Free Tubulin Modulates Mitochondrial Membrane Potential in Cancer Cells

Malignant cancer cells typically display high rates of glycolysis even when fully oxygenated (1, 2). In aerobic tissues, glycolysis contributes only 5% of cellular ATP production. In cancer cells, high glycolytic rates and net formation of pyruvate and lactate persist despite adequate oxidation, as confirmed in cancer patients by positron emission tomography of the glucose analogue ¹⁸fluoro-2-deoxyglucose (3). Enhanced glycolysis in tumor cells accounting for 50% to 70% of ATP formation may be advantageous for rapid growing tumors exceeding their blood and oxygen supply (4). Furthermore, glycolysis although less efficient may still produce ATP faster than oxidative phosphorylation, an advantage to rapidly dividing cells, including strongly glycolytic embryonic and early fetal tissues (5). Additionally, byproducts and intermediates of glycolytic metabolism are precursors for anabolic biosynthesis (6). The Warburg phenomenon seems to be more than an epiphenomenon, as enhancement of oxidative phosphorylation and/or inhibition of glycolysis (e.g., dichloroacetic acid, 3-bromopyruvate) cause tumor cell death both in vitro and in vivo (4).

In mitochondria, transport of respiratory substrates, ATP, ADP, and phosphate across the mitochondrial inner membrane occurs through a variety of specific transporters. In contrast, metabolite exchange across the outer membrane occurs primarily through the voltage-dependent anion channel (VDAC; refs. 7–9), which is a highly conserved approximately 30-kDa protein that forms channels permeable to molecules up to approximately 5 kDa for nonelectrolytes in the fully open state (10, 11). Each VDAC protein forms a barrel composed of a transmembrane alpha helix and 13 or more transmembrane beta strands that enclose an aqueous channel of approximately 3 nm in internal diameter in the open state and 1.8 nm in the closed state (12, 13).

VDAC shows both ion selectivity and voltage dependence. In the open state, selectivity favoring anions over cations is weak. Both positive and negative membrane potentials (±50 mV) close VDAC. It remains controversial if membrane potential regulates VDAC conductance in intact cells (14). Nonetheless, VDAC closure effectively blocks movement of most organic anions, including respiratory substrates and creatine phosphate, and prevents exchange of ADP and Pi for ATP during oxidative phosphorylation (15). Recently, VDAC closure was hypothesized to contribute to suppression of mitochondrial metabolism in the Warburg phenomenon (16).

Other factors regulate VDAC gating, including glutamate (17), NADH (18), VDAC modulator (19), G-actin (20), hexokinase (21–23), and Bcl2 family members (24). Protein kinases, including protein kinase A (PKA), glycogen synthase 3β (GSK-3β) and protein kinase C epsilon, are reported to phosphorylate VDAC (25–27). Purified VDAC1 is a substrate for PKA in vitro, and PKA phosphorylation of VDAC blocks or inhibits association of VDAC with proapoptotic proteins, such as Bax and tBid. Moreover, PKA-dependent VDAC phosphorylation decreases VDAC conductance (28). In contrast, GSK-3β–mediated VDAC phosphorylation seems to promote channel opening and potentiate chemotherapy-induced cytotoxicity (25, 29). VDAC closure has also been reported to occur after acute treatment of hepatocytes with ethanol (16, 30).

Tubulin, the heterodimeric subunit of microtubules, binds to mitochondria specifically at VDAC (31, 32). Nanomolar concentrations of dimeric tubulin close VDAC reconstituted into planar phospholipid membranes (33). Tubulin also decreases outer membrane permeability to adenine nucleotides in isolated brain mitochondria and in permeabilized synaptosomes and cardiac myocytes (33, 34). These effects of tubulin are enhanced when VDAC is phosphorylated by PKA (35).

