Switching from Aerobic Glycolysis to Oxidative Phosphorylation Modulates the Sensitivity of Mantle Cell Lymphoma Cells to TRAIL

Home , Glycolysis

ORIGINAL ARTICLE Switching from aerobic glycolysis to oxidative phosphorylation modulates the sensitivity of mantle cell lymphoma cells to TRAIL

GL Robinson, D Dinsdale, M MacFarlane and K Cain

TRAIL (TNF (tumour necrosis factor)-related apoptosis-inducing ligand) a putative anti-cancer cytokine induces apoptosis through DISC (death-inducing signalling complex)-mediated activation of caspase-8 and/or cleavage of Bid. TRAIL is relatively specific for tumour cells but primary chronic lymphocytic leukaemia and mantle cell lymphoma (MCL) cells are resistant. Herein, we show that cellular metabolism influences cell death and that MCL cells (Z138 cell line) can survive/proliferate in glucose-free media by switching from aerobic glycolysis to ‘coupled’ oxidative phosphorylation. Extracellular flux analysis and mitochondrial inhibitors reveal that in the absence of glycolysis, Z138 cells have enhanced respiratory capacity coupled to ATP synthesis, similar to ‘classical’ state 3 mitochondria. Conversely, 2-deoxyglucose (2DG) blocked glycolysis and partially inhibited glycolytic- dependent oxidative phosphorylation, resulting in a 50% reduction in cellular ATP levels. Also, 2DG sensitised Z138 cells to TRAIL and induced a marked decrease in caspase-8, -3, cFLIPS, Bid and Mcl-1 expression but Bak remained unchanged, altering the Mcl-1/Bak ratio, facilitating cytochrome c release and cell death. Conversely, under glucose-free conditions, Z138 cells were less sensitive to TRAIL with reduced TRAIL-R1/R2 surface receptor expression and impaired DISC formation. Anti-apoptotic proteins Bcl-2 and XIAP were up-regulated while pro-apoptotic BAX was down-regulated. Additionally, mitochondria had higher levels of cytochrome c and ultrastucturally exhibited a condensed configuration with enhanced intracristal spaces. Thus, metabolic switching was accompanied by mitochondrial proteome and ultrastructural remodelling enabling enhanced respiration activity. Cytochrome c release was decreased in glucose-free cells, suggesting that either pore formation was inhibited or that cytochrome c was more tightly bound. Glucose-free Z138 cells were also resistant to intrinsic cell death stimuli (ABT-737 and ionising radiation). In summary, in MCL cells, the anti-glycolytic effects of 2DG and glucose restriction produced opposite effects on TRAIL-induced cell death, demonstrating that mitochondrial metabolism directly modulates sensitivity of tumour cells to apoptosis.

Oncogene (2012) 31, 4996--5006; doi:10.1038/onc.2012.13; published online 6 February 2012 Keywords: TRAIL; apoptosis; oxidative phosphorylation; aerobic glycolysis; mitochondria; 2DG

INTRODUCTION The extrinsic pathway is initiated by the extracellular binding of Otto Warburg1 first described how cancer cells utilise aerobic members of the tumour necrosis factor (TNF) ligand family, such glycolysis rather than oxidative phosphorylation---a phenomena as TNFa, CD95L/FasL or TNF-related apoptosis-inducing ligand now known as the ‘Warburg effect’. In cancer cells, even under (TRAIL), to their cognate transmembrane death receptors.7 TRAIL, normoxic conditions, there is enhanced conversion of glucose to unlike CD95L or TNFa, can be selectively toxic to tumour cells8--11 lactate with decreased mitochondrial respiration. This results and initiates apoptosis through the death-inducing signalling in increased glucose demand, which is used diagnostically with complex (DISC). TRAIL-R1/-R2 ligation leads to recruitment of the FDG-PET scanning whereby the non-metabolisable fluorodeoxy- adaptor molecule Fas-associated death domain (FADD), which gluocose analogue is used to image tumours.2 During neoplasia, then recruits procaspase-8 to form the DISC. Caspase-8, activated extensive metabolic reprogramming supports tumour cell growth within the DISC, initiates a caspase cascade activating the and proliferation3 and although aerobic glycolysis is an inefficient executioner caspases-3/-7.12 --14 Originally described for CD95, method of producing ATP, it allows cancer cells to utilise glucose, different cell types respond to death ligands as either; type I cells, which is in high concentration. This provides key intermediates which rapidly form large amounts of DISC and activate caspase-8 necessary for the pentose phosphate pathway, which provides and -3 directly or type II cells, where comparatively small amounts ribose sugars (nucleotide synthesis), glycerol and citrate (lipid of DISC are formed and the intrinsic pathway is activated by synthesis) and non-essential amino acids. Since the switch from cleaved Bid.15 Type I and II responses have also been described for mitochondrial respiration to glycolysis appears to be a hallmark of TRAIL where blockade of the mitochondrial amplification arm by 16 cancer cells, there is increasing interest in using and developing either over-expression of anti-apoptotic Bcl-2/Bcl-XL, or knock- glycolytic inhibitors to target tumour cells.4 out of caspase-917 or Bax18 attenuates TRAIL-induced apoptosis. Treatments for cancer largely work by inducing apoptosis Modulation of glucose metabolism has been reported to affect through either the extrinsic or intrinsic (mitochondrial) pathway. intrinsic and extrinsic apoptotic cell death. For example, inhibition The latter involves cytotoxic stress (for example, irradiation of doxorubicin-induced apoptosis by the anti-glycolytic analogue or chemical), cytochrome c release, Apaf-1 oligomerisation 2-deoxyglucose (2DG) leads to cellular depletion of ATP and and caspase-9 recruitment to form the apoptosome complex.5,6 abrogation of DNA cleavage.19 Similarly, oligomycin-induced ATP

