Voltage-dependent Anion Channel 1-based Peptides Interact with Bcl-2 to Prevent Antiapoptotic Activity*

The antiapoptotic proteins of the Bcl-2 family are expressed at high levels in many types of cancer. However, the mechanism by which Bcl-2 family proteins regulate apoptosis is not fully understood. Here, we demonstrate the interaction of Bcl-2 with the outer mitochondrial membrane protein, voltage-dependent anion channel 1 (VDAC1). A direct interaction of Bcl-2 with bilayer-reconstituted purified VDAC was demonstrated, with Bcl-2 decreasing channel conductance. Expression of Bcl-2-GFP prevented apoptosis in cells expressing native but not certain VDAC1 mutants. VDAC1 sequences and amino acid residues important for interaction with Bcl-2 were defined through site-directed mutagenesis. Synthetic peptides corresponding to the VDAC1 N-terminal region and selected sequences bound specifically, in a concentration- and time-dependent manner, to immobilized Bcl-2, as revealed by the real-time surface plasmon resonance. Moreover, expression of the VDAC1-based peptides in cells over-expressing Bcl-2 prevented Bcl-2-mediated protection against staurosporine-induced apoptotic cell death. Similarly, a cell-permeable VDAC1-based synthetic peptide was also found to prevent Bcl-2-GFP-mediated protection against apoptosis. These results point to Bcl-2 as promoting tumor cell survival through binding to VDAC1, thereby inhibiting cytochrome c release and apoptotic cell death. Moreover, these findings suggest that interfering with the binding of Bcl-2 to mitochondria by VDAC1-based peptides may serve to potentiate the efficacy of conventional chemotherapeutic agents.

of the protein, exhibiting motion during voltage gating (37). Moreover, such mobility of the N-terminal VDAC1 ␣-helix may modulate the accessibility of apoptosis-regulating proteins of the Bcl-2 family (i.e. Bax and Bcl-x L ) to their binding sites on VDAC1 (38). Recently, we have demonstrated that the antiapoptotic proteins, hexokinase (HK) and Bcl-2, interact with the N-terminal region of VDAC1 to inhibit mitochondria-mediated apoptosis (16,19).
VDAC has been shown to interact with apoptosis-regulating proteins such as Bax/Bak and Bcl-x L (5, 39 -41), and with Bax and Bim (41,42). Cytochrome c release induced by Bax and Bim was found to be inhibited by anti-VDAC antibodies (43), whereas Bid (but not Bax) was shown to modify the conductance of VDAC channels (44). The interaction of Bcl-x L with VDAC1 was also demonstrated using NMR spectroscopy (18,29). Bcl-x L was further shown to modify the oligomerization state of VDAC1, as revealed by chemical cross-linking of micelle-bound VDAC1, shifting the equilibrium from the trimeric to the dimeric state (18,29,67). It has also been suggested that VDAC1 interacts with both Bax and Bcl-x L to form a tertiary complex and that Bcl-x L interacts with VDAC via a putative loop region of VDAC1 (38). In addition, Bcl-2 and Bcl-x L were proposed to interact with VDAC to block As 2 O 3 -induced VDAC dimerization (45). These results indicate that Bcl-2 family proteins regulate VDAC-mediated apoptosis and hence, the release of apoptogenic proteins from mitochondria.
In this study, through various experimental approaches, we demonstrate the interaction of Bcl-2 with VDAC and the subsequent modulation of apoptosis. Purified Bcl-2(⌬C23) reduced channel conductance of native but not mutated VDAC1 reconstituted into a planar lipid bilayer (PLB). Using site-directed mutagenesis, we identified VDAC1 domains that are involved in the interaction of the protein with Bcl-2 to confer protection against apoptosis. These VDAC1 domains were used as templates for creating VDAC1-based recombinant and synthetic peptides. The interaction of such peptides with Bcl-2 was demonstrated using surface plasmon resonance or when peptide expression prevented Bcl-2-mediated protection against cell death. These results thus offer new insight into the function of VDAC in mediating Bcl-2 antiapoptotic activity, actions that promote the survival of tumor cells.
