Mitochondrial bound hexokinase activity as a preventive antioxidant defense: steady-state ADP formation as a regulatory mechanism of membrane potential and reactive oxygen species generation in mitochondria.

Brain hexokinase is associated with the outer membrane of mitochondria, and its activity has been implicated in the regulation of ATP synthesis and apoptosis. Reactive oxygen species (ROS) are by-products of the electron transport chain in mitochondria. Here we show that the ADP produced by hexokinase activity in rat brain mitochondria (mt-hexokinase) controls both membrane potential (Deltapsi(m)) and ROS generation. Exposing control mitochondria to glucose increased the rate of oxygen consumption and reduced the rate of hydrogen peroxide generation. Mitochondrial associated hexokinase activity also regulated Deltapsi(m), because glucose stabilized low Deltapsi(m) values in state 3. Interestingly, the addition of glucose 6-phosphate significantly reduced the time of state 3 persistence, leading to an increase in the Deltapsi(m) and in H(2)O(2) generation. The glucose analogue 2-deoxyglucose completely impaired H(2)O(2) formation in state 3-state 4 transition. In sharp contrast, the mt-hexokinase-depleted mitochondria were, in all the above mentioned experiments, insensitive to glucose addition, indicating that the mt-hexokinase activity is pivotal in the homeostasis of the physiological functions of mitochondria. When mt-hexokinase-depleted mitochondria were incubated with exogenous yeast hexokinase, which is not able to bind to mitochondria, the rate of H(2)O(2) generation reached levels similar to those exhibited by control mitochondria only when an excess of 10-fold more enzyme activity was supplemented. Hyperglycemia induced in embryonic rat brain cortical neurons increased ROS production due to a rise in the intracellular glucose 6-phosphate levels, which were decreased by the inclusion of 2-deoxyglucose, N-acetyl cysteine, or carbonyl cyanide p-trifluoromethoxyphenylhydrazone. Taken together, the results presented here indicate for the first time that mt-hexokinase activity performed a key role as a preventive antioxidant against oxidative stress, reducing mitochondrial ROS generation through an ADP-recycling mechanism.

Brain hexokinase is associated with the outer membrane of mitochondria, and its activity has been implicated in the regulation of ATP synthesis and apoptosis. Reactive oxygen species (ROS) are by-products of the electron transport chain in mitochondria. Here we show that the ADP produced by hexokinase activity in rat brain mitochondria (mt-hexokinase) controls both membrane potential (⌬⌿ m ) and ROS generation. Exposing control mitochondria to glucose increased the rate of oxygen consumption and reduced the rate of hydrogen peroxide generation. Mitochondrial associated hexokinase activity also regulated ⌬⌿ m , because glucose stabilized low ⌬⌿ m values in state 3. Interestingly, the addition of glucose 6-phosphate significantly reduced the time of state 3 persistence, leading to an increase in the ⌬⌿ m and in H 2 O 2 generation. The glucose analogue 2-deoxyglucose completely impaired H 2 O 2 formation in state 3-state 4 transition. In sharp contrast, the mt-hexokinase-depleted mitochondria were, in all the above mentioned experiments, insensitive to glucose addition, indicating that the mt-hexokinase activity is pivotal in the homeostasis of the physiological functions of mitochondria. When mt-hexokinase-depleted mitochondria were incubated with exogenous yeast hexokinase, which is not able to bind to mitochondria, the rate of H 2 O 2 generation reached levels similar to those exhibited by control mitochondria only when an excess of 10-fold more enzyme activity was supplemented. Hyperglycemia induced in embryonic rat brain cortical neurons increased ROS production due to a rise in the intracellular glucose 6-phosphate levels, which were decreased by the inclusion of 2-deoxyglucose, N-acetyl cysteine, or carbonyl cyanide p-trifluoromethoxyphenylhydrazone. Taken together, the results presented here indicate for the first time that mt-hexokinase activity performed a key role as a preventive antioxidant against oxidative stress, reducing mitochondrial ROS generation through an ADP-recycling mechanism.
