1906 Research Article

IntroductionProgrammed cell death – apoptosis – is a genetically predeterminedmechanism elicited by several molecular pathways (Kroemer et al.,2007). Defects in the regulation of apoptosis are often associatedwith disease, such as neuronal degenerative diseases, tumorigenesis,autoimmune disorders, and viral infections (Reed, 2002; Thompson,1995). Mitochondria are potent integrators and coordinators ofapoptosis by releasing pro-apoptotic agents and/or disruptingcellular energy metabolism. Following an apoptotic stimulus,various proteins that normally reside in the intermembrane spaceof the mitochondria, including cytochrome c and apoptosis-inducingfactor (AIF), are released to initiate the activation of pro-caspases,the protease mediators of cell death (Green and Reed, 1998;Halestrap et al., 2000). It remains unclear, however, how theseapoptotic initiators cross the outer mitochondrial membrane (OMM)and are released into the cytosol. Some models predict that releaseis facilitated by swelling of the mitochondrial matrix and subsequentrupture of the OMM, whereas other models predict the formationof protein-conducting channels that are large enough to allow thepassage of cytochrome c and other proteins into the cytosol, withoutcompromising the integrity of the OMM (Halestrap et al., 2000;Shoshan-Barmatz et al., 2008). The finding that cytochrome c canleak from intact mitochondria (Doran and Halestrap, 2000; Martinouet al., 2000; Shoshan-Barmatz et al., 2006) supports the lattermodels. Release of cytochrome c is regulated by different proteins,some of which interact with the OMM protein, VDAC (voltage-dependent anion channel), also known as mitochondrial porin.VDAC is thus believed to participate in the release of cytochromec and to interact with anti- and pro-apoptotic proteins, and isconsidered a probable candidate for such an OMM pore-formingprotein (Crompton, 1999; Halestrap et al., 2000; Kroemer et al.,

2007; Lemasters et al., 1999; Shoshan-Barmatz et al., 2006;Tsujimoto and Shimizu, 2002; Zamzami and Kroemer, 2003).

Indeed, a variety of apoptotic stimuli were shown to triggerapoptosis by modulation of VDAC and implicate VDAC1 as acomponent of the apoptosis machinery (Shoshan-Barmatz et al.,2008; Shoshan-Barmatz et al., 2006; Tsujimoto and Shimizu, 2002;Zheng et al., 2004). By contrast, VDAC proteins have been reportedto be dispensable for Ca2+- and oxidative stress-induced permeabilitytransition pore (PTP) opening (Baines et al., 2007). Nonetheless,recent studies have elucidated that VDAC1 is an indispensableprotein for PTP opening and induction of apoptosis (Tajeddine etal., 2008; Yuan et al., 2008). It was found that reducing VDAC1levels by siRNA attenuates endostatin-induced apoptosis inendothelial cells, and that endostatin-induced PTP opening isaccompanied by upregulation of VDAC1 expression (Yuan et al.,2008). Similarly, siRNA-mediated depletion of VDAC1 stronglyreduced cisplatin-induced cytochrome c release and apoptotic celldeath (Tajeddine et al., 2008). These recent reports and previousaccumulated evidence strongly suggest that VDAC serves a keyfunction in apoptosis (Crompton, 1999; Halestrap et al., 2000;Kroemer et al., 2007; Lemasters et al., 1999; Shoshan-Barmatz etal., 2008; Shoshan-Barmatz et al., 2006; Tsujimoto and Shimizu,2002; Zamzami and Kroemer, 2003). VDAC is the major OMMtransporter and plays an important role in the controlled passage ofadenine nucleotides, Ca2+ and other metabolites into and out ofmitochondria, thereby assuming an important role in energyproduction through the control of metabolite traffic (Colombini,2004; Lemasters and Holmuhamedov, 2006). Recently, the three-dimensional (3D) structures of recombinant human and murineVDAC1 were determined by NMR spectroscopy and X-raycrystallography (Bayrhuber et al., 2008; Hiller et al., 2008; Ujwal

The release of mitochondrial-intermembrane-space pro-apoptotic proteins, such as cytochrome c, is a key step ininitiating apoptosis. Our study addresses two major questionsin apoptosis: how are mitochondrial pro-apoptotic proteinsreleased and how is this process regulated? Accumulatingevidence indicates that the voltage-dependent anion channel(VDAC) plays a central role in mitochondria-mediatedapoptosis. Here, we demonstrate that the N-terminal domainof VDAC1 controls the release of cytochrome c, apoptosis andthe regulation of apoptosis by anti-apoptotic proteins such ashexokinase and Bcl2. Cells expressing N-terminal truncatedVDAC1 do not release cytochrome c and are resistant to

apoptosis, induced by various stimuli. Employing a variety ofexperimental approaches, we show that hexokinase and Bcl2confer protection against apoptosis through interaction with theVDAC1 N-terminal region. We also demonstrate that apoptosisinduction is associated with VDAC oligomerization. Theseresults show VDAC1 to be a component of the apoptosismachinery and offer new insight into the mechanism ofcytochrome c release and how anti-apoptotic proteins regulateapoptosis and promote tumor cell survival.

Key words: Apoptosis, Bcl2, hexokinase, mitochondria, VDAC1

Summary

The VDAC1 N-terminus is essential both for apoptosisand the protective effect of anti-apoptotic proteinsSalah Abu-Hamad*, Nir Arbel*, Doron Calo, Laetitia Arzoine, Adrian Israelson, Nurit Keinan,Ronit Ben-Romano, Orr Friedman and Varda Shoshan-Barmatz‡

Department of Life Sciences and the National Institute for Biotechnology in the Negev, Ben-Gurion University of the Negev, Beer-Sheva, 84105,Israel*These authors contributed equally to this work‡Author for correspondence (e-mail: vardasb@bgu.ac.il)

Accepted 26 February 2009Journal of Cell Science 122, 1906-1916 Published by The Company of Biologists 2009doi:10.1242/jcs.040188

Jour

nal o

f Cel

l Sci

ence

1907(Δ26)VDAC1-mediated apoptotic resistance

et al., 2008). VDAC1was shown to adopt a β-barrel architecturecomprising 19 β-strands and an α-helix located within the pore.The α-helix of the N-terminal segment is oriented against the interiorwall, causing a partial narrowing at the center of the pore. As such,this helix is ideally positioned to regulate the conductance of ionsand metabolites passing through the VDAC pore. These structuraland other studies (De Pinto et al., 2007) further indicate that thehelical region of the VDAC N-terminal domain forms a distortedα-helix that does not span the whole sequence but only part of it.Although these recently determined VDAC1 structures suggest thatthe hydrophilic N-terminus of the protein is nestled within the pore(Bayrhuber et al., 2008; Hiller et al., 2008; Ujwal et al., 2008),other approaches point to the N-terminal α-helix as being exposedto the cytoplasm (De Pinto et al., 2003), to cross the membrane(Colombini, 2004) or to lie on the membrane surface (Reymann etal., 1995). Further evidence suggesting that the N-terminal regionof VDAC1 in fact constitutes a mobile component of the proteincomes from the observations that the N-terminal α-helix exhibitsmotion during voltage gating (Mannella, 1997), that anti-VDACantibodies raised against the N-terminal region of the protein interactwith membranal VDAC (Abu-Hamad et al., 2006; Junankar et al.,1995; Shoshan-Barmatz et al., 2004) and that mobility of the N-terminal α-helix may modulate the accessibility of apoptosis-regulating proteins of the Bcl2 family (i.e. Bax and Bcl-xL) to theirbinding sites on VDAC1 (Shi et al., 2003).

In the present study, the role of the VDAC1 N-terminal domainin regulating VDAC1 activity in cell life and apoptotic cell deathwas investigated. Using Δ(26)mVDAC1, an N-terminal truncatedform of murine VDAC1 expressed in human (h)VDAC1-shRNAcells expressing low levels of endogenous hVDAC1, the N-terminalregion of VDAC1 was shown to be required for the release ofcytochrome c and apoptotic cell death. Furthermore, the VDAC1N-terminal α-helix was shown to mediate the interaction of VDAC1with the anti-apoptotic, pro-survival factors, hexokinase (HK)-I,HK-II and Bcl2. The findings of this study thus indicate that VDAC1is a critical component in mitochondria-mediated apoptosis and itsregulation and point to the N-terminal domain of VDAC1 as beinga critical component in the regulation of cytochrome c release andapoptosis.

