Glucose Deprivation-induced Cytotoxicity and Alterations in Mitogen-activated Protein Kinase Activation Are Mediated by Oxidative Stress in Multidrug-resistant Human Breast Carcinoma Cells*

,

We have previously observed that glucose deprivation activates mitogen-activated protein kinase (MAPK) 1 and induces cell death in multidrug-resistant human breast carcinoma MCF-7/ADR cells (1,2). Since activation of MAPK has been suggested to result from oxidative stress (3)(4)(5)(6), the hypothesis that glucose deprivation-induced cytotoxicity and alterations in MAPK signal transduction could result from metabolic oxida-tive stress was tested. Metabolic oxidative stress is defined as oxidative stress generated by mitochondrial biochemical pathways.
MAPK, also known as extracellular regulated protein kinases (ERKs), are serine/threonine kinases (7). Three MAPK homologs have been identified in mammalian cells as follows: ERK1, ERK2, and ERK3 (7). ERK1 and ERK2 are expressed ubiquitously, and their molecular masses are 44 and 42 kDa, respectively. The MAPK pathway is activated by various growth factors (8,9), alterations in intracellular redox state (3,6), and polyunsaturated fatty acids (10,11). MAPK activation appears to involve extracellular signals including oxidants stimulating the formation of active Ras, GTP (12,13). The GTP-bound Ras protein binds to Raf kinase and initiates a protein kinase cascade that leads to MAPK activation (14,15). Activated MAPK is believed to translocate into the nucleus (16) and phosphorylate nuclear substrates including redox regulated transcription factors believed to be involved with cellular responses to oxidative stress (e.g. ternary complex factor Elk-1, c-Myc, and c-Fos) (9,(17)(18)(19)(20)(21). Therefore, the MAPK pathway appears to be activated in response to oxidative stress and could be involved in cellular responses to oxidative stress.
Glucose metabolism has been shown to be involved with cellular sensitivity to oxidative stress mediated by hydroperoxides (22) presumably via the formation of pyruvate and NADPH. Furthermore, superoxide and hydroperoxides (i.e. H 2 O 2 ) have been shown to be produced as by-products of mitochondrial respiration (23). Since mitochondrial metabolism would be the preferred route of energy production during glucose deprivation, it is reasonable to hypothesize that glucose deprivation would create a metabolic state where superoxide and hydroperoxide concentrations were increased and peroxide scavenging via pyruvate as well as NADPH-dependent reactions was decreased. This would be expected to result in a condition of metabolic oxidative stress characterized by increased levels of pro-oxidant production and increased levels of oxidized glutathione.
In this report, we demonstrate that glucose deprivationinduced cell death in MCF-7/ADR is accompanied by activation of the MAPK pathway as well as oxidative stress as evidenced by significant increases in intracellular oxidized glutathione content and pro-oxidant production. Glucose or glutamate treatment shortened the duration of MAPK activation and rescued cells from cytotoxicity indicating that cellular metabolism was integrally related to both processes. In addition pyruvate (a known peroxide scavenger) as well as glutamate rescued MCF-7/ADR from glucose deprivation-induced cytotox-icity and suppressed pro-oxidant production. N-Acetyl-L-cysteine (NAC), a well known thiol antioxidant, also protected cells from glucose deprivation-induced cytotoxicity and prevented prolonged activation of MAPK, suggesting that oxidative stress was causally related to both processes. These findings support the hypothesis that glucose deprivation causes an increase in the steady state levels of intracellular pro-oxidants (presumably hydroperoxides) resulting in metabolic oxidative stress which triggers MAPK activation in MCF-7/ADR cells. The increase in oxidative stress also appears to account for the cytotoxicity seen during prolonged periods of glucose deprivation in MCF-7/ADR.

