Aspirin induces cell death by directly modulating mitochondrial voltage-dependent anion channel (VDAC)

Aspirin induces apoptotic cell death in various cancer cell lines. Here we showed that silencing of VDAC1 protected HeLa cells from aspirin-induced cell death. Compared to the wild type cells, VDAC1 knocked down cells showed lesser change of mitochondrial membrane potential (Δψm), upon aspirin treatment. Aspirin augmented ATP and ionomycin-induced mitochondrial Ca2+ uptake which was abolished in VDAC1 knocked down cells. Aspirin dissociated bound hexokinase II (HK-II) from mitochondria. Further, aspirin promoted the closure of recombinant human VDAC1, reconstituted in planar lipid bilayer. Taken together, these results imply that VDAC1 serves as a novel target for aspirin. Modulation of VDAC1 is possibly associated with the cell death and anticancer effects of aspirin.


Results
VDAC1 is associated with aspirin-induced cell death. Aspirin is known to induce apoptotic cell death in different cancer cell lines. When HeLa cells were treated with 100 μ M and 500 μ M of aspirin for 16 h, cell viability decreased to 64 ± 5% and 35 ± 7% respectively (Fig. 1A). HeLa cells were also treated with another NSAID, ibuprofen. Interestingly, in similar experimental conditions, cell death caused by ibuprofen was substantially lower as compared to aspirin. Cell viability was about 80% and 70% when treated with 100 μ M and 500 μ M of ibuprofen respectively (Fig. 1A). To test whether aspirin preferentially target cancer cells, its effect on oral cancer cell line, SCC-131 and non-cancerous oral mucosal cell line, FBM were compared. As shown in Fig. 1A, 100 μ M and 500 μ M of aspirin caused significantly lesser death of FBM cells. Further, we checked the release of cytochrome c from mitochondria in HeLa cells to establish the induction of apoptosis by aspirin. Cells were treated with 100 μ M of aspirin for 6 h. Cytosolic and mitochondrial fractions were separated and probed for the presence of cytochrome c by Western blot. In agreement with previous reports 45,46 , aspirin treatment decreased the content of cytochrome c in mitochondria and subsequently it was increased in cytosol, confirming the induction of apoptosis (Fig. 1B). The involvement of VDAC1 in aspirin-induced cell death was studied by suppressing its expression with siRNA. To determine optimum dosage of siRNA, cells were transfected with varying concentrations of siRNA and the VDAC1 level was checked after 72 h, using Western blot. 100 nM of siRNA attenuated the VDAC1 expression significantly, while scrambled siRNA had no effect (Fig. 1C). Effect of aspirin was studied on siRNA transfected [si-VDAC1] and scrambled siRNA (si-Scr) transfected cells. Cells were treated with 100 and 500 μ M aspirin for 16 h. As shown in Fig. 1D, si-VDAC1-treated cells showed significantly higher viability after aspirin treatment, compared to the si-Scr-treated control cells, suggesting a possible role of VDAC1 in aspirin mediated cell death.
Aspirin alters mitochondrial Membrane potential (Δψ m ). Apoptosis is often accompanied by a decrease of Δ ψ m . We studied the effect of aspirin on Δ ψ m in si-Scr-treated and si-VDAC1-treated cells, using JC-1 dye. In healthy mitochondria, the dye aggregates and shows an emission maxima of 590 nM (red). The loss of Scrambled siRNA (100 nM; lane 2) had no effect. Lower panel: β -actin band is shown, as loading control. After developing the blot for VDAC1, it was stripped and probed with anti-β -actin antibody. D. Effects of aspirin on scrambled siRNA (si-Scr) treated control and VDAC1 knocked down (si-VDAC1) cells are compared. si-VDAC1 transfected cells showed significantly higher viability. Values are the mean ± SEM of five independent experiments.
Scientific RepoRts | 7:45184 | DOI: 10.1038/srep45184 Δ ψ m leads to the monomerisation of the dye and emission shifts to 525 nM (green). Thus the decrease in the ratio of red/green fluorescence reflects loss of Δ ψ m . JC-1 loaded cells were treated with 100 μ M aspirin and the images (red and green fluorescence) were captured at 5 min interval for 15 min. The Δ ψ m of control cells (without aspirin) remained unaltered throughout the experimental period. Aspirin decreased the Δ ψ m in both si-Scr-treated and si-VDAC1-treated cells (Fig. 2). Interestingly, the extent of reduction is much lesser in si-VDAC1-treated cells, compared to the si-Scr-treated cells.
