Conversion of 2-deoxyglucose-induced growth inhibition to cell death ...

Cancer Chemother Pharmacol (2013) 72:251–262 DOI 10.1007/s00280-013-2193-y

ORIGINAL ARTICLE

Conversion of 2-deoxyglucose-induced growth inhibition to cell death in normoxic tumor cells Huaping Liu • Metin Kurtoglu • Yenong Cao • Haibin Xi • Rakesh Kumar • Jeffrey M. Axten Theodore J. Lampidis

•

Received: 1 January 2013 / Accepted: 10 May 2013 / Published online: 23 May 2013 Ó Springer-Verlag Berlin Heidelberg 2013

Abstract Background Inhibition of glucose metabolism has recently become an attractive target for cancer treatment. Accordingly, since 2-deoxyglucose (2-DG) competes effectively with glucose, it has come under increasing scrutiny as a therapeutic agent. The initial response of tumor cells to 2-DG is growth inhibition, which is thought to conserve energy and consequently protect cells from its ATP-lowering effects as a glycolytic inhibitor. However, since 2-DG also mimics mannose and thereby interferes with N-linked glycosylation, the question is raised of how this sugar analog inhibits tumor cell growth and whether the mechanism by which it protects cells can be Electronic supplementary material The online version of this article (doi:10.1007/s00280-013-2193-y) contains supplementary material, which is available to authorized users. H. Liu M. Kurtoglu T. J. Lampidis (&) Department of Cell Biology and Sylvester Comprehensive Cancer Center, University of Miami Miller School of Medicine, Miami, FL, USA e-mail: [email protected] H. Liu e-mail: [email protected] M. Kurtoglu e-mail: [email protected] Y. Cao Department of Molecular and Cellular Pharmacology, University of Miami Miller School of Medicine, Miami, FL, USA e-mail: [email protected]

manipulated to convert 2-DG-induced growth inhibition to cell death. Methods Cell growth and death were measured via counting viable and dead cells based on trypan blue exclusion. Markers of ATP reduction and the unfolded protein response (UPR) were detected by Western blot. Protein functions were manipulated through chemical compounds, siRNA and the use of gene-specific wild-type and knock-out mouse embryonic fibroblasts (MEFs). Results At 2-DG concentrations that can be achieved in human plasma without causing significant side effects, we find (a) It induces growth inhibition predominantly by interference with glycosylation, which leads to accumulation of unfolded proteins in the endoplasmic reticulum activating the UPR; (b) Inhibition of PERK (but not ATF6 or IRE1), a major component of the UPR, leads to conversion of 2-DGinduced growth inhibition to cell death and (c) secondarily to Present Address: H. Xi Department of Molecular, Cell and Developmental Biology, University of California, Los Angeles, Los Angeles, CA, USA R. Kumar Biology, GlaxoSmithKline, Collegeville, PA, USA e-mail: [email protected] J. M. Axten Medicinal Chemistry, Protein Dynamics DPU, GlaxoSmithKline, Collegeville, PA, USA e-mail: [email protected]

H. Xi Sheila and David Fuente Graduate Program in Cancer Biology, University of Miami Miller School of Medicine, Miami, FL, USA e-mail: [email protected]; [email protected]

123

252

PERK, inhibition of GCN2, a kinase that is activated in response to low intracellular glutamine, increases 2-DG’s cytotoxic effects in PERK -/- MEFs. Conclusions Overall, these findings present a novel anticancer strategy that can be translated into therapeutic gain as they uncover the metabolic target PERK, and to a lesser degree GCN2, that when inhibited convert 2-DG’s static effect to a toxic one in tumor cells growing under normoxia. Keywords 2-Deoxyglucose Endoplasmic reticulum stress Unfolded protein response PERK GCN2 Abbreviations 2-DG 2-Deoxyglucose UPR Unfolded protein response MEF Mouse embryonic fibroblast AMPK AMP-activated protein kinase ER Endoplasmic reticulum PPS Pentose phosphate shunt ROS Reactive oxygen species

Background Although the ‘‘Warburg effect,’’ which identifies increased glycolysis in tumor versus normal tissue, was reported in the 1930s, it is only recently that much attention has been given to investigating inhibitors of glucose metabolism as a means of exploiting this phenomenon for therapeutic gain. Similarly, the glycolytic inhibitor 2-deoxyglucose (2-DG) was previously used mainly as a probe of glucose uptake in a variety of cell types both in vitro and in vivo [1–4]. More recently, due to its inhibitory activity on glycolysis, 2-DG has been shown to selectively kill tumor cells growing under three different conditions of simulated hypoxia in vitro [5, 6]. In vivo proof of principle that 2-DG kills the chemo- and radio-resistant hypoxic cell population of solid tumors has also been recently reported in a transgenic model of retinoblastoma [5, 7, 8]. Moreover, a phase I trial using this strategy has been completed using 2-DG in combination with docetaxel in patients with various tumor types [9]. Results indicate that 2-DG is well tolerated and that there is evidence of clinical benefit justifying phase II studies in several solid tumors. In contrast to cells treated with 2-DG under hypoxic conditions, most tumor types under normoxia are growth inhibited but do not undergo cell death when treated with moderate concentrations of 2-DG [2]. 2-DG is known to block or interfere with different metabolic pathways which result in a cell switching from an anabolic to a catabolic phenotype to conserve energy and survive. It appears that most of the cell’s response to interference with normal

