection of PKM2 in human thyroid cancer could serve as a biomarker underlying the efficacy of treatment with metformin. Therefore, we assessed expression of PKM2 by immunostaining in a series of metastatic thyroid cancers.
As demonstrated in Fig. 6A, the level of PKM2 was low in normal follicular cells, but was increased in all examined primary thyroid cancers. However, the level of PKM2 expression significantly varied between examined tumors. Strong homogenous immunostaining was detected in 7/16 primary tumors, whereas a different degree of heterogeneity was observed in the remaining nine PTCs. Tumors with mixed follicular and papillary patterns of growth frequently displayed heterogeneous staining with anti-PKM2, and the intensity of immunostaining was higher in the area formed by papillary structures. Interestingly, the levels of PKM2 were also different between neighboring cancer cells forming the same papilla.
We next examined PKM2 in metastatic lesions that developed after initial therapy. Strong homogenous immunostaining was detected in 12/16 cases. The seven metastatic lesions that originated from tumors with strong homogenous PKM2 expression maintained the same pattern of PKM2 staining. Metastases that derived from primary tumors with heterogeneous PKM2 expression were characterized by loss of PKM2 immunoreactivity in one case, persistent heterogeneity in three cases, and increased PKM2 immunostaining in five cases (Fig. 6B).
The average immunohostichemical scores in primary PTCs (mean score 8.4) and metastases (mean score 11) were higher than in normal thyroid tissue mean score (mean score 0.9). As demonstrated in Fig. 6C, PKM2 immunoreactivity in metastatic lesions was significantly higher than in primary tumors (P=0.018).
These results show that most metastatic lesions from patients with progressive thyroid cancers are characterized by high level of PKM2 expression.
To survive in an adverse tumor microenvironment with limited nutrients, cancer cells reprogram canonical biochemical pathways to provide the necessary precursors of proteins, nucleic acids, and membrane lipids via shifts from anaerobic to aerobic glycolysis (Vander Heiden et al. 2009). We have explored molecular mechanisms implicated in the thyroid cancer cell response to glucose deprivation, and to treatment with metformin. We demonstrate that: i) the level of glucose in medium significantly affects the efficacy of metformin as an anti-tumor agent; ii) in low-glucose medium, metformin induces autophagy and oncosis, and these effects are mediated via AMPK-dependent and AMPK-independent mechanisms; iii) metformin induces glycolytic activity in thyroid cancer cell lines, increasing their dependency on glucose in the extracellular milieu, and sensitizing them to a pharmacological inhibitor of glycolysis; iv) metformin inhibits expression of the master regulator of aerobic glycolysis PKM2 in cell lines; and v) expression of PKM2 was elevated in tissue samples from thyroid cancer recurrences.
In our prior report (Klubo-Gwiezdzinska et al. 2013), we demonstrated that supra-physiological concentrations of metformin were required to inhibit thyroid cancer cell proliferation in medium containing 20 mM of glucose. Similar results were found in the current study when treatment with metformin was performed in DMEM medium with 20 mM glucose. The efficacy and sensitivity of treatment with metformin was increased when cells were cultured in 5 mM glucose medium, with micromolar concentrations of metformin being sufficient to inhibit viability of thyroid cancer cells.
Our results corroborate previously reported findings in breast cancer cells, demonstrating pro-apoptotic activity of metformin in low-, but not in high-glucose medium (Wahdan-Alaswad et al. 2013). It also has been shown that metformin led to massive cell death in a glucose-deprived microenvironment (Menendez et al. 2012).