Aspirin disrupts cellular calcium homeostasis. Apoptosis is often preceded by the disruption of cellular calcium homeostasis. Aspirin has been shown to increase cytosolic calcium ([Ca 2+ ] i ) in different cell types 47,48 . We measured ([Ca 2+ ] i ratiometrically using fura-2. As shown in Supplementary Fig. S1, 100 and 500 of μ M aspirin did not alter the [Ca 2+ ] i . We studied the effect of aspirin on ATP and ionomycin-induced Ca 2+ rise. ATP increases Ca 2+ influx by stimulating ionotropic purinergic receptors. It also releases stored Ca 2+ from ER by activating metabotropic purinergic receptors -IP3-IP3 receptors cascade. Thus, ATP-induced [Ca 2+ ] i rise is a combination of Ca 2+ , entered from extracellular solution and released-Ca 2+ from internal store. HeLa cells have been reported to express both ionotropic and metabotropic purinergic receptors 49 . 1 mM ATP increased [Ca 2+ ] i in HeLa cells, as reflected by ~ 4 fold rise of F 340 /F 380 (Fig. 3Ai,iii.). Cells treated with 100 μ M aspirin + ATP showed [Ca 2+ ] i rise to the same extent as ATP alone. Further, we checked if VDAC1 is involved in this process. As shown in Fig. 3Ai and iii, both in si-Scr-treated control cells and si-VDAC1-treated cells, [Ca 2+ ] i increased to the same extent when treated either with ATP or ATP + aspirin. Calcium ionophore, ionomycin caused a robust rise (~ 10 fold) of [Ca 2+ ] i . si-Scr-treated control and si-VDAC1-treated cells showed the rise to the same extent ( Fig. 3Aii and iii) and aspirin (100 μ M) did not alter it.
We measured mitochondrial calcium ([Ca 2+ ] m ) using mitochondrially targeted inverse pericam. The fluorescence intensity (Δ F) of inverse pericam decreases with increasing concentration of [Ca 2+ ] m . Unlike [Ca 2+ ] i , cells treated with 100 μ M aspirin showed significantly higher [Ca 2+ ] m rise in response to ATP (Fig. 3Bi,iii). The Δ F of inverse pericam decreased ~ 40% when 1 mM ATP was applied, whereas the decrease was ~70% in case of aspirin treated cells. In si-VDAC1-treated cells mitochondrial Ca 2+ uptake is impaired, as anticipated. ATP-induced [Ca 2+ ] m uptake reduced significantly in si-VDAC1-treated cells. Δ F decreased ~20% in response to 1 mM ATP indicating reduced but significant rise of mitochondrial Ca 2+ (Fig 3Bi,iii). VDAC1 is known to participate in Ca 2+ flux across the outer membrane of mitochondria, therefore knocking down of VDAC1 attenuated Ca 2+ entry. Interestingly, unlike control (si-Scr-treated), the potentiating effect of aspirin on mitochondrial Ca 2+ entry was not observed in si-VDAC1-treated cells. ATP-induced [Ca 2+ ] m rise (decrease of fluorescence) were same with or without aspirin treatment (Fig. 3Bi,iii). Same trend was observed when ionomycin was used to elevate [Ca 2+ ] m . Control cells showed ~70% decrease of Δ F upon ionomycin treatment, which significantly changed to ~ 90% in aspirin treated cells (Fig. 3Bii,iii). In siVDAC1-treated cells the Δ F decreased ~30% both with ionomycin and ionomycin + aspirin (Fig. 3Bii,iii). It implies that that aspirin potentiates Ca 2+ entry to the mitochondria by acting on VDAC1.
Further, to check the involvement of Ca 2+ in aspirin-induced cell death, cells were incubated with BAPTA-AM, a known chelator of Ca 2+ . Cell death was reduced considerably in BAPTA-treated cells ( Supplementary Fig. S2).
