Transformed cells are highly glycolytic and overexpress hexokinase II (HXK II). HXK II is capable of binding to the mitochondria through an interaction with the voltage-dependent anion channel (VDAC), an abundant outer mitochondrial membrane protein. The binding of HXK II to mitochondria has been shown to protect against loss of cell viability. Akt activation inhibits apoptosis partly by promoting the binding of HXK II to the mitochondria, but the mechanism through which Akt accomplishes this has not been characterized. The present report shows that Akt mediates the binding of HXK II to the mitochondria by negatively regulating the activity of glycogen synthase kinase 3β (GSK3β). On inhibition of Akt, GSK3β is activated and phosphorylates VDAC. HXK II is unable to bind VDAC phosphorylated by GSK3β and dissociates from the mitochondria. Inhibition of Akt potentiates chemotherapy-induced cytotoxicity, an effect that is dependent on GSK3β activation and its attendant ability to disrupt the binding of HXK II to the mitochondria. Moreover, agents that can force the detachment of HXK II from mitochondria in the absence of Akt inhibition or GSK3β activation promoted a synergistic increase in cell killing when used in conjunction with chemotherapeutic drugs. Such findings indicate that interference with the binding of HXK II to mitochondria may be a practicable modality by which to potentiate the efficacy of conventional chemotherapeutic agents.
Over 70 years ago, Warburg emphasized that poorly differentiated and rapidly growing tumors tend to have higher glycolytic rates ( 1). An increased rate of glycolysis would be beneficial to a tumor cell, giving it the ability to generate adequate supplies of ATP anaerobically. However, the production of ATP may not be the only or primary necessity for the increased glycolysis seen in tumor cells, because high rates of glycolysis occur in such cells even under conditions where oxygen is plentiful. Indeed, measurements of the rate of glycolysis indicate that a large proportion of the glycolytic flux is coupled to mitochondrial oxidation, a circumstance called aerobic glycolysis ( 2).
The high glycolytic rate is now known to be due in part to the greatly increased expression of hexokinase II (HXK II) in transformed cells ( 3– 6). In an ATP-dependent process, HXK II mediates the first and a rate-controlling step of glycolysis, phosphorylating glucose to produce glucose-6-phosphate. HXK I and HXK II are capable of binding to the mitochondria ( 7, 8). Both of these isoforms possess a hydrophobic NH2-terminal sequence of 15 amino acids that has features compatible with an amphipathic α-helical structure. This NH2-terminal domain is essential for the mitochondrial binding of the proteins. The specificity of the high-affinity binding of HXK I and HXK II to the outer mitochondrial membrane is due to their interaction with the voltage-dependent anion channel (VDAC) also known as porin. VDAC is a 30-kDa β-barrel pore protein that spans the mitochondrial outer membrane and mediates the permeation of metabolites into and out of the mitochondrial intermembrane space ( 9, 10).
In recent years, it has become apparent that mitochondria are positioned at a critical juncture in the control of apoptosis where proapoptotic signals from diverse upstream pathways converge ( 11, 12). Mitochondria can play this role due to the fact that a cadre of proapoptotic proteins is located in the mitochondrial intermembrane space. Sundry modes of protein release from the intermembrane space are possible and the control of the escape pathways is an ongoing area of controversy.
There is evidence that VDAC is capable of modulating the entry of intermembrane space proteins to the cytosol during apoptosis. Significantly, it has been shown that, under certain circumstances, VDAC can assume a configuration that promotes the release of cytochrome c. For instance, superoxide anions have been shown to cause a massive release of cytochrome c, which was suppressed by agents that promote VDAC closure, such as Koning's polyanion and antibodies that bind to VDAC ( 13). Proapoptotic proteins can also promote the release of mitochondrial intermembrane space constituents through an interaction with VDAC. Electrophysiologic studies in reconstituted systems have shown that Bax, Bak, and Bim can interact with VDAC and induce the formation of a novel large pore ( 14– 16). Conversely, the antiapoptotic protein Bcl-xL prevented Bax from inducing VDAC-dependent pore formation.