MRC Toxicology Unit, University of Leicester, Leicester, UK. Correspondence: Professor M MacFarlane or Professor K Cain, MRC Toxicology Unit, University of Leicester, Hodgkin Building, Lancaster Road, Leicester LE1 9HN, UK. E-mail: [email protected] or [email protected] Received 4 July 2011; revised 6 December 2011; accepted 22 December 2011; published online 6 February 2012 Metabolism modulates MCL sensitivity to TRAIL GL Robinson et al 4997 depletion blocked the intrinsic pathway induced by anti-cancer residual ECAR, suggesting that this residual acidification is most agents but failed to inhibit CD95L-induced apoptosis,20 support- likely due to a rotenone-sensitive ETC-dependent ‘proton leak’ ing earlier studies showing that ATP levels determine whether a from the mitochondria (Figure 1b). cell dies by apoptosis or necrosis.21 ATP is needed for both Previous findings have shown that 2DG potentiates TNFa killing intrinsic and extrinsic-mediated apoptotic cell death, and apopto- in U937 cells23 and CD95 and TRAIL-induced killing in HeLa, MCF-7 tic cells appear to have elevated levels of ATP.22 However, some and melanoma cells.25,28 Similarly, Z138 cells treated with 2DG studies have shown that cells deprived of glucose have decreased were six- to eightfold (Figure 1g) more sensitive to TRAIL-induced ATP levels and are sensitised to death-receptor-mediated apop- cell killing (EC50 ¼ 25--30 ng/ml) compared with control cells 21,23 --26 tosis. We therefore decided to investigate the effect of (TRAIL EC50 ¼ B180--200 ng/ml). Cleavage of caspase-8, -3 and chronic glucose deprivation on cellular metabolism, ATP levels PARP, a caspase-3 substrate, was detectable with lower concen- and TRAIL-induced apoptosis. Using XF (extracellular flux) analysis, trations (25 ng/ml) in 2DG-treated cells compared with 100 ng/ml we have measured mitochondrial oxidative phosphorylation and for control cells (Figure 1h). 2DG alone increased the endogenous aerobic glycolysis in a mantle cell lymphoma (MCL) cell line (Z138). level of cell death from B10 to 20% (Figure 1g) and flow Prolonged exposure of Z138 cells to glucose-free media, cytometry showed that this was due to increased apoptotic and supplemented with pyruvate/glutamine, mediated a metabolic secondary necrotic cell death (Supplementary Figures S1a and b), switch from aerobic glycolysis to oxidative phosphorylation. Under which was largely insensitive to 20 mM zVAD.fmk (Supplementary these conditions, ATP levels were maintained and the cells were Figure S1c). This contrasted with TRAIL-induced caspase-8 significantly less sensitive to TRAIL and ABT-737. Conversely, activation and cell death, which was sensitive to zVAD.fmk inhibiting aerobic glycolysis with 2DG decreased ATP levels and (Figure 1g). TRAIL DISC formation, as determined by FADD and enhanced sensitivity to TRAIL. Glucose-free conditions induced caspase-8 recruitment to TRAIL-R1/R2 was rapid in both control significant changes in pro- and anti-apoptotic proteins. We and 2DG-treated cells (Figure 1i). Interestingly, TRAIL-R1 migrated propose that ATP levels regulate the levels of mitochondrially on SDS--PAGE (sodium dodecyl sulphate polyacrylamide gel associated pro- and anti-apoptotic proteins and suggest that, in electrophoresis) as two separate bands, which after 2DG some cells, oxidative phosphorylation can supply enough ATP to treatment migrated faster (Figure 1i), indicating decreased enable them to survive glucose-free conditions. molecular weight, possibly due to changes in post-translational modification (see later). The effects of 2DG on TRAIL induced caspase-mediated RESULTS apoptosis, but not on cellular ATP levels, were inhibited by the 2DG inhibits glycolysis and sensitises Z138 cells to TRAIL-induced poly caspase inhibitor zVAD.fmk (Figures 1f and g). Thus, the apoptosis metabolic effects of 2DG on ATP production were not due to Since aerobic glycolysis is a hallmark of cancer cells, we first caspase activity. Following 2DG treatment, pro-and anti-apoptotic investigated how glycolytic and oxidative phosphorylation status proteins including caspase-8, caspase-3, BID, Bax, cFLIPS Mcl-1 and affects TRAIL-induced apoptosis. First, we determined the effect of BCL-XL were down-regulated (Figure 1j). Mcl-1 in particular was 2DG on Z138 cells using a Seahorse XF analyser (Billerica, MA, USA) significantly depleted in 2DG-treated cells, in agreement with a to simultaneously measure cellular rates of oxygen consumption recent study using other human cancer cell lines.29 (OCR) and lactate production by extracellular acidification (ECAR; Figures 1a--e). Z138 cells maintained on glucose/pyruvate/ Z138 cells cultured in glucose-free media switch from aerobic glutamine were incubated with or without 5 mM 2DG and real- glycolysis to oxidative phosphorylation and are less sensitive time OCR and ECAR measured (Figures 1a and b). Without 2DG, to TRAIL-induced apoptosis Z138 cells exhibited basal OCR levels of B600 pmol/min/106 cells Since 2DG markedly inhibited cellular ATP levels, we investigated (Figure 1c) coupled with ECAR values of B200 mpH/min/106 cells whether glucose removal from pyruvate and glutamine supple- (Figure 1d). In comparison, incubating Z138 cells with 2DG for 20 h mented culture media would produce similar effects on metabo- reduced basal OCR by 35% to B400 pmol/min/106 cells lism, ATP levels and TRAIL sensitivity. We therefore investigated (Figure 1c), demonstrating that B1/3rd of the basal OCR is the effect of glucose deprivation on Z138 cells by switching the derived from glucose metabolism. The remaining respiratory cells to glucose-free media supplemented with pyruvate and capacity is derived from pyruvate and glutamine via the TCA cycle glutamax (L-alanyl-L-glutamine). Total cell death in cells switched and oxidative phosphorylation. Since 2DG is a competitive to minus glucose media for 1 h showed a small increase in inhibitor of hexokinase,27 it had a much greater effect on the sensitivity to TRAIL as compared with glucose-maintained cells glycolytic rate of Z138 cells where the basal ECAR was reduced by (Supplementary Figure S2a). However, analysis by flow cytometry 80% to B40 mpH/min/106 cells (Figure 1d). Thus, 2DG inhibition showed that this was due to a substantial increase in secondary shifted the OCR/ECAR ratio of the cells to a predominantly necrosis (red bars, Supplementary Figures S2c and e). However, oxidative phosphorylation-dependent metabolism (Figure 1e), after 20 h in glucose-free media, Z138 cells were much less coupled with an B50% reduction in ATP levels (Figure 1f). sensitive to TRAIL-mediated cell death (Supplementary Figure S2b) To define the metabolic profile of 2DG-treated cells, the effects and there was a marked reduction in both the endogenous levels of the protonophore FCCP and the electron transport chain (ETC) of necrotic and apoptotic cell death (Supplementary Figures S2b, complex I inhibitor rotenone were determined (Figures 1a and b). d and f), which is in contrast to the effects observed with 2DG FCCP, in well-coupled mitochondria, depolarises the mitochondrial (Figure 1g; Supplementary Figure S1) We next determined that membrane and stimulates oxygen consumption via the ETC. In Z138 cells maintained on glucose (À) media for a minimum of 1 Z138 cells incubated with or without 2DG, FCCP induced a small week were viable, although the cells proliferated more slowly with increase (20%) in OCR, suggesting that the mitochondria of Z138 a doubling time of 35--39 h compared with glucose-maintained cells were either not tightly coupled or working at near maximal ( þ ) cells (doubling time 24--24 h). XF analysis showed that cells levels of respiration (see Discussion). The small increase in OCR cultured in glucose (À) media had increased levels of oxidative produced by FCCP was observed with and without 2DG phosphorylation (Figure 2a) with basal OCR B60% higher than consistently in three independent experiments. Furthermore, cells cultured in glucose ( þ ) media (Figure 2c). Conversely, ECAR rotenone produced almost complete inhibition (495%) of the activity in glucose ( þ ) cells was B6-fold higher than in the OCR in --2DG cells (Figure 1a) but had very little effect on the ECAR glucose (À) cells (Figures 2b and d), demonstrating almost (Figure 1b), demonstrating that this respiration was almost totally complete elimination of glycolysis. Residual ECAR in glucose (À) derived from the ETC. In 2DG-treated cells, rotenone inhibited the cells (B10 mph/min) may be derived from a small ‘proton leak’

& 2012 Macmillan Publishers Limited Oncogene (2012) 4996 --5006 Metabolism modulates MCL sensitivity to TRAIL GL Robinson et al 4998

- 2DG + 2DG 400 1000 1000 FCCP Rotenone n.s

s) 800 800 300

cell

cells)

6 600 6 600 200

in/10

pmol/min) 400 400 - 2DG

BASAL OCR

OCR ( 100 + 2DG

(pmol/m 200 (pmol/m 200 Basal 0 0 0 0 40 120 200 280 BASAL ECAR (mpH /min/106 cells)

5 100 250

4 80 s) 200

cell 60 6 150 3

mpH/min)

ar ATP 40 in/10 100 2 Basal (AFLUS)

Cellul

BASAL ECAR ECAR ( 20 50 * 1

(mpH /m 0 0 0 0 20 40 60 80 2DG + - Time (min) -2DG +2DG +2-DG+ zVAD.fmk

2DG - + 100

TRAIL (ng/ml) 80 01000 0 1000 WB Pro-Casp-8 60 Caspase 8 p43/41

Death 40

% Cell Pro-Casp-3 20 + 2DG - 2DG Caspase 3 p20 + 2DG + zVAD.fmk p19 0 p17 0 200 400 600 800 1000 p116 TRAIL (ng/ml) PARP p85

-2DG +2DG -2DG +2DG Pro-apoptotic Anti-apoptotic TRAIL ----++++ WB 2DG -+ WB -+ W.B. Pro-Casp-8 cFLIPs TRAIL-R1 * Pro-Casp-3 MCl-1 TRAIL-R2 cFLIPL BCL-2