Aspirin dissociates mitochondrially bound HK-II but not HK-I. Since several pro-apoptotic agents are known to release mitochondrially bound HK by disrupting VDAC-HK interaction 38-40 , we anticipated similar activity of aspirin. Mitochondria and cytosolic fractions were isolated from the control and aspirin treated HeLa cells. Total proteins from both mitochondrial and cytosolic fractions were resolved on SDS-PAGE and probed in Western blot using monoclonal antibody against HK-I and HK-II. VDAC1 and β actin were probed as loading control for mitochondria and cytosol respectively. Figure 4 shows that the amount of HK-II in mitochondrial fraction reduced significantly after aspirin treatment. Consequently, the cytosolic content of HK-II increased, following aspirin treatment. However, aspirin did not alter the content of HK-I either in mitochondria or in cytosol. It clearly demonstrates that aspirin releases mitochondria-associated HK-II.
Aspirin induces closure of VDAC1. To test the direct effect of aspirin on VDAC1, recombinant human VDAC1 was overexpressed and purified. Figure 5A shows the coomassie-stained purified VDAC1 on SDS-PAGE. Purified VDAC was reconstituted in PLB and the channel properties were studied before and after aspirin treatment. Aspirin (100 μ M) induced the closing of VDAC1 when added in the cis side of the PLB. The current traces recorded at − 60 mV and + 10 mV and are shown (Fig. 5B). At 10 mV, VDAC remained fully open, but the current amplitude decreased after addition of aspirin. The single channel conductance (in 1 M KCl) at 10 mV decreased from 4.02 ± 0.24 nS, to 1.2 ± 0.18 nS upon treatment with aspirin. At − 60 mV holding potential, VDAC fluctuates between open state and different closed/sub-conductance states. However after aspirin treatment VDAC showed similar rise of calcium in response to ATP or ATP + aspirin. Cells were pretreated with Aspirin (100 μ M) for 10 min and Aspirin was also co-applied with ATP. In control group, cells were not exposed to aspirin and Ca 2+ was elevated with ATP alone. ii Experimental conditions are same as i, except ionomycin was applied instead of ATP. Ionomycin caused a bigger increase of [Ca 2+ ] i . Aspirin had no effect on ionomycininduced Ca 2+ rise, either on control or si-VDAC1 treated cells.   stabilized in the closed state (5B). In Fig. 5C, the normalized channel conductance is plotted against voltages. As shown in the figure, aspirin reduced the channel conductance at all voltages.

Discussion
Aspirin induces death in several types of cancer cells through apoptosis. Different mechanisms e.g. inhibition of proteasome function, cell cycle arrest and activation of caspases-8 have been shown as underlying mechanisms [12][13][14][15][16][17][18] . Here for the first time we showed that aspirin directly modulates VDAC1, leading to cell death. Aspirin-induced cell death is lesser in si-VDAC1-treated cells compared to si-Scr-treated control cells. VDAC1 plays a crucial role in apoptosis. In the intrinsic pathway of apoptosis, mitochondrial matrix remodeling is followed by the change in mitochondrial shape, reduction of Δ ψ m and the release of cytochrome c 50 . Loss of Δ ψ m is considered as an early event of the induction of apoptosis in many cell types [51][52][53] . We observed a time dependent loss of Δ ψ m upon aspirin treatment. Interestingly, dissipation of Δ ψ m is attenuated in si-VDAC1-treated cells.
The elevated [Ca 2+ ] i is removed from the cytosol by several means, including its uptake by mitochondria and ER. The rise of Ca 2+ in mitochondria over a considerate period leads to apoptosis. We showed that aspirin (100 μ M) augmented both ionomycin and ATP-induced [Ca 2+ ] m rise, and VDAC1 is involved in this process. VDAC1 is the major Ca 2+ entry channel across the outer membrane of mitochondria. Therefore, knocking down of VDAC1 attenuated ATP-induced Ca 2+ entry to mitochondria. Interestingly, the potentiating effect of aspirin on mitochondrial Ca 2+ influx was also abrogated in si-VDAC1-treated cells. It implies that aspirin exerts its effect possibly by modulating VDAC1. To ascertain this, we studied the effect of aspirin on the electrophysiological properties of purified VDAC1, reconstituted in PLB. Aspirin induced the closure of VDAC1. The channel conductance reduced markedly after aspirin treatment. Closure of VDAC1 limits the normal flux of metabolites and ions, resulting in the induction of cell death processes. Additionally, in the closed state as VDAC1 is cation selective, mitochondrial Ca 2+ influx increased which in turn triggers the events associated with apoptosis 54,55 . Several other pro-apoptotic agents have also been reported to induce VDAC1-closure 56,57 .