123

Cancer Chemother Pharmacol (2013) 72:251–262

glucose metabolism converges on inhibition of protein synthesis. To achieve this, several possibilities exist each of which may be contributing to the growth inhibitory effects of 2-DG as follows: First is the well-known AMPactivated protein kinase (AMPK) pathway. It can be activated as a consequence of 2-DG’s activity on glycolysis if intracellular ATP levels are reduced enough, which appears to vary from cell type to cell type as well as depend on environmental growth conditions, that is, aerobic versus anaerobic [10]. Activated AMPK leads to inhibition of protein synthesis thru reduction in mTOR activity, thereby slowing cell growth [11]. Second is as a consequence of 2-DG interfering with oligosaccharide synthesis which results in endoplasmic reticulum (ER) stress–induced activation of the unfolded protein response (UPR) [12, 13]. This in turn leads to activation of the UPR signal transducer PERK which phosphorylates the mRNA translation initiation factor eIF2a resulting in lowering of cyclin D1 levels, cessation of the cell cycle and growth inhibition [14, 15]. Phosphorylation of eIF2a leads to attenuation of mRNA translation, which is considered to be a cell survival mechanism by limiting the influx of new proteins into the ER, and thereby reducing further accumulation of misfolded proteins [16]. eIF2a phosphorylation can also occur thru a PERK independent mechanism via GCN2, another eIF2a kinase, resulting in effective shutdown of mRNA translation [16]. GCN2 is known to be activated in response to glutamine depletion. This raises the possibility that as a consequence of 2-DG’s activity as a glycolytic inhibitor, lower pyruvate may induce mitochondria to use glutamine as an energy source by a mechanism known as glutaminolysis. This would result in depleting cytoplasmic glutamine levels and activating GCN2. Thus, it remains unclear which pathway 2-DG induces to effectively inhibit tumor cell growth under normal oxygen conditions. The realization that growth inhibition represents a mechanism by which cells survive treatment with 2-DG has led us to investigate which of the pathways described above play a prominent role in this biologic activity. By identifying this pathway, we hypothesize that interference with its components may allow for converting 2-DGinduced growth inhibition to cell death in tumor cells growing under normoxia. In this paper, we present data which indicate that ER stress leading to an UPR is the major mechanism responsible for 2-DG’s inhibitory activity on tumor cell growth. We find that interference with the UPR and downstream pathways (eIF2a) can convert 2-DG’s static effect to a toxic one in different cell types growing under normal oxygen conditions. Our results show promise for clinical application as compounds that inhibit components of the UPR (PERK inhibitors) have been synthesized, and data presented here demonstrate their activity in tumor cells.


Materials and methods Cell types Human tumor cell lines 1,469 (pancreatic) and MDA-MB435 (melanoma) were purchased from American Type Culture Collection (ATCC) and maintained in DMEM with 1 g/l of glucose (Mediatech). Mouse embryonic fibroblasts (MEFs) that are PERK or IRE1 knock-out and their wildtype pairs were a kind gift from Dr. David Ron (University of Cambridge, Cambridge, UK). MEFs that have wild-type eIF2a and eIF2a with its serine51 mutated to alanine were a kind gift from Dr. Glen Barber (University of Miami, Miami, FL). All MEFs were maintained in RPMI with 2 g/l of glucose (Invitrogen), except the eIF2a wild-type and mutant cells which were maintained in DMEM with 4.5 g/l glucose (Mediatech). All culture media were supplemented with 10 % fetal bovine serum (FBS) (Invitrogen) and penicillin/streptomycin (Invitrogen). Cells were grown under 5 % CO2 at 37 °C. Drugs and antibodies 2-DG, 2-fluoro-deoxy-D-glucose (2-FDG), mannose and glutamine were purchased from Sigma–Aldrich. The following rabbit primary antibodies were from Cell Signaling: GRP78, PERK, GCN2, phospho-GCN2, phospho-eIF2a and cleaved caspase 3. Mouse anti-CHOP and anti-b-actin antibodies were from Cell Signaling and Sigma-Aldrich, respectively. Horseradish peroxidase (HRP)-conjugated anti-rabbit and anti-mouse IgG were purchased from Promega. PERK inhibitors PERK inhibitors were synthesized at GlaxoSmithKline. Their activity against PERK enzyme was evaluated using GST-PERK cytoplasmic domain (aa 536-1116)-mediated phosphorylation of human eIF2a. GSK A, B, and C inhibit PERK enzyme activity with an IC50 of 0.16, 0.79 and 1,259 nM, respectively. Western blot analysis Cells were seeded onto six-well plates and cultured for 18–22 h to reach 40–70 % confluence. Following drug exposure for the indicated times, cells were harvested and lysed with the lysis buffer (100 mM Tris–HCl at pH 7.4, 1 % SDS, phosphatase inhibitor cocktail 2 and protease inhibitor cocktail from Sigma–Aldrich). Protein concentrations of each sample were determined using a Micro BCA Protein Assay Kit (Thermo Scientific)

253

according to the manufacturer’s directions, and equal amounts of proteins were loaded onto 4–15 % Tris–HCl gradient gels (Bio-Rad). After SDS–PAGE, proteins were transferred onto a polyvinylidene fluoride (PVDF) membrane (Millipore), blocked with 5 % milk and probed with corresponding primary antibodies overnight (except 1 h for b-actin). The membrane was washed and probed with secondary antibodies for 1 h. Membrane was then incubated with SuperSignal West Pico or Femto Chemiluminescent Substrate (Thermo Scientific) and signals were visualized on Blue Lite Autorad Films (ISCBioExpress). All primary antibodies were used at 1:1,000 dilution except for b-actin (1:10,000), and the secondary antibodies were used at 1:10,000. Representative blots from at least two independent experiments were shown unless otherwise indicated. Growth inhibition and cytotoxicity assays Cells were seeded onto 24-well plates and cultured for 18–22 h. After drug exposure for 48 h, attached cells and their respective culture media were collected and centrifuged at 400g for 5 min. The pellets were then resuspended in Hanks Balanced Salt Solution (HBSS) (Mediatech) and analyzed with Vi-Cell cell viability analyzer (Beckman Coulter) based on trypan blue exclusion. Growth inhibition was measured by counting viable cells (trypan blue negative) and presented as the percentage of the viable cell number in the corresponding control group. Cell death was shown as the percentage of dead cells (trypan blue positive) out of total cells counted from each group. Data were the averages of triplicate samples ± standard deviation (SD) from one representative experiment out of at least two independent analyses unless otherwise indicated. siRNA transfection Cells were seeded onto 6- or 24-well plates and cultured for 24 h using antibiotics-free media. Then, cells were transfected with anti-Luc siRNA-1 (targeting luciferase) or ONTARGETplus SMARTpool siRNA against PERK, ATF6 or GCN2 using the DharmaFECT siRNA transfection reagent #2 (Dharmacon). Twenty-four to forty-eight hours after transfection, cells were drug treated for Western blot or cytotoxicity analyses, respectively. Statistical analysis Data were compared using two-tailed Student’s t test, and p value less than 0.05 was considered significant.