Studies suggest that the binding of HXK II to mitochondria may prevent mitochondrial dysfunction during cell stress and injury through its interaction with VDAC. Antibodies against VDAC prevent Bax-induced cytochrome c release in isolated mitochondria and intact cells ( 17). Such a result suggests that a protein that can bind to VDAC, like HXK II, may also prevent proapoptotic proteins from interacting with the mitochondria. Indeed, when expressed in NIH3T3 and Rat1a cell lines, HXK I, an isoform that binds to VDAC like HXK II, was found to increase cell proliferation and inhibit apoptosis ( 18). By contrast, a decrease in the level of mitochondrial bound HXK I worsened the apoptosis of Rat1a cells induced by growth factor withdrawal or UV irradiation. Mitochondrial bound HXK II was also shown to inhibit the binding of Bax to mitochondria and its accompanying damage to mitochondrial function ( 19). Similarly, Akt can protect against Bid-induced cell killing by promoting the binding of HXK II to the mitochondria, wherein the oligomerization Bak is prevented ( 20).
Akt promotes the binding of HXK II to the mitochondria, which may in part account for some of Akt's antiapoptotic properties ( 18, 21, 22). However, the mechanism by which Akt regulates the binding of HXK II to the mitochondria is uncharacterized. Akt is known to localize to the mitochondria along with one of its targets, glycogen synthase kinase 3β (GSK3β; refs. 23, 24). Akt phosphorylates GSK3β on Ser9 resulting in an inhibition of the kinase's activity ( 25). Activation of GSK3β is proapoptotic in several systems and can provoke mitochondrial injury ( 26– 29).
In the present report, it is shown that GSK3β activation induces mitochondrial injury in transformed cells in part by disrupting the binding of HXK II to the mitochondria. When activated, GSK3β induces the dissociation of HXK II from the mitochondria by phosphorylating VDAC. Thus, Akt promotes the binding of HXK II to mitochondria by inhibiting GSK3β. Importantly, inhibition of Akt activity potentiates chemotherapy-induced cell killing through the subsequent activation of GSK3β and the ensuing disruption in the binding of HXK II to mitochondria. Moreover, agents that induce the detachment of HXK II from the mitochondria bring about a synergistic increase in cytotoxicity when used in combination with chemotherapeutic drugs.
Materials and Methods
Treatment of cells. After an overnight incubation, HeLa cells were washed once with PBS and placed in fresh DMEM. The cells were then either left untreated or pretreated for the period of time as indicated in the figures with 100 nmol/L wortmannin or 1 μmol/L Akt inhibitor IV. Alternatively, the cells were exposed to the indicated concentrations of doxorubicin or paclitaxel in the presence or absence of either wortmannin or Akt inhibitor IV.
Measurements of viability. Cell viability was determined by the ability of the cells to exclude trypan blue. Ten microliters of a 0.5% solution of trypan blue were added to the wells. Both viable and nonviable cells were counted for each data point in a total of eight microscopic fields. Cell viability was also monitored using the plasma membrane–impermeable, dimeric cyanine dye, YOYO-1. YOYO-1 fluoresces brightly only when bound to nucleic acids. After treatments, cells are incubated with YOYO-1 before and after the addition of digitonin. Fluorescence before digitonin addition originates from dying cells that have lost plasma membrane integrity and take up YOYO-1. At the end of the experiments, digitonin was added to a final concentration of 100 μg/mL, which permeabilized 100% of the cells.
