Targeting Glucose Consumption and Autophagy in Myeloma with the Novel Nucleoside Analogue 8-Aminoadenosine*

Multiple myeloma, an incurable plasma cell malignancy, is characterized by altered cellular metabolism and resistance to apoptosis. Recent connections between glucose metabolism and resistance to apoptosis provide a compelling rationale for targeting metabolic changes in cancer. In this study, we have examined the ability of the purine analogue 8-aminoadenosine to acutely reduce glucose consumption by regulating localization and expression of key glucose transporters. Myeloma cells counteracted the metabolic stress by activating autophagy. Co-treatment with inhibitors of autophagy results in marked enhancement of cell death. Glucose consumption by drug-resistant myeloma cells was unaffected by 8-aminoadenosine, and accordingly, no activation of autophagy was observed. However, these cells can be sensitized to 8-aminoadenosine under glucose-limiting conditions. The prosurvival autophagic response of myeloma to nutrient deprivation or to nucleoside analogue treatment has not been described previously. This study establishes the potential of metabolic targeting as a broader means to kill and sensitize myeloma and identifies a compound that can achieve this goal.

Multiple myeloma, an incurable plasma cell malignancy, is characterized by altered cellular metabolism and resistance to apoptosis. Recent connections between glucose metabolism and resistance to apoptosis provide a compelling rationale for targeting metabolic changes in cancer. In this study, we have examined the ability of the purine analogue 8-aminoadenosine to acutely reduce glucose consumption by regulating localization and expression of key glucose transporters. Myeloma cells counteracted the metabolic stress by activating autophagy. Co-treatment with inhibitors of autophagy results in marked enhancement of cell death. Glucose consumption by drug-resistant myeloma cells was unaffected by 8-aminoadenosine, and accordingly, no activation of autophagy was observed. However, these cells can be sensitized to 8-aminoadenosine under glucose-limiting conditions. The prosurvival autophagic response of myeloma to nutrient deprivation or to nucleoside analogue treatment has not been described previously. This study establishes the potential of metabolic targeting as a broader means to kill and sensitize myeloma and identifies a compound that can achieve this goal.
Multiple myeloma (MM) 2 is a plasma cell malignancy with a yearly incidence of 14,000 in the United States and accounts for 10% of deaths from hematological malignancies (1). Multiple myeloma remains incurable with a median survival of 3-5 years. Current therapeutic options extend longevity, but patients eventually succumb to the disease due to the development of drug resistance (2).
Nucleoside analogues are antimetabolites that play a pivotal role in the treatment of a spectrum of hematological malignancies. Our laboratory has extensively characterized the novel purine nucleoside analogues 8-chloroadenosine and congener 8-aminoadenosine (8-NH 2 -Ado) (3)(4)(5), which are highly effec-tive in tissue culture models of multiple myeloma as well as a spectrum of other hematological and solid cancers. 8-Chloroadenosine is currently in phase I clinical trial. The general mechanism of action of these antimetabolites involves RNA and/or DNA termination, ATP depletion, and subsequent induction of apoptosis (6), but a comprehensive analysis of their cellular effects has not been reported.
In this study, we describe a novel feature in the mechanism of action of purine nucleoside analogue 8-NH 2 -Ado. We find that 8-NH 2 -Ado inhibits glucose consumption that is associated with an activation of autophagy. This ability to reduce glucose consumption is particularly important, given the dependence of many tumor cells on aerobic glycolysis and the ensuing increase in glucose consumption to meet energetic and biosynthetic demands. Normal cells typically rely on oxygen to metabolize glucose via the glycolytic and oxidative phosphorylation pathways to generate ATP. In contrast, many tumor cells demonstrate enhanced glycolysis and lactate production, utilizing the less efficient mode of ATP production even in the presence of oxygen (7). This phenomenon, termed aerobic glycolysis, was first described by Otto Warburg in the 1930s and was originally thought to be a result of defects in oxidative phosphorylation (OXPHOS) (8). Recent studies have shown that tumor cells do contain functional mitochondria (9) yet still produce excessive lactate, suggesting that the enhanced glycolytic flux may confer a growth advantage. In support of this notion, tumor cells forced to revert to oxidative phosphorylation display reduced tumorigenicity (9). Clinicians have capitalized on the ubiquity of these metabolic alterations in cancer (7), including myeloma (10,11), for diagnostic and prognostic purposes through the use of [ 18 F]fluorodeoxyglucose positron emission tomography. The significance of the glycolytic phenotype in neoplastic pathogenesis has been explored in a variety of studies. Large scale gene expression analyses reveal the selective up-regulation of genes encoding constituents of the glycolytic pathway across numerous forms of cancer, including myeloma (12). An intimate association between transcriptional control of metabolic genes and the activity of classical oncogenes and tumor suppressors, including Myc, p53, and Hif1␣, is well documented (13). Recently, requisite events preceding the switch from OXPHOS to aerobic glycolysis in immortalized cells have been shown to involve expression of the embryonic form of pyruvate kinase, correlating with the ability to form tumors in xenograft studies (14). These studies have stimulated a renewed interest in strategies that target metabolism and cellular bioenergetics unique to cancer cells.