BID FADD BCL-XL

BAX Tubulin Caspase 8 BAK

Beads Supernatant Tubulin

Figure 1. 2DG inhibition of glycolysis sensitises Z138 cells to TRAIL-induced apoptosis. Z138 cells were cultured for 20 h in 11 mM glucose, 1mM pyruvate and 2 mM glutamax supplemented media with (&) or without (K)5mM 2DG. Real-time measurements (mean±s.e.m., n ¼ 3) of (a) oxidative phosphorylation (OCR) and (b) glycolysis (ECAR) in 24-well Seahorse assay plates (4.0 Â 106 cell/well) were carried out as described in Materials and methods. Graphs are representative of three independent experiments and show the effects of mitochondrial inhibitors, FCCP (400 nm) and rotenone (1 mM) injected as indicated. Basal OCR (c) and ECAR (d) values were calculated from the mean basal values of each real-time run normalised to cell number (mean±s.e.m., n ¼ 3). Where indicated, cumulative data were analysed by Student’s t- test, n.s., not signiﬁcant and *Po0.05. The effect of 2DG on the balance between glycolysis (ECAR) and oxidative phosphorylation (OCR) is shown in (e). (f) ATP levels of Z138 cells treated without (open bar), with 2DG (black bar) or with 2DG plus zVAD.fmk (hatched bar) were measured as described in Materials and methods (mean±s.e.m., n ¼ 3). (g) Z138 cells were cultured for 20 h without 2DG (K), with 5 mM 2DG (&) or with 5 mM 2DG þ zVAD.fmk (D) followed by a further 4 h incubation with varying concentrations of TRAIL (0--1000 ng/ml) before analysing for % cell death (mean±s.e.m., n ¼ 3) as described in Materials and methods. Cell pellets from (g) were analysed for caspase-8, -3 and PARP cleavage by immunoblotting with antibodies (W.B.) to the indicated proteins (h). (i) DISC formation was assayed in Z138 cells cultured with or without 5 mM 2DG for 20 h and treated with biotinylated WT-TRAIL (500 ng/ml) for 25 min as described in Supplementary Materials and Methods. Puriﬁed DISC (beads) and supernatants were immunoblotted for the canonical TRAIL DISC components as indicated. * Indicates shift in mobility of two TRAIL-R1 forms (j) Whole cell pellet samples from --2DG and þ 2DG-treated Z138 cells were immunoblotted for the indicated pro- and anti-apoptotic proteins. Tubulin was used as a loading control. Immunoblots are representative of two independent experiments.

Oncogene (2012) 4996 --5006 & 2012 Macmillan Publishers Limited Metabolism modulates MCL sensitivity to TRAIL GL Robinson et al 4999 - Glucose +Glucose - Glucose +Glucose Oligo Olig FCCP Rotenone 1000 200 FCCP Rotenone 400 1000 * 800 800 160

in)

in) 300

cells)

6 - Glucose

0 120 0 600 600

200 in/1 80 400 400 + Glucose

BASAL OCR

BASAL OCR 100 OCR (pmol/m 40 OCR (pmol/m 200 200

(pmol/m

(pmol/min/1 BASAL BASAL 0 0 0 0 0 40 80 120 160 0 20 40 60 80 100 BASAL ECAR (mpH /min/106 cells)

70 150 5 20 60 4 10 50

/min)

cells)

/min) 100

40 0 3 0 30

ular ATP 2

50 (AFLUS) 20 BASAL -10

Cell BASAL ECAR BASAL ** 1 ECAR (mpH ECAR (mpH 10

(mpH /min/1 -20 0 0 0 0 10203040506070 Glucose +- Glucose +- 0 20406080100 Time (min) Time (min)

100 100 - Glucose - Glucose + Glucose + Glucose 80 80

60 60

Death

40 40

% Cell

% Cell Death 20 20

0 0 0 200 400 600 800 1000 0 1 2 3 4 5 6 TRAIL (ng/ml) Time (h)

+ Glucose - Glucose

Time (h) 0 1 1.5 2 4 6 0 1 1.5 2 4 6 WB Pro-Casp-8 p43/41 Caspase 8 p18 *

Pro-Casp-3 Caspase 3 p20 p19 p17 p116 PARP p85 Figure 2. Z138 cells cultured in the absence of glucose undergo a metabolic switch from aerobic glycolysis to oxidative phosphorylation. Z138 cells were cultured for a minimum of 7 days in 2 mM glutamax and 1 mM pyruvate containing media with (&) or without (K)11mM glucose. Real-time measurements (mean±s.e.m., n ¼ 3) of OCR and ECAR were measured as in Figure 1, and oligomycin (400 nm), FCCP (400 nM) and rotenone (1 mM) injected as shown (a, b). In (c, d) basal OCR and ECAR values are shown, calculated as in Figure 1 (mean±s.e.m., n ¼ 3) and cumulative data were analysed by Student’s t-test, n.s., not significant and *Po0.05. The metabolic shift in metabolism from glycolytic to oxidative phosphorylation for cells grown in the absence of glucose is shown in (e). (f) ATP levels of Z138 cells grown in glucose containing ( þ ) or glucose-free (À) were assayed as in Figure 1 (mean±s.e.m., n ¼ 3). (g, h) Primary leukaemic cells obtained from MCL patients were purified as described in Supplementary Materials and Methods. XF analysis was determined as described in panel (a, b) after 6 h culture in media plus and minus glucose. Similar results were obtained in primary cells obtained from another MCL patient (i) Z138 cells grown with (&) or without (K) glucose were treated with TRAIL (0--1000 ng/ml) for 4 h and assessed for cell death. (j) Z138 cells cultured with (&) or without (K) glucose were treated with 400 ng/ml TRAIL and cell death assayed at the indicated times. (k) Cell pellets taken from (j) were analysed by immunoblotting for caspase-8, -3 and PARP cleavage. Immunoblots blots are representative of three independent experiments and * indicates non-specific binding. from the respiring mitochondria as this was inhibited by metabolic dependency/switch to oxidative phosphorylation oligomycin and rotenone (Figure 2b). Without glucose, Z138 cells (Figure 2e). In contrast to the 2DG experiments, this metabolic exhibited a much higher OCR/ECAR ratio, demonstrating a switch to oxidative phosphorylation did not adversely affect