Here, we evaluated the role of tubulin in regulating mitochondrial function in intact HepG2 human hepatoma cells. Our results indicate that microtubule depolymerization with colchicine, nocodazole, and rotenone leads to partial mitochondrial depolarization, whereas a decrease of free tubulin induced by paclitaxel promotes hyperpolarization. Additionally, the PKA agonists dibutyryl-cAMP and 8-CPT-cAMP promote depolarization, whereas PKA inhibition by H89 causes hyperpolarization. In contrast, GSK-3β inhibition with SB216763, SB415286, or 1-azakenpaullone causes depolarization. Microtubule destabilization also causes mitochondrial depolarization in human A549 lung carcinoma and UM-SCC-1 head and neck cancer cells, whereas microtubule stabilization promotes mitochondrial hyperpolarization. In rat hepatocytes, microtubule depolymerization also depolarizes mitochondria but microtubule stabilization does not hyperpolarize. These results are consistent with the conclusion that free tubulin and protein kinases dynamically regulate mitochondrial function in cancer cells but not in nontransformed primary cells, possibly by altering VDAC conductance.

HepG2, A549, and UM-SCC-1 cells (American Type Culture Collection) were grown in Eagle's minimum essential medium, F-12K medium, and Dulbecco's modified Eagle's medium (high glucose), respectively, supplemented with 10% fetal bovine serum, 100 units/mL penicillin, and 100 μg/mL streptomycin in 5% CO₂/air at 37°C. Cancer cells used for experiments were maintained in culture for less than 3 months. Confocal microscopy of cells cultured on glass bottom culture dishes (MatTek) for 48 hours was conducted in 5% CO₂/air at 37°C either in modified Hank's balanced salt solution (HBSS) containing (in mmol/L): NaCl 137, Na₂HPO₄ 0.35, KCl 5.4, KH₂PO₄ 1, MgSO₄ 0.81, Ca₂Cl 0.95, glucose 5.5, NaHCO₃ 25, and HEPES 20, pH 7.4, or complete growth medium.

Rat hepatocytes were isolated from overnight fasted male Sprague–Dawley rats (200–250 g) by collagenase digestion, as described previously (36). Cell viability exceeded 90%. Hepatocytes were cultured overnight in plates or glass bottom dishes coated with type 1 rat tail collagen in Waymouth's MB 752/1 medium containing 27 mmol/L NaHCO₃, 2 mmol/L l-glutamine, 10% fetal calf serum, 100 nmol/L insulin, and 10 nmol/L dexamethasone.

Tetramethylrhodamine methylester (TMRM) accumulates electrophoretically into mitochondria in response to the negative mitochondrial ΔΨ (37). Cells in HBSS or complete growth medium were loaded 30 minutes at 37°C with 200 nmol/l TMRM. After loading and washing, subsequent incubations were conducted with 50 nmol/L TMRM to maintain equilibrium distribution of the fluorophore (38). For simultaneous determination of plasma membrane and mitochondrial ΔΨ, HepG2 cells were coloaded with 500 nmol/L bis-1,3-dibutylbarbituric acid-trimethine oxonol [DiBAC₄(3)] and 200 nmol/L TMRM for 30 minutes and examined without washing.

Cells were imaged with a Zeiss LSM 510 inverted laser scanning confocal microscope (Thornwood) with a 63× 1.4 N.A. planapochromat oil immersion lens using 543-nm HeNe laser (0.5%–1.5% full power) and 488-nm Ar laser (0.2%–0.3% power) excitation, respectively. Emitted red fluorescence of TMRM was detected through a 560-nm long-pass filter and a 1 Airy unit diameter pinhole and emitted green fluorescence of DiBAC₄ (3) was detected through a 500- to 530-nm barrier filter and 10.6 Airy unit pinhole. Focal planes for each field were selected that showed nuclei as black structures with regular borders without superimposed mitochondria.