123

254


a

b

c

d

e

Fig. 1 2-DG inhibits tumor cell growth via ER stress. a MDA-MB435 cell line was treated with the indicated doses of 2-DG for 24 h followed by protein extraction and Western blot analysis of mTOR, AMPK and eIF2a phosphorylation as well as expression of GRP78 and CHOP. Actin was used as a loading control. b MDA-MB-435 cells were treated with either 2-DG or 2-FDG for 48 h at doses indicated. c GRP78 expression was analyzed by Western blot in MDA-MB-435 cells treated with increasing doses of either 2-DG or

2-FDG. Actin was used as a loading control. d MDA-MB-435 cells were treated with indicated doses of 2-DG in the presence or absence of 1 mM exogenous mannose (Man) for 48 h followed by growth inhibition analysis. e Twenty-four hours following the same treatment conditions as in d, proteins were extracted for Western blot analysis of GRP78 expression, while actin was used as a loading control. Each panel is representative of at least 2 experiments. *p \ 0.05; **p \ 0.01

Results

phosphorylated (inhibited) at 2-DG concentrations where growth inhibition as well as increases in UPR markers GRP78 and CHOP is significant (0.2–20 mM) (Fig. 1a, b). Thus, stimulation of either or both ATP and/or ER sensing pathways may be contributing to 2-DG-induced growth inhibition. Due to the lack of a hydroxyl group on its C-2 carbon, 2-DG equally mimics glucose as well as mannose, a critical component involved in N-linked glycosylation. Because of the size similarity between the fluorine group on 2-fluorodeoxy-D-glucose (2-FDG) and the hydroxyl group on glucose, 2-FDG, after converting to its intracellular metabolite 2-FDG-6-phosphate, mimics glucose-6-phosphate better than 2-DG-6-phosphate, and is more potent in allosterically blocking hexokinase and inhibiting glycolysis. However, since the C-2 carbon fluorine group on 2-FDG and hydroxyl group on mannose have opposite conformation, 2-FDG is less potent in mimicking mannose and interfering with glycosylation than 2-DG [2, 17, 18]. As shown in

ER stress is a major mechanism by which 2-DG induces growth inhibition in tumor cells treated under normoxia As indicated in Fig. 1a, 2-DG at low concentrations affects both ATP and ER sensing pathways. At 2-DG concentrations (0.4–6 mM) where growth inhibition is observed (Fig. 1b), the intracellular energy sensor AMPK is detected to be phosphorylated (0.2–20 mM) (Fig. 1a). This is accompanied with a decrease in the phosphorylation of mTOR, which is negatively regulated by AMPK. Similarly, a correlation is found between induction of ER stress, as detected by increased levels of the UPR folding protein chaperone GRP78, and growth inhibition when MDA-MB435 human melanoma cells are treated with 2-DG under aerobic conditions. Moreover, the translation initiation factor, eIF2a, which responds to ER stress, is found to be

123


255

Fig. 1b, 2-FDG displays significantly less growth inhibitory activity than 2-DG, which correlates with less induction of the UPR marker GRP78 at equivalent concentrations, that is, 0.4 mM and 2 mM (Fig. 1c). Moreover, addition of exogenous mannose, which was previously reported to reverse the effects of 2-DG on ER stress but not ATP [10], reverses a significant amount of the growth inhibition and ER stress induced by 2-DG (Fig. 1d, e). Taken together, these results indicate that 2-DG’s induction of ER stress is a major mechanism by which cells are growth inhibited.

the increased proteolysis of these ER stress markers secondary to death in cells treated with the more effective PERK inhibitor, that is, GSK A. Moreover, the doses of 2-DG that are required to induce equivalent amounts of cell death in these cell lines when administered without the PERK inhibitors are more than 10-fold higher and therefore difficult to achieve in patients. Overall, these results clearly demonstrate that PERK inhibitors are effective in converting 2-DG-induced growth inhibition to cell death correlating with their potency to inhibit PERK kinase activity.

Inhibition of PERK kinase converts 2-DG-induced growth inhibition to tumor cell death

siRNA directed against PERK increases UPR markers and UPR-mediated cell death in human melanoma MDA-MB-435 cells treated with 2-DG

It is thought that when activated by ER stress, PERK, a component of the UPR, yields a protective effect by phosphorylating eIF2a, an initiator of protein translation, thereby shutting down further protein synthesis and entry of unfolded proteins into the ER [14, 16]. Thus, we investigated whether by blocking PERK, cells would become hypersensitized to 2-DG. Indeed, at a concentration of 2-DG which is non-toxic in human pancreatic cell line 1,469 (Fig. 2a) and melanoma cell line MDA-MB-435 (Fig. 2b), we find growth inhibition is converted to cell death as a function of PERK inhibition. For these experiments, three PERK inhibitors, GSK A, B and C were used, which display decreasing kinase-blocking activity as determined previously in isolated enzyme assays (refer to methods section for more details) (Fig. 2b). Moreover, as expected, the more potent PERK inhibitor GSK A is able to reduce 2-DG-induced PERK phosphorylation, while GSK B has less and GSK C has no effect on phosphorylated PERK levels at equivalent concentrations (Fig. 2c). It should be noted that in our experimental system, phosphoPERK in human cells proved difficult to be detected via Western blot even when antibodies claimed to be phospho specific were used. However, the antibody we used that recognizes total PERK showed decreased affinity which correlated well with conditions to increase PERK phosphorylation. Therefore, in this study, we use decreased total PERK levels as an indirect measurement of increased phospho-PERK. In agreement with the differential ability of GSK compounds to inhibit PERK phosphorylation, GSK A, as compared to GSK B, is found to block 2-DG-induced phosphorylation of eIF2a better, while GSK C has no effect on it. Interestingly, when GSK A, B or C are combined with 2-DG, increases in the ER stress markers GRP78 and CHOP are detected in the first two but not the latter. It is worth noting that the surprisingly greater increase in these markers induced by GSK B as compared to GSK A in 2-DG-treated cells is accompanied by little or no induction of cell death. This result may be reflective of