Glucose is an essential molecule for maintaining life as it is a major metabolic fuel, which, through its degradation via glycolysis and subsequent oxidative phosphorylation, generates high energy phosphate compounds responsible for driving many cellular processes. Normal levels of glucose and growth factors may protect cells from apoptotic events (1,2). However, despite its clear beneficial roles, in mammals glycemia must be tightly regulated because excessive glucose, such as in diabetes, can be harmful to tissues (3)(4)(5). Although the mechanism of glucose toxicity is not completely understood, many studies have shown that high glucose increases the formation of advanced glycation end products (6), glucose flux through the aldose reductase pathway (7), and also the production of reactive oxygen species (ROS) 1 in different cell lines (8). It has been proposed that the production of ROS by mitochondria via the respiratory chain is a causal link between high glucose and the main pathways responsible for hyperglycemic damage (3,8).
Transport and the first step of glucose utilization within the cells is catalyzed by hexokinase (EC 2.7.1.1), which participates in blood glucose homeostasis. In mammals, there are four isoforms of hexokinase (hexokinase-I to hexokinase-IV), differing in their affinities for glucose and inhibition by glucose 6-phosphate (Glc-6-P) and P i as well as in their subcellular distribution (9,10). It has also been found that hexokinase dissociates from mitochondria in a reversible manner, depending on the levels of Glc-6-P (11,12). The preferential mitochondrial localization of hexokinase in rat brain provided a predominant access to ATP generated by oxidative phosphorylation instead of others sources of ATP (13,14). Hexokinase-I and hexokinase-II bind to mitochondria, and the extent of their associa-tion varies from tissue to tissue, with brain, kidney, cardiac, and skeletal muscle and tumor cells being the sites that display a larger percentage of mitochondrial association (15)(16)(17)(18)(19). It has been suggested that this binding takes place at the voltage-dependent anion channels (VDACs), also known as mitochondrial porins (20,21). Associated with VDAC is the ADP/ATP carrier (or adenine nucleotide translocator, ANT), which allows the exchange of ADP and ATP through the inner mitochondrial membrane (22,23). Interestingly, recent studies indicated that mt-hexokinase activity inhibits not only the Bax-induced cytochrome c release and apoptosis in HeLa cells but also early apoptotic events mediated by Akt/protein kinase B activation (1,24).
The physiological rate of mitochondrial ROS production is inversely proportional to the availability of cytosolic ADP (30). Thus, a diminution in the ADP levels induces an increase in the magnitude of the mitochondrial membrane potential (⌬⌿ m ), which, in turn, decreases the respiratory rate, leading to stimulation of ROS generation due to the highly reduced state of the components of the electron transport chain. When the mitochondrial ADP levels rise, the inverse occurs. This has been attributed to the reduction of the ⌬⌿ m through F 1 F 0 ATP synthase complex activity. In fact, mitochondrial ROS generation is strongly dependent on ⌬⌿ m levels, because high H ϩ gradients increase O 2 . and H 2 O 2 formation (40). On the other hand, a small decrease in ⌬⌿ m levels strongly diminishes H 2 O 2 production (40). Therefore, in the present work the relationship between ⌬⌿ m and ROS production with the ADP generated by the mt-hexokinase activity was evaluated in both isolated rat brain mitochondria and rat embryonic cortical neurons. The results suggested that mt-hexokinase activity was directly involved in the local ADP recycling mechanism, providing a novel physiological antioxidant role in rat neuronal cells.