& 2012 Macmillan Publishers Limited Oncogene (2012) 4996 --5006 Metabolism modulates MCL sensitivity to TRAIL GL Robinson et al 5000 cellular ATP content (Figure 2f), showing that pyruvate and which TRAIL death receptors signal to cell death in a particular cell glutamine metabolism via the respiratory chain and mitochondrial type.30 In Z138 cells, the TRAIL-R1-specific ligand signalled to oxidative phosphorylation can maintain ATP at normal levels. apoptosis as efficiently as WT-TRAIL and there was a correspond- These results are in apparent contrast to other studies, which have ing loss of sensitivity with cells maintained in glucose (À) media reported that glucose limitation for 16 h markedly reduced ATP (Figure 3d). The TRAIL-R2-specific ligand did not induce apoptosis levels in U937 cells.25 Our results show that in Z138 cells grown in irrespective of the presence/absence of glucose (Figure 3d). Thus, glucose (À) media, oxidative phosphorylation maintains ATP levels in glucose (À)-maintained cells, down-regulation of TRAIL-R1 but and supports viable cell growth. not TRAIL-R2, could in part explain both the decreased TRAIL DISC To characterise the mitochondrial coupling status of glucose (À) formation and loss of sensitivity to TRAIL. cells we used oligomycin, which inhibits the ATP synthase Death receptor glycosylation has also been implicated as a complex preventing phosphorylation of ADP to ATP and thus factor in determining sensitivity of tumour cells to TRAIL.31 In Z138 blocks proton re-entry into the mitochondrial matrix. In Z138 cells cells, we found that TRAIL-R1 migrated on SDS--PAGE as two grown in glucose ( þ ) and (À) media, oligomycin inhibited OCR by bands, suggesting post-translational modification. Endo H an B75 and 85%, respectively (Figure 2a), demonstrating that endoglycosidase that cleaves chitobiose core high mannose and respiratory chain activity is tightly coupled to the ATP synthase hybrid-type glycosylated proteins32 did not alter the mobility of complex. FCCP uncoupled the mitochondria, by-passed the the TRAIL-R1 doublet (Supplementary Figure S4a). However, oligomycin block and promoted maximal respiratory chain activity PNGase F (specific for complex glycans) decreased the apparent (OCR), which was inhibited by rotenone (Figure 2a). Thus, molecular weight of both TRAIL-R1 bands (Supplementary Figure mitochondria in Z138 cells are metabolically efficient with S4a), suggesting that they contain ‘complex glycan’ chains. respiration tightly coupled to ATP synthesis. This is reminiscent Importantly, the glycosylation profile of TRAIL-R1 was not altered of mitochondria in the classical high respiration ‘state 3’ when the cells were grown in glucose (À) media (Supplementary configuration (that is, substrate and ADP are not limited). This Figure S4a). Also, treatment of cells with N-acetylglucosamine to ability of Z138 cells to switch to oxidative phosphorylation and up- preserve protein O-linked glycosylation did not affect TRAIL regulate their OCR may be a general property of MCL, because sensitivity (Supplementary Figure S4b), demonstrating that primary leukaemic cells exhibited similar characteristics when glucose (À) conditions did not significantly change the glycosyla- cultured in glucose-free media (Figures 2g and h). tion status of the cells, unlike 2DG, which apparently inhibited In contrast to the effects of 2DG on TRAIL sensitivity, Z138 cells TRAIL-R1 glycosylation (compared with Figure 1i). grown in glucose-free media exhibited a twofold decrease in sensitivity to TRAIL-induced cell death (Figure 2i). Furthermore, in Mitochondrial-mediated cell death is defective in Z138 cells glucose (À) media, TRAIL-induced cell death progressed more cultured without glucose slowly than cells cultured in glucose ( þ ) media (Figure 2j). These Type II cells have lower DISC formation and require generation of findings were supported by further analysis of the time- tBid to engage the intrinsic (mitochondrial-dependent) cell death dependent cleavage of caspase-8, caspase-3 and PARP in TRAIL- pathway, which leads to loss of mitochondrial membrane treated cells (Figure 2k). Cleaved caspase-8 subunits (p43/41) were potential (DcM) and cytochrome c release.33 In Z138 cells, loss not detected in cells grown on glucose (À) media until B4 h post- of DcM was detected in TRAIL-treated cells (Figure 4a) and treatment with 400 ng/ml TRAIL, whereas in cells grown on correlated with the cell death response (Figure 2i). To further glucose ( þ ) media, caspase processing was detected 1 h post- examine the response of glucose (À) cells to mitochondrial- TRAIL treatment (Figure 2k). Caspase-3 cleavage to the active p20/ dependent apoptosis, we treated Z138 cells with ABT-73734 and 19/17 subunits was detected within 1--1.5 h in glucose ( þ ) cells ionising radiation. ABT-737 induced cell death in glucose ( þ ) cells but was delayed (2--4 h) in the absence of glucose. PARP cleavage (data not shown) and produced a corresponding loss of DcM with followed a similar time course and was again delayed in cells an EC B25--30 nM (Figure 4b). In the absence of glucose, ABT- grown in glucose (À)-free media (Figure 2k). 50 737 was 40-fold less potent (EC50 B1000 nM) and similarly, we observed a two- to threefold decrease in the response of glucose- The extrinsic cell death pathway is defective in Z138 cells cultured (À) to ionising radiation compared with glucose ( þ ) cells without glucose (Supplementary Figure S5). The loss of DcM after TRAIL treatment Since Z138 cells in glucose (À) media were less sensitive to TRAIL, was delayed in cells maintained on glucose (À) media (Figure 4c), we determined whether the extrinsic cell death pathway was and was accompanied by delayed Bid cleavage, cytochrome c altered under glucose-free conditions. Immunoblotting of key pro- release and caspase-9 activation (Figure 4d). Interestingly, and anti-apoptotic proteins involved in DISC-mediated apoptosis mitochondrial levels of cytochrome c were much higher in cells were detected, which showed that total cellular levels of FADD, grown in glucose-free media (Figure 4d). However, cytochrome c release, the initiating factor in the intrinsic cell death pathway, was caspase-8 and caspase-3 were increased while, cFLIPS, TRAIL-R1 and -R2 were reduced in cells cultured in glucose (À) media attenuated in cells cultured in glucose (À) media, and apopto- (Supplementary Figure S3). In addition, cellular surface expression some-mediated caspase-9 activation was less extensive with of TRAIL-R1 and -R2 in cells grown on glucose (À) media was reduced cleavage to the p37 and 35 subunits (Figure 4d). The slightly decreased, compared with cells cultured on glucose ( þ ) total cellular content of most key pro- and anti-apoptotic proteins media (Figure 3a). TRAIL DISC formation, as determined by associated with the mitochondrial pathway (Figure 4e) were only receptor, FADD and caspase-8 recruitment, was detected in marginally affected by the glucose (À) conditions. Notably, Bax glucose ( þ ) cells within B20 min of TRAIL treatment whereas levels were decreased in the absence of glucose, while Bcl-2 and without glucose, DISC formation was delayed reaching a XIAP were increased in comparison to the glucose ( þ ) cells. Thus, maximum of B50 min after TRAIL treatment (Figure 3b). Recruit- culturing Z138 cells in the absence of glucose altered the ratio ment of FADD and caspase-8 (and caspase-8 processing) in cells Bax:Bcl-2, which in turn could account for the loss of sensitivity to grown in glucose (À) media was significantly reduced, possibly TRAIL, ABT-737 and ionising radiation. reflecting in part the decreased TRAIL-R1/R2 expression (Figure 3a). In addition, caspase-8 activity of the isolated DISC, Z138 cells cultured in the absence of glucose have more mitochondria as measured by IETD.afc cleavage, was significantly reduced (B3- with increased cristae density and higher cytochrome c content fold lower) when cells were grown in glucose (À) media Changes in some apoptotic proteins associated with the (Figure 3c). TRAIL receptor-specific ligands can be used to identify mitochondrial-mediated pathway were investigated further by

Oncogene (2012) 4996 --5006 & 2012 Macmillan Publishers Limited Metabolism modulates MCL sensitivity to TRAIL GL Robinson et al 5001 TRAIL-R1 TRAIL-R2 50 50

40 40

30 30

Counts 20 20

10 10

0 0 100 101 102 103 103 100 101 102 103 103 FL2-Height FL2-Height + Glucose - Glucose

+ Glucose - Glucose W.B. + Glucose - Glucose TRAIL-R1

0 1020 30 40 50 600 1020 30 40 50 60 Time (min) U.T. U.T. TRAIL-R2

W.B. down TRAIL-R1 FADD

TRAIL-R2 DISC pull Caspase 8

down 1200 FADD 1000

DISC pull

ase Caspase v in) 800 600

(pmol/m 400 cFLIP L DISC IETD 200 0 bTRAIL -+ -+

100 100 100

80 80 80

60 60 60

Death 40 40 40

% Cell 20 + Glucose 20 20 - Glucose 0 0 0 0 200 400 600 800 1000 0 200 400 600 800 1000 0 200 400 600 800 1000 WT-TRAIL (ng/ml) R1-TRAIL (ng/ml) R2-TRAIL (ng/ml) Figure 3. The extrinsic cell death pathway is defective in Z138 cells cultured without glucose. (a) Cell surface receptor expression was determined with PE-conjugated TRAIL-R1 and TRAIL-R2-specific antibodies or IgG (black fill) by FACS analysis as described in Supplementary Materials and Methods. (b) Time-dependent DISC formation was assayed in Z138 cells cultured with ( þ ) or without (À) glucose using biotinylated WT-TRAIL (500 ng/ml) and immunoblotting for TRAIL DISC components as outlined in Supplementary Materials and Methods. DISCs were prepared from cells treated with TRAIL for 25 min and immunoblotted for TRAIL-R1 and -R2, FADD and caspase-8 (upper panel (c)). Isolated DISC preparations were assayed for catalytic activity (IETDase activity) using ac.IETD.afc a fluorimetric tetra-peptide substrate for caspase-8 (lower panel (c)). (d) Z138 cells cultured with ( þ ) or without (À) glucose were incubated for 4 h with varying concentrations of WT-TRAIL, TRAIL-R1 or TRAIL-R2-specific ligands and analysed for cell death. subcellular fractionation. Cytosol, mitochondria and nuclei frac- released with TRAIL treatment (Figure 4d), we carried out an tions were immunoblotted for the indicated proteins (Figure 5a) ultrastructural study of the mitochondria in Z138 cells (Figure 5b). using PARP as a nuclear protein marker, and COXIV and SOD2 as Most mitochondria in glucose-maintained cells exhibited lucent markers for mitochondrial purity/loading. Cells cultured in glucose matrixes and sparse, disorganised cristae characteristic of tumours ( þ ) media had a higher Bax content in all three subcellular and tumour-derived cell lines presumably reflecting their depen- fractions, whereas Bak was only detected in the mitochondria and dence on glycolytic metabolism. Glucose grown cells had on essentially did not vary, irrespective of the growth conditions. The average of six mitochondria/cell with a mean area of BH3-only proteins BAD and Puma were detected in the cytosol but 0.51±0.127 mm2 (n ¼ 6, s.e.m.; Figures 5b and c). However, not in mitochondria. The anti-apoptotic proteins Bfl-A1, Bcl-xL and mitochondria were more numerous (average of 10/cell) in cells Mcl-1 were concentrated in the mitochondrial fractions but were maintained in glucose (À) media, with a smaller mean area of largely unaffected by the growth conditions. In particular Mcl-1, 0.31±0.057 mm2 (n ¼ 10, s.e.m.), and contained more numerous, which was markedly down-regulated after 2DG treatment, was densely packed and regular transverse cristae than those unchanged in glucose-free cells. However, the cytochrome c observed in the glucose ( þ ) cultured cells (Figures 5b and c). content of mitochondria isolated from cells in glucose (À) media The ratio of IMM/OMM was significantly higher in cells grown was significantly higher than that of cells cultured with glucose. without glucose as compared with the glucose ( þ ) cells Since cells maintained in glucose (À) media had high levels of (Figure 5c). Overall, the mitochondrial morphology in the glucose- cytochrome c (Figure 5a), which interestingly was not readily free cells was reminiscent of tightly coupled mitochondria.35