To further determine whether the 2-DG-induced toxic effects we observed when co-treated with PERK inhibitors is due to the inhibition of PERK, we transfected MDAMB-435 cells with siRNA directed against PERK. In Fig. 3a, our results demonstrate that 2-DG-induced GRP78 and CHOP are significantly increased in cells when PERK is knocked down as compared to controls. Moreover, knock-down of PERK also results in a statistically significant albeit slight increase in cell death induced by 2-DG (Fig. 3b). Similar upregulation of GRP78 and cytotoxicity are also obtained in PERK knocked-down cells treated with the classical ER stressor tunicamycin (TM) (Online Resource, Figure S1A and B). Interestingly, although it is attenuated, eIF2a is still phosphorylated by 2-DG in PERK knock-down as compared to wild-type cells. This suggests that a kinase other than PERK may be contributing to 2-DG-induced eIF2a phosphorylation. Nonetheless, these experimental results support our findings with PERK inhibitors and hypothesis that inactivating PERK leads to increased toxicity of 2-DG in tumor cells growing under normal oxygen conditions. Deletion of PERK but not IRE1 or knock-down of ATF6, in mouse embryonic fibroblasts augments the toxic effects of 2-DG, which correlates with increased UPR-mediated cell death It is understood that the three known transducers of the UPR pathway, IRE1, ATF6 and PERK, act in concert to protect a cell from undue ER stress. Among these three, PERK is shown to phosphorylate eIF2a and, thereby reducing the protein load of the ER while the first two act more dominantly on upregulating folding machinery such as molecular chaperones [12, 13]. Similar to tumor cells where PERK is inhibited by small molecules or siRNA, mouse embryonic fibroblasts (MEFs) with deletions in PERK (PERK -/- cells) are found to be more sensitive to

123

256


a

c

b

Fig. 2 PERK inhibitors convert 2-DG-induced growth inhibition into cell death. a The PERK inhibitor GSK A was used to treat 1,469 cells either in the presence or absence of 10 mM of 2-DG for 48 h followed by cell death analysis. b Comparison of three PERK inhibitors to convert 2-DG-induced growth inhibition to cell death in MDA-MB-435 cells. Cells were treated with the indicated doses of each PERK inhibitor (GSK A, B and C) in the presence or absence of 2 mM of 2-DG for 48 h followed by cell death analysis. c MDA-MB435 cells were treated with 100 nM of each PERK inhibitor and

2 mM of 2-DG followed by Western blot analysis of PERK, phosphorylated eIF2a, GRP78, CHOP and cleaved caspase 3. Actin was used as a loading control. Note that the PERK antibody recognizes only non-phosphorylated PERK. Therefore, reduction in the intensity of the PERK band is consistent with its phosphorylation. Numbers below GRP78 band indicate densitometric quantification (ImageJ) of GRP78 normalized to actin. Each panel is representative of at least 2 experiments *p \ 0.05; **p \ 0.01

the toxic effects of 2-DG (Fig. 3c) as well as TM (Online Resource, Figure S1C) than their PERK ?/? counterparts. In contrast, 2-FDG, an analog of glucose that is less potent than 2-DG in inducing ER stress and activating the UPR, yields very little or no toxicity in PERK -/- cells at concentrations that do not induce ER stress but block glycolysis effectively. When the same experiments are performed in IRE1 -/- cells, absence of this UPR transducer is found not to augment 2-DG-induced toxicity (Fig. 3d). Similarly, knock-down of the third UPR transducer, ATF6, does not modulate the cytotoxicity of 2-DG

(Fig. 3e). These results suggest that among the UPR transducers, PERK underlies the conversion of 2-DGinduced growth inhibition into cell death. PERK -/- cells display significantly lower levels of phosphorylated eIF2a than their PERK ?/? counterparts when measured as a function of time of treatment with 2-DG (Fig. 3f). This result is in agreement with our previous findings that PERK plays a major role in 2-DGmediated eIF2a phosphorylation. However, eIF2a phosphorylation is found to increase as a function of time in PERK -/- cells (albeit less than that in PERK ?/?

123

Cancer Chemother Pharmacol (2013) 72:251–262 Fig. 3 Knock-down or knockout of PERK converts 2-DGinduced growth inhibition to cell death. a MDA-MB-435 cells were incubated with siRNA against either luciferase (siLuc, as negative control) or PERK (siPERK) for 24 h before addition of 2-DG. Following 24 h of 2-DG treatment, proteins were extracted for Western blot analysis of PERK, phosphorylated eIF2a, GRP78, CHOP and cleaved caspase 3. Actin was used as a loading control. b MDA-MB-435 cells incubated with either siLuc or siPERK were then treated with 8 mM of 2-DG for 48 h followed by cell death analysis. c PERK wild-type (PERK ?/?) and knock-out (PERK -/-) MEFs were treated with indicated doses of 2-DG or 2-FDG for 48 h followed by cell death analysis. d IRE ?/? and IRE -/- MEFs were treated with indicated doses of 2-DG for 24 h followed by cell death analysis. e MEFs transfected with either siLuc or siATF6 were treated with indicated doses of 2-DG for 48 h followed by cell death analysis. Insect shows the ATF6 knock-down efficiency. f PERK wild-type and knock-out cells were treated with 2-DG for the indicated time followed by Western blot analysis of phosphorylated eIF2a. Actin was used as a loading control. Numbers below p-eIF2a band indicate densitometric quantification (ImageJ) of p-eIF2a normalized to actin. Each panel is representative of at least 2 experiments. *p \ 0.05; **p \ 0.01

257

b

a

c

e

d

f

cells), indicating that 2-DG also induces eIF2a phosphorylation through a mechanism independent of PERK. In this regard, it is known that eIF2a can respond to stress signals induced by lowered amino acid content namely thru phosphorylation of GCN2, another eIF2a phosphorylating kinase. In Fig. 4a, b, data is presented which shows 2-DG cytotoxicity is increased in PERK -/- cells when GCN2 is knocked down with siRNA. Interestingly, knock-down of GCN2 in PERK ?/? cells does not alter 2-DG-induced toxicity (Online Resource, Figure S2A) indicating that PERK is relatively more dominant than

GCN2 in maintaining survival when cells are treated with 2-DG. GCN2 is known to respond to depletion of amino acids, and thereby shutting down further protein synthesis by inhibiting eIF2a. With this in mind, we hypothesized that addition of exogenous glutamine to 2-DG-treated PERK -/- cells would block activation of GCN2. This in turn would allow eIF2a to produce more proteins that would lead to increased ER stress and UPR-mediated cell death. This hypothesis is supported by the results in Fig. 4c, d which show increased toxicity in 2-DG-treated PERK -/-,