Here we report that 8-NH 2 -Ado acutely regulates glucose uptake prior to loss of mitochondrial membrane potential. Glucose deprivation-induced bioenergetic stress is buffered by the cellular activation of autophagy, providing the rationale for combining 8-NH 2 -Ado with inhibitors of autophagy. In cells more resistant to 8-NH 2 -Ado, artificial reduction of glucose consumption robustly sensitizes these cells to 8-NH 2 -Ado without the activation of autophagy. This finding suggests that tumor cells with deficient autophagic responses may be uniquely susceptible to interference with glucose utilization pathways.
Patient Samples-Multiple myeloma cells were freshly isolated from bone marrow aspirates after informed consent. Mononuclear cells were isolated with Ficoll/Histopaque 1077 (Sigma), followed by CD138ϩ enrichment with microbeads and automated magnetic cell sorting using an AutoMacs cell sorter (Miltenyi Biotec, Auburn, CA).
Cell Proliferation Assays-CellTiter 96 AQ ueous Non-Radioactive Cell Proliferation Assay (Promega, Madison, WI) was used to determine cell growth. Cells (10,000 -25,000 cells/ well) were cultured in 96-well plates, and 490-nm absorbance measured was proportional to the number of live cells.
AnnexinV/DAPI Staining-Subsequent to specific treatments, cells were washed in PBS and stained with AnnexinVfluorescein isothiocyanate/APC according to the manufacturer's instructions. Samples were run on Dako CyAn TM ADP analyzer. Data analysis was performed with the FCS Express version 3 (De Novo Software, Los Angeles, CA).
HPLC Analysis of Intracellular 8-NH 2 -ATP and ATP-Nucleotides were extracted using perchloric acid and neutralized with KOH according to the protocol described in Ref. 15. 1 ϫ 10 7 cells were harvested after treatment and washed with PBS, and acid extraction of cell pellet was performed with 220 l of 0.3 M cold perchloric acid. After vortexing tubes were kept on ice for 15 min. Acidic supernatant was isolated by centrifuga-tion at 14,000 rpm for 30 s and neutralized with 100 l of 0.5 N KOH. Samples were stored at Ϫ80°C and eluted isocratically from a TSK gel DEAE-2 SW (240 ϫ 4.6 mm, 5-m particle size, (TOSOH Corp. Tokyo, Japan) with a 0.06 M Na 2 HPO 4 (pH 6.9) acetonitrile buffer (60:40) at a flow of 1.0 ml/min. The eluant was monitored at 261 nm, and concentrations of ATP and 8-NH 2 -ATP were determined from an external standard curve, which was linear from 0.01 to 10 mM for both compounds. Intracellular concentrations were calculated by determination of cell volume. The volume of cells was determined by measuring mean cell diameter using a Beckman Coulter Vi-Cell TM XR cell viability analyzer.
Mitochondrial Membrane Potential-0.5 ϫ 10 6 cells were treated with 3 M 8-NH 2 -Ado for the indicated times. 30 min prior to the harvest time, potentiometric dye tetramethylrhodamine ethyl ester perchlorate was added to all wells at a final concentration of 100 nM. Cells were incubated at 37°C for 30 min, followed by PBS wash, and resuspended in PBS containing 2% fetal bovine serum and 20 nM tetramethylrhodamine ethyl ester perchlorate, and samples were analyzed by flow cytometry.
Oxygen Consumption Measurements in Intact Cells-The respiration rate was measured with an Oxygraph (Hansatech Instruments) in 2 ϫ 10 6 (MM.1S) or 5 ϫ 10 6 (U266) cells/ml of PBS. After obtaining a stable rate, sodium cyanide was added at a concentration of 5 M, and the resulting respiration rate was recorded. Oxygen consumption rates were expressed in nmol of oxygen consumed/10 6 cells.
Glucose Consumption Assays-Cellular glucose consumption correlating to [ 3 H]2-deoxyglucose uptake was determined by modification of the protocol described previously (16). Briefly 5 ϫ 10 6 MM.1S cells were left untreated or treated with 3 M 8-NH 2 -Ado for 5 h. Cells were then harvested and resuspended in glucose-free medium with or without drug. And the end of 5 h, cells were harvested and resuspended in 1 ml of glucose-free RPMI 1640 containing 0.2 Ci/ml 2-deoxy-D-[1-3 H]glucose (specific activity 8 Ci/mmol; GE Healthcare) and 0.5 M 2-deoxyglucose, with or without cytochalasin B (5 M final concentration), for 25 min. The uptake was stopped by pelleting cells and resuspending them in ice-cold PBS supplemented with 5 mM 2-deoxyglucose. Cell pellets were lysed in 3% SDS, and 3 H counts/min in the total lysate were measured using a liquid scintillation counter.
The rate of cellular glucose consumption was also determined using the Amplex Red glucose/glucose oxidase kit (Invitrogen) following manufacturer's instructions. Glucose concentrations in medium samples taken at the start and end of the experiments were used to determine cellular uptake.
Immunoblot Analysis-Cellular lysates were prepared with the Complete Lysis-M buffer (Roche Applied Science) supplemented with protease and phosphatase inhibitor mixture tablets (Roche Applied Science) prepared according to the manufacturer's instructions. Protein concentrations were determined by a Bio-Rad protein assay. Proteins were separated on precast Tris-glycine gels (Invitrogen) transferred to a polyvinylidene difluoride membrane (Immobilon-P; Millipore, Bedford, MA), following which membranes were blocked in casein. Primary antibody incubations were over-night at 4°C, followed by 1-h room temperature incubations with horseradish peroxidase-linked anti-rabbit secondary antibody (Cell Signaling Technology, Inc., Danvers, MA) or anti-mouse secondary antibody (Amersham Biosciences). Blots were developed using the Enhanced Chemiluminescence Plus Western blotting detection reagent (Amersham Biosciences).