MATERIALS AND METHODS
Cell Culture and Survival Determination-Multidrug-resistant human breast carcinoma (MCF-7/ADR) cells (S. H. Kim, Henry Ford Hospital, Detroit, MI) were cultured in McCoy's 5a medium with 10% iron-supplemented bovine calf serum (HyClone, Logan, UT) and 26 mM sodium bicarbonate for monolayer cell culture. Two or three days prior to the experiment, cells were plated into 35-mm Petri dishes or T-25 flasks. The Petri dishes/flasks containing cells were kept in a 37°C humidified incubator with a mixture of 95% air and 5% CO 2 . For survival determination, cells were trypsinized, counted, and plated for clonogenic cell survival assay. After 1-2 weeks of incubation at 37°C, colonies were stained and counted. Data were normalized to shamtreated control plating efficiencies.
Treatment with Glucose Deprivation-Cells were rinsed three times with Hanks' balanced salt solution which took approximately 10 min. Cells were then treated with glucose-free McCoy's 5A medium with 10% dialyzed bovine calf serum (Life Technologies, Inc.).
Drug Treatment-Glucose, glutamate, pyruvate, glutamine, and Nacetyl-L-cysteine were obtained from Sigma. Drug treatment was accomplished by aspirating the medium and replacing it with medium containing the drug.
Western Blot Procedures-Proteins separated by SDS-PAGE were transferred onto a nitrocellulose membrane by electroblotting. The transfer was performed at a current of 0.12 A and 30 V overnight. The membrane was incubated in blocking solution (3% bovine serum albumin) for 1 h at 37°C, washed, and then incubated with the ERK-2 monoclonal antibody (Life Technologies, Inc., 1:500 or Calbiochem, 1:3000 dilution). After incubation with the primary antibody, the membrane was washed and incubated with alkaline phosphatase-conjugated rabbit anti-mouse IgG (diluted 2000; Zymed South San Francisco, CA) for 1-2 h. The membrane was then stained using nitro blue tetrazolium and 5Ј-bromo-4-chloro-3-indolylphosphate.
Measurement of Glutathione Content-Cells were washed with icecold phosphate-buffered saline, scraped into cold phosphate-buffered saline, and centrifuged at 4°C for 5 min at 400 ϫ g to obtain cell pellets which were then frozen at Ϫ80°C. Pellets were then thawed and homogenized in 50 mM potassium phosphate buffer, pH 7.8, containing 1.34 mM diethylenetriaminepentaacetic acid. Total glutathione content was determined by the method of Andersen (25). Reduced (GSH) and oxidized (GSSG) glutathione were distinguished by addition of 2 l of a 1:1 mixture of 2-vinylpyridine and ethanol per 50 l of sample followed by incubation for 1 h and assay as described previously by Griffith (26). All glutathione determinations were normalized to the protein content of whole homogenates using the method of Lowry et al. (27).
Intracellular Hydroperoxide Production-Pro-oxidant production was determined using oxidation-sensitive (C-400, 10 g/ml) and oxidation-insensitive (C-369, 10 g/ml) fluorescent dyes (in Me 2 SO) obtained from Molecular Probes (Eugene, OR). The oxidation-insensitive dye acts as a control for changes in uptake, ester cleavage, and efflux so that any changes in fluorescence seen with the oxidation-sensitive dye can be directly attributed to changes in pro-oxidant production. Cells were treated for the indicated period and then harvested at 37°C using trypsin, resuspended in 37°C glucose-free medium with or without drug, labeled with the fluorescent dyes for 15 min at 37°C, placed on ice, and analyzed by using a FACScan flow cytometer (Becton Dickinson Immunocytometry Systems, Inc., Mountain View, CA) (excitation 488 nm, emission 535 nm). Mean fluorescence intensity of 25,000 cells was calculated for each sample and corrected for autofluorescence obtained from samples of unlabeled cells. Genuine H 2 O 2 (5 M) was used for a positive control. The oxidation-sensitive dye has been shown previously to be sensitive to H 2 O 2 as well as organic hydroperoxides (Molecular Probes Technical Manual). Data analysis was performed with Winlist software (Verity Software House, Topsham, ME). Fig. 1A shows glucose deprivation-induced cytotoxicity in MCF-7/ADR cells. The survival of the cells decreased exponentially to 1.5 ϫ 10 Ϫ3 by 8 h of incubation in glucose-free medium. Fig. 1A also shows that the glucose deprivation-induced cytotoxicity was inhibited by adding glucose into the medium at time 0. Survival was dependent on the concentration of glucose (0.001-1 mM) added. Fig. 1B shows glucose-induced MAPK activation as determined by ERK-2 phosphorylation in MCF-7/ADR cells. Several researchers, including our own group, have shown that ERK-2 is activated via phosphorylation of tyrosine and threonine residues (28,29), and activated ERK-2 changes its mobility on SDS-PAGE (1, 30). Furthermore, we have shown in the MCF-7/ADR model system that changes in mobility on the SDS-

FIG. 1. Effect of the addition of glucose on glucose deprivationinduced cytotoxicity (A) and ERK-2 activation (B) in MCF-7/ ADR cells.