Many cancer cells adapt a survival mechanism in the hostile hypoxic micro-environment by translocating HK-II to the mitochondria. Interaction of HK-II and VDAC1 provides metabolic advantage to the cancer cells by strengthening anaerobic glycolysis. When HK-II was dissociated from mitochondria, cells became sensitive to many apoptotic agents 58 . Many anti-cancer compounds have been shown to release HK-II from mitochondria. We showed that aspirin dissociates HK-II from mitochondria in intact HeLa cells. However, it is not clear if the desorption of HK-II is solely due the direct interaction of aspirin with VDAC1. Interaction of aspirin with HK-II cannot be ruled out. Therefore, aspirin-induced cell death is a cumulative effect of VDAC1-closure and desorption of HK-II from mitochondria. Disruption of mitochondrial calcium homeostasis and dissipation of Δ ψ m, which aid to the cell death process are possibly the outcome of VDAC-closure. In summary, we have reported VDAC1 as a new target for aspirin. Aspirin-induced closure of VDAC1 correlates with the elevation of mitochondrial Ca 2+ , a strong apoptotic signal. Additionally, aspirin dissociated HK-II from mitochondria that cumulatively decreased cell viability. Our observations will be helpful in designing aspirin based anti-cancer drugs.

Materials and Methods
Materials. HeLa and SCC131 cells were obtained from National Centre for Cell Sciences, Pune, India. FBM cell line was kindly gifted by Dr. Milind Vaidya (ACTREC, India). Dulbecco's modified Eagle's medium (DMEM), fetal bovine serum (FBS), penicillin, streptomycin were purchased from Hi Media, India. E. coli M15 bacterial strain and Ni-NTA matrix were purchased from Qiagen. All the salts were purchased from Sigma-Aldrich. Fura 2-AM, and JC-1 were purchased from Molecular probes Inc., USA. Antibodies were procured from Cell Signaling Technologies (CST), USA. Inverse pericam construct was kindly gifted by Dr. Atsushi, Miyawaki Riken, Japan.
Purification and reconstitution of human VDAC1. The plasmid PDS56/RBII-6xHis encoding His tag human VDAC1 (hVDAC1) was transformed in E.Coli M15 (pREP4). The over-expressed protein was purified using Ni-NTA column as described before 43 . Purified VDAC was reconstituted in the PLB, made up of 1,2diphytanoyl-sn-glycero-3 phosphatidyl choline (DPhPC) (Avanti Polar Lipids, Alabaster, AL), following the method described earlier 59 . Briefly, DPhPC (20 mg/ml in n-decane) was painted on the 150 μ m diameter aperture of a polystyrene bilayer cuvette (Warner instrument, USA). Both cis and trans chambers were filled with symmetrical solutions of 1 M KCl, 5 mM MgCl 2 and 10 mM HEPES (pH 7.4). Cis chamber was connected to the ground electrode and trans chamber was connected to the amplifier through PC501A headstage (Warner Instrument, USA). Bilayer formation was monitored by measuring the membrane capacitance. Purified VDAC was added to the cis chamber and the solution was mixed with magnetic stirrer. Channel activity was recorded at different voltages before and after adding aspirin (100 μ M final concentration) to the cis chamber. Currents were low pass filtered at 1 kHz and digitized at 5 kHz. The pClamp software (version 9, Molecular Devices) was used for data acquisition and analysis. siRNA knockdown of VDAC1. Scrambled and human specific hVDAC1 siRNAs were obtained from Sigma Aldrich. HeLa cells were seeded on six-well culture dishes. 50-70% confluent cells were transfected with different amount of hVDAC1 siRNA, using Lipofectamine reagent (Life Technologies), according to the protocol provided by the manufacturer.

SDS-PAGE and Western blotting.
Cells were lysed in PBS, supplemented with protease inhibitor cocktail.