Free and polymerized tubulin fractions were assessed using a Microtubules/Tubulin In Vivo Assay Kit (Cytoskeleton). Cells were homogenized in cell lysis and microtubule stabilization buffer containing (in mmol/L): MgCl₂ 5, EGTA 1, GTP 0.1, ATP 1, and PIPES buffer 100; and (in %): glycerol 30, Nonidet-P40 0.1, Triton X-100 0.1, Tween-20 0.1, beta-mercaptoethanol 0.1, antifoam 0.001, and BME 0.2, pH 7.4, with a protease inhibitor cocktail. Homogenates were centrifuged (100,000 × g, 30 minutes, 37°C) to yield supernatants containing free tubulin and pellets containing microtubules, which were resuspended in cold 200 μmol/L CaCl₂.

Free tubulin and microtubule tubulin fractions were loaded on 4%–12% Bis-Tris gels. Proteins were transferred using an iBlot Dry Blotting System (Invitrogen). Blots were blocked in 5% nonfat milk and probed with 1:1,000 anti-tubulin monoclonal antibody (Cytoskeleton) for 1 hour at room temperature. Immunoblots were detected by 1:3,000 secondary antibodies conjugated to peroxidase (goat anti-mouse IgG.HRP: Sc-2005; Santa Cruz Biotechnology) for 1 hour at room temperature. Detection was conducted using a chemiluminescence kit (Supersignal Westpico Chemiluminescent Substrate). Protein was quantified by the Bradford method (Bio-Rad).

TMRM and DiBAC₄(3) fluorescence was quantified using Zeiss LSM (Carl Zeiss GmbH) and Photoshop CS4 (Adobe Systems) software. Briefly, images were taken from cells grown to 70%–80% confluency. For each preparation and treatment, a minimum of 4 random fields containing about 5 to 10 cells was selected for quantitative analysis. Images shown in the figures are representative of these fields. All cells from each field were outlined, and the mean intensity of fluorescence was determined by histogram analysis of the red or green channel, as appropriate. Background values were obtained from images collected while focusing within the coverslip and were subtracted from the mean cellular fluorescence of each field. Mean intensity of cellular fluorescence after background subtraction indicated relative cellular uptake of red-fluorescing TMRM or green-fluorescing DiBAC₄(3). Control experiments were conducted by treating the cells with the same concentration of vehicle used to deliver drugs. Otherwise, the experimental conditions were identical. A minimum of 3 independent experiments were used to make final calculations.

Differences between groups were analyzed by the Student's t test using P < 0.05 as the criterion of significance. Results were expressed as means ± SEM. Images are representative of 3 or more experiments.

HepG2 cells at approximately 70% confluency were loaded with TMRM and imaged by confocal microscopy. Red fluorescence revealed round and filamentous mitochondria relatively densely packed throughout the cytoplasm (Fig. 1). Addition of myxothiazol (10 μmol/L), a complex III respiratory inhibitor, decreased TMRM fluorescence by 8% indicating a small drop of mitochondrial ΔΨ (Fig. 1). To test the hypothesis that ATP hydrolysis by the mitochondrial F₁F₀-ATP synthase operating in reverse was maintaining mitochondrial ΔΨ in the presence of myxothiazol, oligomycin (10 μg/mL), a specific F₁-F₀ ATP synthase inhibitor, was subsequently added. As expected, oligomycin in the presence of myxothiazol collapsed ΔΨ nearly completely (Fig. 1). Notably, changes of mitochondrial ΔΨ after myxothiazol plus oligomycin did not affect cell shape (Fig. 1). When oligomycin was added first, TMRM fluorescence increased by 93% and then was lost nearly completely after subsequent myxothiazol (data not shown). These results indicate that mitochondria of HepG2 cells are metabolically active and catalyzing ΔΨ formation and ATP synthesis driven by respiration and that ATP hydrolysis after respiratory inhibition can also sustain ΔΨ.

