Selective induction of cancer cell death by VDAC1‐based peptides and their potential use in cancer therapy

Mitochondrial VDAC1 mediates cross talk between the mitochondria and other parts of the cell by transporting anions, cations, ATP, Ca2+, and metabolites and serves as a key player in apoptosis. As such, VDAC1 is involved in two important hallmarks of cancer development, namely energy and metabolic reprograming and apoptotic cell death evasion. We previously developed cell‐penetrating VDAC1‐derived peptides that interact with hexokinase (HK), Bcl‐2, and Bcl‐xL to prevent the anti‐apoptotic activities of these proteins and induce cancer cell death, with a focus on leukemia and glioblastoma. In this study, we demonstrated the sensitivity of a panel of genetically characterized cancer cell lines, differing in origin and carried mutations, to VDAC1‐based peptide‐induced apoptosis. Noncancerous cell lines were less affected by the peptides. Furthermore, we constructed additional VDAC1‐based peptides with the aim of improving targeting, selectivity, and cellular stability, including R‐Tf‐D‐LP4, containing the transferrin receptor internalization sequence (Tf) that allows targeting of the peptide to cancer cells, known to overexpress the transferrin receptor. The mode of action of the VDAC1‐based peptides involves HK detachment, interfering with the action of anti‐apoptotic proteins, and thus activating multiple routes leading to an impairment of cell energy and metabolism homeostasis and the induction of apoptosis. Finally, in xenograft glioblastoma, lung, and breast cancer mouse models, R‐Tf‐D‐LP4 inhibited tumor growth while inducing massive cancer cell death, including of cancer stem cells. Thus, VDAC1‐based peptides offer an innovative new conceptual framework for cancer therapy.