ResultsThe N-terminal region of VDAC1 is not required for cell growthand localization to mitochondriaThe N-terminal domain of VDAC1 has been the object ofconsiderable interest and was proposed to possess functionallyrelevant properties (Colombini et al., 1987; De Pinto et al., 2007;Peng et al., 1992; Stanley et al., 1995; Ujwal et al., 2008; Yehezkelet al., 2007). VDAC provides the major pathway for nucleotidesand other metabolites moving across the OMM. To verify that theN-terminal region of VDAC1 is necessary for such transportactivity, N-terminally truncated murine VDAC1 [Δ(26)mVDAC1]was expressed in human T-REx293 cells in which endogenousVDAC1 expression had been suppressed by ~85% using a singlehVDAC1-shRNA (Abu-Hamad et al., 2006). Such cells proliferatedslowly, compared with normal cells (Fig. 1A), owing to low levelsof ATP and reduced ATP synthesis capacity, both of which couldbe restored upon expression of mVDAC1 under the control oftetracycline (Abu-Hamad et al., 2006). Δ(26)mVDAC1 expressionin hVDAC1-silenced cells (shRNA-hVDAC1) also restored cellgrowth (Fig. 1A), suggesting that the 26 N-terminal residues arenot essential for VDAC1 function in cell growth and that

Δ(26)mVDAC1 can substitute for native VDAC1 in supporting themetabolite exchange between the mitochondria and the cytosol incells expressing low levels of VDAC1 (Abu-Hamad et al., 2006).Expressing N-truncated VDAC1 in porin-less yeast restored theirgrowth to levels of porin-less yeast expressing native VDAC (datanot shown).

The oxygen consumption rates of control T-REx cells, shRNA-hVDAC1 cells with a low VDAC1 expression level, and cells stablyexpressing native or N-terminal-truncated mVDAC1 were measuredas an index of in situ mitochondrial metabolism (Fig. 1B). Theoxygen consumption rate of shRNA-VDAC1 cells was about 67%that of control cells, suggesting that VDAC1 is required for normalmitochondrial respiration. When such cells were transfected toexpress either native or Δ(26)mVDAC1, the respiration rate wasincreased to almost the level obtained by control cells. Rotenoneinhibited oxygen consumption in all cell types (Fig. 1B). Theexpression levels of VDAC, as analyzed by immunoblotting usingpolyclonal anti-VDAC antibodies, clearly show that the level ofVDAC expression in VDAC1-shRNA cells was dramaticallyreduced by over 90%. That value could be restored to control levels

Fig. 1. N-terminally truncated VDAC1 restores growth in T-REx293 cellssilenced for hVDAC1 expression by shRNA-hVDAC1. (A) T-REx293 cells(�) silenced for hVDAC1 expression by shRNA-hVDAC (�) showedinhibited cell growth, which was restored by expression of either native (�) orΔ(26)mVDAC1 (�) under the control of 1 μg/ml tetracycline. Cells werestained with trypan blue and counted under a microscope. (B) Oxygenconsumption was determined polarographically using a Clark oxygenelectrode, as described in Materials and Methods. Oxygen utilization of T-REx293, hVDAC1-shRNA T-REx293 cells expressing stable native orΔ(26)mVDAC1 (2�106) cells was measured in the absence (black bars) andpresence (grey bars) of rotenone (5 μM) for up to 10 minutes each. Rotenonewas added directly into the respiration chamber and measurement of oxygenconsumption was continued. The data represent the means from two differentexperiments. (C) A representative immunoblot analysis of VDAC expressionlevel in the various cell types used in experiment B, using anti-VDACantibodies. (D) hVDAC1-shRNA-T-REx293 expressing GFP, mVDAC1-GFPor Δ(26)mVDAC1-GFP were visualized using confocal microscopy. Scale bar:10 μm. (E) Immunoblot analysis of GFP, mVDAC1-GFP or Δ(26)mVDAC1-GFP expression in T-REx293 cells using anti-GFP antibodies.

Jour

nal o

f Cel

l Sci

ence

1908

by transfecting the cells with plasmids encoding mVDAC1 orΔ(26)mVDAC1 (Fig. 1C). These results suggest thatΔ(26)mVDAC1 is as active as native mVDAC1.

We then demonstrated, by confocal microscopy, that mVDAC1-GFP and Δ(26)mVDAC1-GFP expressed in T-REx293 cells showedsimilarly distributed punctuated fluorescence, which was confinedto the mitochondria (Fig. 1D). The expression of Δ(26)mVDAC1-GFP in these cells was also shown by immunoblot analysis (Fig.1E). The results indicate that both native and truncated mVDAC1were localized to the mitochondria, suggesting that the 26 N-terminal residues do not act as a targeting sequence.

Channel activity of Δ(26)mVDAC1 and its modulation byhexokinase We further demonstrated that Δ(26)mVDAC1 functions as achannel-forming protein. Δ(26)mVDAC1 or mVDAC1 wereexpressed in the mitochondria of porin-less yeast, purified andreconstituted into a planar lipid bilayer (PLB), where VDAC channelactivity was analyzed. Single-channel recording gave a maximalconductance of 4 nS (at 1 M NaCl) for both the native and the N-terminally truncated proteins. However, whereas bilayerreconstituted-mVDAC1 showed a typical voltage-dependenceconductance, with the highest values being obtained attransmembrane potentials between –20 and +20 mV and decreasingat both high negative and positive potentials (Fig. 2A),Δ(26)mVDAC1 showed no voltage dependence and exhibited highconductance at all the tested voltages (Fig. 2A) in agreement withprevious results (De Pinto et al., 2008; Koppel et al., 1998; Poppet al., 1996). Since there is no significant membrane potential acrossthe OMM, the loss of voltage sensitivity of Δ(26)mVDAC1 maynot have physiological significance, as reflected in the finding thatits ability to support cell growth was similar to that of the intactprotein (Fig. 1A).

In our previous studies (Abu-Hamad et al., 2008; Azoulay-Zoharet al., 2004), we demonstrated that hexokinase (HK) interacts withVDAC to decrease it channel activity. The interaction of the VDAC1N-terminal region with purified rat brain HK-I was studied usingbilayer-reconstituted mVDAC1 or Δ(26)mVDAC1. Addition ofHK-I to bilayer-reconstituted native mVDAC1 induced the channel,fluctuating between the fully open and a sub-conducting state, toadopt a stable, long-lived low-conducting state (Fig. 2B). Although

Journal of Cell Science 122 (11)

Δ(26)mVDAC1 assumed a stable highly conductive state, as wouldbe expected for a voltage-insensitive channel (Fig. 2A), HK-I hadno effect on the channel activity of Δ(26)VDAC1. These resultsindicate that the VDAC1 N-terminal region is required for HK-Imodification of channel conductance.

N-terminally truncated mVDAC1-expressing cells do notrelease cytochrome c and do not undergo apoptosisIt has been previously demonstrated that overexpression of human,murine, yeast or rice VDAC1, VDAC1-GFP or mutated VDAC1induces apoptotic cell death (Abu-Hamad et al., 2006; Bae et al.,2003; Zaid et al., 2005). Accordingly, upon overexpression ofmVDAC1 in T-REx293 cells silenced for hVDAC expression byhVDAC1-shRNA (using 2.5 μg/ml tetracycline), growth continuedfor 3 days before cells starting to die, with all cells having died bythe fifth day (Fig. 3A). By contrast, cells expressing Δ(26)mVDAC1continued to grow for >8 days, with only 5-8% of the cellssuccumbing to apoptosis, as reflected by enhanced nuclearfragmentation, visualized by Acridine Orange and ethidium bromidestaining (Fig. 3B). The overexpression of native and N-terminaltruncated mVDAC1 in cells stably expressing these proteins isshown in Fig. 3C.

In the light of the findings that cells overexpressing N-terminal-truncated VDAC1 had lost their ability to undergo mitochondria-mediated apoptosis (Fig. 3), we tested for the involvement of theN-terminal region of VDAC1 in apoptosis, as induced by variousstimuli. hVDAC1-shRNA-T-REx293 cells expressing mVDAC1 orΔ(26)mVDAC1 were challenged with various apoptosis-inducingagents, all of which affect the mitochondria but produce their effectby a different mechanisms (Fig. 4). Staurosporine (STS), curcumin,As2O3 and cisplatin, all induced death in cells expressing nativemVDAC1 but not in those expressing Δ(26)mVDAC1. By contrast,tumor necrosis factor alpha (TNFα), an agent that activates apoptosisvia a mitochondria-independent pathway (Schmitz et al., 2000),induced a similar extent of cell death in native mVDAC1- and Δ(26)-mVDAC1-expressing cells (Fig. 4A).

Similar results were obtained by flow cytometric analysis, usingannexin V-FITC and propidium iodide (PI; Fig. 4B,C). The resultsshowed that induction of mVDAC1 overexpression by a highconcentration of tetracycline (2.5 μg/ml) resulted in an increase incell death of approximately three- to fourfold, but reduced to 1.4-

Fig. 2. Channel properties of bilayer-reconstituted Δ(26)mVDAC1. Δ(26)mVDAC1 functions as a channel but shows no voltage-dependence conductance.mVDAC1 and Δ(26)mVDAC1 were expressed in porin-less yeast mitochondria, purified, and then reconstituted into a PLB. (A) Currents through the VDAC1channel in response to a voltage step from 0 mV to voltages between –60 to +60 mV were recorded. Relative conductance was determined as the ratio ofconductance at a given voltage (G) to the maximal conductance (Go). A representative of four similar experiments is shown, mVDAC1 (�) and Δ(26)mVDAC1(�). (B) HK had no effect on the channel activity of N-terminal truncated VDAC1. Currents through bilayer-reconstituted mVDAC1 or Δ(26)mVDAC1 inresponse to a voltage step from 0 to –40 mV were recorded before and 10 minutes after the addition of HK-I (28 mIU/ml). The dashed lines indicate the zero andthe maximal current levels.