123

258


a

c

b

d

Fig. 4 Inhibition of GCN2 increases 2-DG-induced cell death in PERK knock-out MEFs. a PERK knock-out MEFs (PERK -/-) were incubated with either siLuc or siGCN2 for 24 h before addition of 2-DG. Following 24 h of 2-DG treatment, proteins were extracted for Western blot analysis of phosphorylated and total GCN2. Actin was used as a loading control. b PERK -/- cells incubated with either siLuc or siGCN2 were treated with 8 mM of 2-DG for 48 h followed by cell death analysis. c PERK -/- cells growing in either normal

growth medium (Gln 4 mM) or medium supplemented with exogenous glutamine (Gln 8 mM) were treated with low or high doses of 2-DG for 24 h followed by Western blot analysis of phosphorylated GCN2, GRP78 or CHOP. Actin was used as a loading control. d PERK -/- cells growing in either 4 or 8 mM glutamine were treated with low or high dose of 2-DG for 48 h followed by cell death analysis. Each panel is representative of at least 2 experiments. *p \ 0.05; **p \ 0.01

in contrast to PERK ?/? (Online Resource, Figure S2B) cells, when co-treated with exogenous glutamine as compared to cells treated with normal glutamine levels. When grown in 4 mM glutamine, 2-DG at concentrations of 2 and 8 mM induces phosphorylation of GCN2. However, when exogenous glutamine is increased to 8 mM, 2-DG-induced GCN2 phosphorylation is found to be attenuated (Fig. 4c). Furthermore, the reduction in GCN2 phosphorylation by exogenous glutamine correlates with increased ER stress and UPR-mediated cell death as assayed by GRP78 and CHOP levels (Fig. 4c). Thus, we interpret these results to indicate that although secondary to PERK, GCN2 also plays a protective role against 2-DG-induced cell death.

Additionally, confirmatory data that the PERK inhibitors used to convert 2-DG growth inhibition to cell death are working thru the UPR/PERK/eIF2a pathway is presented in Fig. 5c. Here, we find that PERK inhibitors do not increase 2-DG toxicity in the mutant cell line as compared to the parental cell line. Overall, this latter result supports our hypothesis that 2-DG-induced growth inhibition can be converted to cell death in tumor cells growing under normoxia, which could yield substantial clinical benefit if this concept and inhibitors can be used in patients.

Genetic inhibition of eIF2a phosphorylation in MEFs renders cells hypersensitive to 2-DG

Overwhelming evidence from PET scans makes it clear that cells undergo a fundamental change in metabolizing glucose when they transform from normal to malignant, confirming data and hypotheses originally presented by Warburg in the 1930s [19]. Studies with radioactive as well as carbon-13-labeled glucose have shown that in addition to supplying energy to a cell, the glucose molecule is used as a skeleton to produce ribose and deoxyribose necessary for RNA and DNA synthesis as well as lipids and certain amino acids necessary for cell growth [20]. It has more

Since both PERK and GCN2 appear to protect the cells against 2-DG toxicity by inhibiting eIF2a, we examined the effects of 2-DG in MEFs that contain a mutant form of eIF2a whose active site cannot be phosphorylated (inhibited). In Fig. 5a, b, it can be seen that this mutant cell line is more sensitive to the toxic effects of 2-DG than its parental counterpart expressing normal eIF2a.

123

Discussion


a

b

c

Fig. 5 2-DG induces greater cell death in MEFs with non-phosphorylatable eIF2a. a MEFs with wild-type (S/S) or mutated (A/A) eIF2a were treated with the indicated doses of 2-DG for 24 h followed by Western blot analysis of GRP78, phosphorylated eIF2a and cleaved caspase 3. Actin was used as a loading control. b Wild-type (eIF2a S/S) and mutant (eIF2a A/A) cells were treated with increasing doses of 2-DG for 48 h followed by cell death analysis. c Wild-type (left side of graph) or mutant (right side of graph) MEF cell lines were treated with a PERK inhibitor (GSK A) in the presence or absence of 4 mM of 2-DG for 48 h followed by cell death analysis. Each panel is representative of at least 2 experiments. *p \ 0.05; **p \ 0.01, ***p \ 0.001

recently been appreciated that nearly all the pathways that are responsible for carcinogenesis in one way or another are involved with altered glucose metabolism [21]. Thus, a tumor cell being continuously driven through the cell cycle by oncogenic pathways must have a mechanism (increased glucose metabolism) to procure the necessary precursors,

259

as well as energy, required for rapid cell growth and division. Data continue to emerge demonstrating the oncogenes that drive carcinogenesis are also responsible for increased glucose metabolism. This provides a fundamental target that can be exploited for selective therapeutic advantage using false sugars such as 2-DG and 2-FDG. Glycolysis and N-linked glycosylation are two of the major glucose metabolism pathways that these sugar analogs interfere with [2]. Previous reports suggested that by inhibiting glycolysis, and thereby depleting intracellular ATP, 2-DG suppresses cell proliferation [22–24]. However, data presented here, which compare the growth inhibitory activity of 2-DG and 2-FDG, argue against this idea. We previously reported that 2-FDG is more potent than 2-DG in blocking glycolysis and killing hypoxic cells, which correlates with its closer structural similarity to glucose than 2-DG. In contrast, because of its structural similarity to mannose, 2-DG is known to be more effective than 2-FDG in interfering with N-linked glycosylation [17, 25]. Thus, although 2-DG inhibits both glycolysis and N-linked glycosylation as indicated in Fig. 1a, it is found to be more potent than 2-FDG in growth-inhibiting tumor cells which indicates that N-linked glycosylation interference is the predominant mechanism by which this occurs. Previously, it was shown that mannose and not glucose can reverse N-linked glycosylation interference by 2-DG [18, 26, 27]. Similarly, here, we find that mannose reverses a significant amount of 2-DG-induced growth inhibition, which correlates with its reversal of the UPR marker, GRP78. A possible explanation for why 2-DG interferes with N-linked glycosylation more effectively than glycolysis may be the 50 times lower extracellular concentration of mannose as compared to glucose found in tissue culture medium as well as in human blood [28]. This will stochiometrically favor 2-DG’s interference with N-linked glycosylation as opposed to glycolysis. Interestingly, mannose reverses growth inhibition almost completely when cells are treated with lower concentrations of 2-DG, while at increased 2-DG concentrations, reversal by mannose is incomplete (Fig. 1d, e). This implies that at higher concentrations of 2-DG, inhibition of pathways other than N-linked glycosylation, that is, glycolysis and/or pentose phosphate shunt may be contributing to the growth inhibitory effects of 2-DG. Interestingly, shutdown of protein translation appears to be more significant in maintaining survival after ER stress induced by 2-DG, since knock-out of IRE1 or knock-down of ATF6 do not augment the growth inhibitory or toxic effects of 2-DG on MEF cells. Thus, growth inhibition via PERK-induced phosphorylation of eIF2a appears to be the main mechanism by which cells survive 2-DG-mediated stress. It follows that inhibition of the PERK pathway may