Western blot analysis. Samples were separated on 12% SDS-PAGE gels and electroblotted onto nitrocellulose membranes. Phospho-GSK3β and total GSK3β were detected with rabbit polyclonal antibodies at a dilution of 1:500 (Cell Signaling Technology, Beverly, MA). HXK II was detected by a rabbit polyclonal antibody (Santa Cruz Biotechnology, Santa Cruz, CA). Total VDAC was detected and immunoprecipitated with a mouse monoclonal antibody (Calbiochem, La Jolla, CA). Phosphothreonine-reactive VDAC was detected by first immunoprecipitating VDAC using the mouse monoclonal antibody followed by running the immunoprecipitate out on an SDS-PAGE gel and electroblotting onto a nitrocellulose membrane. The Western blot was then probed with an anti-phosphothreonine-reactive antibody (Zymed Laboratories, Carlsbad, CA). In all cases, the relevant protein was visualized by staining with the appropriate secondary horseradish peroxidase–labeled antibody (1:10,000) and was detected by enhanced chemiluminescence.
RNA interference. The Dharmacon (Lafayette, CO) SMART selection and SMART pooling technologies are used for the RNA interference studies of GSK3β. The small interfering RNAs (siRNA) 1, 2, and 3 all decreased GSK3β expression, with siRNA 3 giving the most consistent results of 95% suppression. Therefore, siRNA 3 was used in subsequent experiments to assess the effects of decreasing GSK3β levels on the sensitivity of the cells to chemotherapeutic agents. The siRNA was delivered by a lipid-based method supplied from a commercial vendor (Gene Therapy Systems, San Diego, CA) at a final siRNA concentration of 100 nmol/L. After 24 hours, the GSK3β levels were decreased by 95% and were maintained at that level for 1 week.
Isolation of mitochondria fraction. Following treatment, the cells were harvested by trypsinization and centrifuged at 600 × g for 10 minutes at 4°C. The cell pellets were washed once in PBS and then resuspended in 3 volumes of isolation buffer in 250 mmol/L sucrose. The cells were disrupted by 40 strokes of a glass homogenizer. The homogenate was centrifuged twice at 1,500 × g at 4°C to remove unbroken cells and nuclei. The mitochondria-enriched fraction was then pelleted by centrifugation at 12,000 × g for 30 minutes. The mitochondria fractions were normalized for protein content (25 μg/lane) and run on 12% SDS-PAGE gels followed by electroblotting onto nitrocellulose membranes.
Akt and glycogen synthase kinase 3β activity assay. Akt or GSK3β protein was immunoprecipitated from total cell extracts. Akt substrate or GSK3β substrate peptides were added and the kinase reaction was started by adding ATP. Samples were incubated for 15 minutes at 37°C. After centrifugation, equal aliquots of the supernatants were transferred to Pierce (Rockford, IL) phosphocellulose unit and washed thrice with binding buffer. The level of phosphorylation of the peptides was then detected with an anti-phosphoserine antibody in an ELISA colorimetric assay.
In vitro phosphorylation of voltage-dependent anion channel. VDAC was obtained by in vitro transcription and translation followed by purification on a nickel resin. Purified VDAC was coincubated with recombinant and active GSK3β (10 ng). Magnesium/ATP cocktail was then added. The reaction was incubated for 30 minutes at 30°C. VDAC was immunoprecipitated with anti-VDAC antibody. Phosphorylated VDAC was then detected with a phosphothreonine-reactive antibody. In some experiments, HXK II was added to the incubation reaction to assess its interaction with VDAC. The reaction was incubated for 30 minutes at 30°C as above followed by immunoprecipitation with anti-VDAC antibody and immunoblotting with anti-HXK II antibody.