Introduction
Cancer cells share a common set of hallmarks, including unlimited proliferation potential, self-sufficiency in terms of growth signals, reprogramed metabolism, and avoiding apoptosis by alterations in the expression levels of pro-and anti-apoptotic proteins, as well as due to reduced caspase function and impaired death receptor signaling (Antonsson and Martinou, 2000).
Overexpression of anti-apoptotic proteins, such as Bcl-2 and Bcl-xL, has been demonstrated in numerous cancers, including colon, thyroid, breast, and endometrial cancer. Moreover, Bcl-2 expression is correlated with the degree of aggressiveness and resistance to chemotherapy-induced apoptosis (Llambi and Green, 2011).
Mitochondria occupy a central position in cell life and death, with mitochondrial bioenergetics, biosynthesis, and signaling being critical for tumorigenesis. Multistep mitochondria-mediated apoptosis that relies on the intrinsic pathway can be initiated by a variety of stimuli involving outer mitochondrial membrane (OMM) permeabilization, leading to the release of apoptotic proteins, such as cytochrome c (Cyto c) and apoptosis-inducing factor (AIF). This results in the activation of a cascade of cysteine proteases (caspases), subsequently leading to organized cell demise (Kroemer et al., 2007).
In addition to its role in apoptosis, VDAC1 plays a crucial role in regulating the metabolic and energetic functions of the cell. VDAC1 assumes a crucial position at the OMM, serving as the main interface between mitochondrial and cellular metabolisms, mediating the transport of anions, cations including Ca 2+ , ATP, and many metabolites between the mitochondria and the cytosol (Shoshan- . Moreover, its location at the boundary between the mitochondria and the cytosol enables VDAC1 to function as an anchoring site for a diverse set of cytosolic proteins, such as HK, tubulin, actin, and several dozen other proteins (Shoshan-Barmatz et al., 2017c), that mediate metabolic and signaling cross talk between cytosol and mitochondria and regulate the integration of mitochondrial functions with other cellular activities , 2017b.
The ability of cancer cells to evade chemotherapyinduced apoptosis is a major cause of treatment failure and aggressive tumorigenesis. Overexpression of antiapoptotic proteins, such as Bcl-2 and Bcl-xL (Llambi and Green, 2011), and downregulation of apoptosispromoting factors, such as Bax and Fas (McCurrach et al., 1997), have been detected in many primary tumors and tumor cell lines, with such alterations having been linked to chemotherapeutic resistance.
At the same time, a small population of cells within a tumor, known as cancer stem cells (CSCs), are resistant to traditional cancer treatment (Weiswald et al., 2015;Zhang et al., 2017). The CSCs hypothesis postulates that a subpopulation of malignant cells constantly supplies the tumor with cancerous cells. These CSCs possess unique markers, a gene expression profile like that of embryonic and somatic stem cells, selfrenewal, and multipotent differentiation abilities (Frank et al., 2010;Lee and Herlyn, 2007). However, in contrast to normal stem cells, CSCs are not regulated, such that their continuous expansion allows them to maintain a tumor. CSCs are, moreover, associated with drug resistance and cancer recurrence. CSCs have been identified in a number of solid tumors, such as breast, lung, and colon cancers, brain tumors, and melanoma (Dawood et al., 2014).
Although selective induction of cancer cell death remains a major concern and a challenge for the development of new anticancer therapies, treatments designed to activate the apoptotic machinery and abolish the ability of tumors to evade apoptosis are still considered to be most effective (Fulda et al., 2010).
Rapidly growing tumor cells typically display altered aerobic glycolysis (the Warburg effect), with this metabolic reprograming being directly related to tumor growth and progression (Vander Heiden et al., 2009). The Warburg effect can be described as a metabolic phenotype characterized by enhanced glycolysis and suppression of mitochondrial metabolism at any level of oxygen. However, although enhanced glycolysis is a prominent feature of most tumor cells, cancer cells in fact use both glycolysis and oxidative phosphorylation (OXPHOS), with the degree each strategy is used being dependent on the prevailing normoxic or hypoxic environmental conditions and the capacity of such cells to express adequate levels of oncogenes and tumor suppressor gene products for cell growth (Majeed et al., 2012). In addition, rapidly growing tumors meet their metabolic demands by increased expression of genes encoding glucose transporters and glycolytic enzymes, including HK. As a key mediator of aerobic glycolysis, HK catalyzes the first and rate-limiting step in glucose metabolism, that is, the conversion of glucose to glucose 6-phosphate (G-6-P). HK thus lies at the apex of the glycolytic pathway that provides the metabolic intermediates required by those biosynthetic pathways on which a transformed cell makes heavy demand (Mathupala et al., 2010;Pedersen, 2007). By binding to VDAC1 at the OMM, HK gains direct access to mitochondrially produced ATP. The interaction of HK with VDAC1 protects against apoptosis, with this protection being abolishing upon dissociation of the HK-VDAC1 complex (Azoulay-Zohar et al., 2004;Pastorino et al., 2002;Shoshan-Barmatz et al., 2017a). Indeed, disrupting the VDAC1-HK interaction is a potential strategy for cancer treatment (Cao et al., 2008;Fulda et al., 2010;Galluzzi et al., 2008;Goldin et al., 2008). In addition, dissociation of HK from VDAC1 regulates Bax translocation to the mitochondria (Pastorino et al., 2002).
In our previous studies, key sequences and amino acid residues of VDAC1 required for Bcl-2, Bcl-xL, and HK binding and protection against cell death were identified Arbel and Shoshan-Barmatz, 2010;Arzoine et al., 2009;Shi et al., 2003;Shoshan-Barmatz et al., 2009). On the basis of these findings, we designed and tested 27 versions of VDAC1-based cell-penetrating peptides representing two different structural regions of VDAC1 to identify the most stable short apoptosis-inducing peptides (Prezma et al., 2013). These cell-penetrating peptides included the D-D(1-14)N-Ter-Antp peptide, composed of the VDAC1 N-terminal region (15-26 amino acids) fused to Antp (Penetrating), a 16-residuelong sequence from the Drosophila Antennapedia homeodomain, both containing amino acids in the Dconfiguration, and Tf-D-LP4, comprising a VDAC1derived cytosol-facing loop sequence, defined as LP4, fused to a human transferrin receptor (hTfR)-recognition sequence, HAIYPRH (Tf) (Daniels et al., 2012), with only the amino acids of the VDAC1-derived sequence being in D-configuration. hTfR is highly expressed in many cancers (Daniels et al., 2012). Additional versions of these VDAC1-based peptides are presented in this study. These VDAC1-based peptides serve as a decoy, targeting the anti-apoptotic proteins, and thereby inhibiting the protection against apoptosis. As such, they induce cell death in cancerous but not in noncancerous cells (Abu-Hamad et al., 2008Arbel and Shoshan-Barmatz, 2010;Arbel et al., 2012;Arzoine et al., 2009;Shi et al., 2003;Shoshan-Barmatz et al., 2009;Zaid et al., 2005).
In this work, we demonstrate that VDAC1-based peptides are able to selectively induce apoptotic cell death in a large panel of cancer cell lines, regardless of their cancer origin or the mutations carried, including mutated p53. Furthermore, we explored the mode of action of the peptides and report on the triple mode of action, namely impairing cell energy homeostasis, prevention of the protective effects of anti-apoptotic proteins, and induction of massive apoptosis. Finally, in subcutaneous glioblastoma (U-87MG), lung cancer (A549), and breast cancer (MDA-MB-231) cell-derived xenograft mouse models, a VDAC1-based peptide inhibited tumor development by inducing massive apoptotic cell death involving enhanced expression of pro-apoptotic proteins, reversal of the metabolic reprograming of the tumor cells, and elimination of cancer stem cells (CSCs).