To further investigate the effect of respiratory inhibitors on mitochondrial ΔΨ, we exposed HepG2 cells to rotenone, an inhibitor of complex I, which like myxothiazol inhibits respiration and oxidative phosphorylation. Unexpectedly, rotenone decreased TMRM fluorescence by about 60% (Fig. 2A). The decrease of ΔΨ plateaued within 30 minutes and further changes after up to an hour did not occur (data not shown). In control experiments, mitochondrial ΔΨ remained unchanged for an hour after vehicle (dimethyl sulfoxide; data not shown). Rotenone also caused cell rounding with partial and sometimes complete detachment of cells. Cell rounding after rotenone paralleled mitochondrial depolarization and did not occur after myxothiazol or vehicle (Fig. 2A compared with Fig. 1, data not shown).

Because previous studies show that rotenone depolymerizes microtubules and increases free tubulin in cells (39, 40), we examined the effects of 2 structurally unrelated microtubule destabilizers, colchicine and nocodazole, on TMRM fluorescence and cell morphology. Colchicine (10 μmol/L) decreased mitochondrial TMRM fluorescence by 60% and promoted cell rounding and detachment similar to rotenone (Fig. 2B). Nocodazole (10 μmol/L) caused virtually identical changes (Fig 2C). Thus, each of the structurally distinct microtubule destabilizers studied caused mitochondrial ΔΨ to decrease.

To assess whether changes of ΔΨ were specifically dependent on free tubulin inside cells, we treated HepG2 cells with the microtubule stabilizer paclitaxel (41). After paclitaxel (10 μmol/L), TMRM fluorescence increased 60% within 30 minutes and cell morphology was preserved (Fig. 2D). Moreover, paclitaxel prevented loss of mitochondrial TMRM fluorescence when cells were subsequently treated with rotenone, colchicine, or nocodazole (Fig. 2A–C).

Changes of tubulin polymerization in HepG2 cells in response to various treatments were determined by Western blots. In untreated cells, the free to polymerized tubulin ratio was approximately 5 (Fig. 3A). After treatment with rotenone, colchicine, and nocodazole, free to polymerize tubulin ratios increased to 60, 78, and 58, respectively. In contrast, after paclitaxel, the ratio decreased to 0.3. Moreover, pretreatment with paclitaxel prevented microtubule depolymerization by rotenone, colchicine, and nocodazole. Myxothiazol unlike rotenone caused almost no change of the free to polymerized tubulin ratio (Fig. 3A). These results, compared with the results of Figure 2A–D, show a strong inverse correlation between the free to polymerized tubulin ratios and mitochondrial ΔΨ measured by TMRM fluorescence (Fig. 3B).

TMRM uptake into mitochondria is driven by both mitochondrial and plasma membrane ΔΨ. To exclude that plasma membrane depolarization was causing mitochondrial TMRM release after microtubule-depolymerizing agents, we measured plasma membrane ΔΨ with DiBAC₄(3), which is an anionic fluorophore that intercalates into membranes enhancing fluorescence. Highly negative plasma membrane ΔΨ excludes DiBAC₄(3) from cells, and fluorescence is low. In contrast, plasma membrane depolarization promotes DiBAC₄(3) uptake increasing fluorescence as evidenced by depolarization induced by high K⁺ (Fig. 4). DiBAC₄(3) fluorescence remained unchanged after exposure of HepG2 cells to colchicine, nocodazole, and rotenone (Fig. 4). Thus, decreased plasma membrane potential cannot account for loss of TMRM fluorescence after exposure of HepG2 cells to microtubule-depolymerizing agents.