To clarify the mechanisms of metformin-inducible cell death in low-glucose medium, we initially examined a caspase-dependent pro-apoptotic cascade. Treatment of thyroid cancer cells with metformin in low-glucose medium was not associated with induction of caspase-3 or PARP cleavage, suggesting a role for non-apoptotic mechanisms. In low-glucose medium, metformin activated autophagy, a process that permits recycling of cellular constituents as internal fuel sources when external nutrient supplies are limited. We observed differences in regard to expression of LC3B-lipidated form between FTC133 and BCPAP cells at baseline and after exposure to metformin. BCPAP cells harbor the BRAFV600E mutation resulting in constitutive activation of MAPK/ERK signaling. In melanomas, oncogenic BRAF has been associated with inhibition of mTOR and upregulation of basal autophagy (Maddodi et al. 2010). It is possible that autophagic activity is constitutively activated in thyroid cancer cells harboring BRAF mutations.
Morphological changes in thyroid cancer cells that were treated with metformin in low-glucose evoked oncosis. Oncosis is considered as a multi-step process and a recent study (Balvan et al. 2015) has demonstrated that oncotic and apoptotic types of cell death can be distinguished using time-lapse microscopy. The real-time imaging demonstrating that metformin induces cellular swelling, followed by cell shrinkage, membranous blebbing, cell detachment, and finally death, suggested that oncosis is a mechanism of metformin-inducible thyroid cancer cell death.
Oncosis is a specific type of cellular damage that was described in cardiomyocytes after profound myocardial ischemia following coronary occlusion (Majno & Joris 1995). In vitro, oncosis was demonstrated in oxygen–glucose deprivation models used to mimic the effects of stroke. It has been postulated that oncosis may be induced by ATP depletion, leading to progressive membrane injury with resultant ion and water leakage and cell swelling. Consequently, the cell membrane becomes leaky to propidium iodide, indicating the development of a non-selective increase in membrane permeability, with eventual physical disruption of the cell membrane and necrosis (Weerasinghe & Buja 2012). The demonstration of ER stress in combination with cellular swelling, followed shortly thereafter by cell detachment and death, suggests that oncosis is a mechanism of metformin-inducible thyroid cancer cell death. There is limited information on molecular mechanisms underlying the crosstalk between specific types of cancer cell death. A recent study (Del Nagro et al. 2014) demonstrated that cancer cell lines with a slower rate of ATP depletion (average t½ of 45 h) activate caspase-3 and show evidence of apoptosis and autophagy, whereas cell lines with rapid depletion of ATP (average t½ of 32 h) do not activate caspase-3 or show signs of apoptosis or autophagy. Oncosis was identified as a predominant form of cell death once the ATP level was depleted by >20-fold. Results of our study suggest that treatment with metformin in a low-glucose environment decreases the ability of thyroid cancer cells to regenerate ATP, leading to induction of oncosis. At the present time, there are limited data on molecular markers of oncosis in cancer cells. Gene expression profiling of oncotic and apoptotic cardiomyocytes revealed that over-expression of perforin was associated with oncosis (Weerasinghe et al. 2013). In our study, treatment with metformin was associated with induction of perforin in FTC133 as well as in BCPAP cells cultured in low-glucose medium.
Together these data support the statement that induction of oncosis can be one of the mechanisms implicated in the anti-cancer effects of metformin.
Since ATP depletion is considered as a triggering event in oncosis, we extended our study by analysis of AMPK, a well-characterized sensor of intracellular ATP deficiency.
We found that culturing thyroid cancer cells in low-glucose for 48 h resulted in AMPK activation. Interestingly, in low-glucose conditions, treatment with metformin for 48 h did not further increase the level of phospho-AMPK but diminished the phosphorylation of pS6, suggesting a role for AMPK-independent mechanisms of action. It has been demonstrated previously that metformin directly inhibits mTOR by enhancing the association of PRAS40 with RAPTOR in an AMPK-independent manner (Liu et al. 2014). It also has been shown that metformin can inhibit mTORC1 signaling via Rag GTPases in the absence of AMPK (Kalender et al. 2010).