123

260

convert the growth inhibitory effect of 2-DG into cytotoxicity. Indeed, as shown in Fig. 2, when 2-DG is combined with PERK inhibitors at concentrations where both drugs alone display no toxicity, significant cell death is detected. Moreover, the degree of cell death correlates with the potency of these inhibitors to block PERK phosphorylation, where compounds A and B are significantly more active than compound C. Further support that PERK inhibition leads to 2-DG-induced cell death is evidenced by increased cytotoxicity in tumor cells when transfected with siRNA directed against PERK as well as in PERK -/MEFs (Fig. 3). Additionally, our data suggest that the conversion of 2-DG-induced growth inhibition to cell death occurs via augmenting ER stress, leading to UPR-mediated cell death as demonstrated by higher levels of GRP78 and CHOP (Figs. 2c, 3a). This conclusion is further corroborated by the observation that the classical ER stress inducer TM results in similar increases in UPR markers and cell death when PERK is compromised (Online Resource, Figure S1). It should be noted that although the pro-apoptotic ER stress effector CHOP is a well-known downstream target of the PERK-eIF2a signaling, it has also been reported to be regulated by the ATF6 and IRE1 branch of the UPR [29– 31]. Therefore, the increased CHOP expression observed with 2-DG treatment when PERK is inhibited or knocked down may be explained by enhanced ATF6 and/or IRE1 activation due to ER stress. Although eIF2a phosphorylation is found to be lower in PERK -/- as compared to PERK ?/? cells when treated with 2-DG, the former cells displayed temporal increases in phosphorylated eIF2a levels. This observation indicates that in response to 2-DG, pathways other than PERK play a role in phosphorylating eIF2a which may also influence cell survival. In this regard, it is known that eIF2a can also be phosphorylated by GCN2, which is activated following intracellular glutamine depletion [32]. Thus, we find that 2-DG-induced cell death is increased in PERK -/- cells by transfecting with siRNA directed against GCN2 or by increasing the amount of glutamine (Fig. 4). The observation that under similar conditions 2-DG toxicity is not increased in PERK ?/? cells (Online Resource, Figure S2) indicates that GCN2 may play a secondary role to PERK in protecting cells against 2-DG-induced cell death. In a recently published report, however, GCN2 was observed to play a toxic role when cells were either deprived of glucose or treated with 2-DG [33]. In that study, cells were treated with much higher concentrations of 2-DG, 50 mM, which most likely are affecting other pathways of glucose metabolism that could account for the differences seen between our results and theirs. One possibility is that at very high concentrations, 2-DG affects the pentose phosphate shunt (PPS). It has previously been

123


reported that 50 mM 2-DG increases intracellular reactive oxygen species (ROS) that play a role in the toxic effects of 2-DG in cells grown under normal oxygen conditions [34– 36]. An explanation offered for increased ROS in these cells is that 2-DG blocks the PPS leading to lower NADPH levels, which in turn leads to decreased levels of reduced glutathione resulting in increased intracellular ROS and cell death [34–36]. However, it was also reported that 2-DG can be used as a substrate by glucose-6-phosphate dehydrogenase to synthesize NADPH that can maintain survival when oxidative stress in a cell is increased [37]. Thus, it appears that 2-DG may inhibit PPS only if its concentration is high enough (i.e., 50 mM) to compete with exogenous glucose for both hexose transporters and hexokinase. Therefore, at the lower concentrations of 2-DG used in our studies, our results indicate that primarily Nlinked glycosylation and to a lesser extent glycolysis but not PPS are inhibited. It follows that at low concentration of 2-DG, pyruvate and not NADPH levels would be decreased, triggering activation of glutaminolysis, thereby decreasing intracellular levels of glutamine and activating GCN2. Our results (Fig. 4) where addition of exogenous glutamine leads to increased cell death in PERK -/- cells treated with low concentrations of 2-DG fit best with this possibility. Future experiments to determine whether at low 2-DG concentrations pyruvate levels are reduced, activating glutaminolysis as well as GCN2 will be useful in verifying this possibility. Although glucose deprivation and synthetic inhibition of glucose metabolism with sugar analogs, such as 2-DG, are often interpreted to be one in the same, our results point to a marked difference between these two conditions. Glucose deprivation simultaneously inhibits PPS, glycolysis and Nlinked glycosylation due to lack of substrate, whereas it appears that the concentration of 2-DG determines which of these pathways are inhibited. Our results suggest a hierarchy of 2-DG’s activity as follows: at low concentrations, it interferes primarily with N-linked glycosylation, while at moderate concentrations glycolysis is inhibited more and only at high concentrations can the PPS be blocked. Thus, the outcomes of 2-DG treatment may be better interpreted in the context of the concentrations used. Our findings indicate that when treated with low-tomoderate concentrations of 2-DG that have been reported to be clinically achievable [9], activation of PERK and GCN2 yield a protective effect on cell survival by phosphorylating eIF2a and inhibiting its activity. Thus, the increased toxicity of 2-DG we find in MEF cells expressing a non-phosphorylatable eIF2a ([60 % cell death) as compared to wild-type MEF cells (\12 %) (Fig. 5) is in accordance with this thinking. Although our data generally agree with previously published results [33], in these same cells treated with 2-DG, they differ in the degree of toxicity


261

2-DG

Interfere with Glycosylation

Inhibition of Glycolysis

blocking of eIF2a phosphorylation can account for the marked difference in toxicity between the mutants and the wild-type. Indeed, the absence of synergism between PERK inhibitors and 2-DG in the non-phosphorylatable eIF2a mutants further supports this interpretation.