& 2012 Macmillan Publishers Limited Oncogene (2012) 4996 --5006 Metabolism modulates MCL sensitivity to TRAIL GL Robinson et al 5002 100 100

80 80

ΔΨ ΔΨ 60

60 f

40 40

% Loss o

20 20 + Glucose + Glucose - Glucose - Glucose 0 0 0 0 200 400 600 800 1000 200 400 600 800 1000 TRAIL (ng/ml) ABT 737 (nM)

+ Glucose - Glucose 100 Time (h) 0 1 1.5 264 0 1 1.5 264 80 W.B p22 ΔΨ 60 BID

40 Mito fraction

% Loss o Cyto c Supernatant 20 + Glucose - Glucose Pro-Casp-9 0 Caspase 9 0 1 2 3 4 5 6 p37 p35 Time (h)

Glucose Glucose +- +-

BID BCL-2

BAX BCL-XL

BAK MCL-1

BAD XIAP PUMA Tubulin Caspase 7 Anti-apoptotic

Caspase 9

p53

Pro-apoptotic Figure 4. The mitochondrial-mediated cell death pathway is also defective in Z138 cells cultured without glucose. (a) Z138 cells cultured with (&) or without (K) glucose were treated with TRAIL (0--1000 ng/ml) for 4 h and assessed for loss of membrane potential (DcM) as described in Supplementary Materials and Methods. (b) Loss of DcM was measured in ( þ ) and (À) glucose Z138 cells after 4 h ABT-737 treatment (0-- 1000 nM). (c) The time course of loss of DcM following treatment with 400 ng/ml TRAIL was measured in ( þ ) and (À) Z138 cells. (d) Whole cell pellets from (c) were analysed for loss of full-length BID, release of cytochrome c from the mitochondria and caspase-9 cleavage as described in Materials and methods. (e) Whole cell pellets taken from Z138 cells cultured with ( þ ) or without (À) glucose were analysed for pro-and anti-apoptotic proteins by immunoblotting. Tubulin was used as a loading control. Immunoblots are representative of three independent experiments.

DISCUSSION death. Strikingly, 2DG treatment of Z138 cells produced almost Cancer cells invariably exhibit a high rate of aerobic glycolysis, complete inhibition of glycolysis but reduced oxidative phosphor- which provides sufﬁcient ATP and intermediate metabolites to ylation (OCR) by only B30% (Figures 1c and d). Z138 cells still support tumour growth. However, many chemotherapeutic agents retained signiﬁcant oxidative phosphorylation activity but impor- induce apoptotic cell death via the mitochondrial cell death tantly cellular ATP levels decreased to 50% of the control (non-2DG- pathway, which is an active form of cell death believed to require treated) cells (Figure 1f). Under these conditions, Z138 cells were high levels of ATP.21 Aerobic glycolysis can be inhibited either by sensitised to TRAIL-induced apoptotic cell death (Figure 1g). In glucose depletion or with glycolytic inhibitors such as 2DG.27 Our contrast, with glucose deprivation we observed increased levels of results show that in Z138 cells, 2DG treatment and glucose oxidative phosphorylation, maintenance of ATP levels and reduced limitation produced opposing effects on both metabolism and sensitivity to TRAIL, ABT-737 (Figures 2 and 4) and ionising apoptotic cell death. Using real-time measurements of oxidative radiation-induced cell death (Supplementary Figure S5). Recent phosphorylation and glycolysis, we have correlated whole cell studies have shown that activated Akt, in the presence of glucose, bioenergetics with ATP levels and their respective sensitivity to cell prevents Mcl-1 loss and protects against growth factor-induced

Oncogene (2012) 4996 --5006 & 2012 Macmillan Publishers Limited Metabolism modulates MCL sensitivity to TRAIL GL Robinson et al 5003 + Glucose - Glucose

W.B.

CytosolMitochondriaNucleus CytosolMitochondriaNucleus Cyto c Bax Bak Pro-apoptotic Bad Puma

Bfl-A1 Bcl-2 Anti-apoptotic Bcl-xl Mcl-1

PARP Nucleus

SOD 2 Mitochondria Cox IV

+ Glucose - Glucose * 3.5 3.5

cell) 3.0 3.0

m 2.5 2.5

μ 2.0 2.0

a ( 1μm 1.5 1.5 1.0 1.0 0.5 0.5 IMM/OMM rati

Mitochondri 0.0 0.0 Glucose + - + -

Figure 5. Z138 cells cultured in the absence of glucose have more mitochondria with increased cristae density and cytochrome c content. (a) Cytosolic, mitochondrial and nuclear fractions from Z138 cells cultured with ( þ ) or without (À) glucose were obtained by subcellular fractionation and analysed for the indicated proteins as outlined in Supplementary Materials and Methods. PARP was used as a nuclear marker while SOD2 and COXIV were used as mitochondrial markers. (b) Z138 cells grown with ( þ ) and without (À) glucose were analysed by electron microscopy (EM) as described in Materials and methods. Two representative images are shown with inset dimension bar. Mitochondria per cell and IMM/OMM ratios were determined as described in Materials and methods (c). apoptosis.36 In this study, suppression of glucose metabolism (with marked decrease in mRNA associated with polysomes (Supple- 2DG) markedly sensitised DLBCL cell lines to ABT-737. In addition, mentary Figure S6a) as compared with cells without 2DG 2DG enhances ABT-263/737 killing of cancer cells in a mouse treatment, indicating a global decrease in protein translation. xenograft model of hormone-independent chemoresistant human Protein expression levels of individual proteins are determined by prostate cancer.37 Also, 2DG-induced toxicity in mouse B-cell the homeostatic balance between protein translation and lymphoma cell lines has been reported to be antagonised by Bcl- degradation. We observed marked differences in the levels of 2, which in turn can be overcome with ABT-737.38 In the latter pro- and anti-apoptotic proteins during 2DG treatment. For study, 2DG up-regulated Bim, which was identiﬁed as a critical example, cFLIPL levels in the cell were maintained but the anti- target in 2DG-mediated endoplasmic reticulum stress. We have apoptotic short form of FLIP was markedly down-regulated observed a similar synergistic effect of 2DG on ABT-737 cell killing (Figure 1j), which is consistent with published studies showing 40 in Z138 cells (data not shown), but it should be noted that Z138 that cFLIPS has a markedly shorter half-life than cFLIPL. This cells do not express Bim39 and we did not detect Bim in whole cell change in the ratio of long and short isoforms of cFLIP would in immunoblots (data not shown). This suggests that in Z138 cells, turn be predicted to enhance DISC-mediated caspase-8 proces- 2DG is enhancing cell killing by a mechanism, which involves sing and caspase-3 (Figures 1h and 6). Pro- and anti-apoptotic another BH3 protein. proteins associated with the mitochondrial cell death pathway In Z138 cells, TRAIL induces cell death by direct activation of were also markedly affected by 2DG treatment. The major caspase-8, which cleaves/activates the effector caspases, and by consequence of these changes is that Mcl-1 levels were markedly cleavage of Bid to yield to tBId, thereby activating the reduced but Bak was unaffected (Figures 1j and 6). Mcl-1 has a mitochondrial cell death pathway (Figure 6). With 2DG, direct very short half-life of B1 h, whereas other Bcl-2 family members caspase activation via DISC assembly was accelerated and cell have much-longer half-lives (for example, Bcl-2 is B10 h).41 death enhanced. The inhibition of glycolysis resulted in decreased Knockdown of Mcl-1 but not Bcl-xL overcomes resistance to ATP levels, which is required for protein translation. Sucrose TRAIL, CD95/FasL and TNFa in Bax-deﬁcient cells,42 indicating that density centrifugation of 2DG-treated cells lysates revealed a the Mcl-1/Bak ratio is the major determinant of TRAIL sensitivity in