PKA phosphorylates VDAC, and conductance of VDAC phosphorylated by PKA is more sensitive to inhibition by tubulin (28, 35). Accordingly, we assessed the effect on TMRM fluorescence of the PKA agonists dibutyryl-cAMP (1 mmol/L) and 8-pCPT-cAMP (50 μmol/L). Dibutyryl-cAMP and 8-pCPT-cAMP decreased TMRM fluorescence by 45% and 41%, respectively, within 20 minutes without causing cell rounding or detachment (Fig. 5A, B, and D). Okadaic acid, a phosphatase inhibitor that indirectly promotes protein phosphorylation, also decreased TMRM fluorescence by 50% (Fig. 5C and D). In contrast, H89 (1 μmol/L) at a concentration that selectively blocks PKA (42, 43) increased mitochondrial TMRM fluorescence by 71% (Fig. 5D). H89-treated cells remained hyperpolarized even after subsequent addition of cAMP analogues, showing that H89 blocked the cAMP-dependent effect (Fig. 5A, B, and D). In addition to blocking depolarization by dibutyryl-cAMP and 8-pCPT-cAMP, H89 also reversed mitochondrial depolarization when added after the cAMP analogues (data not shown). Inhibition and activation of PKA did not alter plasma membrane ΔΨ, because DiBAC₄(3) fluorescence remained unchanged after exposure of cells to dibutyryl-cAMP (1 mmol/L), 8-pCPT-cAMP, and H89 (Supplementary Fig. S1). DiBAC₄(3) fluorescence also remained unchanged after GSK-3β inhibition (see below; Supplementary Fig. S1).

In contrast to PKA, GSK-3β is reported to promote VDAC opening (25). Accordingly, we evaluated the effect of the GSK-3β inhibitors SB216763 (400 nmol/L), 1-azakenpaullone (1 μmol/L), and SB415286 (1 μmol/L) on mitochondrial TMRM fluorescence in HepG2 cells (Fig. 6A). Each inhibitor caused a 45% to 55% decrease in TMRM fluorescence (Fig. 6B). These results are consistent with the conclusion that GSK-3β regulates mitochondrial ΔΨ oppositely to PKA.

To assess the possibility that effects of free tubulin on mitochondrial ΔΨ are a consequence of cell incubation in serum-, amino acid-, and growth factor–free HBSS, we evaluated the effect of microtubule destabilization/stabilization in HepG2 cells incubated in complete growth medium (Eagle's minimum essential medium with 10% fetal bovine serum). In complete growth medium, nocodazole decreased TMRM fluorescence (23%), although to a somewhat lesser extent than in HBSS (Fig. 7A and C). However, dibutyryl-cAMP had little effect on TMRM fluorescence (data not shown). In contrast, paclitaxel increased TMRM fluorescence by 108% over baseline even in the presence of dibutyryl-cAMP (Fig. 7B and C). Similarly, PKA inhibition by H89 increased TMRM fluorescence by 61% (Supplementary Fig. S2). In control experiments, vehicle (dimethyl sulfoxide) did not change mitochondrial ΔΨ (Supplementary Fig. S3). These findings indicate that free tubulin-dependent inhibition of mitochondrial ΔΨ formation in HepG2 cells occurs in both HBSS and complete growth medium.

After confirming that free tubulin dynamically modulates ΔΨ in HepG2 cells, we extended our study to A549 and UM-SCC-1 cancer cell lines incubated in their respective growth media. In the 2 cell lines, nocodazole decreased TMRM fluorescence by 19% and 23%, respectively (Fig. 7A and C), but dibutyryl-cAMP had virtually no effect (data not shown). In contrast, paclitaxel after db-cAMP pretreatment increased TMRM fluorescence by 36% and 22%, respectively (Fig. 7B and C). Immunoblots confirmed the expected effects of nocodazole and paclitaxel on tubulin polymerization (Supplementary Fig. S4).

To determine whether free tubulin modulates mitochondrial ΔΨ in nontransformed cells, we repeated our studies in primary rat hepatocytes incubated in culture medium. Similar to cancer cell lines in growth medium, nocodazole decreased TMRM fluorescence of primary hepatocytes by 22% and dibutyryl-cAMP had little effect. However, in contrast to cancer cells, paclitaxel and H89 did not cause mitochondrial hyperpolarization in hepatocytes (Fig. 7A–C). The lack of response to paclitaxel may reflect higher polymerized tubulin and lower free tubulin in hepatocytes than in the cancer cell lines studied (Supplementary Fig. S4; see also Fig. 3). Thus, although tubulin depolymerization impaired mitochondrial ΔΨ formation in primary hepatocytes, increased tubulin polymerization with paclitaxel and PKA inhibition did not increase ΔΨ. These findings were consistent with the conclusion that free tubulin is not dynamically controlling mitochondrial ΔΨ in primary rat hepatocytes.