ER Stress

Glutamine

Conclusions

Glutamine

p-PERK

p-GCN2

p-eIF2

Growth Inhibition

Survival Fig. 6 2-DG competes with both glucose and mannose to inhibit glycolysis and N-linked glycosylation, respectively. Since extracellular as well as intracellular concentrations of mannose are at least 10-fold less than glucose, 2-DG at clinically achievable doses interferes with N-linked glycosylation more readily than glycolysis. Therefore, 2-DG-induced growth inhibition in tumor cells is predominantly a consequence of interference with N-linked glycosylation that results in ER stress. When ER function is perturbed, eIF2a becomes phosphorylated via PERK to prevent further protein loading into ER, which suppresses proliferation and allows the cell to recover from ER stress. Therefore, inhibition of PERK increases 2-DG-induced ER stress and consequently converts growth inhibition to cell death. When the PERK pathway is inhibited in tumor cells treated with 2-DG, eIF2a can still be phosphorylated by GCN2. We hypothesize that by inhibiting glycolysis, 2-DG lowers intracellular glutamine and thereby activating GCN2. Overall, inhibition of PERK and to a lesser extent GCN2 converts 2-DG-induced growth inhibition to cell death in tumor cells growing under normal oxygen levels

between wt and mutant eIF2a MEFs. In the previous study, at high 2-DG concentrations, the difference in toxicity between the wt and mutant cell lines are shown to be statistically non-significant although the average % cell death values are slightly greater in the mutant cell line (70 %) versus the wt cells (58 %). Here, again at the high concentration of 2-DG (50 mM) used in their study, mechanisms other than absence of eIF2a phosphorylation may be contributing to the smaller differences in 2-DG-induced toxicity between eIF2a mutants and wild-type. However, at the moderate-to-low concentrations used in our studies,

In summary, we find that 2-DG interferes with both glycolysis and N-linked glycosylation, activating respective response pathways involved with growth inhibition and survival of tumor cells when treated under normoxic conditions. However, our data strongly implicate interference with N-linked glycosylation rather than glycolysis as the predominant means by which 2-DG induces growth inhibition mediated through the UPR. Therefore, by blocking a critical component of the UPR, PERK, a major mechanism responsible for cancer cell protection from ER stress is overcome and 2-DG’s growth inhibitory effects become toxic (Fig. 6). Overall, our data indicate that use of PERK inhibitors combined with low (clinically achievable) concentrations of 2-DG offers a novel treatment strategy which may be universally applicable in a wide variety of tumor cell types growing under aerobic conditions. Acknowledgments We thank Dr. David Ron (University of Cambridge, Cambridge, UK) for the PERK and IRE1 wild-type and knockout MEF cell pairs. We also want to thank Dr. Glen Barber (University of Miami, Miami, FL) for providing us with the eIF2a wild-type and serine51 mutant MEFs. This study is supported by the National Cancer Institute grant CA37109 and Pap Corps award to TJL. Conflict of interest RK and JMA are employees of GlaxoSmithKline. HL, MK, YC, HX and TJL declare that they have no competing interests.

References 1. Burckhardt D, Stalder GA (1975) Cardiac changes during 2-deoxy-D-glucose test. A study in patients with selective vagotomy and pyloroplasty. Digestion 12(1):1–8 2. Kurtoglu M, Maher JC, Lampidis TJ (2007) Differential toxic mechanisms of 2-deoxy-D-glucose versus 2-fluorodeoxy-D-glucose in hypoxic and normoxic tumor cells. Antioxid Redox Signal 9(9):1383–1390 3. Duke WW, Hirschowitz BI, Sachs G (1965) Vagal stimulation of gastric secretion in man by 2-deoxy-D-glucose. Lancet 2(7418):871–876 4. Maher JC, Savaraj N, Priebe W, Liu H, Lampidis TJ (2005) Differential sensitivity to 2-deoxy-D-glucose between two pancreatic cell lines correlates with GLUT-1 expression. Pancreas 30(2):e34–e39 5. Boutrid H, Jockovich ME, Murray TG, Pina Y, Feuer WJ, Lampidis TJ, Cebulla CM (2008) Targeting hypoxia, a novel

123

262

6.

7.

8.

9.

10.

11.

12. 13. 14.

15.

16.

17.

18.

19. 20. 21.