Inhibition of Akt promotes the activation of glycogen synthase kinase 3β resulting in a disruption of the binding of hexokinase II to the mitochondria. HeLa cells were treated with either wortmannin, an inhibitor of phosphatidylinositol-3-kinase (PI3-kinase), the upstream activator of Akt, or a selective Akt inhibitor (Akt inhibitor IV). Akt-1 was immunoprecipitated from HeLa cell lysates and its activity is determined by its ability to phosphorylate a substrate peptide containing the Akt phosphorylation consensus sequence. As can be seen in Fig. 1A (left) , treatment of HeLa cells with 100 nmol/L wortmannin caused a decrease in basal Akt activity that reached a minimum by 4 hours. Treatment with the selective Akt inhibitor IV caused a more rapid decrease of Akt activity than wortmannin, reaching a minimum after only 2 hours of exposure. When phosphorylated on Ser9 by Akt, GSK3β is inactivated ( 30). Therefore, the activity of GSK3β was measured in cells when Akt was inhibited. GSK3β was immunoprecipitated and its activity was determined by its ability to phosphorylate the GSK3β peptide substrate. As shown in Fig. 1A (right), as anticipated, GSK3β activity was markedly increased in wortmannin- and Akt inhibitor IV–treated cells, reaching a peak 4-fold increase over that of control levels at 2 hours. Importantly, the activation of GSK3β was associated with its translocation to the mitochondria. As shown in Fig. 1B, treatment of HeLa cells with wortmannin or Akt inhibitor IV resulted in a decrease of inactive Ser9 phosphorylated GSK3β in the cytosol and a concomitant increase of active nonphosphorylated GSK3β at the mitochondria.
The increase of activated GSK3β at the mitochondria on inhibition of Akt was accompanied by a disruption in the binding of HXK II to the mitochondria. As shown in Fig. 1C (A), at 4 hours, inhibition of Akt activity by wortmannin or Akt inhibitor IV resulted in a loss of mitochondrial bound HXK II. The increased GSK3β activity seen during inhibition of Akt was essential for the accompanying detachment of HXK II from the mitochondria. GSK3β activity was inhibited by pretreating HeLa cells with a selective GSK3β inhibitor peptide or by the use of siRNA targeting GSK3β. As seen in Fig. 1B and C, inhibition of GSK3β activity by either method preserved the binding of HXK II to the mitochondria during inhibition of Akt activity brought about by either wortmannin or Akt inhibitor IV.
Glycogen synthase kinase 3β phosphorylates voltage-dependent anion channel and prevents it from binding to hexokinase II. VDAC was immunoprecipitated from HeLa cells and its phosphorylation state was determined by immunoblotting with a phosphothreonine-reactive antibody. As shown in Fig. 2A (1, lane 1) , control untreated HeLa cells displayed a low level of threonine phosphorylated VDAC. However, in cells where Akt activity was inhibited by treatment with wortmannin (lane 2) or the Akt inhibitor (lane 3), there was a marked elevation in the levels of threonine phosphorylated VDAC. The phosphorylation of VDAC brought about by inhibiting Akt was dependent on GSK3β activation. As shown in Fig. 2A (lanes 4 and 5), inhibition of GSK3β activity through the use of siRNA against GSK3β resulted in an abrogation of VDAC phosphorylation brought about by treatment with wortmannin or Akt inhibitor IV. Importantly, none of the treatments altered the total level of VDAC ( Fig. 2A, 2). Purified VDAC was incubated with active recombinant GSK3β. As seen in Fig. 2B (1, lane 2), recombinant GSK3β phosphorylated purified VDAC. Thr51 of VDAC lies in a GSK3β phosphorylation consensus sequence. In Fig. 2B (1, lane 4), incubation of a mutant VDAC, where Thr51 was mutated to an alanine residue, abolished the phosphorylation of VDAC by GSK3β. Importantly, the ability of GSK3β to phosphorylate VDAC was inhibited by coincubation with active Akt-1 or the GSK3β inhibitor peptide ( Fig. 2B, 1, lanes 5 and 6). In another series of experiments, purified HXK II was added to the incubation mixture containing VDAC and GSK3β followed by immunoprecipitation of VDAC. As seen in Fig. 2B (2, lane 1), HXK II became bound to VDAC when the two where incubated in kinase buffer. However, as shown in Fig. 2B (2, lane 2), HXK II was unable to bind to VDAC when GSK3β was present. Conversely, VDAC mutated at Thr51 to alanine was able to bind HXK II even in the presence of GSK3β ( Fig. 2B, 2, lane 4), thus illustrating the importance of VDAC phosphorylation in the ability of HXK II to bind VDAC. Addition of active Akt-1 or the GSK3β inhibitor peptide, agents that inhibit the kinase activity of GSK3β, restored the binding of HXK II to VDAC in the presence of GSK3β ( Fig. 2B, 2, lanes 5 and 6).