Peptide synthesis and solution preparation
Custom VDAC1-based peptides were synthesized by GL Biochem (Shanghai, China) to a level of > 85% purity. The peptides were dissolved in 5% or 20% dimethyl sulfoxide and stored in aliquots at À20°C. Peptide concentrations were determined using absorbance at 280 nm and the specific molar excitation coefficient, as calculated based on amino acid composition. For all experiments, the final concentration of DMSO in control and peptide-containing samples was ≤ 0.5%. The major peptides used in this study, with the cell-penetrating sequence underlined, the amino acids added to form a loop-shaped tryptophan zipper in italics, and amino acids in D-configuration in bold, are listed as follows: Antp  Table S1.

Isolation of peripheral blood mononuclear cells (PBMCs)
PBMCs were isolated from the venous blood of healthy donors by Ficoll-Paque PLUS density gradient centrifugation. After informed consent, venous blood (10 mL) was drawn from normal adult donors. Blood was collected into heparin tubes and was diluted 1 : 1 with balance solution composed of two stock solutions, A (1% D-glucose, 50 mM CaCl 2 , 0.98 mM MgCl 2 , 5.4 mM KCl, and 0.145 M Tris/HCl, pH 7.6) and B (0.14 M NaCl) in a 1 : 9 ratio. The resulting mix was carefully layered onto Ficoll-Paque PLUS (GE Healthcare, Uppsala, Sweden) (10 mL of diluted blood to 15 mL Ficoll) in 50-mL conical tubes. The tubes were centrifuged at 400 g (with minimal acceleration and deceleration) at 18-20°C for 40 min. The fine layer of mononuclear cells was transferred to a new centrifuge tube, washed three times with balance solution, and resuspended in culture medium appropriate to the application. Cell viability was analyzed by assaying trypan blue (0.25%) exclusion, as counted with a Countess automated cell counter (Invitrogen).

Determination of cellular ATP and Ca 2+ levels
Cellular ATP levels were estimated using a luciferasebased assay (CellTiter-Glo; Promega). HeLa cells (3 9 10 5 mL À1 ) were incubated with the indicated concentrations of Tf-D-LP4 or R-Tf-D-LP4 peptides for 3 h, washed twice with PBS, and transferred to 96well white plates at densities of 1 9 10 5 cellsÁmL À1 . ATP levels were assayed according to the manufacturer's protocol, and luminescence was recorded using an Infinite M1000 plate reader (Tecan, M€ annedorf, Switzerland).