Plasmids and Site-directed Mutagenesis-Five mVDAC1based peptide-encoding sequences (AP-W (Trp 63 -Asn 78 ), BP-W (Pro 105 -Lys 119 ), CP-W (Glu 117 -Thr 134 ), DP-W (Lys 199 -Asn 215 ), and the peptide corresponding to Met 1 -Gly 26 N-terminal residues) were generated by standard PCR and cloned into the BamHI/EcoRI sites of the tetracycline-inducible pcDNA4/TO vector (Invitrogen), as described previously (19). The four VDAC1 loop-shaped peptides, i.e. AP-W to DP-W, were designed to be flanked by a tryptophan zipper motif, namely the SWTWE amino acid sequence, at the N terminus of the peptide and the KWTWK sequence at the C terminus of the peptide. The presence of these sequences induces the formation of a stable ␤-hairpin. See supplemental data for additional information.
Cell Lines and Growth-T-REx-293 cells, corresponding to transformed primary human embryonal kidney cells expressing the tetracycline repressor (Invitrogen), were maintained in a humidified atmosphere at 37°C with 5% CO 2 . The cells were maintained in Dulbecco's modified Eagle's medium, supplemented with 10% fetal calf serum, 2 mM L-glutamine, 100 units/ml penicillin, 100 g ml streptomycin, and 5 g/ml blastocydin. hVDAC1-shRNA-T-REx-293 cells represent T-REx-293 cells silenced for the expression of endogenous human (h)VDAC1 using a shRNA-expressing vector (33).
Cell Transfection-Logarithmically growing T-REx-293 cells were transiently transfected with plasmid pEGFP-Bcl-2 alone or with a pcDNA4/TO plasmid encoding for one of the mVDAC1-peptides, or for native or mutated mVDAC1. The transfections were performed using the metafectene transfection reagent, according to the manufacturer's instructions or using calcium phosphate. Expression was induced for 48 -72 h in cells grown in the presence of tetracycline (1-1.5 g/ml). Cells were exposed to STS (1.25 M) for the times indicated in Figs. 3, 5, and 7.
VDAC Purification-VDAC was purified from sheep liver mitochondria after solubilization with N,N-lauryl-(dimethyl)amineoxide and purified using hydroxyapatite and methylcellulose columns chromatography as described previously (21). For recombinant VDAC1 purification, mVDAC1 and E72Q-mVDAC1 were expressed in the Saccharomyces cerevisiae
Reconstitution of Purified VDAC into a PLB-Single and multiple channel current recording and data analysis were carried out as described previously (47). Briefly, a PLB was prepared from soybean asolectin dissolved in n-decane (50 mg/ml). Purified VDAC (ϳ1-10 ng) was added to the chamber defined as the cis side containing 1 or 0.5 M NaCl. After one or a few channels were inserted into the PLB, excess protein was removed by prefusing the cis chamber with ϳ20 volumes of solution to prevent further incorporation. Following several recordings of channel activity at different voltages, purified Bcl-2(⌬C23) (ϳ1-2 g) was added, and currents through the channel were again recorded. Recordings were made under voltage clamp using a Bilayer Clamp BC-535B amplifier (Warner Instruments, Hamden, CT). Currents were measured with respect to the trans side of the membrane (ground). The currents were low pass-filtered at 1 kHz and digitized on-line using a Digidata 1440-interface board and Clampex 10.2 software (Axon Instruments, Union City, CA).