Cell monolayers were washed free of glucose (which took 10 min) and then placed into media containing the indicated concentrations of glucose. Panel A, cell survival after treatment with glucose-free medium (f) or various concentrations (0.001-1 mM) of glucose-supplemented medium for various intervals (up to 16 h). Panel B, ERK-2 activation and deactivation during various treatment times (5-180 min) with various concentrations (0.01-10 mM) of glucose-supplemented medium. Control cultures (marked C) were untreated. Lysates from equal amounts of protein (30 g) were separated by SDS-PAGE, transferred onto a nitrocellulose membrane, and processed for immunoblotting with mouse monoclonal antibodies against ERK-2. The activated ERK-2 is marked by the arrowhead and the inactive form is marked by the arrow. PAGE gel are accompanied by increased activation of MAPK as determined directly by an in-gel kinase assay (1). ERK-2 was ac tivated within 5 min in MCF-7/ADR, and its activation was maintained for 3 h during treatment with glucose-free medium (Fig. 1B). Similar results were observed in ERK-1 by treatment with glucose-free medium (data not shown). Fig. 1B also shows the effect of the addition of glucose on the time course of activation and deactivation of ERK-2. The duration of sustained ERK-2 activation was shortened in glucose-supplemented medium in a dose-response fashion similar to that seen for glucose deprivation-induced cytotoxicity. For example, the activated (phosphorylated) ERK-2 became inactivated (presumably by dephosphylation) within 30 min by adding 1 mM glucose and by 60 min with 0.1 mM glucose. Table I indicate that oxidative stress is induced by glucose deprivation as evidenced by a disruption in glutathione metabolism. This was determined by measuring the intracellular concentrations of oxidized, reduced, and total glutathione as an index of oxidative stress during glucose deprivation. Table I shows that total and reduced glutathione concentrations were increased after 4 h of glucose deprivation. This suggests that glutathione synthesis was induced by glucose deprivation during the same time frame as cytotoxicity was occurring. Table I also shows a significant increase in the steady state levels of oxidized glutathione (GSSG) induced by 4 h of glucose deprivation. These results are consistent with the hypothesis that glucose deprivation induced intracellular oxidative stress during the same time frame as cytotoxicity and alterations in signal transduction through the MAPK pathway were occurring. Fig. 2 and Table I show intracellular pro-oxidant production determined in the cells deprived of glucose for 2 h using the oxidation-sensitive probe and flow cytometry. Fig. 2, panels B-D, shows that mean fluorescence intensity is greater in the glucose-deprived cells relative to the glucose-competent cells (compare panels C and D). This increase is roughly equivalent to the increase in fluorescence intensity caused by treatment with 5 M genuine H 2 O 2 (panel B). This probe is not specific and has been shown to be sensitive to H 2 O 2 as well as organic hydroperoxides. Therefore, increases in mean fluorescence intensity are interpreted to indicate increases in intracellular pro-oxidant production, and hydroperoxide involvement is inferred by suppressing fluorescence with hydroperoxide scavengers such as pyruvate (31). Table II shows that the increase in fluorescence intensity caused by glucose deprivation is suppressed by pyruvate as is glucose deprivation-induced cytotoxicity; this strongly suggests the involvement of intracellular hydroperoxide production. Taken together the results of the glutathione analysis as well as the hydroperoxide measurements support the hypothesis that glucose deprivation induces an increase in intracellular hydroperoxide production which leads to an increase in oxidized glutathione causing oxidative stress and cytotoxicity during the same time frame that acti-vation of MAPK signal transduction is occurring.