Jour

nal o

f Cel

l Sci

ence

1909(Δ26)VDAC1-mediated apoptotic resistance

fold or less (Fig. 4B,C) when Δ(26)mVDAC1 was overexpressed.Exposure of cells expressing mVDAC to As2O3 resulted in 47.5%annexin V-FITC-positive and PI-positive cells (i.e. late apoptoticcells), as compared with the 22.5% value obtained with cellsexpressing Δ(26)mVDAC1 (Fig. 4C). These results indicate that incells expressing Δ(26)mVDAC1, apoptosis is inhibited.

To test whether the apoptotic resistance of Δ(26)mVDAC1-expressing cells is due to an inability of the cells to releasecytochrome c, cells expressing native mVDAC1 or Δ(26)mVDAC1were exposed to As2O3, and the localization of cytochrome c, beforeand after exposure to the apoptosis inducers, was analyzed byimmunocytochemistry (Fig. 5A). Although in cells expressing nativemVDAC1 the punctuate mitochondrial localization of cytochromec (i.e. colocalized with MitoTracker, a dye that specifically labelsmitochondria in living cells) changed upon exposure to apoptosisstimuli to a diffused cytoplasmic location (Fig. 5B), no such changein cytochrome c distribution was observed in cells expressingΔ(26)mVDAC1 (Fig. 5C). Cytochrome c release into the cytosolwas also analyzed by western blotting using anti-cytochrome cmonoclonal antibodies (Fig. 5D). STS or cisplatin inducedcytochrome c release in cells expressing native mVDAC1 but notthose expressing Δ(26)mVDAC1. These findings clearly show thatthe N-terminal region of VDAC1 is essential for cytochrome crelease.

The Δ(26)mVDAC1 mutant has a dominant-negative effectTo test whether the resistance to apoptosis conferred by N-terminal-truncated mVDAC1 has a dominant negative effect, Δ(26)mVDAC1was expressed in three different cell lines expressing endogenous

VDAC, and apoptosis was induced by various stimuli. As with cellspossessing low level of VDAC1 (Fig. 4), all cell types showedinsensitivity to apoptosis induction by reagents that act throughactivation of cytochrome c release, but not by TNFα, whenexpressing Δ(26)mVDAC1 (Fig. 6). These results show thatalthough the cells express endogenous hVDAC1, the presence ofthe N-terminal-truncated mVDAC1 mutant, completely preventedthe induction of apoptosis.

A synthetic peptide corresponding to the VDAC1 N-terminalregion interacts with HK and Bcl2 and, when expressed incells, prevents their anti-apoptotic effectsHK-I has been shown to interact with VDAC and protect againstapoptotic cell death as induced by STS or VDAC1 overexpression(Abu-Hamad et al., 2008; Arzoine et al., 2009; Azoulay-Zohar etal., 2004; Zaid et al., 2005). As shown in Fig. 2B, purified HK-Idid not modify the channel conductance of bilayer-reconstituted

Fig. 3. Overexpression of native but not N-truncated mVDAC1 triggers celldeath. (A) In T-REx293 cells silenced for hVDAC1 expression, overexpressionof native mVDAC1 (�) or Δ(26)mVDAC1 (�) was induced by 2.5 μg/mltetracycline. A representative of three similar experiments is shown.(B) Apoptotic cell death visualized by Acridine Orange and ethidium bromidestaining. Arrow indicates cells in an early apoptotic state, reflected bydegraded nuclei (stained with Acridine Orange). Arrowhead indicates lateapoptotic state (stained with Acridine Orange and ethidium bromide). Scalebars: 15 μm. (C) Western blot analysis of VDAC levels in control cells andcells transfected to express mVDAC1 or Δ(26)mVDAC1 (15 μg) usingpolyclonal anti-VDAC antibodies. For loading controls, actin levels in thesamples (15 μg) were compared using anti-actin antibodies.

Fig. 4. Cells expressing Δ(26)mVDAC1 are resistant to apoptotic induction.Cells overexpressing native but not N-terminal-truncated VDAC1 undergomitochondria-mediated apoptosis. (A) T-REx293 cells silenced for hVDAC1expression and expressing mVDAC1 (black bars) or Δ(26)mVDAC1 (greybars), as induced by tetracycline (1 μg/ml), were treated with curcumin(200 μM, 48 hours), As2O3 (60 μM, 48 hours), STS (1.25 μM, 5 hours),cisplatin (50 μM, 30 hours) or TNFα (25 ng/ml, 7 hours). In addition,overexpression (as induced by tetracycline, 2.5 μg/ml) of mVDAC1 but not ofΔ(26)mVDAC1 led to apoptosis. Statistical analysis of apoptosis in thedifferent treatments was performed by ANOVA and t-tests; P<0.01. Data aremeans ± s.e.m. (n=4). (B) FACS analysis of apoptotic cell death, as induced byoverexpression of mVDAC1 or Δ(26)mVDAC1 or by As2O3 was carried outusing annexin V-FITC and propidium iodide (PI), as described in Materialsand Methods. A representative of three similar FACS analyses of unstainedand double stained cells with annexin V-FITC and PI for each treatment isshown. (C) Quantitative analysis of apoptosis in the FACS experiments shownin B for cells expressing mVDAC1 (hatched bars) or Δ(26)mVDAC1 (whitebars). The same experiments were analyzed by Acridine Orange and ethidiumbromide staining for cells expressing mVDAC1 (black bars) orΔ(26)mVDAC1 (grey bars). Data shown in C are the mean ± s.e.m. of threeindependent experiments.

Jour

nal o

f Cel

l Sci

ence

1910

Δ(26)mVDAC1. Thus, to further assess the interaction of HK-I andBcl2 with the VDAC1 N-terminal region, HK-I (Fig. 7A) andrecombinant Bcl2(Δ23) (Fig. 7B) or, as a control, rabbit IgG, werecoupled to surface plasmon resonance (SPR) biosensor chips.Increasing concentrations of a synthetic peptide corresponding tothe N-terminal region of VDAC1 were injected onto the sensorchips, and binding to purified immobilized HK-I (Fig. 7C),Bcl2(Δ23) (Fig. 7D) or IgG was monitored. The N-terminal peptidebound to immobilized HK-I or Bcl2(Δ23) in a concentration- andtime-dependent manner, rapidly associating with and dissociatingfrom the immobilized proteins. The binding of the VDAC1 N-terminal peptide to HK-I and Bcl2(Δ23) was specific, since no signalwas obtained with the IgG-immobilized chip. Moreover, a differentVDAC1-based peptide (designated BP), in a loop configuration (18amino acids, E157 to T174), did not interact with either HK-I (Fig.7A) or Bcl2(Δ23) (Fig. 7B). The results thus demonstrate the directand specific interaction of VDAC1-N-terminal peptide with HK-Iand Bcl2.

The interaction of the VDAC1-N-terminal peptide with HK-I wasfurther demonstrated at the cellular level by expressing the HK-I–GFP fusion protein alone or together with the N-terminal peptide(1-26 amino acids) or other VDAC1-based peptides (i.e. peptidesA, 63W-N76, and B, 157E-T174) and then visualizing HK-I–GFPcellular distribution (Fig. 8A). Confocal fluorescence microscopyshowed that in transformed control cells, HK-I–GFP fluorescencewas punctuated, as expected for a mitochondria distribution.However, HK-I–GFP fluorescence in cells expressing the N-

Journal of Cell Science 122 (11)

terminal peptide was diffuse throughout the cytosol (Fig. 8A).Similar results were obtained when peptide A (AP), but not peptideB (BP) was expressed (Fig. 8A). These results indicate that the N-terminal peptide can detach or prevent HK-I binding to themitochondria.

Next, we tested whether expression of the VDAC1 N-terminal 26-amino-acid peptide in cells overexpressing Bcl2-GFP would interferewith Bcl2- or HK-mediated protection against apoptosis, as inducedby STS. Upon exposure of control cells to STS for 5 hours, about50% of the cells underwent apoptosis (Fig. 8B). By contrast, cellstransformed to overexpress Bcl2-GFP (Fig. 8D) had completeprotection against STS-induced apoptosis (Fig. 8B). However, in cellsoverexpressing both Bcl2-GFP and the VDAC1 N-terminal peptide,Bcl2-GFP did not confer protection against STS-induced cell death(Fig. 8B). Since an inducible plasmid for VDAC1 N-terminal peptideexpression was used, inhibition of the anti-apoptotic effect of Bcl2-GFP by the peptide was observed only when its expression wasinduced by tetracycline (Fig. 8B). As with Bcl2, overexpression ofHK-I or HK-II (Fig. 8C,E) also conferred protection against cell deathinduced by STS. This protection was prevented upon expression ofthe VDAC1 N-terminal peptide (Fig. 8C).