Real-time Surface Plasmon Resonance-Surface plasmon resonance (SPR), using the ProteOn-XPR36 system (Bio-Rad) was employed to study the interactions of VDAC1-based synthetic peptides with purified Bcl-2(C⌬23). Purified Bcl-2(C⌬23) and rabbit IgG were immobilized onto a GLM sensor surface, according to the manufacturer's instructions. The peptides were diluted in running buffer (150 mM NaCl, 0.005% Tween 20, 4% (v/v) dimethyl sulfoxide, phosphate-buffered saline, pH 7.4) and injected onto the sensor chip at varying concentrations, at a flow rate of 40 l/min. All experiments were carried out at 25°C. Responses (resonance units) were monitored using the ProteOn imaging system and related software tools. Signals were normalized using appropriate controls.
Cell Death Analysis-To determine the extent of apoptosis, cells were subjected to staining with acridine orange and ethidium bromide, as described previously (46). Cells were visualized by fluorescence microscopy (Olympus IX51), and images were recorded with an Olympus DP70 camera, using a superwide band filter. In each independent experiment, ϳ300 live, early, and late apoptotic cells were counted. Apoptosis was also analyzed by flow cytometric analysis using propidium iodide and Annexin V-FITC. Cells (2-4 ϫ 10 6 ) were exposed to an apoptosis inducer, collected (1,500 ϫ g for 5 min), washed twice with phosphate-buffered saline and resuspended in 400 l of binding buffer (10 mM Hepes/NaOH, pH 7.4, 140 mM NaCl, and 2.5 mM CaCl 2 ). Annexin V-FITC was added to a final concentration of 4.5 g/ml, and the cells were incubated in the dark for 15 min. The cells were then washed twice and resuspended in 400 l of binding buffer, to which propidium iodide was added immediately before analysis using ModFIT-lt2.0 software (FACScan, Beckton-Dickinson, San Jose, CA).

DP-Antp Treatment of Bcl-2-overexpressing Cells-T-REx-293 cells were transiently transfected with plasmid pcDNA3.1
Bcl-2 or the control plasmid, pcDNA3.1. Control or 48 h posttransfection cells were incubated for 90 min with the DP-Antp (Antennapedia homeodomain protein) peptide at a final concentration of 10 M in serum-free Dulbecco's modified Eagle's medium and then exposed to STS (1.25 M, 3 h). Cells were analyzed for cell death by FACS, using Annexin V-FITC and propidium iodide.

RESULTS
Members of the Bcl-2 family of proteins, such as Bcl-2, Bclx L , and Bax, have been proposed to interact with VDAC (5, 16 -19). In this study, we addressed the interaction of Bcl-2 with VDAC using several approaches, including site-directed mutagenesis, modulation of VDAC channel activity, through the use of synthetic and recombinant VDAC1-based peptides, and by assessing the modulation of Bcl-2 antiapoptotic activity by such peptides.
Bcl-2 Interacts with VDAC to Induce Channel Closure-First, we demonstrated that purified recombinant Bcl-2 interacts directly with purified planar lipid bilayer-reconstituted VDAC VDAC1-based Peptides Interact with Bcl-2 FEBRUARY 26, 2010 • VOLUME 285 • NUMBER 9 JOURNAL OF BIOLOGICAL CHEMISTRY 6055 to reduce channel conductance (Fig. 1). Recombinant Bcl-2(⌬C23) was expressed and purified from E. coli BL21 cells using a Hi-trap nickel column (Fig. 1D). Purified mitochondrial VDAC (Fig. 1D) was reconstituted into a PLB, and channel activity was studied under voltage-clamp conditions (47,51). We found that the interaction of Bcl-2(⌬C23) with VDAC requires exposure of VDAC to high voltages, as demonstrated in the following experiment (Fig. 1A). When reconstituted VDAC was exposed to Bcl-2(⌬C23) at a holding potential of Ϫ10 mV, no effect on VDAC channel activity was observed (even after a 27-min incubation). However, when the voltage was stepped to Ϫ60 mV for 2 min, and channel conductance was recorded 3 min later at Ϫ10 mV, a decrease in channel conductance was observed, with the channel being stabilized in a low conducting state (Fig. 1A). Such voltage dependence suggests that Bcl-2 interacts with a high voltage-stabilized VDAC conformation. Thus, in all experiments involving the Bcl-2 interaction with bilayer-reconstituted VDAC, a 2-min expo-sure to high voltage was applied before current recording. Upon addition of purified Bcl-2(⌬C23) to bilayer-reconstituted VDAC, the current produced in response to voltage steps from a holding potential of 0 mV to Ϫ10 mV was recorded before and 10 -15 min after the addition of Bcl-2(⌬C23). Channel conductance was reduced by Bcl-2(⌬C23), and the channel was stabilized in a low conducting state (Fig. 1B). Moreover, removing Bcl-2(⌬C23) upon replacement of bath solution with Bcl-2(⌬C23)-free solution restored channel conductance, as demonstrated in single channel (Fig. 1B) or multichannel (Fig. 1C) recordings. Bcl-2(⌬C23)based modification of VDAC conductance was only observed when Bcl-2(⌬C23) was added to the cis side of the bilayer in which VDAC was reconstituted, indicating that Bcl-2(⌬C23) interacts with the cytosolic face of VDAC.