Approximately 50 μ g of total protein was resolved on 12% SDS-PAGE and then transferred to polyvinylidene fluoride (PVDF) membrane (Bio-Rad, USA) 60 . The blocking was done with 5% BSA for 1 h at room temperature and then the blots were incubated overnight at 4 °C with different primary antibodies. Antibodies against hVDAC1 (CST catalogue #4866 S) and HK-II, HK-I (CST catalogue # C64G5, C35C4 respectively) were diluted to 1:750; for cytochrome c (CST catalogue # 136F3), 1:1000 dilution and for β actin, 1:2000 dilution were used. After several washes with TBS-Tween-20 solution, the blots were incubated for 1 h with HRP-conjugated secondary antibody at 1:5000 dilution. The blots were treated with super signal west pico-chemiluminescent substrate (Thermo Scientific, USA) and then visualized on a Chemidoc XRS (Bio-Rad).

Measurement of mitochondrial membrane potential (Δψ m ).
Cells were incubated with 0.5 μ M of JC-1 dye for 15 min at 37 °C and then washed with PBS. Glass coverslip containing cells was placed in an imaging chamber and perfused continuously with the bathing solution (pH 7.4) containing (in mM): NaCl 126, KCl 4, NaHCO 3 26, CaCl 2 1.8, NaH 2 PO 4 1.5, MgSO 4 1.5 and Glucose 10. Using appropriate filter set up, cells were excited at 488 nm and the emission was captured at 534 nm & 596 nm. Images were taken at 5 min interval with Andor EMCCD camera, attached with an inverted microscope (Olympus IX71). The intensities of the red (R) and green (G) fluorescence were calculated from the background subtracted images. Fluorescence ratio (R/G) was plotted against time.
Measurement of cytosolic and mitochondrial Ca 2+ . Cytosolic Ca 2+ was estimated ratiometrically using fura-2AM as described before 61 . Briefly, cells were incubated with 10 μ M fura 2-AM (Invitrogen, USA) in bath solution at room temperature for 30 min. Cells were washed for 30 min in fura-free solution. Coverslip containing fura-loaded cells were placed in a small glass bottom recording chamber, mounted on the stage of the Olympus inverted microscope (IX71). Cells were illuminated alternatively with 340 nm & 380 nm, with the help of Lambda-DG4 (Sutter instruments, USA), and the emission was set to 510 nm. Images were acquired at every 5 second interval. F 340 /F 380 was calculated from the background subtracted images using Andor IQ software.
Mitochondrial Ca 2+ was measured in the cells transfected with GFP based Ca 2+ sensor, inverse pericam 62 . Fluorescence intensity (Δ F) of the inverse pericam decreases with increasing concentration of mitochondrial Ca 2+ . Images were acquired every 5 seconds and the intensity was calculated off line using Andor IQ program. Ca 2+ rise was triggered with 1 mM ATP or 10 μ M ionomycin. To see the effect of aspirin, cells were pre-treated with 100 μ M aspirin for 10 min.

Isolation of mitochondria from HeLa cells.
Mitochondria were isolated from HeLa cells as described earlier 63 . Briefly, HeLa cells, grown in flask were rinsed with PBS. Then cells were lifted by scrapping. Cells were centrifuged at 2000× g for 5 min and the pellet obtained was homogenized with 200 μ l IBC buffer (2: 225-mM mannitol, 75-mM sucrose and 30-mM Tris-HCl pH 7.4) in ice using a Dounce homogenizer. The sample was centrifuged at 4500× g for 10 min at 4 °C. The supernatant was collected and again centrifuged at 10000× g for 20 min at 4 °C. The mitochondrial pellet was stored at − 80 °C for future use (for immunoblot experiments).
MTT assay. Cell viability was estimated by MTT assay. The cells were treated with 3-(4, 5-Dimethylthiazol-2-yl)− 2, 5-diphenyltetrazolium bromide (MTT) for 3 h at room temperature in dark. The dye was solubilized with acidified isopropanol, followed by centrifugation. The absorbance of the supernatant was monitored at 570 nm.
Statistical analysis. Student's t-test was used to compare two groups. One way ANOVA was performed for comparing several groups. P values less than 0.05 were considered as significant difference.