Animals and Mitochondrial Isolation-Adult male Wistar rats weighing 200 -230 g fasted overnight prior to being killed by decapitation. Mitochondria were isolated by conventional differential centrifugation from forebrains as described elsewhere (41) and kept at 4°C throughout the isolation procedure. Briefly, brains from two rats were rapidly removed to an ice-cold isolation buffer containing 0.32 M sucrose, 1 mM EDTA, 1 mM EGTA, and 10 mM Tris-HCl (pH 7.4). After five washes to remove contaminating blood, the tissue was sliced into little pieces in isolation buffer. The tissue was manually homogenized during two cycles of 10 s in a Teflon glass potter. The homogenate was centrifuged at 1.330 ϫ g for 3 min in a Hitachi Himac SCR20B RPR 20-2 rotor. The supernatant was carefully removed and centrifuged at 21,200 ϫ g for 10 min. The pellet was re-suspended in isolation buffer (10 ml/g tissue originally homogenized) and divided in two equal volumes. To obtain the mt-hexokinase-depleted mitochondria, one fraction of the 21,200 ϫ g pellet was mixed with 2 mM Glc-6-P (42). The other fraction was used as a control. Both fractions were incubated in an ice bath for 30 min and then centrifuged at 21,200 ϫ g for 10 min. The supernatants were gently removed and stored for hexokinase activity. Alternatively, 2 mM Glc-6-P was included in the isolation buffer from the beginning of the isolation procedure and kept throughout all steps. The pellets obtained were re-suspended in 15% Percoll. The discontinuous density gradient was prepared manually by layering 3-ml fractions of the resuspended pellet on two preformed layers consisting of 3.5 ml of 23% Percoll above 3.5 ml of 40% Percoll. Tubes were centrifuged for 5 min at 30,700 ϫ g with slow brake deceleration. The material equilibrating near the interface between the 23 and 40% Percoll layers was gently diluted 1:4 with isolation buffer and then centrifuged at 16,700 ϫ g for 10 min. A firm pellet was obtained and gently resuspended in the isolation buffer in which sucrose was substituted by 0.32 M mannitol and supplemented with 0.2 mg/ml fatty acid-free bovine serum albumin. After centrifugation at 6,900 ϫ g for 10 min, the supernatant was rapidly decanted, and the pellet resuspended in the same buffer using a fine Teflon pestle. Protein was determined by the Folin-Lowry method using bovine serum albumin as standard (43). This procedure yielded about 5 mg of protein per rat brain. Mitochondrial associated hexokinase activity was essentially present in control mitochondrial pellets (ϳ75%) whereas, as expected (11), pre-incubation with Glc-6-P promoted a release of Ͼ80% of the mt-hexokinase from mitochondria. All of the experiments with isolated mitochondria were carried out at 37°C with continuous stirring in a respiration buffer containing 10 mM Tris-HCl, pH 7.4, 0.32 M mannitol, 8 mM inorganic phosphate, 4 mM MgCl 2 , 0.08 mM EDTA, 1 mM EGTA, 0.2 mg/ml fatty acid-free bovine serum albumin, and 50 M Ap 5 A.
Determination of Hexokinase Activity-The activity of mitochondrial bound hexokinase was determined based on a previously described method with minor modifications (44). Briefly, mitochondrial protein used in this assay varied from 0.03-0.08 mg/ml, and the activity of hexokinase was determined by NADH formation following the absorbance at 340 nm. The assay medium contained 10 mM Tris-HCl, pH 7.4, 0.1 mM Ap 5 A as an inhibitor of adenylate kinase, 5 mM D-glucose, 10 mM MgCl 2 , 1 mM ATP, 1 mM NAD ϩ , and 1 unit/ml Glc-6-P dehydrogenase (Leuconostoc mesenteroides) in a final volume of 1 ml. The reaction temperature was 37°C. When ATP generated intramitochondrially by oxidative phosphorylation was used to measure the mt-hexokinase activity, the standard respiration buffer supplemented with 2 mM succinate and 0.1 mM ADP was employed. In the experiments using exogenous yeast hexokinase, the units of enzyme added were calculated considering the endogenous rat brain mt-hexokinase activity in control mitochondria in the same reaction mixture described above as a reference. The mitochondrial protein in these assays was 0.1-0.2 mg/ml.
Oxygen Uptake Measurements-Oxygen uptake was measured in an oxymeter fitted with a water-jacketed Clark-type electrode (Yellow Springs Instruments Co., model 5300). Mitochondria (0.2 mg/ml) were incubated with 1.5 ml of the standard respiration buffer described above. The cuvette was closed immediately before starting the experiments. Each experiment was repeated at least three times with different mitochondrial preparations, and Fig. 1 shows representative experiments. Other additions are indicated in the Fig. 1 legend. Respiratory control ratio values were obtained with isolated mitochondria by using both pyruvate and malate as complex I substrates or succinate as a complex II substrate and were in good agreement with previous reported values (41).