Cell treatment with VDAC1-based peptides and cell death analysis
Leukemia cells, U-937, MEC-1 cells (4 9 10 5 or 8 9 10 5 cells/sample, respectively), as well as other leukemia or suspension cell lines, were incubated in 200 lL serum-free medium with various concentrations of the peptide for 90 min at 23-25°C, collected by centrifugation (500 g, 5 min), washed with PBS, and analyzed for cell death. Adherent cells (6 9 10 5 mL À1 at 70-80% confluence) were incubated with the peptide of interest in 500 lL serum-free recommended culture medium for 3-6 h at 37°C and 5% CO 2 . The cells were then trypsinized, centrifuged (1500 g, 5 min), washed with PBS, and analyzed for the desired activity. Following peptide treatment, cells were subjected to cell death analysis. Propidium iodide (PI) staining was performed by addition of PI (6.25 lgÁmL À1 ) to the cells and immediate analysis by the EC800 Flow cytometer Analyzer- For PI and annexin V-FITC staining, A549 or MDA-MB-231 cells, untreated or treated with VDAC1based peptides, were collected (1500 g for 5 min), washed, and resuspended in 200 lL binding buffer (10 mM HEPES/NaOH (pH 7.4), 140 mM NaCl, and 2.5 mM CaCl 2 ). Annexin V-FITC staining was performed according to the recommended protocol. Cells were then washed once with binding buffer and resuspended in 200 lL binding buffer, to which PI was added immediately before flow cytometric analysis. At least 10 000 events were collected, recorded on a dot plot, and analyzed by flow cytometer. Acridine orange/ ethidium bromide staining was carried as described previously (Zaid et al., 2005), with cells visualized by fluorescence microscopy (Olympus LX2-KSP, Tokyo, Japan), and images were captured with a CCD camera. About 200 cells were counted for each experiment.

Mitochondria-bound HK detachment
Peptide-induced HK detachment from mitochondria to the cytosol was analyzed by following the cellular localization of HK-I-GFP before and after incubation with Tf-D-LP4 or D-Δ(1-14)N-Ter-Antp peptides. HeLa cells were grown on coverslips and transfected with plasmid pEGFP-HK-I. Twenty-four hours posttransfection, the cells were incubated for 3 h with a solution containing 0.07% DMSO or Tf-D-LP4 or D-Δ(1-14)N-Ter-Antp peptide to the indicated final concentration, fixed for 15 min with 4% paraformaldehyde in PBS, rinsed with PBS, permeabilized with 0.3% Triton X-100, and stained with DAPI. Cells imaging was carried out by confocal microscopy (Olympus 1X81).

Recombinant HK-I and II expression, purification, and activity assay
Human HK-I and HK-II were cloned into plasmid pET-His DNA and expressed in Escherichia coli BL21 upon IPTG induction. HK was purified from lysed cells using a HiTrap nickel column (GE Healthcare). Following dialysis, the imidazole-eluted protein was further purified using a Blue HiTrap column (GE Healthcare). HK was eluted using 10 mM glucose, 20 mM Tris/HCl (pH 8.5), 20% glycerol, and 1.5 mM glucose-6-phosphate. Fractions were analyzed by SDS/PAGE and tested for HK activity. The purified active fractions were combined and stored in aliquots at À80°C.

Xenograft mouse model
U-87MG cells (3 9 10 6 /mouse), A549 cells (5 9 10 6 / mouse), or MDA-MB-231 cells (3 9 10 6 /mouse) were inoculated subcutaneously (s.c.) into the hind leg flanks of athymic 8-week-old male nude mice (Envigo, Huntingdon, UK). Tumor size was measured using a digital caliper and volume was calculated, and when reached 50-100 mm 3 for each tumor type, mice were randomly divided into two groups (five mice per group). One group was intratumorally injected with HBSS (5.33 mM KCl, 0.44 mM KH 2 PO 4 , 138 mM NaCl, 4 mM NaHCO 3 , 0.3 mM Na 2 HPO4, and 5.6 mM glucose, pH 7.3) containing 0.05% DMSO and the second group with the peptide in HBSS containing 0.05% DMSO, with the R-Tf-D-LP4 peptide at a final concentration of 40 or 60 lM. The xenografts were injected (two points, 20 lL per tumor) every two days. Beginning on the day of inoculation, mouse weight and tumor volume were monitored twice a week. Experiments were terminated after 14, 29, or 20 days for U-87MG-, MDA-MB-231-, or A549-based tumors, respectively. The mice were sacrificed and tumors were excised. Half of each tumor was fixed in 4% buffered formaldehyde, paraffinembedded, and processed for IHC, while the second half was frozen in liquid nitrogen for immunoblotting and qPCR analyses. These experimental protocols were approved by the Institutional Animal Care and Use Committee of Ben-Gurion University.