Activation of glycogen synthase kinase 3β potentiates chemotherapy induced cytotoxicity by disrupting the binding of hexokinase II to mitochondria. We determined if detaching HXK II from mitochondria had any effect on the potency of chemotherapeutic drug–induced cell killing. A NH2-terminal peptide comprising the NH2-terminal 15 amino acids of HXK II (N-HXK II-peptide) has been shown to force detachment of HXK II from the mitochondria ( 19, 21). In Fig. 3A (1 and 2, lane 2) , it is shown that treatment of mitochondria isolated from HeLa cells with N-HXK II-peptide causes the majority of the HXK II bound to the mitochondria to be released. Importantly, incubation of the mitochondria with active GSK3β also resulted in the release of HXK II from the mitochondria ( Fig. 3A, 1 and 2, lane 3). However, unlike treatment with the N-HXK II-peptide, the release of HXK II by GSK3β was accompanied by the phosphorylation of VDAC ( Fig. 3A, 3, lane 2 versus lane 3). The detachment of HXK II from mitochondria induced by GSK3β or the N-HXK II-peptide resulted in an enhanced sensitivity to Bax-induced cytochrome c release ( Fig. 3A, 4 and 5). In Fig. 3B, HeLa cells were treated with doxorubicin or paclitaxel. Importantly, the concentrations of the agents employed were chosen so that, when they were used alone, only a low level of cytotoxicity was exhibited and there was no inhibition of Akt activity or stimulation of GSK3β. As shown in Fig. 3B (1 and 2), doxorubicin at 1.0 μmol/L (squares) and paclitaxel at 10 nmol/L (squares) induced a 10% to 15% loss of cell viability in HeLa cells after 24 hours of treatment. Inhibition of Akt activity by pretreatment with either 100 nmol/L wortmannin or 1 μmol/L Akt inhibitor IV produced no detectable loss of cell viability after 24 hours of exposure ( Fig. 3B, 1 and 2, respectively, inverted triangles). However, inhibition of Akt activity markedly potentiated the cell killing induced by suboptimal doses of doxorubicin and paclitaxel. As shown in Fig. 3B (1), treatment of cells with 1.0 μmol/L doxorubicin, concomitantly with Akt inhibition, resulted in a marked loss of cell viability, with 67% to 75% cell death after 24 hours of exposure to 1.0 μmol/L doxorubicin in combination with either wortmannin or Akt inhibitor IV ( Fig. 3B, 1, circles and triangles, respectively). A similar potentiation of cell killing was seen when the cells were treated with 10 nmol/L paclitaxel in the presence of wortmannin or Akt inhibitor IV ( Fig. 3B, 2, circles and triangles, respectively).
Importantly, the potentiation of chemotherapeutic drug–induced cell killing brought about by inhibition of Akt was dependent on the activation of GSK3β. Levels of GSK3β were depleted in HeLa cells by treatment with siRNA targeting GSK3β. As shown in Fig. 3B (3 and 4), down-regulation of GSK3β eliminated the cell killing induced by 1.0 μmol/L doxorubicin or 10 nmol/L paclitaxel when combined with either wortmannin or Akt inhibitor IV (circles and squares, respectively). If the protective effect of GSK3β inhibition is mediated by preserving the binding of HXK II to mitochondria, then detachment of HXK II from the mitochondria by exposure to the N-HXK II-peptide should overcome it. Treatment of the cells with 20 μmol/L cell-permeable N-HXK II-peptide alone resulted in only a 7% to 10% loss of cell viability after 24 hours of exposure. However, as shown in Fig. 3B (3 and 4), detachment of HXK II from mitochondria by pretreatment of the cells with 20 μmol/L N-HXK II-peptide eliminated the protective effect furnished by GSK3β inhibition against the cytotoxicity induced by chemotherapeutic agents when combined with Akt inhibition. Pretreatment of HeLa cells with 20 μmol/L N-HXK II-peptide resulted in 75% to 