& 2012 Macmillan Publishers Limited Oncogene (2012) 4996 --5006 Metabolism modulates MCL sensitivity to TRAIL GL Robinson et al 5004 TRAIL TRAIL Glucose-free

2-DG Glucose Glucose Casp-3 Casp-8 Hexokinase Hexokinase Casp-8 Casp-3

Glucose-6-Phosphate Glucose-6-Phosphate

Glycosylation Fructose-6-Phosphate Fructose-6-Phosphate Glycosylation

Bid

Lactate Glycolytic Pathway Glycolytic Pathway Lactate Bid

tBid PYRUVATE

Hexokinase Bak TCA Protein Translation Mcl-1 Cycle

Ox phos Bax tBid Bcl-2

Cell DEATH

Protein Translation

GLUTAMATE Cell DEATH Figure 6. Scheme showing how metabolic changes induced by 2DG and glucose deprivation modulate metabolism, key apoptotic proteins and TRAIL-induced cell death. In this scheme, Z138 cells are shown are as type II cells that undergo TRAIL-induced cell death by DISC assembly and caspase-8 activation. Caspase-8 can then activate caspase-3 and cell death directly or alternatively can cleave Bid to produce tBid, which induces cytochrome c release and apoptosome activation. In glucose (À) cells, glycolysis is limited and the cells respond by up-regulating oxidative phosphorylation, maintaining ATP levels and anti-apoptotic proteins (Bcl-2/Bcl-XL/Mcl-1) while down-regulating the pro-apoptotic protein Bax. In the case of 2DG, there is a marked inhibition of glycolysis and a decrease in oxidative phosphorylation, resulting in diminished levels of ATP. Protein translation is inhibited and Mcl-1 is rapidly degraded, but levels of its antagonist Bak remain high and this can trigger apoptotic cell death. In addition, direct caspase-8 activation of the caspase cascade is accelerated, resulting in enhanced sensitivity to TRAIL.

the absence of Bax. Significantly, 2DG also caused a marked cells utilise pyruvate/glutamine as a substrate and in this situation decrease in Bax levels but had proportionately less effect on Bcl- ADP levels must be sufficiently high to maintain state 3-like xL/Bcl-2, which normally neutralise Bax-mediated killing. Thus, respiration and ATP production. Rising AMP levels decrease the 2DG facilitates DISC-mediated caspase-8 activation, which is ATP/AMP ratio and activate the AMPK pathway, switching on ATP- augmented by translational-dependent down-regulation of the generating catabolic pathways and switching off ATP-consuming anti-apoptotic protein Mcl-1 and activation of the mitochondrial anabolic pathways such as protein translation.45 The XF data show cell death pathway as suggested recently29 (Figure 6). that oxidative metabolism in Z138 cells is up-regulated to Removal of glucose or inhibition by 2DG is generally believed to generate ATP from ADP and this would not be expected to produce essentially the same effects on cellular metabolism and activate AMPK (Figure 6). This conclusion is supported by the fact 29,43 sensitivity to apoptotic agents. Our study shows that this is that cellular Mcl-1, Bcl-XL and Bcl-2 levels were similar or higher in not necessarily true for all cell types. In Z138 cells maintained in glucose (À) as compared with glucose ( þ ) cells (Figure 2e). glucose-free conditions, TRAIL-R1 and -R2 expression were slightly Significantly, mitochondria isolated from glucose (À) cells did not down-regulated and TRAIL-R1 induced DISC formation and activity show significant changes in the anti-apoptotic Bcl-2 proteins, (IETDase) attenuated (Figures 3a--d), which could in part explain although there was a decrease in total cellular and mitochondrial the slower onset of apoptosis (Figure 4c). However, using XF Bax. This could explain why Z138 cells were not only less sensitive analysis we show that Z138 cells conditioned for 7 days in to TRAIL-induced apoptotic cell death (Figures 2g and h) but also glucose-free media had low glycolytic and enhanced oxidative to ABT-737 (Figure 4b) and ionising radiation (Supplementary phosphorylation activity, and had similar ATP levels to cells grown Figure S5), indicating that the mitochondrial apoptotic pathway is on glucose. Thus, mitochondrial oxidative phosphorylation can predominantly affected by the switch from aerobic glycolysis to meet the metabolic demand imposed by decreased glycolytic flux oxidative phosphorylation. and maintain ATP levels to support cell viability (Figure 2f). In anti-glycolytic conditions, the major effects appear to be on Mitochondria in Z138 cells were well coupled and functionally the mitochondrial pathway and therefore it is logical to assume competent, because FCCP on its own produced only a small that metabolic changes have affected the relative balance increase in respiration (Figure 1a), whereas when mitochondrial between anti- and pro-apoptotic Bcl-2 family members. respiration was first blocked by oligomycin, there was a much This argument works well for the effect of 2DG, as clearly key greater response to FCCP (Figure 2a). This is analogous to tightly proteins such as Mcl-1 are degraded. But this model only partly coupled isolated mitochondria respiring in classical state 3 mode44 explains the phenomenon for glucose deprivation as Bax and Bcl- where the respiratory chain is running at near maximal levels and 2 seem to be the only protein most significantly affected by the cannot be further stimulated by FCCP. Thus, without glucose, Z138 switch to oxidative metabolism. Another important contributory

Oncogene (2012) 4996 --5006 & 2012 Macmillan Publishers Limited Metabolism modulates MCL sensitivity to TRAIL GL Robinson et al 5005 factor may be that the enhanced oxidative phosphorylation Cell metabolism assays results in enhanced retention of cytochrome c within the Rates of glycolysis and oxidative phosphorylation were determined by mitochondria, which exhibited a more extensive well-defined measuring lactic acid release (ECAR) and OCR using an XF24 XF analyser cristae network and denser matrix (Figure 5b). This is typically (Seahorse Bioscience, Billerica, MA, USA) as previously described for murine defined as the classical condensed configuration and was B cells.49,50 Briefly, 4.0 Â 105 Z138 cells/well were adhered to XF 24-well originally observed in mouse liver mitochondria undergoing state cell culture microplates (Seahorse Biosciences) using BD Cell-Tak (BD 35 3 respiration. Electron microscopy and ultrastructural tomogra- Biosciences, Oxfordshire, UK) and plates incubated at 37 1C for 1 h before 46 phy studies (reviewed in Benard and Rossignol ) show the OCR and ECAR analysis. FCCP and rotenone were loaded into drug delivery condensed state to be a reversible phenomenon, probably related ports and added sequentially at the indicated time points. XF assay to ionic and osmotic changes related to ATP synthesis. Interest- medium was low-buffered bicarbonate-free DMEM (pH 7.4) and replicated ingly, mitochondria in HeLa cells grown under glucose (À) the glucose and pyruvate/glutamax composition of the respective conditions have enhanced oxidation phosphorylation, increased experimental conditions. mitochondrial protein expression, a condensed configuration with increased cristae density but without changes in mitochondrial mass.47 Thus, it is possible that the increased cytochrome c ATP analysis retention we observed is a result of these structural rearrange- Cellular ATP content was determined using a bioluminescence assay ments in the mitochondria, induced by changes in the mitochon- (Promega, Hampshire, UK) as outlined in the manufacturer’s protocol. drial proteome. Alternatively, this may be a dynamic phenomenon Essentially, Z138 cells were equilibrated in opaque-walled 96-well plates controlled by the availability of glucose and pyruvate/glutamine. for 30 min at room temperature. Equal volumes of CellTiter-Glo reagent Structurally, cytochrome c is compartmentalised between the were then added and mixed for 2 min followed by 10 min incubation at intracristal space and the intermembrane space and it is possible room temperature. ATP content was measured using a Wallac Victor2 1420 that when cell metabolism switches to oxidative phosphorylation, Multilabel counter (Perkin Elmer, Cambridge, UK). there is a compensatory increase in intracristal space, which traps cytochrome c in a less mobile compartment. Whole cell protein expression In Z138 cells, we have shown that adaptation to a glucose-free Glucose-containing, glucose-free, 2DG-untreated and 2DG-treated Z138 environment result in altered responses to TRAIL and ABT-737. 6 Similar experiments with BJAB, Jurkat and HeLa cells have cells (5.0 Â 10 ) were centrifuged at 250 g for 3 min at 4 1C and washed produced varying cell type-specific responses (unpublished three times in ice-cold phosphate-buffered saline. Cell pellets were results). In some cases, the cell lines do not survive and grow in resuspended in SDS-sample buffer, sonicated and analysed by SDS--PAGE (--) glucose conditions, but in the UPN1 (MCL) cell line, growth is and immunoblotting. maintained and the cells are viable. Interestingly, in the latter cells, the cell death response to TRAIL is different with glucose altering Electron microscopy the mode of cell death (the balance between necrosis and Cells were fixed in 2% glutaraldehyde in 0.1 M sodium cacodylate buffer apoptosis). Thus, the cell death response to altered glucose (pH 7.4) at 4 1C overnight and post-fixed with 1% osmium tetroxide/1% metabolism may be cell type and tissue specific, and may reflect potassium ferrocyanide for 1 h at room temperature. Fixed cells were differences in compensatory changes in transcriptional and stained en bloc with 5% aqueous uranyl acetate overnight at room translational expression of key mitochondrial and glycolytic temperature, dehydrated and embedded in Taab epoxy resin (Taab enzymes. Glucose-free Z138 cells showed less dramatic changes Laboratories Equipment Ltd., Aldermaston, UK). Electron micrographs of in apoptotic protein expression (Figure 4e) than 2DG-treated cells ultrathin sections were recorded using an ES1000W CCD camera and (Figure 1) and protein translation as judged by sucrose density Digital Micrograph software (Gatan, Abingdon, UK) in a Zeiss 902A electron polysome analysis was much less affected than was seen with 2DG microscope (Zeiss, Cambridge, UK). Mitochondria were counted in 100 cell (Supplementary Figure S6a and data not shown). However, profiles, selected at random from each sample. The inner and outer microarray analysis comparing cells grown with and without mitochondrial membranes were traced and measured on 90 images glucose showed substantial changes with 126 genes down- recorded at a magnification of Â 7000, using Image-J (NIH, Bethesda, regulated and 48 genes up-regulated (Supplementary Figure S6b). MD, USA) and a Wacom Intuos4 pen tablet (Wacom Technology These changes may contribute to the switched metabolism, Corporation, Vancouver, WA, USA). The resulting data were analysed, retention of cytochrome c and reduced sensitivity to ABT-737/ using Microsoft Excel, to establish the mean area and the IMM/OMM ratio TRAIL, and are currently being investigated further. in each sample.