Cancer cells generate the bulk of their ATP by glycolysis even in the presence of oxygen, although mitochondria isolated from tumor cells are functional (44). In confirmation, our work showed that polarized mitochondria of HepG2, A549, and UM-SCC-1 cells took up the potential-indicating fluorophore TMRM. Moreover, respiration and ATP hydrolysis were each individually sufficient to maintain mitochondrial polarization, as the combination of myxothiazol and oligomycin, but neither agent alone, was required to depolarize mitochondria and release TMRM fluorescence from HepG2 cells consistent with expectations of chemiosmotic theory (Fig. 1).

Although myxothiazol alone did not depolarize HepG2 mitochondria, rotenone, another respiratory inhibitor, did depolarize and caused cell rounding (Fig. 2A). Rotenone, unlike myxothiazol, inhibits microtubule polymerization in addition to causing respiratory inhibition (39, 40). To determine whether rotenone-induced mitochondrial depolarization was related to microtubule depolymerization, we destabilized microtubules with colchicine and nocodazole in HepG2 cells. Both colchicine and nocodazole caused cellular rounding and TMRM release to virtually the same extent as rotenone (Fig. 2B and C). In contrast, the microtubule-stabilizing agent paclitaxel increased TMRM uptake and prevented loss of TMRM fluorescence after rotenone, colchicine, and nocodazole treatment (Fig. 2A–C). Measurement of free and polymerized tubulins confirmed a very strong inverse relationship between free to polymerized tubulin ratios and mitochondrial ΔΨ (Fig. 3B).

Plasma membrane ΔΨ also drives TMRM uptake into cells, increasing TMRM uptake into mitochondria. To determine whether fluctuations of plasma membrane ΔΨ changed TMRM fluorescence after microtubule destabilization/stabilization, we examined cells loaded with DiBAC₄(3), a plasma membrane potential indicator. Rotenone, colchicine, nocodazole, or paclitaxel did not alter DiBAC₄(3) fluorescence, although it increased as expected after plasma membrane depolarization by high K⁺ (Fig. 4). Thus, microtubule destabilization/stabilization did not change plasma membrane ΔΨ but specifically altered mitochondrial ΔΨ.

Microtubules are involved in diverse cellular functions, including motility, maintenance of cell shape, cell division, and organelle distribution (45). Microtubules form by side-to-side self-association of tubulin, a heterodimer of α-tubulin and β-tubulin (46). Specific interactions of microtubules with mitochondria mediate intracellular movement of mitochondria, for example, during axoplasmic transport (31, 32). Recent studies show that tubulin reversibly inhibits VDAC reconstituted into planar phospholipid membranes and decreases respiration in isolated mitochondria. This inhibition seems mediated by insertion of a negatively charged extended C-terminal tail of tubulin into the VDAC channel (33). Our findings showing that free tubulin regulates mitochondrial ΔΨ are consistent with the conclusion that free tubulin is also inhibiting VDAC in situ and regulating the supply of respiratory substrates and/or ATP required for mitochondrial polarization. Our data also suggest that tubulin is dynamically regulating VDAC in cancer cells, because an increase of free tubulin caused depolarization whereas a decrease caused hyperpolarization.

Previous studies indicate that PKA-mediated phosphorylation leads to VDAC closure whereas GSK-3β activity promotes VDAC opening (25,28). Thus, to further assess whether changes of VDAC activity were modulating mitochondrial polarization, we determined the effect of various kinase agonists and antagonists on mitochondrial TMRM uptake. Consistent with a role of VDAC in regulating mitochondrial ΔΨ in situ, PKA activation with 2 different cell-permeant cAMP analogues decreased TMRM fluorescence in HepG2 cells incubated in HBSS whereas H89, a highly specific PKA inhibitor, blocked and reversed the effect of the cAMP analogues hyperpolarizing mitochondria (Fig. 5A–C). In contrast, 3 different inhibitors of GSK3-β caused depolarization (Fig. 6A–C). These results also support the conclusion that VDAC regulates mitochondrial ΔΨ in cancer cells.