Gel electrophoresis and immunoblotting
Cells or tumor tissues were lysed using lysis buffer (50 mM Tris/HCl (pH 7.5), 150 mM NaCl, 1 mM EDTA, 1.5 mM MgCl 2 , 10% glycerol, and 1% Triton X-100, supplemented with a protease inhibitor cocktail (Calbiochem, San Diego, CA, USA). The lysates were then centrifuged at 12 000 g (10 min at 4°C), and aliquots (10-40 lg of protein) were subjected to SDS/ PAGE and immunoblotting using various primary antibodies (sources and dilutions are provided in Table S2), followed by incubation with appropriate HRP-conjugated secondary antibodies (i.e., anti-mouse, anti-rabbit, or anti-goat). Blots were developed using enhanced chemiluminescence (Biological Industries). Band intensities were analyzed by densitometry using FUSION-FX (Vilber Lourmat) software, and the values were normalized to the intensities of the appropriate b-actin signal that served as a loading control.

IHC of tumor tissue sections
Formalin-fixed, paraffin-embedded sections of control and peptide-treated tumors (peptide-TTs) derived from U-87MG, A549, and MDA-MB-231 cells were hematoxylin-eosin-stained and probed with appropriate antibodies for IHC staining. IHC staining was performed on 5-lm-thick formalin-fixed and paraffin-embedded tumor tissue sections. The tissue sections were deparaffinized by placing the slides at 60°C for 1 h and using xylene, followed by rehydration with a graded ethanol series (100-50%). Antigen retrieval for some proteins (AIF, ATP synthase 5a, CD44, citrate synthase, Cyto c, cytochrome c oxidase subunit IVc, GAPDH, Glut-1, KLF4, Nestin, p53, S100b, SOX2, VDAC1) was performed in 0.01 M citrate buffer (pH 6.0). For HK-I and Ki-67, antigen retrieval was performed in 10 mM Tris/ EDTA (pH 9) or 0.5 M Tris/EDTA (pH 10) for 30 min at 95-98°C. Sections were washed with PBS containing 0.1% Triton X-100 (pH 7.4), and nonspecific antibody binding was reduced by incubating the sections in 10% normal goat serum for 2 h overnight at 4°C with primary antibodies (sources and dilutions used detailed in Table S2) and washing with PBST. Endogenous peroxidase activity was blocked by incubating the sections in 3% H 2 O 2 for 15 min. Sections were washed thoroughly with PBST, incubated with the appropriate secondary antibodies (Table S2) for 2 h, and washed five times in PBST, and the peroxidase reaction was subsequently visualized by incubating with 3,3-diaminobenzidine (DAB) (ImmPact-DAB). After rinsing in water, the sections were counterstained with hematoxylin and mounted with mounting medium. Finally, the sections were observed under a microscope (Leica DM2500, Allendale, NJ, USA) and images were collected at 203 magnification with the same light intensity and exposure time. Nonspecific control experiments were carried out using the same protocols but omitting incubation with the primary antibodies. Hematoxylin-eosin (H&E) staining was performed as described previously (Elsam, 2008).

TUNEL assay
Paraffin-embedded-fixed tumor sections were processed for a TUNEL assay using the DeadEnd Fluorometric TUNEL system according to the manufacturer's instructions. Sections were deparaffinized, equilibrated in PBS, permeabilized with proteinase K (20 lgÁmL À1 in PBS), postfixed in 4% paraformaldehyde, and incubated in TdT reaction mix for 1 h at 37°C in the dark. Slides were then washed in 29 saline-sodium citrate (SSC) buffer and counterstained with propidium iodide (1 lgÁmL À1 ), and coverslipped with Vectashield mounting medium (Vector Laboratories, Burlingame, CA, USA). Fluorescent images of apoptotic cells (green) and cell nuclei (red) were captured using a confocal microscope (Olympus 1X81).

RNA preparation and quantitative real-time PCR (qPCR)
Total RNA was isolated from untreated and peptidetreated tumors (three mice each) using an RNeasy Mini Kit (Qiagen, Hilden, Germany) according to the manufacturer's instructions. Total RNA quality was analyzed using the Agilent RNA 6000 Nano Kit. The obtained samples' RNA integrity values for total RNA were 8-10. qPCR was performed using specific primers (KiCqStart Primers; Sigma-Aldrich) in triplicate, using Power SYBR Green Master Mix (Applied Biosystems, Foster City, CA, USA). The levels of target genes were normalized relative to b-actin mRNA levels. Samples were amplified by a 7300 Real-Time PCR System (Applied Biosystems) for denaturation step of 5 min at 95°C and 40 cycles using the following PCR parameters: 95°C for 15 s, 60°C for 1 min, and 72°C for 1 min. The copy numbers for each sample were calculated by the CT-based calibrated standard curve method. The mean fold changes (AESE) of three replicates were calculated. The genes examined and primers used are listed in Table S3.