MATERIALS AND METHODS Subcellular fractionation Reagents and antibodies Cells were washed in phosphate-buffered saline, pelleted and resuspended

These are described in Supplementary Materials and Methods. in homogenisation buffer (10 mM KCl, 1.5 mM, MgCl2,10mM HEPES-KOH, pH 7.5) and left on ice for 10 min before homogenisation in a glass Dounce homogeniser. Disrupted cells were diluted in 2.5 volumes of 2.5 Â Cell lines and culture conditions concentrated mannitol/sucrose buffer (525 mM mannitol, 174 mM sucrose, 48 The MCL cell line Z138 was obtained as previously described and glucose 2.5 mm EDTA, 12.5 mM Hepes-KOH, pH 7.5) and centrifuged (1000 g/ ( þ ) cells were cultured in RPMI 1640 media (11 mM glucose) supplemen- 10 min) to remove nuclei. The supernatant was then centrifuged at ted with 10% FCS, 2 mM glutamax (L-alanyl-L-glutamine provides both 10 000 g for 10 min to pellet mitochondria and remove the cytosol glutamine and alanine) and 1 mM sodium pyruvate and incubated at 37 1C (supernatant). Mitochondria were finally resuspended in mannitol/sucrose under 5% CO2 in a humidified atmosphere. Glucose (À) Z138 cells were resuspension buffer (200 mM mannitol, 70 mM Sucrose, 1 mM EGTA, cultured in RPMI 1640 (without glucose), supplemented with 10% FCS, 10 mM HEPES-KOH, pH 7.5 þ protease inhibitors (Roche complete 2mM glutamax, and 1 mM sodium pyruvate. Treatment with 2DG was mini-EDTA, Roche, Surrey, UK) and analysed by SDS--PAGE and immuno- performed in Z138 cells cultured in glucose-containing media supple- blotting. mented with 5 mM 2DG for 20 h. In cells grown on glucose-free media, the final residual glucose concentration (coming from the FCS) was calculated to be 0.28 mM and from metabolic calculations (Eyal Gottlieb, personal Cell death assays, TRAIL receptor expression and DISC analysis communication) would be rapidly metabolised within 5 h of cell culture. These procedures are described in Supplementary Methods.51