Glucose Deprivation of MCF-7/ADR Results in Oxidative Stress as Determined by Oxidized Glutathione Content and Increased Hydroperoxide Production-Data from
Effect of the Addition of N-Acetyl-L-cysteine on Glucose Deprivation-induced Cytotoxicity and ERK-2 Activation-Previous studies have shown that glucose deprivation promotes the production of free radicals in neuronal cells (32). Free radicals as well as H 2 O 2 have been suggested to activate MAPK (3,4,6). The hypothesis that oxidative stress induced by glucose deprivation could be responsible for cytotoxicity and alterations in signal transduction in MCF-7/ADR was further tested by adding the free radical scavenging thiol antioxidant N-acetyl-Lcysteine (NAC) to MCF-7/ADR cells during glucose deprivation. Data from Fig. 3A show that NAC protected the cells from glucose deprivation-induced cytotoxicity in a dose-response fashion, and addition of 1 mM NAC completely protected against glucose deprivation-induced cytotoxicity in MCF-7/ ADR cells. Fig. 3B shows that prolonged activation of ERK-2 was also inhibited by treatment with NAC. These results indicate that treatment with a thiol antioxidant was capable of inhibiting both the cytotoxicity and the alterations in MAPK signal transduction induced by glucose deprivation in the MCF-7/ADR cells. These results provide strong support for the   hypothesis that glucose deprivation-induced oxidative stress (seen in Fig. 2 and Tables I and II) was responsible for both the cytotoxicity and alterations in signal transduction seen in the MCF-7/ADR cell line.

TABLE II Survival and hydroperoxide production in 2-h glucose-deprived MCF-7/ADR cells supplemented with glutamate and pyruvate
The Effect of Glutamine or Glutamate on Glucose Deprivation-induced Cytotoxicity and ERK-2 Activation-To examine if glucose deprivation-induced cytotoxicity could be inhibited by adding glutamine, which is known in some cell types to be interchangeable with glucose in generating ATP, MCF-7/ADR cells were incubated in glucose-free medium for 8 h in the presence of an excess amount of glutamine (0.15-150 mM). Data from Fig. 4A show that there was no significant protection from glucose deprivation-induced killing by adding glutamine. Also, addition of glutamine did not alter the duration of ERK-2 activation (Fig. 4B). In striking contrast, glucose deprivationinduced cytotoxicity in MCF-7/ADR cells was significantly reduced by adding glutamate (Fig. 5A), and 10.5 mM glutamate was capable of completely inhibiting the cytotoxicity associated with exposure to glucose-free medium. In addition, the duration of sustained ERK-2 activation was shortened by addition of glutamate (Fig. 5B). Finally, adding glutamate was also capable of inhibiting glucose deprivation-induced increases in pro-oxidant production (Table II). These results suggest that glutamate, by virtue of being able to enter the tricarboxylic acid cycle through the glutamate dehydrogenase conversion to ␣-ketoglutarate, can protect against glucose deprivation-induced cytotoxicity and alterations in signal tranduction caused by oxidative stress. Although glutamine can be converted into glutamate by the enzyme glutaminase, the contrasting results obtained in Figs. 4 and 5 suggest that MCF-7/ADR cells are deficient in this enzyme activity. DISCUSSION The major finding of this work is that glucose deprivationinduced oxidative stress appears to account for the cytotoxicity and alterations in MAPK signal transduction which were observed in MCF-7/ADR cells. The source of the increased prooxidant production caused by glucose deprivation is currently unknown, but several observations support the hypothesis that intracellular hydroperoxide production (i.e. hydrogen peroxide or organic hydroperoxides) by mitochondrial metabolism may be involved. First, glucose deprivation-induced oxidative stress clearly involves a disruption in the steady state levels of total and reduced glutathione levels in the MCF-7/ADR cells (Table  II). This suggests that cells exposed to glucose deprivation are increasing glutathione synthesis in an attempt to counteract the increased production of intracellular pro-oxidant(s). In addition, it is clear that the increase in glutathione content is not sufficient to counteract the pro-oxidant production. This is evident by significant increases in oxidized glutathione and hydroperoxide production during the first 4 h of deprivation and significant cytotoxicity during the same time. Second, the most direct pathway to obtain elevated levels of oxidized glutathione is through the action of glutathione peroxidases. These enzymes metabolize H 2 O 2 and organic hydroperoxides ROOH at the expense of reduced glutathione, with two molecules of GSH being consumed and one molecule of GSSG formed for each molecule of hydroperoxide which is decom-

FIG. 3. Effect of the addition of NAC on glucose deprivationinduced cytotoxicity (A) and ERK-2 activation (B) in MCF-7/ ADR cells.