Results
In a previous study (Prezma et al., 2013), over 27 versions of cell-penetrating VDAC1-based peptides were designed and screened for apoptosis induction in chronic lymphocytic leukemia (CLL)-derived lymphocytes. Three of these peptides were tested in vivo in glioblastoma (Shteinfer-Kuzmine et al., 2017). In this study, we further optimized these cell-permeable VDAC1-based peptides, testing them against a panel of cancer cell lines differing in their origin and mutational status so as to explore the mode of action of selected peptides. We validated the activity of one such peptide on tumors established from three different cancer types in mouse models, demonstrating that the peptide not only inhibits tumor growth but also reverses unique properties of the cancer cells.

VDAC1-based peptides induce cancer cell death regardless of cancer type
Forty-one cancerous and five noncancerous cell lines were used in this study to examine the cell-killing activity of the VDAC1-based peptides Antp-LP4, Tf-LP4, Tf-D-LP4, R-Tf-D-LP4, N-Ter-Antp, and others (Figs 1-3 and Tables 1-3). For each line, cells were incubated with different peptide concentrations, and cell death was determined by PI staining and FACS analysis (Figs 1-3). In addition, for each cell line, the concentration of the peptide inducing 50% cell death (EC 50 ) was determined (Tables 1-3).
The results show that the cell death-inducing activity of the peptides was cell type-dependent. Accordingly, we subgrouped the cell lines into three groups according to their sensitivity to N-Ter-Antp (Fig. 1A) as highly sensitive (Group 1: EC 50 = 0.9-4.5 lM), sensitive (Group 2: EC 50 of 6.5-10 lM), and less sensitive (Group 3: EC 50 > 18 lM) (Table 1). Representative experiments assessing the cell death induction activity of the Antp-LP4 and N-Ter-Antp peptides are shown for the B-cell chronic lymphocytic leukemia (MEC-1) and histiocytic leukemia (U-937) cell lines, assigned to the highly sensitive group, where an EC 50 ≤ 3 lM was measured (Fig. 1D,E). Melanoma (A375) and lung cancer cell lines (A549 and H358), representative of the sensitive group, presented an EC 50 ≤ 6 lM ( Fig. 1B,C,F,G), while the PANC1 pancreatic cancer cell line, representative of the much less sensitive group, presented an EC 50 ≥ 18 lM (Fig. 1H).
Transformed noncancerous T-Rex-293 cells were assigned to the less sensitive group, and showed EC 50 values of 18 and 29 lM for N-Ter-Antp and Antp-LP4, respectively ( Fig. 1I and Table 1). Thus, for both the Antp-LP4 and N-Ter-Antp peptides, most of the cancer cell lines exhibited concentration-dependent cell death following a 6-h incubation, reaching a maximal level of cell death of over 90%. However, for the less sensitive group (Group 3), cell death was increased from about 10% in the absence of peptides to a   1086 steady-state level of about 50% in their presence (Fig. 1H,I).