Bcl-2(⌬C23) decreased VDAC conductance in multichannel experiments at all voltages tested and stabilized VDAC at a low conductance state, regardless of the voltage gradient applied. Channel conductance was restored close to the level noted before Bcl-2(⌬C23) addition upon removal of the added protein (Fig. 1C). These results suggest that Bcl-2 interacts reversibly and in a voltage-dependent manner with VDAC to modify VDAC conductance.
Mapping the Bcl-2-binding Site of VDAC1-To identify VDAC1 amino acid residues and domains involved in the interaction with the Bcl-2 protein, site-directed mutagenesis of VDAC1 and analysis of the antiapoptotic effect of Bcl-2 on cells expressing the mutations was carried out. The VDAC1 mutations E65Q, E72Q, and E202Q were selected based on their roles of these residues in preventing hexokinase isoforms, HK-I-and HK-II-mediated protection against apoptosis (16,19,67). The location of the selected mutated amino acids in both the previous (20) and newly proposed VDAC1 topology models (28 -30) is presented in Fig. 2. The effects of these mVDAC1 mutations on Bcl-2-mediated protection against cell death were then analyzed by expression of native or mutated mVDAC1 in hVDAC1-shRNA T-REx-293 cells, where the endogenous VDAC1 level was suppressed (by ϳ 85%) or in the same cells overexpressing Bcl-2 (Fig. 3).
hVDAC1-shRNA-T-REx-293 cells were transfected to express native, E65Q-, E72Q-, or E202Q-mVDAC1 under the FIGURE 1. Bcl-2(⌬C23) reduces bilayer-reconstituted VDAC conductance. A, VDAC was reconstituted into a PLB and current in response to a voltage step from 0 to Ϫ10 mV was recorded 2 and 25 min after the addition of purified Bcl-2(⌬C23) (1 nM). Thereafter, the voltage was switched for 2 to Ϫ60 mV, and current was recorded at Ϫ10 mV (total incubation time with Bcl-2(⌬C23) was 32 min). B, VDAC was reconstituted into a PLB, and currents through VDAC in response to a voltage step from 0 to Ϫ10 mV were recorded before and 15 min after the addition of purified Bcl-2(⌬C23) (1 nM) and after exposing the protein to high negative voltage for 2-3 min. Following current recording, the cis compartment was perfused to highly decrease the concentration of Bcl-2(⌬C23) by severalfold. The dashed lines indicate the zero current levels. C, shown are multichannel recordings as a function of the voltage (4 s at each voltage), and the average steady-state conductance of mVDAC1 before (F) and 30 min after (E) the addition of Bcl-2(⌬C23) (1 nM) and after exchanging the solution with Bcl-2(⌬C23)free medium (‚) are presented. Relative conductance was determined as the ratio of the conductance at a given voltage (G) to the maximal conductance (Go). D, a Coomassie-stained purified VDAC and Bcl-2 proteins are shown.