Immunofluorescence Microscopy-Cells were washed once in PBS, and ϳ65,000 cells were spun onto coated microscope slides (Shandon Cytoslide TM ) using a Shandon Cytospin TM 2 cytocentrifuge (Thermo Electron Corp., Pittsburgh, PA). Slides were fixed in 3% paraformaldehyde, permeabilized with 0.03% saponin in PBS, and blocked with 10% normal goat serum. All primary and secondary antibody dilutions were made in PBS containing 10% normal goat serum and 0.03% saponin. Cells were stained overnight with 1:100 dilutions of antibodies to human GLUT1, GLUT4, LC3, or Golgin-97 with detection using anti-rabbit IgG-Alexa Fluor 594 or anti-mouse IgG-Alexa 488 (Invitrogen) for 60 min. Cells were mounted with Ultra Cruz mounting medium (Santa Cruz Biotechnology, Inc., Santa Cruz, CA) containing DAPI for counterstaining. Cells were visualized with a 63ϫ (1.4 numerical aperture) oil objective LSM-510 Meta Carl Zeiss confocal microscope. Image analysis was performed using the Zeiss Axiovision LE image browser.
Acridine Orange Staining-Acidic vesicular organelles (AVOs) were detected by incubating cells with acridine orange (17). In stained cells, acidic compartments fluoresced in bright red, quantified using fl3 mode (Ͼ650 nm), and base-line green/ fluorescein isothiocyanate fluorescence was measured utilizing the fl1 channel (500 -550 nm). To assess the percentage of cells forming AVOs, cells were incubated with 0.1 g/ml acridine orange (Molecular Probes) for 15 min at 37°C just prior to harvesting at the end of a treatment period.
Statistical Analyses-All data presented are derived from at least three independent determinations, unless otherwise noted. Two-tailed p values were calculated using paired or unpaired t tests as indicated, with GraphPad Prism software (San Diego, CA). p values of Ͻ0.05 were considered to be statistically significant.

RESULTS
Sensitivity of Myeloma Cell Lines to 8-NH 2 -Ado-Sensitivity to 8-NH 2 -Ado was determined by assessment of cellular proliferation ( Fig. 1, A1) in response to a range of concentrations of 8-NH 2 -Ado in the MM.1S, RPMI 8226, U266 (myeloma) and ARH 77 (B lymphoblastoid) cell lines. The MM.1S cells were most sensitive to the growth-inhibitory effects of 8-NH 2 -Ado followed by ARH 77, RPMI 8226 and U266 cells. Studying two differentially sensitive cell lines represents an ideal system to investigate unappreciated aspects of the mechanism of action of 8-NH 2 -Ado, particularly those that directly influence cell death. Therefore, we selected U266 and MM.1S cells for further study and verified the differential toxicity of 8-NH 2 -Ado by directly measuring cell death with the same dose range of 8-NH 2 -Ado ( Fig. 1, A2). The IC 50 in MM.1S and U266 cells is 1.5 and 8.88 M, respectively. We have previously shown in MM.1S cells that 8-NH 2 -Ado rapidly decreases intracellular ATP with a concomitant accumulation of 8-NH 2 -ATP (5). Studies in adenosine kinase-deficient cells indicate that the toxic effects of 8-NH 2 -Ado require adenosine kinase activity to convert 8-NH 2 -Ado to 8-NH 2 -ATP (5). To ensure that decreased accumulation of the active compound was not responsible for the relative resistance of the U266 cell line, we first measured ATP and 8-NH 2 -ATP levels following treatment in both cell lines.
Reduction in Intracellular ATP Is More Pronounced in the MM.1S Cells, whereas 8-NH 2 -ATP Accumulation Is Equivalent-We determined intracellular 8-NH 2 -ATP and ATP levels in a time course assay of MM.1S and U266 cells treated with 3 M 8-NH 2 -Ado. Although the intracellular concentration of 8-NH 2 -ATP reaches similar levels in both cell types, we observed a greater drop in ATP levels in the MM.1S cells (Fig. 1, B1 and B2). The similar accumulation of 8-NH 2 -ATP in the less sensitive U266 cells rules out the possibility that deficient uptake of pro-drug and subsequent conversion to 8-NH 2 -ATP accounts for lower toxicity. However, we were interested in determining the cause of the steeper drop in ATP levels in the MM.1S cells. Competition between phosphoderivatives of 8-NH 2 -Ado and their corresponding endogenous counterparts (e.g. ADP and AMP) for enzymes involved in adenosine phosphorylation probably accounts for the decreased ATP levels in U266 cells. The observation that 8-NH 2 -ATP accumulation is mirrored in the MM.1S cells while ATP levels decline to a greater extent suggested to us that a bioenergetic pathway may be additionally and selectively impaired in the MM.1S cells. Therefore, we investigated the effects of 8-NH 2 -Ado on two other ATP-generating pathways: OXPHOS and glycolysis.