Our initial experiments were conducted in HepG2 cells incubated in HBSS, a simple glucose-supplemented electrolyte solution. In complete growth medium, microtubule destabilization by nocodazole and microtubule stabilization with paclitaxel also depolarized and hyperpolarized mitochondria. However, in contrast to incubation in HBSS, the response to nocodazole in growth medium was weaker and the response to paclitaxel enhanced (Figs. 3B and 7A–C). Moreover, cAMP did not depolarize mitochondria of HepG2 cells in complete growth medium (data not shown), whereas PKA inhibition with H89 induced similar hyperpolarization (Fig. 5D and Supplementary Fig. S2). Similar changes occurred in A549 and UM-SCC-1 cancer cells in their respective growth media. The decreased response to cAMP in culture medium suggests that PKA may be in a higher activation state in culture medium than in HBSS. Nonetheless, other factors may regulate mitochondrial sensitivity to free tubulin in cancer cells incubated in complete medium. Overall, the results are consistent with the conclusion that mitochondrial suppression by free tubulin is greater in growth medium than in HBSS.

In contrast to all cancer cell lines studied, free tubulin did not inhibit mitochondrial ΔΨ formation in primary rat hepatocytes. Although nocodazole depolarized mitochondria, paclitaxel did not hyperpolarize (Fig. 7A–C). Similarly, PKA inhibition with H89 did not hyperpolarize mitochondria in primary rat hepatocytes in marked contrast to the hyperpolarization that occurred in HepG2 cells (Supplementary Fig. S2). Thus, suppression of mitochondrial function by free tubulin may be a unique characteristic of cancer cell metabolism. Lack of hyperpolarization after paclitaxel may be related to higher tubulin polymerization in hepatocytes than in cancer cell lines (Supplementary Fig. S4).

The present findings suggest that gating of VDAC by tubulin, PKA, and GSK-3β dynamically and globally regulates mitochondrial metabolism in cancer cells and that VDAC has a “governator” function limiting mitochondrial metabolism in aerobic glycolysis (16). Similarly, VDAC gating to a more closed state may promote selective acetaldehyde oxidation in the livers of ethanol-treated livers and slow mitochondrial ATP hydrolysis and free radical release during hypoxia/reoxygenation and oxidative stress (16, 47–49). In aerobic glycolysis (Warburg phenomenon), tumor cells prefer glycolysis to mitochondrial oxidative phosphorylation. Increased expression of glucose transporters and glycolytic enzymes help explain the enhancement of glycolysis, but the suppression of oxidative phosphorylation in cancer cells remains poorly understood. The present findings suggest that elevated levels of free tubulin in cancer cells augmented by PKA agonism are promoting VDAC closure, thereby limiting mitochondrial exchange of respiratory substrates, ATP, ADP, and phosphate. Thus, VDAC closure may account, at least in part, for suppression of mitochondrial metabolism of the Warburg phenomenon. In contrast, in nontransformed rat hepatocytes, VDAC seems not to be rate-limiting for mitochondrial metabolism. An implication of this conclusion is that inhibitors of tubulin- and PKA-mediated VDAC closure might restore normal aerobic metabolism in cancer cells and suppress cancer cell proliferation. Future studies will be needed to determine how free tubulin affects other aspects of mitochondrial function and cellular metabolism, what VDAC isoforms underlie these effects, whether tubulin regulation occurs in vivo, and the feasibility of using VDAC as a therapeutic target.

No potential conflicts of interest were disclosed.

We thank Dr. Martin Brand for pointing out that rotenone is a microtubule-depolymerizing agent.

This work was supported, in part, by grants 2-R01 DK37034, 1 R01 DK073336, and 1 R01 DK070195 from the National Institutes of Health. Imaging facilities were supported, in part, by NIH Center grant 1P30 CA138313.

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.