VDAC1-based Peptides Interact with Bcl-2
control of tetracycline, alone or together with Bcl-2-GFP (Fig.  3). STS induced apoptosis in ϳ65% of the control cells. This was almost completely prevented upon expression of Bcl-2-GFP. Protection against STS-induced apoptosis by Bcl-2-GFP was observed only in cells expressing native or E65Q-mVDAC1 but not in cells expressing E72Q-or E202Q-mVDAC1 (Fig. 3A). These results indicate that Bcl-2 interacts with native mVDAC1 but not with certain mVDAC1 mutations.
Next, the interaction of Bcl-2 with mutated E72Q-mVDAC1 was addressed using bilayer-reconstituted recombinant native or mutated mVDAC1. Native or E72Q-mVDAC1 were expressed in porin-less yeast, purified from their mitochondria, and reconstituted into a PLB (Fig. 4). The currents produced in response to voltages stepped from a holding potential of 0 mV to Ϫ10 were recorded before and 5 min after the addition of Bcl-2(⌬C23) (Fig. 4A). Upon addition of Bcl-2(⌬C23) to native mVDAC1, channel conductance was reduced, and the channel was stabilized in a low conducting state. On the other hand, Bcl-2(⌬C23) addition had no effect on the conductance of E72Q-mVDAC1 assayed at the single channel (Fig. 4B) or mul-tichannel levels at all voltages tested (Ϫ60 to ϩ60 mV) (Fig. 4C). These results suggest that Glu 72 is essential for the interaction of VDAC1 with Bcl-2(⌬C23).

VDAC1-based Peptides Prevent Bcl-2-mediated Protection against
Apoptosis-Next, based on the results obtained with mutated VDAC1 (Figs. 3 and 4), we constructed and expressed several VDAC1-based peptides and tested their effects on the antiapoptotic effects of Bcl-2 (Fig. 5). VDAC1based peptides were expressed alone or together with Bcl-2, and apoptosis, as induced by STS, was analyzed in the various transfected cells (Fig. 5). Bcl-2 overexpression prevented apoptotic cell death as induced by STS. When cells expressing Bcl-2 were also transfected to express VDAC1-based peptides, designed according to the previously proposed VDAC1 topology model (19) (see Fig. 2), it was observed that the N-terminal, PA-W, PB-W, or PD-W but not the PC-W peptide prevented the protection (25-70%) afforded by Bcl-2 against STS-induced cell death (Fig. 5A).
Bcl-2 Interaction with VDAC1-based Peptides-Based on the results obtained with mutated VDAC1 (Figs. 3 and 4) and the expressed peptides (Fig. 5), we generated the N-terminal, PA, PB, PD, and PC peptide sequences as synthetic peptides and analyzed their direct interactions with purified Bcl-2(⌬C23) using SPR technology (Fig. 6). Purified Bcl-2(⌬C23) was coupled to a SPR biosensor, the ProteOn GLM chip, and increasing concentrations (50 -400 M) of the VDAC1-based peptides were injected onto the sensor chips. Their binding to the immobilized Bcl-2(⌬C23) was then monitored. The N-terminal peptide and peptides PB and PD strongly bound to immobilized

VDAC1-based Peptides Interact with Bcl-2
Bcl-2(⌬C23) in a concentration-and time-dependent manner (Fig. 6, B, D, and E). In contrast, the two other VDAC1-based peptides, i.e. peptides PA and PC, did not interact with Bcl-2(⌬C23) (Fig. 6, A and C). The binding of the PB, PD, and N-terminal peptides to Bcl-2(⌬C23) was specific, as no signal was obtained with a IgG-immobilized control chip (Fig. 6F). All Bcl-2-interacting VDAC1-based peptides associated relatively strongly with the immobilized Bcl-2. However, whereas the PB and N-terminal peptides showed rapid dissociation, the PD peptide showed slow dissociation from the immobilized Bcl-2 (⌬C23). The results thus demonstrate the direct and specific interaction of selected VDAC1-based peptides with Bcl-2(⌬C23).