Two Indices of Mitochondrial Function Reveal No Alterations at Early Time Points following 8-NH 2 -Ado Treatment-To ask whether a selective inhibition of oxidative phosphorylation may account for the steeper drop in ATP levels in the MM.1S cells, we evaluated mitochondrial functionality by measuring mitochondrial membrane potential (⌬ m ) during a time course of 8-NH 2 -Ado treatment. At 5 h post-drug treatment, a time point at which there is significant loss of intracellular ATP in both cell lines, there was no observed reduction of ⌬ m (Fig. 1C). At later time points, there was a selective decrease in ⌬ m in the MM.1S cells, which was probably a secondary effect of cell death caused by mitochondrial outer membrane permeabilization and cytochrome c release. The observation that ATP concentrations decrease before ⌬ m declines suggests that loss of mitochondrial function is not the cause of the bioenergetic deficit. To validate this conclusion, we also measured oxygen consumption in MM.1S and U266 cells treated with 8-NH 2 -Ado.  Additionally, to confirm that there is no change in mitochondrial respiration in the MM.1S and U266 cells, we measured oxygen consumption following cyanide treatment. Cyanide is a potent inhibitor of mitochondrial respiration, and any oxygen consumption measured post-cyanide treatment is considered non-mitochondrial. Utilizing cyanide, we calculated the ratio of mitochondrial oxygen consumption to total oxygen consumption for untreated and 8-NH 2 -Ado-treated cells. This ratio was not significantly different between treated and untreated cells (Fig. 1, D2 and D4) in either MM.1S or U266 cells. These results confirm that changes in mitochondrial respiration are not responsible for the greater decline in intracellular ATP seen in the MM.1S cells.
An alternative mechanism that could explain this occurrence could be changes in flux through the glycolytic pathway. Glucose metabolism via glycolysis is thought to provide a rapid (albeit inefficient) supply of ATP to proliferating tumor cells. To investigate whether 8-NH 2 -Ado elicits changes in glucose metabolism, we measured glucose consumption in MM.1S and U266 cells treated with 8-NH 2 -Ado.
8-NH 2 -Ado Decreases Glucose Uptake in the MM.1S Cell Line-To evaluate glucose uptake, we treated MM.1S cells with 3 M 8-NH 2 -Ado for 5 h. Glucose consumption was measured by the uptake of [ 3 H]2deoxyglucose. We observed that 5 h of 8-NH 2 -Ado treatment caused a 50% reduction in glucose consumption ( Fig. 2A) in comparison with untreated MM.1S cells. Cytochalasin B treatment nonspecifically blocks glucose transporter activity and is included to measure the contribution of transporter-independent glucose uptake, which is minimal.
Myeloma Cells Less Sensitive to 8-NH 2 -Ado Resist Changes in Glucose Consumption-We assessed the 8-NH 2 -Ado-induced regulation of glucose consumption in the panel of cell lines previously characterized for their sensitivity to 8-NH 2 -Ado. Glucose uptake was analyzed in the four

Targeting Glucose Metabolism in Myeloma
cell lines at 5 and 17 h after 8-NH 2 -Ado treatment. Glucose consumption was determined by measuring glucose depletion from media by cells cultured in the absence or presence of 8-NH 2 -Ado. A 40% decline in glucose consumption was observed in MM.1S cells treated with 3 M 8-NH 2 -Ado for 5 h (Fig. 2, B1), thus corroborating previous results evaluating [ 3 H]2-deoxyglucose uptake. RPMI 8226 and ARH 77 cells (Fig.  2, B2 and B4), which are more resistant to 8-NH 2 -Ado, required higher concentrations of 8-NH 2 -Ado (i.e. 6 M) to induce decreases in glucose consumption comparable with those seen in the sensitive MM.1S cells. Even at the higher 6 M dosage of 8-NH 2 -Ado, the U266 cells exhibited only a slight decrease in glucose consumption, which diminished over time (Fig. 2, B3), consistent with the observed resistance to 8-NH 2 -Ado. Given the connections between ATP generation and glucose consumption within the context of malignancy, these observations suggest that 8-NH 2 -Adoinduced glucose limitation could facilitate the greater decline in ATP in the sensitive MM.1S cells.
Regulation of Glucose Uptake Is Unique to 8-NH 2 -Ado-To determine if the reduction in glucose consumption is a generalized phenomenon preceding myeloma cell death or represents a unique property of 8-NH 2 -Ado, we examined glucose uptake in the presence of clinically relevant therapeutics that act through distinct mechanisms. We examined the effect of the myeloma therapeutic bortezomib (Velcade; PS-341) and the nucleoside analogue gemcitabine on glucose transport rates in comparison with 8-NH 2 -Ado. At the concentrations used, these compounds elicit higher toxicity than 8-NH 2 -Ado (18, 19) but do not demonstrate an inhibition of glucose consumption comparable with that seen with 8-NH 2 -Ado (Fig. 2C), thereby suggesting that regulation of glucose consumption is not a generalized phenomenon associated with cytotoxic therapeutics.