& 2012 Macmillan Publishers Limited Oncogene (2012) 4996 --5006 Metabolism modulates MCL sensitivity to TRAIL GL Robinson et al 5006 CONFLICT OF INTEREST 26 Wood TE, Dalili S, Simpson CD, Hurren R, Mao X, Saiz FS et al. Anovelinhibitorofglucose The authors declare no conflict of interest. uptake sensitizes cells to FAS-induced cell death. Mol Cancer Ther 2008; 7: 3546 - 3555. 27 Parniak M, Kalant N. Incorporation of glucose into glycogen in primary cultures of rat hepatocytes. Can J Biochem Cell Biol 1985; 63: 333 --340. ACKNOWLEDGEMENTS 28 Liu H, Jiang CC, Lavis CJ, Croft A, Dong L, Tseng HY et al. 2-Deoxy-D-glucose This work was supported by the UK Medical Research Council. enhances TRAIL-induced apoptosis in human melanoma cells through XBP-1- mediated up-regulation of TRAIL-R2. Mol Cancer 2009; 8:122. 29 Pradelli LA, Beneteau M, Chauvin C, Jacquin MA, Marchetti S, Munoz-Pinedo C REFERENCES et al. Glycolysis inhibition sensitizes tumor cells to death receptors-induced 1 Warburg O. On respiratory impairment in cancer cells. Science 1956; 124: 269 --270. apoptosis by AMP kinase activation leading to Mcl-1 block in translation. 2 Gambhir SS. Molecular imaging of cancer with positron emission tomography. Oncogene 2009; 29: 1641 --1652. Nat Rev Cancer 2002; 2: 683 --693. 30 MacFarlane M, Kohlhaas SL, Sutcliffe MJ, Dyer MJ, Cohen GM. TRAIL receptor- 3 DeBerardinis RJ, Lum JJ, Hatzivassiliou G, Thompson CB. The biology of cancer: selective mutants signal to apoptosis via TRAIL-R1 in primary lymphoid metabolic reprogramming fuels cell growth and proliferation. Cell Metab 2007; 7: malignancies. Cancer Res 2005; 65: 11265 --11270. 11 --20. 31 Wagner KW, Punnoose EA, Januario T, Lawrence DA, Pitti RM, Lancaster K et al. 4 Pelicano H, Martin DS, Xu RH, Huang P. Glycolysis inhibition for anticancer Death-receptor O-glycosylation controls tumor-cell sensitivity to the proapoptotic treatment. Oncogene 2006; 25: 4633 --4646. ligand Apo2L/TRAIL. Nat Med 2007; 13: 1070 --1077. 5 Adrain C, Martin SJ. The mitochondrial apoptosome: a killer unleashed by the 32 Marino K, Bones J, Kattla JJ, Rudd PM. A systematic approach to protein cytochrome seas. Trends Biochem Sci 2001; 26: 390 --397. glycosylation analysis: a path through the maze. Nat Chem Biol 2010; 6: 713 --723. 6 Cain K, Bratton SB, Cohen GM. The Apaf-1 apoptosome: a large caspase-activating 33 Heiskanen KM, Bhat MB, Wang HW, Ma J, Nieminen AL. Mitochondrial complex. Biochimie 2002; 84: 203 --214. depolarization accompanies cytochrome c release during apoptosis in PC6 cells. 7 Wallach D, Varfolomeev EE, Malinin NL, Goltsev YV, Kovalenko AV, Boldin MP. J Biol Chem 1999; 274: 5654 --5658. Tumor necrosis factor receptor and Fas signaling mechanisms. Annu Rev Immunol 34 van Delft MF, Wei AH, Mason KD, Vandenberg CJ, Chen L, Czabotar PE et al. The 1999; 17: 331 --367. BH3 mimetic ABT-737 targets selective Bcl-2 proteins and efficiently induces 8 Wiley SR, Schooley K, Smolak PJ, Din WS, Huang CP, Nicholl JK et al. Identification apoptosis via Bak/Bax if Mcl-1 is neutralized. Cancer Cell 2006; 10: 389 --399. and characterization of a new member of the TNF family that induces apoptosis. 35 Hackenbrock CR. Ultrastructural bases for metabolically linked mechanical activity Immunity 1995; 3: 673 --682. in mitochondria. I. Reversible ultrastructural changes with change in metabolic 9 Pitti RM, Marsters SA, Ruppert S, Donahue CJ, Moore A, Ashkenazi A. Induction of steady state in isolated liver mitochondria. J Cell Biol 1966; 30: 269 --297. apoptosis by Apo-2 ligand, a new member of the tumor necrosis factor cytokine 36 Coloff JL, Macintyre AN, Nichols AG, Liu T, Gallo CA, Plas DR et al. Akt-dependent family. J Biol Chem 1996; 271: 12687 --12690. glucose metabolism promotes Mcl-1 synthesis to maintain cell survival and 10 Walczak H, Miller RE, Ariail K, Gliniak B, Griffith TS, Kubin M et al. Tumoricidal resistance to Bcl-2 inhibition. Cancer Res 2011; 71: 5204 --5213. activity of tumor necrosis factor-related apoptosis-inducing ligand in vivo. Nat 37 Yamaguchi R, Janssen E, Perkins G, Ellisman M, Kitada S, Reed JC. Efficient Med 1999; 5: 157 --163. elimination of cancer cells by deoxyglucose-ABT-263/737 combination therapy. 11 Dyer MJ, MacFarlane M, Cohen GM. Barriers to effective TRAIL-targeted therapy of PLoS One 2011; 6: e24102. malignancy. J Clin Oncol 2007; 25: 4505 --4506. 38 Zagorodna O, Martin SM, Rutkowski DT, Kuwana T, Spitz DR, Knudson CM. 12 Ashkenazi A. Targeting death and decoy receptors of the tumour-necrosis factor 2-Deoxyglucose-induced toxicity is regulated by Bcl-2 family members and is superfamily. Nat Rev Cancer 2002; 2: 420 --430. enhanced by antagonizing Bcl-2 in lymphoma cell lines. Oncogene 2011. Advance 13 Falschlehner C, Schaefer U, Walczak H. Following TRAIL’s path in the immune online publication, 10 October 2011; doi:10.1038/onc.2011.454. system. Immunology 2009; 127: 145 --154. 39 Tagawa H, Karnan S, Suzuki R, Matsuo K, Zhang X, Ota A et al. Genome-wide array- 14 Hughes MA, Harper N, Butterworth M, Cain K, Cohen GM, MacFarlane M. based CGH for mantle cell lymphoma: identification of homozygous deletions of Reconstitution of the death-inducing signaling complex reveals a substrate switch the proapoptotic gene BIM. Oncogene 2005; 24: 1348 --1358. that determines CD95-mediated death or survival. Mol Cell 2009; 35: 265 --279. 40 Poukkula M, Kaunisto A, Hietakangas V, Denessiouk K, Katajamaki T, Johnson MS 15 Scaffidi C, Schmitz I, Zha J, Korsmeyer SJ, Krammer PH, Peter ME. Differential et al. Rapid turnover of c-FLIP short is determined by its unique C-terminal tail. modulation of apoptosis sensitivity in CD95 type I and type II cells. J Biol Chem J Biol Chem 2005; 280: 27345 --27355. 1999; 274: 22532 --22538. 41 Merino R, Ding L, Veis DJ, Korsmeyer SJ, Nunez G. Developmental regulation of 16 Fulda S, Meyer E, Debatin KM. Inhibition of TRAIL-induced apoptosis by Bcl-2 the Bcl-2 protein and susceptibility to cell death in B lymphocytes. EMBO J 1994; overexpression. Oncogene 2002; 21: 2283 --2294. 13: 683 --691. 17 Samraj AK, Keil E, Ueffing N, Schulze-Osthoff K, Schmitz I. Loss of caspase-9 42 Gillissen B, Wendt J, Richter A, Muer A, Overkamp T, Gebhardt N et al. Endogenous provides genetic evidence for the type I/II concept of CD95-mediated apoptosis. Bak inhibitors Mcl-1 and Bcl-xL: differential impact on TRAIL resistance in Bax- J Biol Chem 2006; 281: 29652 --29659. deficient carcinoma. J Cell Biol 2010; 188: 851 --862. 18 LeBlanc H, Lawrence D, Varfolomeev E, Totpal K, Morlan J, Schow P et al. Tumor- 43 El Mjiyad N, Caro-Maldonado A, Ramirez-Peinado S, Munoz-Pinedo C. Sugar-free cell resistance to death receptor-induced apoptosis through mutational inactiva- approaches to cancer cell killing. Oncogene 2011; 30: 253 --264. tion of the proapoptotic Bcl-2 homolog Bax. Nat Med 2002; 8: 274 --281. 44 Chance B, Williams GR. The respiratory chain and oxidative phosphorylation. Adv 19 Thakkar NS, Potten CS. Inhibition of doxorubicin-induced apoptosis in vivo by 2- Enzymol Relat Subj Biochem 1956; 17: 65 --134. deoxy-D-glucose. Cancer Res 1993; 53: 2057 --2060. 45 Hardie DG. New roles for the LKB1--4AMPK pathway. Curr Opin Cell Biol 2005; 2: 20 Ferrari D, Stepczynska A, Los M, Wesselborg S, Schulze-Osthoff K. Differential 167 --173. regulation and ATP requirement for caspase-8 and caspase-3 activation during 46 Benard G, Rossignol R. Ultrastructure of the mitochondrion and its bearing on CD95- and anticancer drug-induced apoptosis. J Exp Med 1998; 188: 979 --984. function and bioenergetics. Antioxid Redox Signal 2008; 10: 1313 --1342. 21 Leist M, Single B, Castoldi AF, Kuhnle S, Nicotera P. Intracellular adenosine 47 Rossignol R, Gilkerson R, Aggeler R, Yamagata K, Remington SJ, Capaldi RA. Energy triphosphate (ATP) concentration: a switch in the decision between apoptosis and substrate modulates mitochondrial structure and oxidative capacity in cancer necrosis. J Exp Med 1997; 185: 1481 --1486. cells. Cancer Res 2004; 1: 64:985 --64:993. 22 Zamaraeva MV, Sabirov RZ, Maeno E, Ando-Akatsuka Y, Bessonova SV, Okada Y. 48 Boyd RS, Jukes-Jones R, Walewska R, Brown D, Dyer MJ, Cain K. Protein profiling of Cells die with increased cytosolic ATP during apoptosis: a bioluminescence study plasma membranes defines aberrant signaling pathways in mantle cell with intracellular luciferase. Cell Death Differ 2005; 12: 1390 --1397. lymphoma. Mol Cell Proteomics 2009; 8: 1501 --1515. 23 Halicka HD, Ardelt B, Li X, Melamed MM, Darzynkiewicz Z. 2-Deoxy-D-glucose 49 Capasso M, Bhamrah MK, Henley T, Boyd RS, Langlais C, Cain K et al. HVCN1 enhances sensitivity of human histiocytic lymphoma U937 cells to apoptosis modulates BCR signal strength via regulation of BCR-dependent generation of induced by tumor necrosis factor. Cancer Res 1995; 55: 444 --449. reactive oxygen species. Nat Immunol 2010; 11: 265 --272. 24 Nam SY, Amoscato AA, Lee YJ. Low glucose-enhanced TRAIL cytotoxicity is 50 Wu M, Neilson A, Swift AL, Moran R, Tamagnine J, Parslow D et al. Multiparameter mediated through the ceramide-Akt-FLIP pathway. Oncogene 2002; 21: 337 --346. metabolic analysis reveals a close link between attenuated mitochondrial 25 Munoz-Pinedo C, Ruiz-Ruiz C, Ruiz de AC, Palacios C, Lopez-Rivas A. Inhibition of bioenergetic function and enhanced glycolysis dependency in human tumor glucose metabolism sensitizes tumor cells to death receptor-triggered apoptosis cells. Am J Physiol Cell Physiol 2007; 292: C125 --CC36. through enhancement of death-inducing signaling complex formation and apical 51 Harper N, MacFarlane M. Recombinant TRAIL and TRAIL receptor analysis. procaspase-8 processing. J Biol Chem 2003; 278: 12759 --12768. Methods Enzymol 2008; 446: 293 --313. Supplementary Information accompanies the paper on the Oncogene website (http://www.nature.com/onc)

Oncogene (2012) 4996 --5006 & 2012 Macmillan Publishers Limited