80% cell killing after 24 hours of exposure to 1.0 μmol/L doxorubicin or 10 nmol/L paclitaxel in the presence of either wortmannin or Akt inhibitor IV ( Fig. 3B, 3 and 4, inverted triangles and triangles, respectively), even in cells were GSK3β activity was inhibited by siRNA. Importantly, the detachment of HXK II from the mitochondria greatly potentiated the cytotoxicity of low doses of doxorubicin and paclitaxel even in the absence of Akt inhibition. Treatment of the cells with 1.0 μmol/L doxorubicin or 10 nmol/L paclitaxel together with the N-HXK II-peptide resulted in a 71% to 83% loss of cell viability after 24 hours of exposure. Such a result shows that the forced displacement of HXK II from the mitochondria, even in the absence of Akt inhibition and subsequent activation of GSK3β, causes a synergistic increase in cytotoxicity when combined with suboptimal concentrations of doxorubicin or paclitaxel. Therefore, a disruption in the binding of HXK II to the mitochondria is the critical factor that accounts for the potentiation of cell death seen in cells treated with low doses of doxorubicin or paclitaxel in the context of Akt inhibition and GSK3β activation. The cell killing induced by all of the treatment conditions examined was through an apoptotic pathway. As an example, in Fig. 3C, it is shown that doxorubicin in combination with either wortmannin (left) or N-HXK II-peptide (right) resulted in a marked increase in DNA fragmentation.
Cell killing induced by doxorubicin or paclitaxel treatment alone is dependent on activation of glycogen synthase kinase 3β and a disruption in the binding of hexokinase II to the mitochondria. We next wanted to determine if doses of doxorubicin or paclitaxel that are capable of inducing extensive cell killing by themselves also required activation of GSK3β to bring about a loss of cell viability. This required a 10-fold increase of the doxorubicin or paclitaxel concentrations compared with that used previously to show a potentiation of cell killing induced by inhibition of Akt or dissociation of HXK II from the mitochondria. As shown in Fig. 4 (left) , treatment of HeLa cells with 10 μmol/L doxorubicin resulted in a progressive loss of cell viability, with 84% of the cells dead at 24 hours of treatment. Similarly, 100 nmol/L paclitaxel resulted in a 67% loss of cell viability after 24 hours of exposure ( Fig. 4, right). The cell killing induced by doxorubicin and paclitaxel was accompanied by a disruption in the binding of HXK II to the mitochondria. As shown in Fig. 4B (1), mitochondria were progressively depleted of HXK II over a 4-hour time course following exposure to 10 μmol/L doxorubicin. The depletion of mitochondrial bound HXK II induced by doxorubicin treatment was accompanied by a parallel accumulation of GSK3β at the mitochondria ( Fig. 4B, 2) and phosphorylation of VDAC ( Fig. 4B, 3). Importantly, the depletion of mitochondrial bound HXK II occurred at a time point before the extensive loss of cell viability that followed. Similar results were seen with paclitaxel. The doxorubicin- and paclitaxel-induced cell killing was preceded by a decrease of Akt activity. As shown in Fig. 4C (left), basal Akt activity dropped in the first 6 hours of exposure to 10 μmol/L doxorubicin or 100 nmol/L paclitaxel. Moreover, Akt activity in the chemotherapeutic drug–treated cells became resistant to stimulation by growth factors. HeLa cells were exposed to 10 μmol/L doxorubicin or 100 nmol/L paclitaxel for 4 hours followed by stimulation with epidermal growth factor (EGF). As shown in Fig. 4C (right), the doxorubicin- and paclitaxel-treated cells exhibited a severely blunted Akt activation in response to EGF stimulation compared with nontreated control cells.