8-NH 2 -Ado Treatment Causes Aggregation of GLUT4 in MM.1S
Cells-To examine the functional role of diminished glucose consumption in the toxicity of 8-NH 2 Ado, we first investigated potential mechanisms that may be responsible for the decline in glucose consumption in the MM.1S cell line. We examined expression and subcellular localization of GLUT1 and GLUT4, two key glucose transporters present in B cells. Immunoblot analysis of whole cell lysates of MM.1S cells treated with 8-NH 2 -Ado for 5 and 24 h indicates a timedependent decrease in GLUT1 expression at 5 h, whereas at 24 h there was a down-regulation of both transporters (Fig. 3,  A1 and A2). Given the rapid, robust inhibition of glucose uptake and the relatively modest reduction in GLUT1 protein levels at early time points, we hypothesized that another regulatory mechanism may be involved. We then determined the subcellular localization of GLUT1 and -4 by immunofluorescence microscopy, since the trafficking of these proteins to the cell  SEPTEMBER 25, 2009 • VOLUME 284 • NUMBER 39 surface also impacts their functionality. Untreated MM.1S (Fig.  3, B1) and U266 cells (Fig. 3, B2) showed a homogenous cytoplasmic and plasma membrane distribution of GLUT1 and -4. Treatment of MM.1S cells with 8-NH 2 -Ado caused the selective aggregation of GLUT4 within the trans-Golgi network, as evidenced by co-localization of GLUT4 with anti-Golgin-97, a trans-Golgi network protein (20) (Fig. 3, B1). A corresponding reduction in the plasma membrane and cytoplasmic distribution of GLUT4 in MM.1S cells was also detected. Conversely, we detected minimal translocation of GLUT1 (Fig. 3, B1) in the MM.1S cells. The localizations of GLUT1 and -4 were not regulated by 8-NH 2 -Ado in the U266 cells at either time (Fig. 4, A1,  A2, and B2). Therefore, abrogation of GLUT4 plasma membrane localization is temporally associated with the selective decrease in glucose uptake seen in the MM.1S cell line, suggesting a causal relationship between these events.

Restoration of Glucose Consumption during 8-NH 2 -Ado Treatment in MM.1S Cells via GLUT1 Overexpression Elicits
Only a Mild Cytoprotective Effect-To directly assess the role of the 8-NH 2 -Ado-induced decrease in glucose consumption in the MM.1S cells, we overexpressed GLUT1 to reverse this effect during treatment. We chose to exogenously express GLUT1 to restore glucose transport rates due to the sustained plasma membrane localization of this protein during 8-NH 2 -Ado treatment. If GLUT4 were overexpressed, this protein would presumably aggregate similarly to the endogenous protein and not affect glucose uptake. Furthermore, we utilized a lower concentration of 8-NH 2 -Ado, which was more suitable for the extended duration of these experiments. Increased GLUT1 expression (Fig. 4A) effectively reversed the 8-NH 2 -Ado-induced decline in glucose consumption (Fig. 4B). Given the strong connections in the literature between glycolytic inhibition and tumor cell death, we anticipated that restoration of glucose consumption during 8-NH 2 -Ado treatment would partially abrogate the toxicity of the drug in the MM.1S cells. This result would have indicated that regulation of glucose transporter functionality accounts for the relative sensitivity of the MM.1S cells compared with the U266 cells. Surprisingly, the protective effect of GLUT1 overexpression and restoration of glucose consumption during 8-NH 2 -Ado treatment was much less than anticipated and did not reach statistical significance (Fig. 4C). Considering that the maintenance of glucose metabolism is critical to the growth and survival of tumor cells, we hypothesized that the deleterious effects of glucose deprivation during 8-NH 2 -Ado treatment of MM.1S cells may be counteracted by a cellular response to metabolic stress. One process known to be activated under conditions of such stress is autophagy. Therefore, we sought to determine whether an increase in autophagic activity may be involved.
Decreased Glucose Consumption Elicited by 8-NH 2 -Ado Activates Autophagy-Nutrient deprivation can be associated with the activation of the process of macroautophagy (herein referred to as autophagy) (21) as a survival mechanism allowing cells to catabolize intracellular components to sustain critical bioenergetic and synthetic needs. This protective response results in resistance to the toxic effects of drugs that elicit cell death by creating metabolic stress (22). To determine whether an autophagic response is associated with 8-NH 2 -Ado treatment, we examined 8-NH 2 -Ado-treated MM.1S cells for changes in the LC3 (microtubule-associated protein-1 light chain-3) protein (23). Upon induction of autophagy, cytoplasmic LC3-I is post-translationally processed to a membranebound form, LC3-II, which accumulates in autophagosomes. Therefore, LC3-II levels represent an ideal marker for the initiation of autophagy. After 2.5 h of 3 M 8-NH 2 -Ado treatment, we detected the formation of LC3-II protein by immunoblot analysis (Fig. 5A) and by the appearance of LC3-II punctuate staining, characteristic of membrane-bound LC3-II (Fig. 5B). As an additional indicator of autophagic activity, we stained with acridine orange (a lysosomal marker), which fluoresces bright red under the acidic conditions in autophagolysosomes (17). Treatment of MM.1S cells with 1 M 8-NH 2 -Ado over a longer period caused the appearance of a population with a high AVO content (Fig. 5, C1 and C2), indicating the activation of  autophagy. To confirm that the AVO-high population demonstrates activation of autophagy, we sorted the AVO-high and AVO-low populations and determined expression levels of p62 protein, which is decreased upon activation of autophagy (24). p62 is responsible for the aggregation of polyubiquitylated proteins with LC3, thereby enabling their degradation in the autophagosome (24). The AVO-high population induced with 8-NH 2 -Ado exhibits a reduction in p62 expression in comparison with the AVO-low population (as shown in Fig. 5, D1 and D2). We also treated cells with inhibitors of autophagy. 3-Methyladenine (3-MA) inhibits Class III phosphatidylinositol 3-kinase Vps34, the activity of which is required for autophagosome formation. Consistent with its activity at proximal steps in the pathway, this compound completely blocked the appearance of the AVO-high population during 8-NH 2 -Ado treatment (Fig. 6A). Chloroquine (CQ), an inhibitor of lysosomal acidification, acts further downstream in the autophagic pathway by inhibiting the fusion of the autophagosomes with lysosomes (25) or by inhibiting the acid-dependent degradation of autophagosomal contents (26). In accordance with its mechanism of action, we observed an increase in the number of cells in the AVO-high population during CQ co-treatment, indicative of impaired processing of autophagosomes (Fig. 6A). In summation, the aforementioned studies suggest that the AVO-high population represents cells actively undergoing autophagy in response to incubation with 8-NH 2 -Ado.