Determination of ⌬⌿ m -Mitochondrial membrane potential was measured by using the fluorescence signal of the cationic dye safranine O, which is accumulated and quenched inside energized mitochondria (45). Mitochondria (0.2 mg protein/ml) were incubated in the standard respiration buffer supplemented with 10 M safranine. FCCP (5 M) was used as a positive control to collapse ⌬⌿ m . Fluorescence was detected with an excitation wavelength of 495 nm (slit 5 nm) and an emission wavelength of 586 nm (slit 5 nm) using a Hitachi (Tokyo, Japan) model F-3010 spectrofluorometer. In different laboratories it has been shown that the addition of ADP promoted a 5-10% decrease of the ⌬⌿ m (37). Under our conditions we found that the addition of 0.2 mM ADP promoted a 20 -25% decrease of the ⌬⌿ m , which indicated that the effect of ADP was even more potent in our preparations than in previously reports. Data are reported as arbitrary fluorescence units. Each experiment was repeated at least three times with different mitochondrial preparations, and Cortex Cell Cultures-Cortices from 14-day-old Wistar rat embryos were dissected and cultured as described previously (47) with minor modifications. Cells were plated on glass coverslips coated previously with 1.5 g/ml polyornithine in neurobasal medium supplemented with B27. After 96 h of culture, the neurobasal medium was replaced by RPMI 1640 medium, which contains 10 mM glucose supplemented with 2 M CM-H 2 DCFDA (Molecular Probes) to assess intracellular ROS formation. To investigate the effects of glucose, several aliquots of the medium were supplemented with different solutions to achieve the following final concentrations: 40 mM glucose; 40 mM glucose, plus 30 mM 2-DOG; and 40 mM glucose plus 5 mM FCCP. The cells were kept for 40 min at 37°C and 4% CO 2 , and randomly chosen fields were examined under a Nikon Eclipse TE300 epifluorescence microscope at a fixed exposure time. Additionally, control cells were preloaded with 1 mM NAC in neurobasal medium and kept for 45 min at 37°C with 4% CO 2 . The medium was then replaced to RPMI plus 2 M CM-H 2 DCFDA plus 40 mM glucose plus 1 mM NAC. Control experiments were also conducted in the presence of 10 mM glucose plus 2 M CM-H 2 DCFDA plus 100 M H 2 O 2 in the presence or absence of 1 mM NAC. All solutions employed in cell culture experiments were prepared and kept at all times under sterile conditions. Five independent fields were counted for each experimental condition (which were carried out in triplicate). Essentially identical results were obtained in at least three repeated experiments using neurons from different animals.
Determination of Intracellular Glc-6-P Content-The intracellular levels of Glc-6-P were measured by a spectrophotometric assay with minor modifications (48). To quantify intracellular Glc-6-P content, cultures of rat cortical neurons were kept as described above and scraped in 200 l of 6% (v/v) trichloroacetic acid per plate (35 mm). The extract was neutralized by adding 80 l of 1 M Tris solution. Glc-6-P levels were determined enzymatically by coupling to Glc-6-P dehydrogenase activity and monitoring this activity by NADH formation at 340 nm. Intracellular Glc-6-P levels were expressed as nmol/mg protein.

Hydrogen Peroxide Formation in Control and
Hexokinasedepleted Mitochondria-Most of the experiments of the present work were designed to investigate the possible involvement of mt-hexokinase activity on ROS production. To accomplish this, we first attempted to produce two different preparations of rat brain mitochondria to investigate the effects of glucose on mitochondrial metabolism in the presence or the absence of mthexokinase. In the first approach, we processed the mitochondria without any treatment (control); in the second approach, we incubated the mitochondria previously with Glc-6-P, a procedure that is known to detach mt-hexokinase from mitochondria (11,12,42,49). Glucose accelerated the rate of oxygen consumption in control mitochondria, whereas in mt-hexokinase-depleted mitochondria it did not affect the rate of oxygen consumption (Fig. 1, A and B). More importantly, mt-hexokinase removal did not change the respiratory control by ADP, indicating that the oxidative phosphorylation apparatus is preserved. Identical results were obtained when we utilized the complex I substrates pyruvate and malate in both mitochondrial preparations (data not shown).