Panel A, cell survival after treatment with glucose-free medium (f) or various concentrations (0.01-1 mM) of NAC-supplemented medium for 6 h before and during glucose deprivation. Panel B, ERK-2 activation and deactivation during treatment with various concentrations (0.01-1 mM) of NAC-supplemented medium. Western blot analysis was done as described in Fig. 1. The activated ERK-2 is marked by the arrowhead, and the inactive form is marked by the arrow. posed. Glutathione reductase then uses the reducing equivalents in NADPH to reduce GSSG back to GSH. In most mammalian cells exhibiting steady state metabolism of O 2 , the major source of superoxide and H 2 O 2 is thought to be derived from mitochondrial respiration (23). Therefore, the most logical explanation for the increases in oxidized glutathione content seen in MCF-7/ADR cells is that the intracellular mitochondrial production of hydroperoxides exceeds the metabolic capabilities of the glutathione peroxidase, glutathione reductase, NADPH system to maintain glutathione in the reduced form, and therefore, oxidized glutathione accumulates. This idea was tested by measuring pro-oxidant production with an oxidationsensitive dye, and glucose deprivation was shown to lead to an increase in pro-oxidant production (Table II). Furthermore, the increase in pro-oxidant production and cytotoxicity during glucose deprivation was suppressed by addition of an ␣-ketoacid (pyruvate) which has previously been shown to be an excellent peroxide scavenger (31).

FIG. 4. Effect of the addition of glutamine on glucose deprivation-induced cytotoxicity (A) and ERK-2 activation (B) in MCF-7/ADR cells. Panel
During any condition of oxidative stress the combination of glutathione peroxidase, glutathione, NADPH, and glutathione reductase represents a major pathway available to cells for reducing hydroperoxide concentration. NADPH production is integrally related to glucose metabolism because glucose-6phosphate dehydrogenase is the entry point for the pentose phosphate cycle to generate NADPH. Since MCF-7/ADR have been shown to have a 3-fold elevation in glycolysis when compared with parental MCF-7 (33,34) and would be expected to shift to mitochondrial energy production during glucose deprivation, they would be expected to generate substantial amounts of hydroperoxides as a by-product of respiratory activity and use up glucose and NADPH stores rapidly. When the combination of reduced glutathione and NADPH was no longer capable of effectively decomposing hydroperoxides, oxidative stress and cytotoxicity would result from the build-up of intracellular hydroperoxides and oxidized glutathione would accumulate. Our data are therefore completely consistent with the hypothesis that glucose deprivation results in oxidative stress via increased steady state concentrations of hydroperoxides from mitochondrial metabolism. In support of this hypothesis other workers (22) have previously provided convincing evidence demonstrating that glucose deprivation results in dramatic cellular sensitization to the cytotoxic effects of exogenous H 2 O 2 exposure.
The ability of the thiol antioxidant, N-acetylcysteine (NAC), to completely block both the cytotoxicity and alterations in signal transduction induced by glucose deprivation also supports the assumption that glucose deprivation-induced oxidative stress and disruption of thiol metabolism are causally related to the biological effects which were observed. Several possibilities exist for how NAC could be blocking oxidative stress during glucose deprivation. NAC can function directly as an antioxidant to scavenge radicals generated by hydroperoxides undergoing Fenton chemistry. NAC could also increase intracellular cysteine pools which could directly scavenge radicals as well as provide a vital substrate for increased synthesis of glutathione. Finally, by increasing intracellular reduced thiol pools NAC could block oxidative stress-induced changes in signal transduction by maintaining redox-sensitive proteins in their reduced form (35).