VDAC oligomerization is enhanced by STS and byoverexpression of native and Δ(26)mVDAC1In a previous study, we employed chemical cross-linking with themembrane-permeable cross-linker, EGS, to show that VDACassumes an oligomeric form in isolated mitochondria (Zalk et al.,

Fig. 5. Cytochrome c release as induced by As2O3,cisplatin or staurosporine in cells expressing native but notN-terminal truncated VDAC1. Cells overexpressingΔ(26)mVDAC1 are resistant to mitochondria-mediatedapoptosis. T-REx293 cells (A), shRNA-hVDAC1-T-REx293 cells stably expressing native (B) orΔ(26)mVDAC1 (C) under the control of 1 μg/mltetracycline were grown on poly-D-lysine (PDL)-coatedcoverslips. After 48 hours, cells were treated with As2O3(60 μM, 24 hours). Then, cells were treated with Mito-Tracker and anti-cytochrome c antibodies as described inMaterials and Methods and visualized using confocalmicroscope (Olympus 1X81). Scale bar: 10 μm.(D) Immunoblot analysis of cytochrome c released to thecytosol in T-REx293 cells silenced for hVDAC1expression and expressing mVDAC1 or Δ(26)mVDAC1,as triggered by STS (1.25 μM, 5 hours) or by cisplatin(40 μM, 24 hours). Actin levels in cells confirmed thatequal amounts of cells were used. (E) Western blotanalysis of VDAC levels in control cells and cellstransfected to express mVDAC1 or Δ(26)mVDAC1(20 μg) using polyclonal anti-VDAC antibodies. Forloading controls, actin levels in the samples (20 μg) werecompared using anti-actin antibodies.

Jour

nal o

f Cel

l Sci

ence

1911(Δ26)VDAC1-mediated apoptotic resistance

2005). Fig. 9 shows that upon apoptosis induction by STS, thedynamic equilibrium between monomeric and oligomeric forms ofVDAC was shifted towards the formation of dimers, trimers,tetramers and multimers. VDAC oligomerization was, moreover,dramatically increased (up to sixfold) in both HEK293 and B-16cells upon exposure to STS (Fig. 9A,B).

VDAC1 overexpression not only encouraged apoptotic celldeath (Figs 3 and 4), but also led to enhanced VDAColigomerization. Upon mVDAC1 overexpression, VDAC oligomers(dimers to hexamers and multimers) became apparent even in theabsence of chemical cross-linking (Fig. 9D). The appearance oftrimers, tetramers (and possibly hexamers and multimers) uponoverexpression of mVDAC1 was, nonetheless, apparent uponchemical cross-linking with EGS (Fig. 9C). Similar results wereobtained upon expression of Δ(26)mVDAC1, as visualized usingpolyclonal anti-VDAC antibodies (Fig. 9C). The Δ(26)mVDAC1dimers (as expected, with a lower molecular mass) and higher

oligomeric states were obtained upon cross-linking with EGS. Usingmonoclonal anti-VDAC1 antibodies that specifically interact withthe N-terminal region of VDAC1, no VDAC cross-linked productswere detected in cells expressing Δ(26)mVDAC1 (Fig. 9D). Theseresults suggest that the N-terminal region of VDAC1 is not requiredfor VDAC1 oligomerization but is essential for apoptosis inductioninvolving a mechanism proposed below.

DiscussionWe have shown that cells expressing a N-terminal deletion mutantof VDAC1 are resistant to mitochondria-mediated apoptosis. Whilethe channel still localizes to mitochondria and functions normally,releases of cytochrome c and apoptotic cell death were inhibited.Moreover, the anti-apoptotic proteins HK and Bcl2 were shown toconfer protection against apoptosis through interaction with theVDAC1 N-terminal region. Finally, we demonstrated that apoptosisinduction is associated with VDAC oligomerizaion. In thisDiscussion section, we consider these findings in light of a modelwe propose for VDAC1 and its N-terminal function in apoptosisinduction and regulation.

N-terminal VDAC1 is essential for cytochrome c release andapoptosisLying in the OMM, VDAC appears to be a convergence point fora variety of cellular survival and death signals. Here, we have shownthat Δ(26)mVDAC1 functions as a channel and, like nativemVDAC1, could restore the growth and oxygen consumption ofcells silenced for hVDAC1 expression by hVDAC1-shRNA,suggesting that the N-terminal region of VDAC1 is not essentialfor VDAC1 transport activity and, hence, for mitochondrial energyproduction.

However, the N-terminal α-helix of VDAC1 is required forcytochrome c release and apoptosis when induced by various stimuli(Figs 3-6). Overexpression of human, murine, yeast or rice VDAC1,VDAC1-GFP and mutated VDAC1 was shown to induce apoptoticcell death, regardless of the cellular host (Abu-Hamad et al., 2006;Bae et al., 2003; Godbole et al., 2003; Muller et al., 2000; Zaid et

Fig. 6. Cells expressing both endogenous VDAC and N-truncated mVDAC1are resistant to apoptosis induction, a dominant negative effect. Expression ofN-terminal-truncated mVDAC1 in different cell lines expressing endogenoushVDAC1, prevented apoptosis induction. (A-C) HEK293, HeLa and MCF-7cells were transfected to express mVDAC1 (black bars) or Δ(26)mVDAC1(grey bars) and treated with STS (1.25 μM, 5 hours), cisplatin (50 μM, 30hours) or TNFα (25 ng/ml, 7 hours). Apoptotic cell death was analyzed byAcridine Orange and ethidium bromide staining, as in Fig. 3. Statisticalanalysis of apoptosis as a result of the different treatments was performed byANOVA and t-tests; P<0.01. Data are means ± s.e.m. (n=4).

Fig. 7. VDAC1 N-terminal peptide binding to HK-I andBcl2(Δ23). Synthetic peptide corresponding to the VDAC1 N-terminal region interacts with HK and Bcl2. Interaction of purifiedHK-I and Bcl2(Δ23) with VDAC1-based peptides was revealedusing real-time surface plasmon resonance. Differentconcentrations (40, 100, 200 μM) of the VDAC1 N-terminal orVDAC1 inter-membranal loop (BP) peptides were run in parallelover surface-strips of HK-I (A) or Bcl2(Δ23) (B) and analyzedusing ProteOn software. As a control for non-specific binding, theinteraction of the peptide (200 μM) with IgG was analyzed.Responses (resonance units, RU), as a function of peptideconcentration, were monitored using the ProteOn imaging systemand related SW-tools. (C,D) Coomassie Blue staining of purifiedHK-I (C) and Bcl2(Δ23) (D) used in these experiments.

Jour

nal o

f Cel

l Sci

ence

1912

al., 2005), suggesting that VDAC1 is a conserved mitochondrialelement of the death machinery. In accordance with previous results,overexpression of mVDAC dramatically triggers cell death. Bycontrast, overexpression of Δ(26)mVDAC1, although distributedto the mitochondria (Fig. 1B), triggered no such apoptotic cell death

Journal of Cell Science 122 (11)

(Figs 3 and 4). Moreover, induction of apoptosis by STS, curumin,cisplatin and As2O3, each relying on different pathways (Duvoixet al., 2005; Schmitz et al., 2000; Zaid et al., 2005; Zhang et al.,2005), all ultimately led to cytochrome c release (Fig. 5) and celldeath (Figs 3, 4 and 6) in cells expressing native but notΔ(26)mVDAC1. Thus, by expressing N-terminal-truncated VDAC1,we have constructed an apoptosis-resistant cell line, namely, a linefor which a highly active apoptosis stimulus did not inducecytochrome c release or cell death.

Recently, it has been shown that overexpression of hVDAC1 inCOS cells increased the number of cells with depolarizedmitochondria, a phenomenon associated with apoptosis, and thatthis effect could not be obtained upon overexpression of(Δ19)hVDAC1 (De Pinto et al., 2007). These findings are consistentwith our results showing that the VDAC1 N-terminal region isnecessary for mitochondria-mediated apoptosis.

The VDAC1 N-terminal region interacts with HK and Bcl2 toprevent their anti-apoptotic effectsHK-I and HK-II are highly expressed in cancer cells, with over70% of the protein being bound to mitochondria (Arora et al., 1990;Pedersen et al., 2002), or more precisely, to VDAC (Abu-Hamadet al., 2008; Arzoine et al., 2009; Azoulay-Zohar et al., 2004;Pastorino et al., 2002). Similarly, Bcl2 is overexpressed in morethan half of all human cancers, with its broad expression in tumorsbeing linked to a conferring of resistance to chemotherapy-inducedapoptosis (Ding et al., 2000). Both HK (Abu-Hamad et al., 2008;Arzoine et al., 2009; Azoulay-Zohar et al., 2004; Pastorino et al.,2002) and members of the Bcl2 family (Pastorino et al., 2002; Shiet al., 2003; Shimizu et al., 1999; Tsujimoto and Shimizu, 2002)are though to act via interaction with VDAC.