Control mitochondria generated H 2 O 2 in appreciable amounts in state 4 respiration induced by succinate (Fig. 1C,  trace 1) and, as expected, the addition of ADP transiently blocked the H 2 O 2 formation during state 3 respiration (50). Activation of mt-hexokinase by glucose substantially decreased H 2 O 2 formation (Fig. 1C, traces 2 and 3). The mt-hexokinasedepleted mitochondria were able to generate H 2 O 2 in state 4 respiration in a similar fashion to control mitochondria and were also sensitive to ADP (Fig. 1D, trace 4). However, the mt-hexokinase-depleted mitochondria were unresponsive to glucose (Fig. 1D, traces 2 and 3).
Effect of Glucose Analogues on Hydrogen Peroxide Formation in Control Mitochondria-To support the notion that mthexokinase activity was indeed responsible for the effects described above, two different glucose analogues, 2-DOG and 6-DOG, were tested on the mitochondrial H 2 O 2 production. 2-DOG is a substrate of hexokinase that is phosphorylated, but instead of Glc-6-P its reaction product is 2-deoxyglucose-6-P, which does not inhibit hexokinase activity in the same range as Glc-6-P (51,52). In contrast, 6-DOG is not phosphorylated by hexokinase (53). The simultaneous addition of ADP and 2-DOG completely abolished the H 2 O 2 generation, whereas 6-DOG was ineffective (Fig. 2, A and C). Mitochondrial associated hexokinase activation by the addition of glucose and ADP in-creased the blockage time of ROS production (Fig. 2B). With 2-DOG, the blockage time of ROS generation persisted throughout the experimental measurement period (Ͼ20 min) (Fig. 2B). Interestingly, the rates of H 2 O 2 generation during the state 3-state 4 transition were expressively reduced in ADP plus glucose than in ADP alone (Figs. 2C and 1C). These results suggested that full activation of mt-hexokinase by glucose results in Glc-6-P accumulation, which, in turn, leads to a progressive mt-hexokinase inhibition and the restoration of mitochondrial ROS production (Fig. 2). The possibility that 2-DOG could be acting as a ROS scavenger was excluded because the addition of Glc-6-P restored the H 2 O 2 formation during the state 3 respiration induced by succinate and 2-DOG plus ADP in isolated mitochondria (data not shown).
Mitochondrial Localization of Hexokinase Activity Is Important For Controlling Hydrogen Peroxide Production-The possibility that an external soluble hexokinase activity that controls mitochondrial ROS production was evaluated by using mt-hexokinase-depleted mitochondria and a yeast hexokinase isoform, which is not able to bind to the VDAC and also is not inhibited by Glc-6-P (54). The addition of yeast hexokinase to mt-hexokinase-depleted mitochondria was much less efficient in reducing the rate of mitochondrial H 2 O 2 generation in the state 3-state 4 transition than was the endogenous rat brain mt-hexokinase (Fig. 3). This indicated that the localization of mt-hexokinase on the outer mitochondrial membrane favored a tight control of ROS production.
mt-Hexokinase Activity Regulates ⌬⌿ m -To assess whether  (Fig. 4A). Interestingly, in the presence of glucose, mitochondria were able to maintain low ⌬⌿ m levels for longer times than without glucose. Noteworthy was the increased time of low ⌬⌿ m values with 200 M ADP plus 1 mM glucose (Fig. 4A, closed circles). Curiously, under this condition glucose phosphorylation alone was not able to sustain a stable low ⌬⌿ m , because a return to the maximal ⌬⌿ m values was observed after 7 min. This suggests that the Glc-6-P formed by the mt-hexokinase reaction could be inhibiting its activity, leading to an increase in ⌬⌿ m . To investigate this possibility, we used two different approaches: (i) including in the reaction medium a mixture composed of glucose plus a Glc-6-P draining system (GGDS); or (ii) adding the glucose analogue 2-DOG (Fig. 4B). The addition of a GGDS efficiently sustained low ⌬⌿ m values (Fig. 4B, trace 1). 2-DOG exhibited a decrease in ⌬⌿ m similar to that seen with the GGDS addition (Fig. 4B, traces 3 and 1). In contrast, the addition of Glc-6-P, even in the presence of GGDS, shortened the time of decreased ⌬⌿ m (Fig. 4B, trace 4). These observations suggested that Glc-6-P accumulation strongly inhibited mt-hexokinase, leading to restoration of ⌬⌿ m levels. Mitochondrial associated hexokinase-depleted mitochondria were insensitive to GGDS, showing only the expected response of ⌬⌿ m to the conversion of ADP to ATP by F 1 F 0 ATP synthase (Fig. 4B, trace 5). The ⌬⌿ m response was not affected by the presence of glucose in mt-hexokinase-depleted mitochondria (Fig. 4C, trace 8).