The inactivation of Akt by doxorubicin and paclitaxel treatment was accompanied by a stimulation of GSK3β. As shown in Fig. 5A , GSK3β activity quickly increased over a 4-hour time course following exposure to 10 μmol/L doxorubicin or 100 nmol/L paclitaxel. Moreover, the activation of GSK3β was required for the expedited induction of cell killing by these agents. Suppression of GSK3β levels with siRNA inhibited the cytotoxicity induced by 10 μmol/L doxorubicin or 100 nmol/L paclitaxel. Treatment with 10 μmol/L doxorubicin or 100 nmol/L paclitaxel caused only an 11% to 16% loss of cell viability after 24 hours of exposure in cells where GSK3β levels were suppressed by siRNA ( Fig. 5B, left and right, circles). The maintenance of cell viability brought about by inhibition of GSK3β was accompanied by a preservation of the binding of HXK II to the mitochondria in doxorubicin- and paclitaxel-treated cells. In cells treated with a nontarget control siRNA, the binding of HXK II to the mitochondria was disrupted after 4 hours of treatment with 10 μmol/L doxorubicin or 100 nmol/L paclitaxel ( Fig. 5C, 1 and 2, lane 1). Identical results were seen in naive nontransfected cells. By contrast, in cells where GSK3β levels were suppressed with siRNA, HXK II remained bound to the mitochondria after 24 hours of exposure to 10 μmol/L doxorubicin or 100 nmol/L paclitaxel ( Fig. 5C, 1 and 2, lane 2). Moreover, the protection afforded by GSK3β inhibition against doxorubicin- and paclitaxel-induced cell killing was dependent on a preservation of the binding of HXK II to the mitochondria. HeLa cells in which GSK3β was suppressed by siRNA were treated with 10 μmol/L doxorubicin or 100 nmol/L paclitaxel in the presence of 20 μmol/L N-HXK II-peptide. Treatment with the N-HXK II-peptide forced the dissociation of HXK II from the mitochondria in doxorubicin- or paclitaxel-treated cells even when GSK3β levels were suppressed ( Fig. 5C, 1 and 2, lane 3). Importantly, as shown in Fig. 5B (left and right, triangles), treatment with the N-HXK II-peptide also abolished the protective effect afforded by inhibition of GSK3β against cell death induced by high doses of doxorubicin and paclitaxel.
In the present report, it is shown that inhibition of the PI3-kinase/Akt pathway results in activation of mitochondrial GSK3β. In turn, GSK3β provokes a disruption in the binding of HXK II to mitochondria by phosphorylating VDAC. Inhibition of Akt activity potentiated chemotherapy-induced cell killing, an effect mediated by the subsequent stimulation of GSK3β and the resulting detachment of HXK II from the mitochondria. The disruption of HXK II binding to mitochondria was an essential element by which inhibition of Akt potentiated chemotherapy-induced cytotoxicity. This was shown by the ability of forced displacement of HXK II from the mitochondria to negate the protective effects of GSK3β inhibition. Moreover, the forced dissociation of HXK II from the mitochondria induced by the N-HXK II-peptide in and of itself promoted a synergistic increase in the cell killing brought about by suboptimal doses of doxorubicin and paclitaxel even in the absence of Akt inhibitors. Additionally, doses of doxorubicin and paclitaxel that are capable of provoking cell killing by themselves actuate an inhibition of Akt activity, thereby stimulating mitochondrial GSK3β that then promotes detachment of HXK II from the mitochondria. Suppression of GSK3β prevented doxorubicin- and paclitaxel-induced cell killing, an effect dependent on the preservation of HXK II binding to the mitochondria.
In highly malignant tumors, the level of HXK II expression may be elevated by up to 10-fold compared with normal tissue expression ( 31). Due to poor vascularization, tumors become hypoxic resulting in a decrease in mitochondrial oxidative phosphorylation. Therefore, the increased levels of HXK II and glycolysis would benefit tumor cell survival by providing ATP through anaerobic glycolysis. However, tumor cells exhibit increased glycolytic flux even when oxygen supplies are plentiful, a condition known as aerobic glycolysis. Aerobic glycolysis depends on the binding of HXK I or HXK II to the mitochondria through an interaction with VDAC ( 32, 33).