Recently, those VDAC1 amino acid residues important for HK-I binding and conferring the anti-apoptotic effects of the proteinwere identified (Abu-Hamad et al., 2008). Clearly, the resultspresented here indicate that the N-terminal region of VDAC1 isessential for the HK-mediated reduction in channel conductance(Fig. 2B) and for HK-I and HK-II protection against apoptosis (Fig.8C). Both HK-I and Bcl2 interact with a synthetic peptidecorresponding to the VDAC1 N-terminus, as revealed using SPRtechnology (Fig. 7A,B). Moreover, expression of this N-terminalpeptide in cells overexpressing HK-I, HK-II or Bcl2 prevented theirprotective effects against apoptosis, as induced by STS (Fig. 8).Our findings suggest that the expressed peptide competes withVDAC1 for binding to Bcl2, HK-I and HK-II and thus interfereswith their binding to VDAC1, an interaction that prevents apoptosis.It has been shown that the N-terminal region of VDAC1 modulatesthe accessibility of VDAC1 to Bax and Bcl-xL (Shi et al., 2003).Our observations, therefore, suggest that the anti-apoptotic activityof HK and Bcl2 is mediated through interactions with the VDAC1N-terminal region, pointing to the N-terminal region of VDAC1 asparticipating in the regulation of apoptosis by anti-apoptoticproteins. This novel mechanism is presented in Fig. 10.

The mobile VDAC1 N-terminal α-helix participates in theformation of a cytochrome c release pore: a proposed modelNumerous studies have focused on the importance of the N-terminalα-helical segment in channel function. These studies have yieldeda wide range of predictions as to the functional disposition of thisdomain, ranging from it forming a segment of the channel wall toacting as the voltage sensor (Colombini et al., 1996; Koppel et al.,1998), to regulating the conductance of ions and metabolites

Fig. 8. Expression of the VDAC1-N-terminal peptide in T-REx293 cellsprevents HK-I-, HK-II- and Bcl2-mediated protection against STS-inducedcell death. Expression of VDAC1 N-terminal peptide inhibited the anti-apoptotic effect of Bcl2, HK-I and HK-II. (A) T-REx293 cells or shRNA-hVDAC1–T-REx293 cells stably expressing native or Δ(26)mVDAC1 underthe control of 1 μg/ml tetracycline were transfected with HK-I–GFP or co-transfected with HK-I–GFP and pcDNA4TO-AP, pcDNA4TO-BP orpcDNA4TO-NP. After 40 hours cells were visualized using confocalmicroscopy. (B) T-REx293 cells and cells expressing Bcl2-GFP and/or the N-terminal 26-amino-acid peptide (NP) under the control of tetracycline wereexposed for 5 hours to 1.25 μM STS, and apoptotic cell death was followed.(C) An experiment similar to that in B was carried out with T-REx293 cellsthat had been transformed to express HK-I, HK-II and/or NP. Apoptosis in thedifferent cell types was analyzed quantitatively with the aid of AcridineOrange and ethidium bromide staining, as in Fig. 2. Data are means ± s.e.m.(n=2-4). (D,E) Western blot analysis of Bcl2-GFP (D) or HK-I or HK-II (E)levels in control cells and cells transfected to overexpress either of theseproteins and/or NP. Aliquots (50 μg) were analyzed for HK-I and HK-II levels,using polyclonal anti-HK-I and anti-HK-II antibodies, respectively, and forBcl2-GFP, using anti-GFP antibodies. For control loading, actin levels in thesamples were compared, using anti-actin antibodies.

Jour

nal o

f Cel

l Sci

ence

1913(Δ26)VDAC1-mediated apoptotic resistance

passing through the VDAC1 pore, based on its position in the pore(Bayrhuber et al., 2008; Ujwal et al., 2008). In addition,conformational instability of the N-terminal part of hVDAC1 wassuggested, with the domain being predicted to switch betweendifferent conformations, during which hydrogen bonds within theN-terminus and the strands of the barrel are transiently formed andbroken (Bayrhuber et al., 2008; Ujwal et al., 2008). Therefore, theN-terminal domain may adopt different conformations, dependingon various factors, such as protein-protein interactions, ioninteractions and lipid environment. Re-arrangement andcoordination of the N-terminal helix is suggested here to beinvolved in VDAC1 function in apoptosis.

The precise mechanisms regulating cytochrome c release, a keyinitial step in the apoptotic process, remain unknown. Similarly,the molecular architecture of the cytochrome-c-conducting channelhas yet to be determined. We hypothesize that a protein-conductingchannel is formed within a VDAC1 homo-oligomer or a hetero-oligomer containing VDAC1 and pro-apoptotic proteins (Shoshan-Barmatz et al., 2008; Shoshan-Barmatz et al., 2006; Zalk et al.,2005; Zheng et al., 2004). Indeed, further support for the supra-molecular organization of VDAC, comprising monomers totetramers, hexamers and higher oligomers, has come from recentstudies applying atomic force microscopy (Goncalves et al., 2007;Hoogenboom et al., 2007) and NMR (Bayrhuber et al., 2008; Maliaand Wagner, 2007). It was also demonstrated that VDAColigomerization is encouraged in the presence of cytochrome c (Zalket al., 2005). The function of this VDAC oligomerization in

apoptosis is demonstrated here, with the demonstration that thesupramolecular assembly of VDAC in cultured cells is highlyenhanced upon apoptosis induction (Fig. 9). Apoptosis inductionby STS or by overexpression of mVDAC1 resulted in several-foldincrease in VDAC oligomerization. Similarly, the apoptosis-inducing effect of As2O3 was attributed to an induction of VDAChomodimerization that was prevented by overexpression of the anti-apoptotic protein, Bcl-XL (Yu et al., 2007; Zheng et al., 2004).However, since (Δ26)mVDAC1 is able to undergo oligomerization(Fig. 9) but not to induce apoptosis (Figs 3, 4 and 6), it appearsthat the N-terminal region is not required for VDAC oligomerizationbut is important for cytochrome c release and subsequent apoptosis.

Together, these observations can be explained by the followingmodel (Fig. 10). Once formed, an oligomeric VDAC1intermolecular pore can serve as a conduit for a protein. Oligomersof VDAC1 β-barrels would, however, form a hydrophobic channelthat would not allow charged proteins, such as cytochrome c topass. However, if the amphipathic N-terminal region (containingthree positive and two negative residues, but with an overall chargeof zero) of each VDAC1 molecule is moved to line the externalsurface of the β-barrel, a hydrophilic pore would form (Fig. 10).In such a scenario, the now hydrophilic pore, lined by several α-helix N-terminal regions, would permit protein transport across theOMM (Fig. 10B). In this regard, it should be noted that a recentstudy suggested that the N-terminus of VDAC in aqueous mediumis uncoiled, requiring a hydrophobic and negatively chargedenvironment to gain its α-helical structure (De Pinto et al., 2007).

Fig. 9. Apoptosis induction by STS or VDAC1overexpression induces VDAC oligomerization ofnative and Δ(26)mVDAC1. STS-induced VDAColigomerization revealed by EGS cross-linking ofHEK293 (A) or B-16 (B) cells before and after 1.5 or3 hours of incubation with STS (1.25 μM). Cells werewashed with PBS and incubated with the indicatedconcentration of EGS at 30°C for 15 minutes, followedby SDS-PAGE (2.6–13% gel gradient or 9%acrylamide for A and B, respectively), and westernblotted using monoclonal anti-VDAC. Anti-actinantibodies were used for the loading control. Cross-linking of isolated mitochondria (Mito) is also shown.In C and D, mVDAC1 and Δ(26)mVDAC1 wereoverexpressed in T-REx293 cells under the control oftetracycline (2.5 μg/ml). The cells were exposed toEGS (125 or 250 μM) and analyzed for VDAC-containing cross-linked products using polyclonal (C)or monoclonal (directed to the N-terminal) anti-VDACantibodies (D). A single and double black asteriskindicate native endogenous VDAC andΔ(26)mVDAC1 dimers, respectively.

Jour

nal o

f Cel

l Sci

ence

1914 Journal of Cell Science 122 (11)

With N-terminally truncated VDAC1, such a hydrophilic protein-conducting pore could not be formed, and hence, cytochrome cwould not be released (Fig. 5). The finding that the VDAC1 N-terminal α-helix specifically interacts with cytochrome c (Mannellaet al., 1987) supports the proposed model.

The ability of a synthetic peptide, corresponding to the N-terminalregion of VDAC1, to bind both HK-I and Bcl2 (Fig. 7), to detachmitochondrial-bound HK-I–GFP (Fig. 8A) and to prevent Bcl2-,HK-I- and HK-II-mediated protection against apoptosis (Fig. 8B,C),suggests that the function of the α-helix N-terminal portion ofVDAC1 in apoptosis can be modulated by anti-apoptotic proteins.According to our proposal, the anti-apoptotic activities of HK-I,HK-II and Bcl2 result from their interaction with the mobile N-terminal region of VDAC1 and preventing its reorientation into thecytochrome c-releasing pore (Fig. 10C).

Taken together, the requirement for the VDAC1 N-terminalregion for cytochrome c release and apoptosis induction, as wellas its interaction with anti-apoptotic proteins, point to the VDAC1N-terminal region as being a key element in regulating apoptosisand the target for anti-apoptotic proteins (Fig. 10).