In our study, the level of AMPK activation inversely correlated with the concentration of glucose in the extracellular milieu, and metformin increased the rate of glucose uptake from the medium. These data suggested that metformin induced a metabolic switch toward glycolysis and increased the dependency of thyroid cancer cells on glucose. Rescue experiment at the early stage of metformin-inducible oncosis showed that supplementation with glucose was sufficient to reverse morphological changes and prevent cell death. Our results were consistent with previously reported findings in leukemia cells, demonstrating that metformin increases glycolysis, while decreasing mitochondrial electron transport chain complex I activity (Scotland et al. 2013).
We demonstrate that the glucose concentration in the extracellular milieu modulates metformin's anti-cancer activity, suggesting a possible clinical correlate: reduced systemic glucose levels in patients with thyroid cancer could increase the effect of treatment with metformin. Serum glucose levels in non-diabetic individuals is consistently maintained between 4 and 5.5 mM (70–100 mg/dl) and in conditions of prolonged nutrient deprivation, serum glucose level may drop to 3.0–3.5 mM (55–65 mg/dl) (Haymond et al. 1982). Blood glucose levels fall by nearly 25% early in a ketogenic diet and are sustained at ∼15% below baseline over time (Phinney et al. 1983). It also has been shown that the rates of glucose uptake, as well as rates of flux into the glycolytic pathway, are reduced in rats fed a ketogenic diet. In vivo experiments demonstrated that metformin inhibition of breast tumor growth was enhanced when the serum glucose level was reduced via a low carbohydrate ketogenic diet (Zhuang et al. 2014).
In the current study, the ‘low’ glucose level used (5 mM) corresponds to 90 mg/dl which lies inside the normal range of glucose levels seen in non-diabetic individuals. On the other hand, the ‘high’ glucose level of 20 mM used corresponds to the highly supraphysiologic level of 360 mg/dl that can only be encountered in the blood of diabetic individuals. As a result, and by extrapolating our results to humans, the current study does not only apply to diabetic patients, but also to non-diabetic individuals.
To further target substrate utilization, we employed a ‘tumor cell hyper-sensitization’ strategy combining a mitochondrial inhibitor with an inhibitor of glycolysis. We examined the combination of metformin with 2-DG, which is phosphorylated by hexokinase and subsequently inhibits ATP generation via the glycolytic pathway. Treatment with 2-DG dramatically increases thyroid cancer cell sensitivity to metformin. The combination of 2-DG (5 mM) with low micromolar concentrations of metformin (25 μM) led to massive cell death in all examined cell lines. Results of our study are consistent with previously reported findings in experimental breast cancer models. In addition, it has been shown that the combination of a mitochondria-targeted drug (Mito-CP) with 2-DG can induce significant tumor regression in the absence of significant morphological changes in kidney, liver and heart (Cheng et al. 2012). Our data suggested a potential utility of combination of metformin with 2-DG for the treatment of thyroid cancer, and additional animal studies are needed to confirm these findings.
The identification of molecular determinants underlying cancer cells sensitivity to the treatment with metabolic agents is important for development of personalized treatment of patients with cancer. A recent study has identified that PKM2 is one of the critical enzymes underlying cancer cell sensitivity to glucose deprivation and to treatment with biguanides (Birsoy et al. 2014). We demonstrated that thyroid cancer cell lines were characterized by over-expression of genes controlling glycolysis, and that metformin decreased the protein level of PKM2 in both FTC133 and BCPAP cells. Increased expression of PKM2 was demonstrated in human papillary thyroid cancers, and its overexpression was associated with advanced tumor stages and lymph node metastasis (Feng et al. 2013). Our data demonstrating a high level of PKM2 in recurrent lesions corroborate previously reported findings, and suggest potential utility of metformin for patients with progressive PTCs.
Clinical trials for glioma have demonstrated that a combination of 2-DG (65 mg/kg per day) with radiotherapy is well tolerated without any acute toxicity or late radiation damage to the normal brain tissue (Dwarakanath et al. 2009). Currently, 2-DG treatment is undergoing clinical trials for hormone-refractory prostate cancer and advanced solid tumors (lung, breast, pancreatic, gastric, head, and neck cancers). Reducing glucose availability also has been attempted as a cancer treatment. At this time, there are 12 clinical trials (ClinicalTrials.gov) examining the potential utility of a ketogenic diet in combination with chemotherapy or radiation for the treatment of various types of cancer.