The coupling of the adenine nucleotide translocator and VDAC with HXK II occurs at contact sites, regions where the inner and outer mitochondrial membranes appose one another ( 34, 35). Contact sites may be critical in the antiapoptotic effects of HXK II. Bax translocates to the mitochondria in a biphasic manner, first localizing to mitochondrial contact sites, whereupon an interaction between Bax and VDAC occurs ( 36). Similarly, Bid has also been found to localize to mitochondrial contact sites ( 37). HXK II prevents mitochondrial injury induced by both Bax and Bid ( 19, 20, 38). Like Bax, the BH3-only subfamily proapoptotic protein Bim interacts with VDAC ( 39). Significantly, both Bax and Bim have been implicated in the cell killing brought about by doxorubicin and paclitaxel, respectively ( 40– 43). Such findings reinforce the notion that the binding of HXK II to VDAC can interfere with the killing mechanisms engendered by doxorubicin and paclitaxel.
Akt/protein kinase B controls a vast array of cellular functions and can exert antiapoptotic effects by several mechanisms. One of these mechanisms may be through promotion of HXK II binding to mitochondria. It was shown recently that Akt activity increases the basal levels of mitochondrial bound HXK I and HXK II and that forced displacement of HXK II from the mitochondria reduces the ability of Akt to inhibit apoptosis ( 21). Indeed, Akt localizes to the mitochondria on growth factor stimulation ( 23). However, the exact mechanism whereby Akt promotes the binding of HXK II to the mitochondria was uncharacterized.
Akt constrains the activity of GSK3β by phosphorylating it on Ser9 ( 30). When Akt is inhibited, GSK3β becomes stimulated. Inhibition of GSK3β has been shown to protect neuronal cells against apoptosis brought about by several conditions ( 44, 45). A large proportion of GSK3β is found in the cytosol, but pools are also found in the nucleus and mitochondria. GSK3β has been implicated in mediating mitochondrial dysfunction. DNA damage induced an association between GSK3β and p53 at the mitochondria, where inhibition of GSK3β blocked release of cytochrome c and caspase activation ( 46). GSK3β has also been shown to phosphorylate Bax and promote its localization to the mitochondria during neuronal apoptosis ( 47). Inhibition of GSK3β activity protects cardiomyocytes from hypoxia/reoxygenation injury by suppressing onset of the mitochondrial permeability transition ( 28). Intriguingly, HXK II has been shown to inhibit opening of the permeability transition pore ( 35).
VDAC possesses a GSK3β phosphorylation consensus motif at amino acids 51 to 55. Mutation of Thr51 to alanine abolished the ability of GSK3β to phosphorylate VDAC. Significantly, this region is thought to be on the cytoplasmic side of the outer mitochondrial membrane where it could interact with GSK3β. This region is highly conserved in mitochondrial porins and is in close proximity to Glu72, a residue identified as being critical to the ability of VDAC to bind HXK I and HXK II.
The present report shows the synergistic increase in cytotoxicity induced by chemotherapeutic drugs when the binding of HXK II to mitochondria is disrupted. The Akt signaling pathway preserves the interaction of HXK II with VDAC by restraining the activity of GSK3β ( Fig. 6 ). When the PI3-kinase/Akt pathway is inhibited, this triggers the activation of GSK3β resulting in phosphorylation of VDAC with an attendant disruption in the binding of HXK II to the mitochondria. Inhibition of GSK3β prevented the cell killing induced by treatment with doxorubicin or paclitaxel. Moreover, the ability of GSK3β inhibition to prevent cytotoxicity was dependent on a preservation of HXK II binding to mitochondria, as forced displacement of HXK II abrogated the protective effect imparted from a restraining of GSK3β activity. Moreover, forced detachment of HXK II from the mitochondria also potentiated the cell killing brought about by suboptimal doses of doxorubicin and paclitaxel.
Grant support: NIH grants K01-AAA-00330-1 and 1-R01-AAA-12897-01A2.
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
- Received June 2, 2005.
- Revision received August 1, 2005.
- Accepted August 25, 2005.
- ©2005 American Association for Cancer Research.