MM.1S Cells Activate Autophagy as a Survival Response to 8-NH 2 -
Ado Exposure-To determine whether autophagy was influencing the toxicity of 8-NH 2 -Ado in the MM.1S cells, we examined the effect of CQ and 3-MA co-treatment on cell death. As shown in Fig. 6, A and B, we observed an enhancement of cell death when autophagy was blocked in the 8-NH 2 -Ado-treated cells. As expected, the AVO-high population treated with CQ demonstrates increased AnnexinV positivity (Fig. 6, A and B). Increased Annex-inV staining was also seen with 3-MA co-treatment, although this manifested in the AVO-low population due to the inhibitory effect of 3-MA on autophagosome formation (Fig. 6, A and B). These results confirm that the activation of autophagy in response to 8-NH 2 -Ado is a prosurvival response.

Inhibition of Glucose Consumption by 8-NH 2 -Ado Is Required for Induction of Autophagy and Sensitization to CQ in
MM.1S Cells-To connect the inhibition of glucose uptake by 8-NH 2 -Ado and the autophagic response of MM.1S cells, we again restored glucose uptake during 8-NH 2 -Ado and assessed markers of autophagy. This time we employed a pharmacological approach to reverse the decline in glucose consumption by pretreating cells with metformin. This anti-diabetic drug increases glucose consumption in muscle cells by promoting plasma membrane localization of GLUT1 and GLUT4 (27). Cotreatment with metformin restored glucose uptake in MM.1S cells to base-line levels (Fig. 7A). Metformin treatment also reversed the appearance of the LC3-II protein (Fig. 7, B1 and B2) and the AVO-high population (Fig. 7, C1 and C2) induced by 8-NH 2 -Ado treatment. Importantly, metformin treatment reverses the enhancement in cell death induced by CQ co-treatment with 8-NH 2 -Ado (Fig. 7D). Therefore, blocking the activation of autophagy enhances the efficacy of 8-NH 2 -Ado by potentiating the stress of glucose deprivation. Overexpression of GLUT1 in MM.1S cells also reversed the enhancement by CQ of 8-NH 2 -Ado-induced cell death (Fig. 7E). Taken together, these results confirm that activation of autophagy is a direct survival response to the inhibition of glucose consumption elicited by 8-NH 2 -Ado.
Limiting Glucose Consumption Sensitizes Resistant U266 Cells to 8-NH 2 -Ado-We were interested in determining the effect of artificially reducing glucose consumption in the resistant U266 cells and evaluating sensitivity to 8-NH 2 -Ado. We cultured U266 cells in glucose-limiting medium (glucose-free medium with 10% serum) and assayed for sensitivity to 8-NH 2 -Ado. The effective glucose concentration in medium containing 10% serum was determined to be 0.7 mM, far lower than the 11 mM glucose of standard medium. Indeed culture in low glucose-containing medium sensitizes U266 cells to 8-NH 2 -Ado with a 1.75-fold enhancement of drug-induced cell death (Fig.  8A). The U266 cells did not demonstrate activation of autophagy based on the lack of an AVO-high cell population and the insignificant regulation of p62 expression (Fig. 8, B1, B2, C1,  and C2). Treatment of U266 cells in complete or low glucosecontaining medium with 8-NH 2 -Ado does not sensitize them to CQ or 3-MA treatments, consistent with the fact that no autophagic response is detected in these cells (Fig. 8D). We assayed a variety of compounds known to activate autophagy (including rapamycin and valproic acid) for use as positive controls in the U266 cells, but all autophagy markers were negative (data not shown). This suggests that the lack of activation of autophagy in the U266 cells may compromise the ability to buffer metabolic stresses.
Primary Multiple Myeloma Cells Exhibit Enhanced Sensitivity to 8-NH 2 -Ado with CQ Co-treatment-We further substantiated the utility of combining 8-NH 2 -Ado with CQ in myeloma patient samples. CD138ϩ plasma cells obtained from myeloma patients were treated with a dose range of 8-NH 2 -Ado with and without CQ co-treatment. Analogous to the MM.1S cell line, CQ co-treatment was found to enhance sensitivity of myeloma primary cells to 8-NH 2 -Ado (Fig. 9). This observation underscores the therapeutic potential of targeting cellular bioenergetics and the ensuing process of autophagy in myeloma.