In summary, we have demonstrated that glucose concentration in the cellular milieu is a factor modulating the anti-cancer activity of metformin. Results of our in vitro experiments suggest that the combination of metformin with inhibitors of glycolysis could represent a new strategy for the treatment of thyroid cancer.
This is linked to the online version of the paper at http://dx.doi.org/10.1530/ERC-15-0402.
Declaration of interest
The authors declare that there is no conflict of interest that could be perceived as prejudicing the impartiality of the research reported. The views presented in this manuscript are those of the authors; no endorsement by the USUHS or Department of Defense has been given or should be inferred.
This work was supported by the Uniformed Services University of the Health Sciences, Department of Pediatrics grant number QP86GI13-S25 2014.
We thank Dr Ildy Katona, Chair, Department of Pediatrics, USUHS for her support for this project.
ChenGXuSRenkoKDerwahlM2012Metformin inhibits growth of thyroid carcinoma cells, suppresses self-renewal of derived cancer stem cells, and potentiates the effect of chemotherapeutic agents. Journal of Clinical Endocrinology and Metabolism97E510–E520. (doi:10.1210/jc.2011-1754).
CheongJHParkESLiangJDennisonJBTsavachidouDNguyen-CharlesCWa ChengKHallHZhangDLuY2011Dual inhibition of tumor energy pathway by 2-deoxyglucose and metformin is effective against a broad spectrum of preclinical cancer models. Molecular Cancer Therapeutics102350–2362. (doi:10.1158/1535-7163.MCT-11-0497).
DwarakanathBSSinghDBanerjiAKSarinRVenkataramanaNKJalaliRVishwanathPNMohantiBKTripathiRPKaliaVK2009Clinical studies for improving radiotherapy with 2-deoxy-d-glucose: present status and future prospects. Journal of Cancer Research and Therapeutics5 (Suppl 1) S21–S26. (doi:10.4103/0973-1482.55136).
FengCGaoYWangCYuXZhangWGuanHShanZTengW2013Aberrant overexpression of pyruvate kinase M2 is associated with aggressive tumor features and the BRAF mutation in papillary thyroid cancer. Journal of Clinical Endocrinology and Metabolism98E1524–E1533. (doi:10.1210/jc.2012-4258).
Klubo-GwiezdzinskaJCostelloJJrPatelABauerAJensenKMeteMBurmanKDWartofskyLVaskoV2013Treatment with metformin is associated with higher remission rate in diabetic patients with thyroid cancer. Journal of Clinical Endocrinology and Metabolism983269–3279. (doi:10.1210/jc.2012-3799).
PhinneySDBistrianBREvansWJGervinoEBlackburnGL1983The human metabolic response to chronic ketosis without caloric restriction: preservation of submaximal exercise capability with reduced carbohydrate oxidation. Metabolism32769–776. (doi:10.1016/0026-0495(83)90106-3).
SchweppeREKlopperJPKorchCPugazhenthiUBenezraMKnaufJAFaginJAMarlowLACoplandJASmallridgeRC2008Deoxyribonucleic acid profiling analysis of 40 human thyroid cancer cell lines reveals cross-contamination resulting in cell line redundancy and misidentification. Journal of Clinical Endocrinology and Metabolism934331–4341. (doi:10.1210/jc.2008-1102).
ZordokyBNBarkDSoltysCLSungMMDyckJR2014The anti-proliferative effect of metformin in triple-negative MDA-MB-231 breast cancer cells is highly dependent on glucose concentration: implications for cancer therapy and prevention. Biochimica et Biophysica Acta18401943–1957. (doi:10.1016/j.bbagen.2014.01.023).