Hyperglycemic Shock Increases Glc-6-P Levels and Induces ROS Formation in Embryonic Rat Cortical Neurons-It has been reported that hyperglycemia induces intracellular ROS formation in different cell cultures (3,8,55). Thus, the intracellular ROS levels in primary cultures of embryonic rat cortical neurons (Fig. 5A) were determined to evaluate whether this effect could be a result of the imbalance of the mt-hexokinase activity. Indeed, we found that hyperglycemia provoked a marked increase in the intracellular levels of ROS (Fig. 5, B and C). The addition of 2-DOG to the hyperglycemic medium markedly reduced the intracellular ROS production as indicated by lower CM-H 2 DCFDA fluorescence (Fig. 5D). Similarly, an uncoupler of oxidative phosphorylation (FCCP) blocked the probe fluorescence, indicating that mitochondria were the sites where ROS was produced (Fig. 5E). ROS formation induced by hyperglycemia was prevented by the previous loading of the cells with the antioxidant NAC (1 mM) (Fig. 5F). As expected, a significant increase in the probe fluorescence was observed in control cultures incubated in low glucose medium (10 mM) with 100 M H 2 O 2 (Fig. 5G), and this was suppressed by the antioxidant NAC (1 mM) (Fig. 5H). The high glucose medium significantly increased the intracellular Glc-6-P levels, which was attenuated by 2-DOG (Fig. 5I). Dissipation of ⌬⌿ m by FCCP in hyperglycemic medium also lowered the Glc-6-P levels (Fig. 5I).

DISCUSSION
Increasing evidence in the literature indicates a relationship between oxidative stress and several pathological conditions such as diabetes, Parkinson's disease, and Alzheimer's disease (56 -59). In addition, the association between oxidative stress and apoptosis is important in the pathogenesis of neurodegenerative disorders (60,61). Besides that fact, it has been very well established that mitochondrial ROS formation is associated with high ⌬⌿ m (40), which can trigger early events of apoptosis (3,62,63). It has been postulated that hyperglycemia results in mitochondrial dysfunction, neuronal injury, and apoptosis due to mitochondrial membrane hyperpolarization and ROS formation (3,5,8,64).
Mitochondrial associated hexokinase activity has been shown to protect HeLa and human embryonic kidney cells from entering apoptosis (24,65). This protection was related to the blockade of the interaction of the pro-apoptotic protein Bax with the VDAC at the contact site between the outer and inner mitochondrial membrane (65). It was proposed that the association of mt-hexokinase with VDAC is important for the integration of glycolysis with mitochondrial energy metabolism, contributing to the survival advantage of many cell types including tumor cells (66). This interaction places the mt-hexokinase in a pivotal position to integrate glycolytic metabolism with mitochondrial ROS production and apoptosis (66). In addition, elevated levels of mt-hexokinase II protect both human lung and renal epithelial cells against oxidative injury (65,67). The glucose dependence of all these effects suggests an important adaptive role for glucose metabolism and, for mt-hexokinase activity in particular, in the maintenance of cell survival. However, the relationship between mt-hexokinase activity and mitochondrial ROS generation had not yet been established.
In the present study, we demonstrated that mt-hexokinase activity was critical for sustaining a constant ADP steady-state cycling which, in turn, sets ⌬⌿ m down to lower levels and consequently decreases mitochondrial ROS formation in rat brain cells. This is the first report that demonstrates the important roles of mt-hexokinase activity as a preventive mechanism against mitochondrial free radical formation.