DISCUSSION
Nucleoside analogues have been used extensively in the treatment of a spectrum of hematological malignancies and more recently for solid tumors (28). Continued understanding of the mechanisms of action and pathways of resistance to these first line cancer chemotherapeutics can elucidate cancer-specific molecular targets. In this study, we set out to determine whether cellular bioenergetics could account for the greater sensitivity of a representative cell line to 8-NH 2 -Ado. Although we found that glucose transport inhibition by 8-NH 2 -Ado was unrelated to this increased sensitivity, this phenomenon did enable us to identify strategies to increase the activity of this compound in two different cellular contexts.
We previously established that 8-NH 2 -Ado toxicity is associated with a substantial decrease in intracellular ATP levels (5). Given the more pronounced decline in ATP levels in the sensitive MM.1S cells compared with the resistant U266 cells, we examined several parameters of cellular bioenergetics in these two cell lines following 8-NH 2 -Ado treatment. We first measured changes in ⌬ m and mitochondrial oxygen consumption in these cell lines in response to 8-NH 2 -Ado as indicators of OXPHOS activity. We noted no difference in these parameters between MM.1S and U266 cells during the first 5 h following treatment. In the absence of changes in cellular respiration, a potential upstream metabolic pathway that might account for a differential decline in intracellular ATP levels between the two cell lines could be glycolysis.
Our data demonstrate the specific regulation of glucose uptake in myeloma cell lines. We have determined aggregation of GLUT4 within the endosomal trans-Golgi network that may potentially account for the decrease in glucose consumption induced by 8-NH 2 -Ado. Although the exact role for localization of GLUT4 in the trans-Golgi network is unknown, studies in muscle and adipose tissue indicate that transit into the trans-Golgi network may allow for storage and/or routing of GLUT4 into an insulin-responsive compartment (29,30). We cannot rule out the roles of other downstream effectors of glycolysis and glucose consumption (i.e. allosteric regulation of phosphofructokinase 1 by phosphorylation of phosphofructokinase-2) (31) and hexokinase II (32) contributing to 8-NH 2 -Ado-induced reduction of glucose consumption.
The overexpression of GLUT1 and restoration of glucose consumption in the MM.1S cells during 8-NH 2 -Ado treatment led to an insignificant reversal of cell death elicited by 8-NH 2 -Ado. Metabolic stress can be counteracted by the activation of autophagy, which aids in the release of critical biosynthetic intermediates by catabolizing cytoplasmic macromolecules. The decrease in glucose consumption stimulated by 8-NH 2 -Ado treatment exerts energetic stress in the MM.1S cell line, which precipitates an activation of autophagy. The glucose deprivation-induced activation of autophagy in MM.1S cells promotes resistance, as evidenced by the increased cell death seen when 8-NH 2 -Ado is administered in combination with inhibitors of autophagy. Using two different approaches (i.e. pretreatment with metformin or overexpression of GLUT1), we were able to demonstrate reversal of CQ-induced hypersensitivity to 8-NH 2 -Ado. The stimulatory effect of metformin on tumor cell glucose transport described herein is of direct clinical significance, since this effect could potentially antagonize the activity of therapeutics, such as genotoxic agents and mTOR inhibitors, that have been shown to be more efficacious in tumor cells exhibiting decreased glycolytic flux (33)(34)(35).
Our data suggest two distinct responses elicited upon regulation of glucose consumption in different myeloma cell types, as outlined in Fig. 10. The sensitive MM.1S cells demonstrate a rapid decline in glucose consumption that does not facilitate cell death and is only responsible for activating autophagy. This active process of protecting cells against nutrient deprivation can be taken advantage of only when induced (i.e. the protective effects are afforded only in the context of a decline in glucose). Therefore, the metabolic stress associated with glucose deprivation can only translate to enhanced cell death upon concomitant inhibition of autophagy. Restoring glucose consumption to basal rates does not increase cell death, because the cell has not experienced nutrient deprivation or activated autophagy. Future work will focus on determining the particular metabo-lites generated by autophagy that can compensate for glucose restriction. Interestingly, we have determined that methyl pyruvate is incapable of compensating for 8-NH 2 -Ado-induced glucose deprivation. This suggests that tricarboxylic acid cycle stimulation is not the critical downstream effector of glucose metabolism in our system. It is tempting to speculate that increased abundance of glutamine following protein catabolism in autophagolysosomes could be playing a role, given the overlapping ability of both glucose and glutamine to stimulate a distinct set of metabolic reactions. For example, alternative pathways exist for the generation of NADPH, biosynthesis of nucleotides, and supply of tricarboxylic acid cycle intermediates that can be stimulated by either molecule (36).
In contrast to the above studies, culture of the less sensitive U266 cells in glucose-limiting medium sensitizes these cells to the cytotoxic effects of 8-NH 2 -Ado. Several studies in the IL-3dependent FL5.12 cell line have demonstrated the protective effects of glucose metabolism during IL-3 withdrawal (37)(38)(39). In the aforementioned study, reduction in glucose consumption was a cause of, rather than a consequence of, declining mitochondrial function, and IL-3-deprived cells were rescued from growth factor withdrawal-induced apoptosis by overexpression of GLUT-1. The magnitude of glycolytic inhibition, which was proportional to growth factor withdrawal or glucose deprivation, correlated to the induction of apoptosis (37). These studies highlight two key observations pertinent to this study in that glucose consumption is not entirely driven by homeostatic demands and the decline in glucose consumption preceded the onset of apoptosis. Studies in the U266 cells indicate that a deficiency in autophagic activity may render cells incapable of buffering glucose deprivation-induced metabolic stress. The fact that many malignancies develop mechanisms to suppress autophagy may provide a unique opportunity for sensitization to classical chemotherapeutics with inhibitors of glucose metabolism (e.g. lonidamine or 2-deoxyglucose).