For further confirmation of those VDAC1 sequences that interact with Bcl-2, a cellulose-bound peptide array consisting of overlapping peptides derived from human VDAC1 was used. The immobilized peptide array was exposed to purified Bcl-2(⌬C23), and interactions were revealed by immunostaining using anti-Bcl-2 antibodies. Interactions of Bcl-2(⌬C23) with three peptides corresponding to the N-terminal region and to sequences overlapping with L14-15 and L16-17 were noted (see Fig. 2), with the strongest interaction being with the N-terminal region peptide (Fig. 7A). These results are in correlation with the findings attained using SPR technology (Fig. 6).
Next, to allow passive penetration of PD (80% overlap with the L14-15 sequence) into the cell, we synthesized the PD-Antp peptide in which PD was flanked by a tryptophan zipper motif (see "Materials and Methods") and attached at the N terminus to a 16-amino acid sequence representing the Antennapedia peptide Antp. Antp is a well known, nontoxic cell-penetrating peptide, which is able to facilitate the translocation of fused peptides across cell membrane (48,49,52,53). STS treatment induced cell death in control but not in cells overexpressing Bcl-2-GFP. However, Bcl-2-mediated protection against STSinduced apoptosis was significantly reduced upon incubation with the PD-Antp peptide (Fig. 7, B and C). Although Bcl-2-GFP protected against STS-induced apoptosis by about 95% in cells exposed to the PD-Antp peptide, protection was reduced by 50% in the presence of the PD-Antp peptide.

DISCUSSION
A number of studies have elucidated the roles of pro-and antiapoptotic Bcl-2 family members in tumor pathogenesis, as well as expanding our understanding of how Bcl-2-like proteins maintain or perturb mitochondrial integrity (54). Moreover, high levels of antiapoptotic Bcl-2 family members are associated with resistance of many tumors to chemotherapy (6,12,13). Based on these and other studies, antiapoptotic Bcl-2 family members offer an attractive but challenging target for the development of anti-cancer agents. Indeed, numerous attempts have been made toward developing rational design-based anti-cancer therapies that directly target Bcl-2-regulated events at the level of mitochondria (55)(56)(57)(58). However, while interfering with the prosurvival functions of Bcl-2 has long been considered an attractive manner to kill tumor cells, Bcl-2 proteins were simply deemed nondeliverable by pharmacological means. On the other hand, antisense-based strategies to reduce expression of Bcl-2 or Bcl-x L , such as Genasense (G3139/ Oblimersen), an 18-mer phosphorothioate oligonucleotide targeted to Bcl-2 mRNA, were found to effectively inhibit the proliferation of

VDAC1-based Peptides Interact with Bcl-2
human lymphoma cells (55,57,58). However, these antisense compounds were also reported to trigger inflammatory responses (57) and can only target one, or at best, two highly homologues Bcl-2 proteins at a time. Certain green tea catechins and black tea theaflavins were reported to be potent inhibitors of Bcl-x L and Bcl-2 (59). Other Bcl-2 inhibitors, known as "BH3 mimetics," are nonpeptidic compounds rationally designed on the basis of protein-protein interactions between anti-and proapoptotic proteins of the Bcl-2 family. Such a putative BH3 mimetic, TW-37, was designed based on the structure of gossypol. A Bim-derived BH3 peptide (TW-37) bound to Bcl-2 was found to displace Bid from its binding to Bcl-2, Bcl-x L , and Mcl-1 and is proposed to kill in a Bak/Bax-dependent manner (56). The peptide was, however, also found to induce activation of the proapoptotic functions of p53 in melanoma cell lines (60). TW-37 was found to disrupt heterodimer formation between Bax or truncated-Bid and antiapoptotic proteins in the order Mcl-1 Ͼ Bcl-2 Ͼ Ͼ Bclx L ) and proposed to be administrated together with standard chemotherapy as an effective strategy in the treatment of B-cell lymphoma (61).