Finally, mitochondria-mediated apoptosis plays a crucial role inthe pathophysiology of many diseases, including heart attack andstroke, neurodegenerative disorders, such as Parkinson’s disease andAlzheimer’s disease, as well as mitochondrial encephalomyopathiesand cancer (Mattson, 2000; Thompson, 1995). Although somediseases, such as cancer, are caused by suppression of apoptosis,other diseases, such as Alzheimer’s disease, are characterized byincreased apoptosis. The VDAC1 N-terminal region offers atempting target for the development of therapeutic agents designedto inhibit apoptosis. In cancer therapeutics, targeting VDAC1 N-terminal peptide to tumor cells overexpressing anti-apoptoticproteins, such as Bcl2 and HK-I, would minimize the self-defense

mechanisms of the cancer cells, thereby promoting apoptosis andincreasing sensitivity to chemotherapy.

Materials and MethodsMaterialsStaurosporine (STS), poly-D-lysine (PDL) and propidium iodide were purchasedfrom Sigma (St Louis, MO). Mito-Tracker red dye CMXPos was from MolecularProbes. Annexin V-FITC kit was purchased from Beender MedSystem. Monoclonalanti-VDAC antibodies raised to the N-terminal region of VDAC1 came fromCalbiochem-Novobiochem (Nottingham, UK), and rabbit polyclonal anti-VDACantibodies, prepared against residues 150-250 of human (h)VDAC1, were fromAbcam. Monoclonal anti-cytochrome c antibodies for immunoblots (1:2000; cat.no. 556433) and for immunostaining (1:800; cat. no. 556432) were obtained fromBD Biosciences Pharmingen. Monoclonal anti-GFP antibodies were obtained fromSanta Cruz Biotechnology. Metefectene was purchased from Biotex (Munich,Germany). RPMI 1640 and DMEM growth media and the supplements, fetal calfserum, L-glutamine and penicillin-streptomycin were purchased from BiologicalIndustries (Beit Haemek, Israel). Blasticidin and zeocin were purchased fromInvivoGen (San Diego, CA). Puromycin was purchased from ICN Biomedicals(Eschwege, Germany).

Plasmids and site-directed mutagenesisVectors expressing N-terminally truncated murine (m)VDAC1 or the N-terminal 26-amino-acid peptide were generated by PCR and cloned into the BamHI-NotI orBamHI-EcoRI sites of the tetracycline-inducible pcDNA4/TO vector (Invitrogen),using mVDAC1 cDNA as template and the following primers: 5�-CGGGATCCAT-GATAAAACTTGATTTGAAAACG-3� (F) and 5�-GCGGCCGCTTATGCTTG AA -ATTCCAGTCC-3� (R) for (Δ26)mVDAC1 and 5�-GCC AGATCTATGGCTGTG -CCACCCACGTA-3� (F) and 5�-CGAATTCTC ATAAGCCAAATCCATAGCCCTT-3� (R) for the 26 amino acid peptide.

The pcDNA3–HK-II and pcDNA3–HK-I plasmids were kindly provided by J. E.Wilson (Michigan State University). HK-I–GFP was generated as describedpreviously(Abu-Hamad et al., 2008).

For construction of C-terminally truncated Bcl2, Bcl2-GFP cDNA was sub-clonedfrom plasmid pcDNA3 using the primers: 5�-CTCCATGGCGATG GCGCACGC -TGGGAGAACG-3� (F) containing a NcoI restriction site and 5�-CGGAATTC -TTACAGAGACAGCCAGGAGAAATC-3� (R), containing a stop codon and anEcoRI restriction site. The generated product, encoding Bcl2(Δ23), was then clonedinto a pHISparallel vector for expression in E. coli BL21 cells.

Cell cultureMCF7 cells are a human breast carcinoma cell line, HEK293 cells are a transformedprimary human embryonal kidney cell line, T-REx293 cells are tetracycline repressor-expressing HEK293 cells and HeLa cells are a cervical cancer-derived cell line. Theseand hVDAC1-shRNA T-REx293 cells were grown in Dulbecco’s modified Eagle’smedium (Biological Industries), supplemented with 10% fetal calf serum, 2 mM L-glutamine, 100 IU/ml penicillin, 100 μg/ml streptomycin and nonessential aminoacids (all from Biological Industries) and maintained in a humidified atmosphere, at37°C with 5% CO2. hVDAC1-shRNA T-REx293 cells stably expressing thepSUPERretro vector encoding shRNA-targeting hVDAC1 were transfected withplasmids pcDNA4/TO-mVDAC1 or pcDNA4/TO-Δ(26)mVDAC1 and grown asdescribed previously (Abu-Hamad et al., 2006). T-REx293 cells were transiently co-transfected with plasmids pEGFP-Bcl2, pcDNA3.1–HK-I or pcDNA3.1–HK-II andpcDNA/4TO encoding the N-terminal 26 amino acids and grown for 48 hours withtetracycline (1 μg/ml), prior to a 5 hour exposure to STS (1.25 μM). Overexpressionof mVDAC1- or Δ(26)mVDAC1-pcDNA4/TO was induced by tetracycline (2.5μg/ml) for 96-100 hours. Apoptosis was analyzed by Acridine Orange and ethidiumbromide staining (Zaid et al., 2005) or by FACS.

Oxygen consumptionOxygen utilization was measured polarographically with a thermostatically controlled(37°C) Clark oxygen electrode (Strathkelvin 782 Oxygen System). T-REx293,hVDAC1-shRNA T-REx293 cells expressing stable native or Δ(26)mVDAC1(2�106) were washed, suspended in a serum-free DMEM medium [1000 mg/lglucose pyridoxine, HCl and NaHCO3, without phenol red (Sigma D-5921)],supplemented with L-glutamine (4 mM) and L-pyruvate (1 mM), and then addedto a 1 ml water-jacketed chamber. Cells respiration was determined in absence andpresence of rotenone (5 μM, added directly into the respiration chamber) for up to10 minutes.

Detachment of mitochondria-bound HK-I–GFP by VDAC1-basedpeptidesCells (5�104) were grown on PDL-coated coverslips in a 24-well dish and transfectedwith plasmid pEGFP–HK-I alone or with DNA encoding the N-terminal peptide (NP),peptide A (amino acids 63W-78N, AP, previously named LP1) and peptide B (157E-174T, BP, previously named LP3) that were generated and cloned into the tetracycline-inducible pcDNA4/TO vector (Invitrogen), as described previously (Arzione et al.,

Fig. 10. Model for VDAC1 N-terminal region-mediated cytochrome c release.(A) Side-view across a membranal VDAC1 with the amphipathic α-helix N-terminal region in various locations: cytoplasmically exposed (De Pinto et al.,2003), membrane-spanning (Colombini, 2004), lying on the membrane surface(Reymann et al., 1995) and positioned in the pore (Bayrhuber et al., 2008;Hiller et al., 2008; Ujwal et al., 2008). (B) Upon an apoptotic signal, VDAColigomerization is enhanced and the amphipathic α-helix N-terminal region ofeach VDAC molecule flips inside the hydrophobic pore formed by the β-barrels, making the pore hydrophilic and capable of conducting cytochrome crelease. (C) Interaction of anti-apoptotic proteins (HK, Bcl2) with the N-terminal region of VDAC1 prevents its translocation and thus the formation ofthe hydrophilic pore, so inhibiting cytochrome c release.

Jour

nal o

f Cel

l Sci

ence

1915(Δ26)VDAC1-mediated apoptotic resistance

2009). Peptides A and B were flanked by a tryptophan zipper motif, i.e. the SWTWEamino acid sequence, at the N-terminus of the peptide, and the KWTWK sequenceat the C-terminus, to form a loop (Arzione et al., 2009).

Cells were fixed for 15 minutes with 4% paraformaldehyde and rinsed with PBSfor 30 minutes prior to imaging by confocal microscopy (using an Olympus 1X81microscope).

Cytochrome c releaseImmunoblot analysisFor immunoblot analysis, cells expressing native or Δ(26)mVDAC1 were grown for72 hours, exposed to STS, cisplatine or As2O3 and analyzed for cytochrome c release(Abu-Hamad et al., 2008). Aliquots of the cytosolic fraction were separated on Tris-Tricine gels (13% polyacrylamide) and immunoprobed using anti-cytochrome cantibodies. To rule out mitochondrial contamination, the cytosolic fraction wasimmunoprobed for VDAC using anti-VDAC antibodies, and then with HRP-conjugated anti-mouse IgG as a secondary antibody (1:10,000). An enhancedchemiluminiscent substrate (Pierece Chemical, Rockford, IL) was used for detectionof HRP.