Multiple myeloma is characterized by a deregulation of glycolytic genes (12). This observation and the effectiveness of [ 18 F]fluorodeoxyglucose positron emission tomography in the detection of myeloma (10, 11) suggest a dependence of myeloma on increased glucose consumption and glycolysis. Glucose can contribute to maintaining mitochondrial integrity via the generation of the NADH-and FADH 2 -reducing equivalents (40,41). Glucose also helps maintain the association of hexokinase II with the mitochondria, thereby preventing the release of cytochrome C (32,42). Besides generating bioenergetic equivalents, such as ATP, glucose also serves to generate  . B1 and B2, U266 cells do not exhibit increased acridine orange staining during 8-NH 2 -Ado treatment regardless of glucose availability. U266 cells were treated or not with 3 M 8-NH 2 -Ado in full or glucose-limiting medium for 24 h before incubation with acridine orange (AO) and flow cytometry analysis. One representative experiment of three is shown. C1 and C2, p62 protein levels remain unaltered following 8-NH 2 -Ado treatment in U266 cells. Cells were treated as in B1 and B2 and lysed for immunoblot analysis of p62 expression. The decline in expression was not statistically significant (n ϭ 3, ϮS.E.). D, neither CQ nor 3-MA co-treatment affects toxicity of 8-NH 2 -Ado in U266 cells cultured in full or low glucose medium. U266 cells were treated for 24 h with the indicated combinations of drugs in complete medium (11 mM glucose; black bars) or glucose-limiting medium (0.7 mM glucose; gray bars). CQ was used at 10 M, whereas 3-MA was used at 2 mM. Cells were harvested and stained with AnnexinV/DAPI and analyzed by flow cytometry (n ϭ 3, ϮS.E.).
biosynthetic intermediates for nucleotide and fatty acid synthesis (36) and is also able to regulate various effectors controlling cell death (i.e. prosurvival Mcl-1 (43), proapoptotic BAD (44), and proapoptotic Bax (38)). The association of glycolysis and glucose consumption in immortalized B cells (37), leukemia and lymphoma cell lines, and other cancer cell lines (33,45) is well established. The contribution of the glycolytic phenotype to increased resistance to apoptosis-inducing agents (7,46) supports the utility in targeting glucose consumption.
Our results also suggest that further delineation of the role of autophagy in various therapeutic settings is warranted. It is important to determine whether activation of this pathway in tumor cells following drug treatment is associated with cell death or survival, since it may provide an opportunity to enhance the selectivity and potency of a given therapeutic. Indeed, chloroquine has been reported to increase the activities of doxorubicin, cyclophosphamide, and inhibitors of Akt in certain types of cancer (47)(48)(49). This notion is particularly intriguing when applied to myeloma, a malignancy characterized by progressive chemoresistance. Alternatively, compounds that have been shown to induce autophagic cell death could be prime candidates for use in malignancies harboring well defined mutations in components of the apoptotic pathway, thereby necessitating the engagement of alternative pathways to cell death.
Our study has demonstrated the effectiveness of 8-NH 2 -Ado as a pleiotropic compound that is able to elicit cell death along with an unrelated inhibition of glucose transport, a feature that appears to be unique to this compound among other myeloma therapeutics tested. The activation of autophagy can be self-destructive in time or provide an addi- FIGURE 9. Primary myeloma cells exhibit increased sensitivity to 8-NH 2 -Ado when co-treated with chloroquine. CD138ϩ cells were isolated from bone marrow aspirates of three myeloma patients. Next, cells were seeded in 96-well plates with the indicated concentrations of 8-NH 2 -Ado with and without 10 M CQ. Cells were incubated for 48 h before determining live cell numbers by an MTS assay. Results are expressed as a percentage of control, untreated cells (n ϭ 1). FIGURE 10. Schematics depicting pathways contributing to cell death following 8-NH 2 -Ado treatment in MM.1S and U266 cells. A, MM.1S cells exhibit decreased glucose uptake during 8-NH 2 -Ado treatment, which causes metabolic stress. This stress is alleviated by autophagy activity, thereby blocking the contribution of glucose deprivation to cell death. However, autophagy inhibitors block this protective response, resulting in much higher levels of cell death upon co-treatment. Furthermore, autophagy induction following 8-NH 2 -Ado treatment can be blocked by restoring glucose consumption via GLUT1 overexpression or metformin cotreatment. B, U266 cells exhibit no decrease in glucose consumption in response to 8-NH 2 -Ado. However, artificial reduction of glucose consumption sensitizes these cells to 8-NH 2 -Ado due to the lack of protective autophagic activity in this cell line. tional strategy to sensitize myeloma cells to 8-NH 2 -Ado. Future studies will focus on defining metabolic signatures in myeloma in order to predict responsiveness to genotoxic drugs and inhibitors of autophagy as well as to enable identification of specific effectors of glucose deprivation-mediated cell death. Understanding cellular metabolism in myeloma is a promising avenue that could yield potent combinatorial strategies based on interfering with glucose utilization during administration of existing therapeutics.