Evaluation of cell death induction by modified VDAC1-based peptides
To increase VDAC1-based peptide stability and specific targeting to cancer cells, we designed new versions of the LP4-based peptide (Figs 2 and 3, and S1 and Tables 2 and 3). To protect the peptides from proteolytic degradation, non-native D-form amino acids were used to create the D-Antp-LP4 and D-N-Ter-Antp peptides. The selectivity of these peptides for cancer cells was addressed by employing the transferrin receptor (hTfR)-targeting peptide HAIYPRH (Tf) as a cell-penetrating sequence. Tf was identified by phage display and shown to internalize conjugated compounds into the cytoplasm via the human hTfR (Oh et al., 2009), overexpressed in many cancers (Daniels et al., 2012) (Fig. 3B). Other modifications included a shortening of the VDAC1-derived sequences from the C' or N' terminus ( Table 2).
The ability of the N-Ter-Antp, Antp-LP4, D-Antp-LP4, and Tf-LP4 peptides to induce cell death and discriminate between cancer and noncancer cells was studied with the PC-3 and DU-145 prostate cancer cell lines and the noncancerous prostate epithelium PWR-1E and T-Rex-293 cell lines (Fig. 2). The four tested peptides exhibited concentration-dependent cell death induction in the prostate cancer cell lines. Sensitivity to the peptides was remarkably reduced in the noncancerous PWR-1E and T-Rex-293 cell lines (Fig. 2). D-Antp-LP4 peptide was about fourfold more effective in inducing cell death of PC-3 and DU-145 cells than was the same peptide composed of L-amino acids (compare Fig. 2A,C to B,D). This may have been due to increased protease protection of the former version of the peptide.
Tf-LP4 was as active in inducing cell death as Antp-LP4, yet was less effective than the N-Ter-Antp peptide ( Fig. 2A,C,E-H). However, Tf-LP4 showed almost no cell death induction in the noncancerous PWR-1E and T-Rex-293 cell lines (Fig. 2E,G), pointing to its selectivity toward cancer cells. Similarly, N-Ter-Antp induced cell death in both the PC-3 and DU-145 cells but only to lesser extent in the noncancerous cell lines (Fig. 2F,H). The specificities of the various peptides toward prostate cancer cells are summarized in Fig. 2I.
To address changes in orientation due to the D-configuration of the amino acids, we also designed a retro-inverso peptide. Retro-inverso peptides are peptides in which the D-amino acid sequence of the peptide is reversed, such that the a-center chirality of the amino acids is also inverted. The reverse sequence helps maintain side-chain topology similar to that of the original L-amino acid peptide (Chorev, 2005;Holm et al., 2011;Liu et al., 2016).
To demonstrate the involvement of TfR in peptide cell penetration, we preincubated the cells with anti-TfR antibodies prior to cells incubation with the R-Tf-D-LP4 peptide (Fig. 3C). The results showed an about twofold decrease in Tf-D-LP4-induced cell death in anti-TfR antibody-treated cells (Fig. 3C), suggesting the requirement of TfR for Tf-D-LP4-mediated cell death.
Cell death induced by Tf-D-LP4 and R-Tf-D-LP4 was tested in several cell lines. In MDA-MB-231 and MCF-7 human breast cancer cells, R-Tf-D-LP4 was less effective than Tf-D-LP4 (Fig. 3D,E), and in GL-261MG mouse glioblastoma, they were similarly effective (Fig. 3F). In THP-1 human acute monocytic leukemia cells, R-Tf-D-LP4 was more effective than Tf-D-LP4 (Fig. 3G). The differences between the two versions of the peptide with respect to solubility and stability (Fig. S3) and, therefore, in their use in vivo will be described.
The results showed that these peptides induced cell death in all cancer cell lines tested (Figs 3D-G and S1), and to a lesser extent in noncancerous MDCK dog kidney epithelial cells (Fig. 3H) and in peripheral blood mononuclear cells (PBMCs) obtained from healthy individuals (Fig. 3I).