ImmunostainingFor immunostaining, cells (5�104) were grown on poly-D-lysine (PDL)-coatedcoverslip in a six-well plate. After 48 hours, cells were treated with As2O3 (60 μM,24 hours). Then, the cells were treated with MitoTracker red dye (25 nM) for 15minutes in an incubator (37°C, 5% CO2), washed three times with PBS, fixed for 15minutes with 4.0% paraformaldehyde and rinsed three times with PBS. Cells weretreated with methanol for 15 minutes (–20°C), then incubated for 1 hour with blockingsolution (5% normal goat serum (NGS) and 0.2% Triton X-100), followed byincubation with anti-cytochrome c antibodies (cat. no. 556432, 1:800, diluted inblocking solution) for 1 hour. After washing with PBS, cells were incubated for 1hour with Alexa-Fluor-488-conjugated antibodies (1:500, diluted in blocking solution)and washed with PBS. Coverslips were mounted on glass slides with mountingmedium and sealed with nail polish. Cells were visualized by confocal microscopy(Olympus 1X81).

Flow cytometryApoptotic cells were estimated using a commercial annexin V-FITC and PI detectionkit or following PI uptake and FACS analysis (FACScan, Beckton-Dickinson, SanJose, CA) and ModFIT-lt2.0 software.

Protein purificationmVDAC1 and Δ(26)mVDAC1 were expressed in the Saccharomyces cerevisiaepor1-mutant strain M22-2 (Zaid et al., 2005) and purified from mitochondria(Shoshan-Barmatz and Gincel, 2003). HK-I was purified from rat brain mitochondria(Azoulay-Zohar et al., 2004). Bcl2(Δ23) was expressed in E. coli BL21, inducedwith isopropyl β-D-1-thiogalactopyranoside. Cells were grown for 30-60 minutesat 20°C, sonicated, and the soluble fraction (5-10% of the total) was purified by Ni-NTA chromatography.

Cross-linking experimentsFor cross-linking, T-REx293, HEK293 or B-16 cells (2-3 mg/ml) were incubatedwith different concentrations of EGS [ethylene glycol bis(succinimidylsuccinate)] inPBS, pH 8.3 (15 minutes, 30°C). Samples (50 μg) were subjected to SDS-PAGE andimmunoblotting using anti-VDAC antibodies.

Single-channel recording and analysisReconstitution of purified native mVDAC1 and Δ(26)mVDAC1 into a planar lipidbilayer (PLB), channel recording, and analysis were carried out as describedpreviously (Gincel et al., 2001).

Real-time surface plasmon resonancePurified rat brain HK-I, recombinant Bcl2 and rabbit IgG were immobilized on aCM5 sensor surface according to the manufacturer’s instructions. N-terminal VDAC1peptide was diluted in running buffer [150 mM NaCl, 0.005% Tween 20, 4% (v/v)DMSO, PBS, pH 7.4] and injected onto the sensor chip at varying concentrations(flow rate of 40 μl/minute, 25°C). Experiments were carried out using the SPRProteOn-XPR36 system (Bio-Rad). Peptides were synthesized by the oligonucleotideand peptide synthesis unit, Weizmann Institute, Israel.

This research was supported, in part, by grants from the Israel ScienceFoundation, the Israel Cancer Association and by the Johnson andJohnson Corporate Office of Science and Technology. The authors thankBio-Rad, Israel, for providing their ProteOn instrument and ZafrirBravman for helping with the SPR experiments and Dorit Ben-Shaharfor helping with oxygen consumption experiments.

ReferencesAbu-Hamad, S., Sivan, S. and Shoshan-Barmatz, V. (2006). The expression level of the

voltage-dependent anion channel controls life and death of the cell. Proc. Natl. Acad.Sci. USA 103, 5787-5792.

Abu-Hamad, S., Zaid, H., Israelson, A., Nahon, E. and Shoshan-Barmatz, V. (2008).Hexokinase-I protection against apoptotic cell death is mediated via interaction with thevoltage-dependent anion channel-1: mapping the site of binding. J. Biol. Chem. 283,13482-13490.

Arora, K. K., Fanciulli, M. and Pedersen, P. L. (1990). Glucose phosphorylation intumor cells. Cloning, sequencing, and overexpression in active form of a full-lengthcDNA encoding a mitochondrial bindable form of hexokinase. J. Biol. Chem. 265, 6481-6488.

Arzoine, L., Zilberberg, N., Ben-Romano, R. and Shoshan-Barmatz, V. (2009). Voltage-dependent anion channel-1-based peptides interact with hexokinase to prevent its anti-apoptotic activity. J. Biol. Chem. 284, 3946-3955.

Azoulay-Zohar, H., Israelson, A., Abu-Hamad, S. and Shoshan-Barmatz, V. (2004).In self-defence: hexokinase promotes voltage-dependent anion channel closure andprevents mitochondria-mediated apoptotic cell death. Biochem. J. 377, 347-355.

Bae, J. H., Park, J. W. and Kwon, T. K. (2003). Ruthenium red, inhibitor of mitochondrialCa2+ uniporter, inhibits curcumin-induced apoptosis via the prevention of intracellularCa2+ depletion and cytochrome c release. Biochem. Biophys. Res. Commun. 303, 1073-1079.

Baines, C. P., Kaiser, R. A., Sheiko, T., Craigen, W. J. and Molkentin, J. D. (2007).Voltage-dependent anion channels are dispensable for mitochondrial-dependent cell death.Nat. Cell Biol. 9, 550-555.

Bayrhuber, M., Meins, T., Habeck, M., Becker, S., Giller, K., Villinger, S., Vonrhein,C., Griesinger, C., Zweckstetter, M. and Zeth, K. (2008). Structure of thehuman voltage-dependent anion channel. Proc. Natl. Acad. Sci. USA 105, 15370-15375.

Colombini, M. (2004). VDAC: the channel at the interface between mitochondria and thecytosol. Mol. Cell Biochem. 256-257, 107-115.

Colombini, M., Yeung, C. L., Tung, J. and Konig, T. (1987). The mitochondrial outermembrane channel, VDAC, is regulated by a synthetic polyanion. Biochim. Biophys.Acta 905, 279-286.

Colombini, M., Blachly-Dyson, E. and Forte, M. (1996). VDAC, a channel in the outermitochondrial membrane. Ion Channels 4, 169-202.

Crompton, M. (1999). The mitochondrial permeability transition pore and its role in celldeath. Biochem. J. 341, 233-249.

De Pinto, V., Messina, A., Accardi, R., Aiello, R., Guarino, F., Tomasello, M. F.,Tommasino, M., Tasco, G., Casadio, R., Benz, R. et al. (2003). New functions of anold protein: the eukaryotic porin or voltage dependent anion selective channel (VDAC).Ital. J. Biochem. 52, 17-24.

De Pinto, V., Tomasello, F., Messina, A., Guarino, F., Benz, R., La Mendola, D., Magri,A., Milardi, D. and Pappalardo, G. (2007). Determination of the Conformation of thehuman VDAC1 N-terminal peptide, a protein moiety essential for the functional propertiesof the pore. Chembiochem 8, 744-756.

De Pinto, V., Reina, S., Guarino, F. and Messina, A. (2008). Structure of the voltagedependent anion channel: state of the art. J. Bioenerg. Biomembr. 40, 139-147.

Ding, Z., Yang, X., Pater, A. and Tang, S. C. (2000). Resistance to apoptosis is correlatedwith the reduced caspase-3 activation and enhanced expression of antiapoptotic proteinsin human cervical multidrug-resistant cells. Biochem. Biophys. Res. Commun. 270, 415-420.

Doran, E. and Halestrap, A. P. (2000). Cytochrome c release from isolated rat livermitochondria can occur independently of outer-membrane rupture: possible role of contactsites. Biochem. J. 348, 343-350.

Duvoix, A., Blasius, R., Delhalle, S., Schnekenburger, M., Morceau, F., Henry, E.,Dicato, M. and Diederich, M. (2005). Chemopreventive and therapeutic effects ofcurcumin. Cancer Lett. 223, 181-190.

Gincel, D., Zaid, H. and Shoshan-Barmatz, V. (2001). Calcium binding and translocationby the voltage-dependent anion channel: a possible regulatory mechanism inmitochondrial function. Biochem. J. 358, 147-155.

Godbole, A., Varghese, J., Sarin, A. and Mathew, M. K. (2003). VDAC is a conservedelement of death pathways in plant and animal systems. Biochim. Biophys. Acta 1642,87-96.

Goncalves, R. P., Buzhynskyy, N., Prima, V., Sturgis, J. N. and Scheuring, S. (2007).Supramolecular assembly of VDAC in native mitochondrial outer membranes. J. Mol.Biol. 369, 413-418.

Green, D. R. and Reed, J. C. (1998). Mitochondria and apoptosis. Science 281, 1309-1312.

Halestrap, A. P., Doran, E., Gillespie, J. P. and O’Toole, A. (2000). Mitochondria andcell death. Biochem. Soc. Trans. 28, 170-177.

Hiller, S., Garces, R. G., Malia, T. J., Orekhov, V. Y., Colombini, M. and Wagner, G.(2008). Solution structure of the integral human membrane protein VDAC-1 in detergentmicelles. Science 321, 1206-1210.

Hoogenboom, B. W., Suda, K., Engel, A. and Fotiadis, D. (2007). The supramolecularassemblies of voltage-dependent anion channels in the native membrane. J. Mol. Biol.370, 246-255.