These various approaches target the Bcl-2 family of proteins and either affect their expression levels or interactions between them. In this study and others (5, 16 -19), we have reported another strategy for modifying the antiapoptotic activity of Bcl-2, namely interfering with the interaction of Bcl-2 with VDAC1. Initially, the interaction of Bcl-2 with VDAC1 was confirmed using several approaches. Purified Bcl-2 directly interacted with purified VDAC, as reflected in the decrease in conductance of bilayer-reconstituted native but not mutated VDAC1 in the presence of purified Bcl-2(⌬C23) (Figs. 1 and 4). Bcl-2 also interacted with immobilized VDAC1-derived peptides (Fig. 7A) (and VDAC1-based peptides interacted with immobilized Bcl-2), as revealed by surface plasmon resonance technology (Fig. 6). In addition, a VDAC1-based peptide linked to a cell-penetrating peptide derived from Antp, added to facilitate translocation across the cell membrane, was found to attenuate Bcl-mediated protection against STS-induced cell death (Fig. 7, B and C).
Based on the results obtained with site-directed mutations of VDAC1 eliminating protection against apoptosis induced by Bcl-2 (Fig. 3) and VDAC1-based peptide expression preventing Bcl-2 antiapoptotic activity (Figs. 5 and 7), possible interaction sites of VDAC1 with Bcl-2 were localized to sequences represented by peptides. These corresponding to residues Trp 209 -Gln 224 , located in ␤-strand 15, and residues Leu 241 -Gly 253 , located in ␤-strand 17, as well as the VDAC1 N-terminal domain (see Fig. 7 and Ref. 16). Interestingly, the peptide corresponding to Leu 241 -Gly 253 represents VDAC1 sequences within ␤-strand 17, which, according to both the previous (20) and the revised (28 -30) VDAC1 topology models, span the membrane (see Fig. 2). Because Bcl-2 is associated with the membrane and is suggested to be inserted into the membrane   FEBRUARY 26, 2010 • VOLUME 285 • NUMBER 9

VDAC1-based Peptides Interact with Bcl-2
via its hydrophobic C-terminal domain (4), it is possible that Bcl-2 interacts with membrane-embedded VDAC1 sequences within ␤-strands 15-17. When studying the interaction of Bcl-2(⌬C23), which does not include the transmembrane domain, with VDAC1 or VDAC1-based peptides, the only possible interactions with VDAC1 would involve those channel amino acid residues facing the cytosolic surface. Thus, Bcl-2 interaction involves membranal and cytosolic VDAC1 sequences.
This study also explored the N-terminal region of VDAC1 as the interaction site of Bcl-2 (Figs. 5, 6, and 7C). This is of considerable of interest, given recent VDAC1 structural studies proposing that the N-terminal region of VDAC1 (residues 1-23) is nestled within the pore (28 -30). These and other studies (62) further indicate that only part of the N-terminal is in the form of an ␣-helix. By contract, other approaches point to the N-terminal ␣-helix as being exposed to the cytoplasm (32), crossing the membrane (20), or lying on the membrane surface (36). Further evidence suggests that the N-terminal region is involved in channel voltage gating (63,64) and exhibits motion during voltage changes (37). Thus, the N-terminal of VDAC1 might adopt different conformations and locations with respect to the VDAC1 ␤-strands and pore, depending on the apoptosis signaling conditions. This proposal is supported by the finding that exposure of bilayer-reconstituted VDAC to high voltages is required for Bcl-2-mediated modification of channel conductance. This suggest that the mobility of the VDAC1 N-terminal ␣-helix may modulate the accessibility of other VDAC1 domains for interaction with Bcl-2 ( Fig. 1) and/or other apoptosis-regulating proteins of the Bcl-2 family (i.e. Bax and Bcl-x L ) (16,38). In addition, we have shown that the N-terminal region of VDAC1 is required for apoptosis induction, suggesting the it adopts different conformations in response to apoptotic signals (16). Our results demonstrating Bcl-2 interaction with the N-terminal region of VDAC1 (Figs. 5, 6, and 7C) suggest that such an interaction would prevent the N-terminal domain of VDAC1 from adapting an apoptosis-induced position, thus inhibiting induction of apoptosis.