Previous works demonstrated the increasing mitochondrial respiration induced by glucose in rat brain and other tissues (14,17,42). The mechanistic explanation for this effect was based on the ADP recycling system by mt-hexokinase. In accordance with these data, we showed that mt-hexokinase ac- tivity controlled both respiration and ROS production (Fig. 1). Korshunov et al. have shown that small changes in ⌬⌿ m values led to dramatic alteration in the mitochondrial ROS production (40). In agreement with these observations, our data showed that the prevention of ROS generation by mt-hexokinase activity is caused by a decrease in ⌬⌿ m . This was confirmed in Fig.  4A, where mt-hexokinase activation led to a reduction of the ⌬⌿ m . It is noteworthy that glucose phosphorylation per se was not able to completely block ROS formation (Fig. 1C, traces 2 and 3 and Fig. 2A, trace 2) or stabilize low ⌬⌿ m values for longer times (Fig. 4, A and B, trace 2). These data indicate that the strong inhibition of mt-hexokinase by Glc-6-P may indirectly regulate ROS formation through fine adjustments of ⌬⌿ m (Fig. 4B, trace 4). The effect of 2-DOG on the rate of ROS production ( Fig. 2A, trace 1, and B and C) and on ⌬⌿ m (Fig. 4B,  trace 3) supports this hypothesis for the following reasons. (i) 2-DOG is efficiently phosphorylated by mt-hexokinase, but the product (2-DOG-6P) is not a strong inhibitor of the enzyme (51). (ii) The addition of a Glc-6-P removing system stimulates the mammalian mt-hexokinase activities (68).
Several lines of evidence indicate that, in addition to its activity, the localization of mt-hexokinase is also relevant to mitochondrial respiration (42,69). This possibility would be explained by the mt-hexokinase binding site in the VDAC-ANT complex. According to the data presented here, mt-hexokinase localization seems to be critical for performing its preventive antioxidant activity, as the ADP would be rapidly delivered through the VDAC-ANT complex to the F 1 F 0 ATP synthase, which phosphorylates it at the expenses of ⌬⌿ m . The data presented in Fig. 3 are in line with these observations, because the replacement of endogenous mt-hexokinase by a soluble yeast hexokinase in mt-hexokinase-depleted mitochondria was much less efficient in decreasing mitochondrial ROS generation than was the endogenous rat brain mt-hexokinase. This result also agrees with previous data (42), which demonstrated that in rat brain mitochondria the mt-hexokinase activity is more effective in stimulating respiration in the presence of glucose than its nonbindable chymotrypsin-treated mt-hexokinase form (42,69).
In hyperglycemia, mitochondrial ROS generation can be drastically increased, causing damage to mitochondria and further toxic effects on other cellular processes, because different antioxidants and uncouplers of oxidative phosphorylation protect the cells against oxidative injury (3-5, 8, 70). ROS can be generated by non-mitochondrial events such as O 2 . generation by glucose auto-oxidation (71). However, our data indicate that the hyperglycemia-induced ROS formation in embryonic rat cortical neurons was mainly associated with a progressive Glc-6-P accumulation (Fig. 5). This resulted from an imbalance between the rate of glucose phosphorylation and the capacity of other metabolic pathways to use the excessive Glc-6-P (i.e. glycolysis, pentose phosphate pathways, and glycogen synthesis). Glc-6-P accumulation would lead to a back inhibition of mt-hexokinase, which, in turn, could lead to the disruption of the ADP-recycling system (mt-hexokinase-VDAC-ANT complex) in mitochondria (Fig. 5I). In fact, it has recently been shown that a primary event in neuronal apoptosis in hyperglycemia is mitochondrial hyperpolarization (3). The data presented in Fig. 4 clearly demonstrate that the product of mthexokinase activity (Glc-6-P) was able to hyperpolarize isolated rat brain mitochondria. In accord with these observations, 2-DOG significantly reduced the intracellular ROS generation FIG. 6. Schematic representation of the proposed mechanism by which mt-hexokinase regulates oxygen consumption, ⌬⌿ m , and ROS production in mitochondria. In brain and other tissues, hexokinase is bound to mitochondria through an association with the VDAC and the ANT. The figure represents mitochondria under three different conditions by which mt-hexokinase would be regulating oxygen consumption, membrane potential, and ROS generation. These conditions are normoglycemia (A), chronic hyperglycemia (B), and mt-hexokinase-depleted mitochondrial states (C). Bold arrows and solid lines indicate a high flux of metabolites, whereas dashed lines represent low flux. Numbers represent the complexes of respiratory electron chain. Q represents ubiquinone, Cyt c the cytochrome c, and GPx the glutathione peroxidase. The term "other ligands" in panel C refers to pro-apoptotic proteins such as Bax or Bid for instance, or high levels of Glc-6-P. SOD, superoxide dismutase. induced by hyperglycemia simultaneously with a reduction in the intracellular Glc-6-P levels (Fig. 5, D and I, respectively). In contrast to Glc-6-P, 2-DOG-6 phosphate is not a strong inhibitor of mt-hexokinase activity (51) and, hence, contributed to the maintenance of low ⌬⌿ m and the impairment of ROS production through the ADP re-cycling system. Although 2-DOG may also compete with glucose for access to its transporters and its further glycolytic metabolism (72), the overall mt-hexokinase activity would not change independently of the hexose utilized (glucose or 2-DOG), because phosphorylation of 2-DOG would be able to promote mitochondrial ADP recycling.