Junankar, P. R., Dulhunty, A. F., Curtis, S. M., Pace, S. M. and Thinnes, F. P. (1995).Porin-type 1 proteins in sarcoplasmic reticulum and plasmalemma of striated musclefibres. J. Muscle Res. Cell Motil. 16, 595-610.

Koppel, D. A., Kinnally, K. W., Masters, P., Forte, M., Blachly-Dyson, E. andMannella, C. A. (1998). Bacterial expression and characterization of the mitochondrialouter membrane channel. Effects of n-terminal modifications. J. Biol. Chem. 273, 13794-13800.

Jour

nal o

f Cel

l Sci

ence

Kroemer, G., Galluzzi, L. and Brenner, C. (2007). Mitochondrial membranepermeabilization in cell death. Physiol. Rev. 87, 99-163.

Lemasters, J. J. and Holmuhamedov, E. (2006). Voltage-dependent anion channel(VDAC) as mitochondrial governator-thinking outside the box. Biochim. Biophys. Acta1762, 181-190.

Lemasters, J. J., Qian, T., Bradham, C. A., Brenner, D. A., Cascio, W. E., Trost, L.C., Nishimura, Y., Nieminen, A. L. and Herman, B. (1999). Mitochondrial dysfunctionin the pathogenesis of necrotic and apoptotic cell death. J. Bioenerg. Biomembr. 31, 305-319.

Malia, T. J. and Wagner, G. (2007). NMR structural investigation of the mitochondrialouter membrane protein VDAC and its interaction with antiapoptotic Bcl-x(L).Biochemistry 46, 514-525.

Mannella, C. A. (1997). Minireview: on the structure and gating mechanism of themitochondrial channel, VDAC. J. Bioenerg. Biomembr. 29, 525-531.

Mannella, C. A., Ribeiro, A. J. and Frank, J. (1987). Cytochrome c binds to lipid domainsin arrays of mitochondrial outer membrane channels. Biophys. J. 51, 221-226.

Martinou, J. C., Desagher, S. and Antonsson, B. (2000). Cytochrome c release frommitochondria: all or nothing. Nat. Cell Biol. 2, E41-E43.

Mattson, M. P. (2000). Apoptosis in neurodegenerative disorders. Nat. Rev. Mol. Cell.Biol. 1, 120-129.

Muller, A., Gunther, D., Brinkmann, V., Hurwitz, R., Meyer, T. F. and Rudel, T. (2000).Targeting of the pro-apoptotic VDAC-like porin (PorB) of Neisseria gonorrhoeae tomitochondria of infected cells. EMBO J. 19, 5332-5343.

Pastorino, J. G., Shulga, N. and Hoek, J. B. (2002). Mitochondrial binding of hexokinaseII inhibits Bax-induced cytochrome c release and apoptosis. J. Biol. Chem. 277, 7610-7618.

Pedersen, P. L., Mathupala, S., Rempel, A., Geschwind, J. F. and Ko, Y. H. (2002).Mitochondrial bound type II hexokinase: a key player in the growth and survival ofmany cancers and an ideal prospect for therapeutic intervention. Biochim. Biophys. Acta1555, 14-20.

Peng, S., Blachly-Dyson, E., Forte, M. and Colombini, M. (1992). Large scalerearrangement of protein domains is associated with voltage gating of the VDAC channel.Biophys. J. 62, 123-131; discussion 131-135.

Popp, B., Court, D. A., Benz, R., Neupert, W. and Lill, R. (1996). The role of the Nand C termini of recombinant Neurospora mitochondrial porin in channel formation andvoltage-dependent gating. J. Biol. Chem. 271, 13593-13599.

Reed, J. C. (2002). Apoptosis-based therapies. Nat. Rev. Drug Discov. 1, 111-121.Reymann, S., Florke, H., Heiden, M., Jakob, C., Stadtmuller, U., Steinacker, P., Lalk,

V. E., Pardowitz, I. and Thinnes, F. P. (1995). Further evidence for multitopologicallocalization of mammalian porin (VDAC) in the plasmalemma forming part of a chloridechannel complex affected in cystic fibrosis and encephalomyopathy. Biochem. Mol. Med.54, 75-87.

Schmitz, I., Kirchhoff, S. and Krammer, P. H. (2000). Regulation of death receptor-mediated apoptosis pathways. Int. J. Biochem. Cell Biol. 32, 1123-1136.

Shi, Y., Chen, J., Weng, C., Chen, R., Zheng, Y., Chen, Q. and Tang, H. (2003).Identification of the protein-protein contact site and interaction mode of human VDAC1with Bcl-2 family proteins. Biochem. Biophys. Res. Commun. 305, 989-996.

Shimizu, S., Narita, M. and Tsujimoto, Y. (1999). Bcl-2 family proteins regulate therelease of apoptogenic cytochrome c by the mitochondrial channel VDAC. Nature 399,483-487.

Shoshan-Barmatz, V. and Gincel, D. (2003). The voltage-dependent anion channel:characterization, modulation, and role in mitochondrial function in cell life and death.Cell Biochem. Biophys. 39, 279-292.

Shoshan-Barmatz, V., Zalk, R., Gincel, D. and Vardi, N. (2004). Subcellular localizationof VDAC in mitochondria and ER in the cerebellum. Biochim. Biophys. Acta 1657, 105-114.

Shoshan-Barmatz, V., Israelson, A., Brdiczka, D. and Sheu, S. S. (2006). The voltage-dependent anion channel (VDAC): function in intracellular signalling, cell life and celldeath. Curr. Pharm. Des. 12, 2249-2270.

Shoshan-Barmatz, V., Arbel, N. and Arzoine, L. (2008). VDAC, the voltage-dependentanion channel: function, regulation and mitochondrial signaling in cell life and death.Cellscience 4, 74-118.

Stanley, S., Dias, J. A., D’Arcangelis, D. and Mannella, C. A. (1995). Peptide-specificantibodies as probes of the topography of the voltage-gated channel in the mitochondrialouter membrane of Neurospora crassa. J. Biol. Chem. 270, 16694-16700.

Tajeddine, N., Galluzzi, L., Kepp, O., Hangen, E., Morselli, E., Senovilla, L., Araujo,N., Pinna, G., Larochette, N., Zamzami, N. et al. (2008). Hierarchical involvementof Bak, VDAC1 and Bax in cisplatin-induced cell death. Oncogene 27, 4221-4232.

Thompson, C. B. (1995). Apoptosis in the pathogenesis and treatment of disease. Science267, 1456-1462.

Tsujimoto, Y. and Shimizu, S. (2002). The voltage-dependent anion channel: an essentialplayer in apoptosis. Biochimie 84, 187-193.

Ujwal, R., Cascio, D., Colletier, J. P., Faham, S., Zhang, J., Toro, L., Ping, P. andAbramson, J. (2008). The crystal structure of mouse VDAC1 at 2.3 A resolution revealsmechanistic insights into metabolite gating. Proc. Natl. Acad. Sci. USA 105, 17742-17747.

Yehezkel, G., Abu-Hamad, S. and Shoshan-Barmatz, V. (2007). An N-terminalnucleotide-binding site in VDAC1: involvement in regulating mitochondrial function.J. Cell Physiol. 212, 551-561.

Yu, J., Qian, H., Li, Y., Wang, Y., Zhang, X., Liang, X., Fu, M. and Lin, C. (2007).Therapeutic effect of Arsenic Trioxide (As2O3) on cervical cancer in vitro and in vivothrough apoptosis induction. Cancer Biol. Ther. 6, 580-586.

Yuan, S., Fu, Y., Wang, X., Shi, H., Huang, Y., Song, X., Li, L., Song, N. and Luo, Y.(2008). Voltage-dependent anion channel 1 is involved in endostatin-induced endothelialcell apoptosis. FASEB J. 22, 2809-2820.

Zaid, H., Abu-Hamad, S., Israelson, A., Nathan, I. and Shoshan-Barmatz, V. (2005).The voltage-dependent anion channel-1 modulates apoptotic cell death. Cell Death Differ.12, 751-760.

Zalk, R., Israelson, A., Garty, E. S., Azoulay-Zohar, H. and Shoshan-Barmatz, V. (2005).Oligomeric states of the voltage-dependent anion channel and cytochrome c release frommitochondria. Biochem. J. 386, 73-83.

Zamzami, N. and Kroemer, G. (2003). Apoptosis: mitochondrial membranepermeabilization - the (w)hole story? Curr. Biol. 13, R71-R73.

Zhang, H., Vollmer, M., De Geyter, M., Durrenberger, M. and De Geyter, C. (2005).Apoptosis and differentiation induced by staurosporine in granulosa tumor cells is coupledwith activation of JNK and suppression of p38 MAPK. Int. J. Oncol. 26, 1575-1580.

Zheng, Y., Shi, Y., Tian, C., Jiang, C., Jin, H., Chen, J., Almasan, A., Tang, H. andChen, Q. (2004). Essential role of the voltage-dependent anion channel (VDAC) inmitochondrial permeability transition pore opening and cytochrome c release inducedby arsenic trioxide. Oncogene 23, 1239-1247.

Journal of Cell Science 122 (11)1916

Jour

nal o

f Cel

l Sci

ence