We have also demonstrated that certain mutations in VDAC1, some which are localized to the membranal domain of a ␤-strand, such as E72Q (according to the new model, see Fig. 2), diminished Bcl-2 binding to VDAC1 (Fig. 4), whereas other mutations prevented the antiapoptotic effects of Bcl-2 (Fig. 3). This may be explained by the location of these amino acid residues in the Bcl-2-binding site, involving the membrane-inserted sequence of the protein. However, these mutations may modify the Bcl-2recognized VDAC1 conformation. Indeed, NMR studies showed that the N-terminal domain of VDAC1 and the first four strands of the ␤-barrel are not stable in detergent and that such structural instability might be influenced by Glu 73 (i.e. Glu 72 , when not counting the initial methionine) (28). Using hydrogen/deuterium exchange measurements coupled to NMR spectroscopy, it was demonstrated that the ␤-barrel becomes more stable by substitution of Glu 73 with valine (28). Substitution of Glu 73 with glutamine had a marked effect on VDAC1, abolishing ruthenium red-and HK I-mediated inhibition of VDAC1 channel activity and protection against apoptosis (19,46,65). Because the E72Q mutation affects ruthenium red, HK-I, HK-II, as well as Bcl-2-mediated protection against apoptosis, it is most likely that E72Q-VDAC1 adapts a conformation not recognized by these agents.
In summary, because high levels of antiapoptotic Bcl-2 family members are associated with resistance of many tumors to chemotherapy (6,12,13), targeting these proteins presents a possible strategy for cancer therapy. Solution of the structure of antiapoptotic Bcl-2 family member proteins has led to the design of novel small molecule inhibitors. Although many such  (n ϭ 3). C, cellulose-bound peptide arrays consisting of overlapping peptides derived from VDAC1 sequence were incubated overnight with recombinant Bcl-2(⌬C23) (30 g/ml) and blotted with anti-Bcl-2 antibodies, followed by incubation with HRP-conjugated anti-mouse IgG and detected using a chemiluminescence kit. Dark spots represent binding of Bcl-2(⌬C23) to the specific peptide corresponding to the N-terminal domain (N-Ter), and to peptide sequences overlapping with the L14-15 and L16-17 sequences (see Fig. 2). A representative blot of three similar experiments is shown.

VDAC1-based Peptides Interact with Bcl-2
molecules have been synthesized, rigorous verification of their specificity is often lacking (66).
Further studies have revealed that many putative Bcl-2 inhibitors are not specific and have other cellular targets, resulting in nonmechanism-based toxicity. Thus, based on the results obtained with site-directed mutations affecting the antiapoptotic effects of Bcl-2, using VDAC1-based peptides interacting with Bcl-2 and preventing the antiapoptotic activity of Bcl-2 and by considering the interaction of VDAC1-based peptides with purified Bcl-2, we suggest the use of such VDAC1-based peptides as a novel strategy for overcoming apoptosis resistance by interfering with the ability of Bcl-2 to prevent apoptosis and promote survival of tumor cells. Targeting of VDAC1-based peptide to tumor cells overexpressing antiapoptotic proteins, such as Bcl-2, would minimize the self-defense mechanisms of the cancer cells, thereby promoting apoptosis and increasing sensitivity to chemotherapy. Thus, VDAC1-based peptides offer an attractive target for further development as anti-cancer agents.