Cancer Chemother Pharmacol (2013) 72:251–262 treatment for advanced retinoblastoma. Invest Ophthalmol Vis Sci 49(7):2799–2805 Maher JC, Krishan A, Lampidis TJ (2004) Greater cell cycle inhibition and cytotoxicity induced by 2-deoxy-D-glucose in tumor cells treated under hypoxic vs aerobic conditions. Cancer Chemother Pharmacol 53(2):116–122 Houston SK, Pina Y, Murray TG, Boutrid H, Cebulla C, Schefler AC, Shi W, Celdran M, Feuer W, Merchan J, Lampidis TJ (2011) Novel retinoblastoma treatment avoids chemotherapy: the effect of optimally timed combination therapy with angiogenic and glycolytic inhibitors on LH(BETA)T(AG) retinoblastoma tumors. Clin Ophthalmol 5:129–137 Pina Y, Houston SK, Murray TG, Boutrid H, Celdran M, Feuer W, Shi W, Hernandez E, Lampidis TJ (2010) Focal, periocular delivery of 2-deoxy-D-glucose as adjuvant to chemotherapy for treatment of advanced retinoblastoma. Invest Ophthalmol Vis Sci 51(12):6149–6156 Raez L, Papadopoulos K, Ricart A, Chiorean G, DiPaola R, Stein M, Rocha Lima C, Schlesselman J, Tolba K, Langmuir V, Kroll S, Jung D, Kurtoglu M, Rosenblatt J, Lampidis TJ (2013) A phase I dose-escalation trial of 2-deoxy-D-glucose alone or combined with docetaxel in patients with advanced solid tumors. Cancer Chemother Pharmacol 71(2):523–530 Xi H, Kurtoglu M, Liu H, Wangpaichitr M, You M, Liu X, Savaraj N, Lampidis TJ (2011) 2-Deoxy-D-glucose activates autophagy via endoplasmic reticulum stress rather than ATP depletion. Cancer Chemother Pharmacol 67(4):899–910 Kimura N, Tokunaga C, Dalal S, Richardson C, Yoshino K, Hara K, Kemp BE, Witters LA, Mimura O, Yonezawa K (2003) A possible linkage between AMP-activated protein kinase (AMPK) and mammalian target of rapamycin (mTOR) signalling pathway. Genes Cells 8(1):65–79 Schroder M, Kaufman RJ (2005) ER stress and the unfolded protein response. Mutat Res 569(1–2):29–63 Parodi AJ (2000) Protein glucosylation and its role in protein folding. Annu Rev Biochem 69:69–93 Harding HP, Zhang Y, Bertolotti A, Zeng H, Ron D (2000) Perk is essential for translational regulation and cell survival during the unfolded protein response. Mol Cell 5(5):897–904 Brewer JW, Diehl JA (2000) PERK mediates cell-cycle exit during the mammalian unfolded protein response. Proc Natl Acad Sci USA 97(23):12625–12630 Shang J, Gao N, Kaufman RJ, Ron D, Harding HP, Lehrman MA (2007) Translation attenuation by PERK balances ER glycoprotein synthesis with lipid-linked oligosaccharide flux. J Cell Biol 176(5):605–616 Lampidis TJ, Kurtoglu M, Maher JC, Liu H, Krishan A, Sheft V, Szymanski S, Fokt I, Rudnicki WR, Ginalski K, Lesyng B, Priebe W (2006) Efficacy of 2-halogen substituted D-glucose analogs in blocking glycolysis and killing ‘‘hypoxic tumor cells’’. Cancer Chemother Pharmacol 58(6):725–734 Schwarz RT, Datema R (1982) The lipid pathway of protein glycosylation and its inhibitors: the biological significance of protein-bound carbohydrates. Adv Carbohydr Chem Biochem 40:287–379 Warburg O (1956) On the origin of cancer cells. Science 123(3191):309–314 Lane AN, Fan TW, Higashi RM (2008) Stable isotope-assisted metabolomics in cancer research. IUBMB Life 60(2):124–129 Hanahan D, Weinberg RA (2011) Hallmarks of cancer: the next generation. Cell 144(5):646–674

123

22. Karczmar GS, Arbeit JM, Toy BJ, Speder A, Weiner MW (1992) Selective depletion of tumor ATP by 2-deoxyglucose and insulin, detected by 31P magnetic resonance spectroscopy. Cancer Res 52(1):71–76 23. McComb RB, Yushok WD (1964) Metabolism of ascites tumor cells. III. Effect of 2-deoxyglucose phosphorylation on phosphorus metabolism. Cancer Res 24:193–197 24. Sridhar R, Stroude EC, Inch WR (1979) Cytotoxicity of glucose analogues in V79 multicell spheroids. In Vitro 15(9):685–690 25. Kurtoglu M, Gao N, Shang J, Maher JC, Lehrman MA, Wangpaichitr M, Savaraj N, Lane AN, Lampidis TJ (2007) Under normoxia, 2-deoxy-D-glucose elicits cell death in select tumor types not by inhibition of glycolysis but by interfering with N-linked glycosylation. Mol Cancer Ther 6(11):3049–3058 26. Kurtoglu M, Philips K, Liu H, Boise LH, Lampidis TJ (2010) High endoplasmic reticulum activity renders multiple myeloma cells hypersensitive to mitochondrial inhibitors. Cancer Chemother Pharmacol 66(1):129–140 27. Schwarz RT, Schmidt MF, Datema R (1979) Inhibition of glycosylation of viral glycoproteins. Biochem Soc Trans 7(2):322–326 28. Etchison JR, Freeze HH (1997) Enzymatic assay of D-mannose in serum. Clin Chem 43(3):533–538 29. Yoshida H, Okada T, Haze K, Yanagi H, Yura T, Negishi M, Mori K (2000) ATF6 activated by proteolysis binds in the presence of NF-Y (CBF) directly to the cis-acting element responsible for the mammalian unfolded protein response. Mol Cell Biol 20(18):6755–6767 30. Ma Y, Brewer JW, Diehl JA, Hendershot LM (2002) Two distinct stress signaling pathways converge upon the CHOP promoter during the mammalian unfolded protein response. J Mol Biol 318(5):1351–1365 31. Sailaja GS, Bhoopathi P, Gorantla B, Chetty C, Gogineni VR, Velpula KK, Gondi CS, Rao JS (2013) The secreted protein acidic and rich in cysteine (SPARC) induces endoplasmic reticulum stress leading to autophagy-mediated apoptosis in neuroblastoma. Int J Oncol 42(1):188–196 32. Wek RC, Jiang HY, Anthony TG (2006) Coping with stress: eIF2 kinases and translational control. Biochem Soc Trans 34(Pt 1):7–11 33. Muaddi H, Majumder M, Peidis P, Papadakis AI, Holcik M, Scheuner D, Kaufman RJ, Hatzoglou M, Koromilas AE (2010) Phosphorylation of eIF2alpha at serine 51 is an important determinant of cell survival and adaptation to glucose deficiency. Mol Biol Cell 21(18):3220–3231 34. Coleman MC, Asbury CR, Daniels D, Du J, Aykin-Burns N, Smith BJ, Li L, Spitz DR, Cullen JJ (2008) 2-Deoxy-D-glucose causes cytotoxicity, oxidative stress, and radiosensitization in pancreatic cancer. Free Radic Biol Med 44(3):322–331 35. Lin X, Zhang F, Bradbury CM, Kaushal A, Li L, Spitz DR, Aft RL, Gius D (2003) 2-Deoxy-D-glucose-induced cytotoxicity and radiosensitization in tumor cells is mediated via disruptions in thiol metabolism. Cancer Res 63(12):3413–3417 36. Simons AL, Mattson DM, Dornfeld K, Spitz DR (2009) Glucose deprivation-induced metabolic oxidative stress and cancer therapy. J Cancer Res Ther 5(Suppl 1):S2–S6 37. Le Goffe C, Vallette G, Charrier L, Candelon T, Bou-Hanna C, Bouhours JF, Laboisse CL (2002) Metabolic control of resistance of human epithelial cells to H2O2 and NO stresses. Biochem J 364(Pt 2):349–359