A summary of the effects of the ADP cycling system provided by mt-hexokinase activation is schematically depicted in Fig. 6. In normoglycemia (Fig. 6A), mt-hexokinase is activated by glucose and ATP derived from oxidative phosphorylation, generating Glc-6-P and ADP, which is exchanged by ATP through the ANT. The Glc-6-P formed is channeled to other metabolic pathways. Once the ADP reaches the mitochondrial matrix, it is used as a substrate by F 1 F 0 ATP synthase to produce ATP at the expenses of ⌬⌿ m . This ATP is further exchanged with another external ADP molecule to be utilized again by mthexokinase, generating a cycling of ADP/ATP that keeps the ADP at steady-state levels and low ⌬⌿ m values. This cycling accelerates the respiration rates, which, in turn, diminish the electron leak, producing fewer O 2 . radicals that are further converted to H 2 O 2 by superoxide dismutase. Hydrogen peroxide may be further decomposed to H 2 O and O 2 by catalase. However, in chronic hyperglycemia (Fig. 6B) Glc-6-P accumulates intracellularly (Fig. 5I) because other metabolic pathways cannot use all of this metabolite, leading to mt-hexokinase inhibition and the impairment of ADP cycling. This induces an increase in ⌬⌿ m values ( Fig. 4; see also Ref. 3) and the rate of ROS generation. If mt-hexokinase is displaced from mitochondria by other VDAC ligands, such as pro-apoptotic proteins (Fig. 6C), then normal glucose levels would not be able to reduce ⌬⌿ m values ( Fig. 4; see also Ref. 24) and ROS production ( Figs. 1 and 3). Under this condition, mitochondria would rely only on other reactions to maintain ADP/ATP cycling. Thus, the endogenous antioxidant defenses present in these mitochondria could not be sufficient to scavenge all ROS generated upon the increase of ⌬⌿ m , establishing an oxidative stress situation. In this regard, it is conceivable that mt-hexokinase and uncoupling protein activities may synergistically act as preventive antioxidant systems in mitochondria because both mechanisms target the control of ⌬⌿ m , promoting mitochondrial depolarization and leading to a reduction of ROS generation.
In conclusion, the mt-hexokinase-VDAC-ANT complex, together with the F 1 F 0 ATP synthase, would represent the core of an ADP recycling system that uses glucose and ATP as fuels and whose function is to maintain low ⌬⌿ m and prevent mitochondrial ROS generation. One might speculate whether tissues that express a mt-hexokinase would be more adapted to oxidative stress situations, such as in hyperglycemia. This would be true if the mechanisms involved in the removal of the intracellular excess of Glc-6-P were activated. Therefore, this ADP recycling system may complement other protective mechanisms (such as the classical antioxidant enzymes superoxide dismutase and glutathione peroxidase) or catalase and nonenzymatic compounds (such as reduced glutathione) against oxidative stress, placing the mt-hexokinase activity as an important enzyme in the mitochondrial physiology.