The results of this work showing that glutamate was capable of reversing the glucose deprivation-induced oxidative stress and alterations in signal transduction in the MCF-7/ADR cells can also be interpreted in light of the aforementioned hypothesis. It is well known that the energy metabolism of cancer cells is quite different from that of normal cells. Whereas normal cells depend primarily on oxidative phosphorylation to synthesize ATP, in normal steady state conditions, cancer cells demonstrate relatively high rates of glycolysis in the presence of oxygen as well as demonstrating slightly reduced levels of oxidative phosphorylation (36). Moreover, since MCF-7/ADR cells are known to have high rates of glycolysis (33,34), glucose deprivation would be expected to rapidly impact upon the process of ATP generation as well as NADPH production in these cells due to their rapid utilization of glucose. During glucose deprivation energy generation would be expected to become increasingly dependent upon the tricarboxylic acid cycle, the common central pathway for the degradation of the two-carbon acetyl residues derived not only from carbohydrates but also from fatty acids and amino acids, as well as ATP production by oxidative phosphorylation. As previously stated, if metabolism in the glucose-deprived condition was shifted to primarily respiration, the steady state levels of superoxide and hydroperoxides would be expected to increase. Glutamate could therefore rescue the cells from glucose deprivation by directly entering the tricarboxylic acid cycle leading to increased NADH production and increased ATP levels. In addition excess glutamate would be expected to increase the formation of ␣-ketoacids such as pyruvate which have been shown to be excellent scavengers of hydroperoxides (31). In fact, our data show that ␣-ketoacids are capable of suppressing pro-oxidant production as well as cytotoxicity during glucose deprivation (Table II). Glutamate is also required for the first committed step in the synthesis of glutathione catalyzed by ␥-glutamylcysteine synthetase. By using these different pathways glutamate could rapidly contribute to increased glutathione synthesis, increased peroxide scavenging, as well as ATP and NADH production.
Another major finding of this work is that oxidative stress associated with glucose deprivation appears to be capable of causing alterations in signal transduction through the ERK1/ ERK2 MAPK pathways that are also reversed by the thiol antioxidant NAC and glutamate. This supports the speculation that alterations in signal transduction caused by hydroperoxide-induced metabolic oxidative stress could represent one possible mechanism by which signal transduction and gene expression are mechanistically linked to metabolism resulting in a diversity of biological effects, including induction of protective responses as well as possible triggers for the cell death signal. In support of this assumption, oxidative stress has been shown by other investigators to trigger signal transduction pathways (3)(4)(5)(6), and recent studies also suggest that reactive oxygen species such as superoxide anion can play a role in the Ras signaling pathway (13,37). Collectively, our studies show that glucose deprivation induces activation of ERK1/ERK2 (this report), Ras, Lyn kinase, and c-Jun N-terminal kinase (JNK1) cascades (38), all of which could be involved with alterations in gene expression seen during glucose deprivation. The activation of MAPK is clearly associated with oxidative stress during glucose deprivation, and MAPK activation has previously been associated with regulation of glucose transport to supply energy for cell growth and division (14,39). Our observations therefore suggest that the prolonged activation of MAPK (and the resulting changes in gene expression) caused by metabolic oxidative stress could represent an attempt by the cells to induce protective mechanisms to try and alleviate the cytotoxic effects of glucose deprivation. Likewise, activation of other kinases such as JNK1 could contribute to the induction of cytotoxic responses during glucose deprivation.
Since tumor cells in general have been suggested to have a disruption in intracellular oxidation/reduction reactions as well as glycolytic metabolism (40 -42), the current results linking glucose metabolism to MAPK signal transduction as well as cytotoxicity via oxidation/reduction reactions may have far reaching implications to understanding the contribution of metabolism to the phenotypic changes associated with cancer as well as suggesting metabolic targets for the design of novel therapeutic interventions.