Characterization of the mode of action of VDAC1-based peptides leading to cell death
The mode of action of the VDAC1-based peptides from the second stage of optimization was examined as was that of the first versions of the peptides ( These efforts showed that in untreated cells, HK-I-GFP fluorescence was punctuated, indicating mitochondrial HK-I-GFP distribution, while in cells incubated with the Tf-D-LP4 or D-D(1-14)N-Ter-Antp peptides, the fluorescence was diffuse, reflecting displacement of HK from the mitochondria and its presence in the cytosol (Fig. 4D). HK detachment induced by the Tf-D-LP4 and R-Tf-D-LP4 peptides was also demonstrated by cell fractionation and immunoblotting, showing an about twofold increase in HK in the cytosolic fraction derived from peptide-treated cells (Fig. 4E).
The effects of the Tf-D-LP4 and R-Tf-D-LP4 peptides on cellular ATP and intracellular Ca 2+ ([Ca 2+ ]i) levels were also analyzed ( Fig. 4F-I). In peptide-treated cells, cellular ATP levels were decreased by about 40% (Fig. 4F), suggesting that mitochondrial energy Table 3. The Tf-D-LP4 and R-Tf-D-LP4 peptides induce cell death in various cancer cell lines. Summary of the half-maximal cell death activity (EC 50 ) values (lM) (calculated from the maximal extent of cell death obtained) for cell death induction by the Tf-D-LP4 and R-Tf-D-LP4 peptides in various cancer cell lines. Cell lines were incubated with different peptide concentrations for 90 min (for cells in suspension) or 6 h (for adherent cells). Cell death was determined by PI staining and FACS analysis. Data represent mean AE SE (n = 3-4). Mutation data taken from COSMIC (http://www.sanger.ac.uk/genetics/CGP/CellLines/). ND, not determined; NR, not reported.   production was impaired by the peptides. As apoptosis induction by various stimuli was shown to disrupt cellular Ca 2+ homeostasis (Florea and Busselberg, 2009;Keinan et al., 2013;Weisthal et al., 2014), the effect of the R-Tf-D-LP4 peptide on [Ca 2+ ]i was tested using Fluo-4 (Fig. 4G,I). R-Tf-D-LP4 increased [Ca 2+ ]i in a concentration-and time-dependent manner (Fig. 4G,  I). Moreover, a linear correlation between the increase in [Ca 2+ ]i (Fig. 4G) and cell death (Fig. 4H) was obtained (Fig. 4I). Next, we analyzed the effects of the VDAC1-based peptides on several apoptosis-associated features, such as VDAC1 oligomerization, Cyto c release, and apoptotic cell death (Fig. 5). Apoptosis induction by various apoptosis inducers has been shown to lead to VDAC1 oligomerization, regardless of the cell type or apoptosis inducer used (Keinan et al., , 2013. Similarly, cell treatment with the peptides led to VDAC1 oligomerization (Fig. 5A). Peptide-induced Cyto c release was demonstrated by immunofluorescence using anti-Cyto c antibodies and confocal microscopy (Fig. 5B). Representative images of control cells show punctuated fluorescence, suggesting mitochondrial distribution of Cyto c, and while in peptide-treated cells, the fluorescence was more diffuse, pointing to the release of Cyto c (Fig. 5B). Cell fractionation and immunoblotting analysis demonstrated Cyto c to be in the cytosolic fraction of only peptide-treated cells (Fig. 5C).
Finally, apoptotic cell death was analyzed using acridine orange/ethidium bromide staining, as well as annexin V-FITC and PI staining and FACS analysis (Figs 5D-G and S2). Cell incubation with the Tf-D-LP4 or R-Tf-D-LP4 peptides induced membrane blebbing, chromatin condensation, and orange staining of nuclei, reflecting a late apoptotic stage (Figs 5D and S2A). Quantitation of these results showed that the peptides induced over 70% apoptotic cell death in A549 cells (Fig. 5E). Apoptosis was also revealed by annexin V-FITC and PI staining, with representative FACS results and their quantitation showing that the R-Tf-D-LP4 peptide induced over 90% apoptotic cell death (Figs 5F,G and S2B,C).
These results indicate that the peptides induced VDAC1 oligomerization, Cyto c release, and subsequently apoptosis, as shown previously for other versions of VDAC1-based peptides in studies of CLL and GBM (Prezma et al., 2013;Shteinfer-Kuzmine et al., 2017).
Due to the advantageous properties of R-Tf-D-LP4 relative to Tf-D-LP4, including inhibition of purified HK-I activity (Fig. S3A,B), increased solubility (Fig. S3C), and more stability with respect to temperature and incubation time (Fig. S3D,E), the former was selected for in vivo experiments.

VDAC1-based peptides inhibit in vivo growth of glioblastoma, lung, and breast cancer tumors
Next, we tested the effects of the R-Tf-D-LP4 peptide in vivo, as it was more soluble than the Tf-D-LP4 peptide and inhibited HK activity (Fig. S3). The effects of the R-Tf-D-LP4 peptide on glioblastoma-derived U-87MG, lung cancer-derived A549, or breast cancerderived MDA-MB-231 cell xenograft mouse models were tested according to an established protocol (Fig. 6A). Nude mice were s.c. injected with the cells, and following tumor formation to a volume of 50-100 mm 3 , the mice were split into two tumor volumematched groups. The control group was treated intratumorally with HBSS, while the second group was treated with the R-Tf-D-LP4 peptide (40 or 60 lM), with the calculated DMSO final concentration in both groups being 0.05%, and tumor growth was followed (Fig. 6B-D). In the three cancer models, untreated tumor volumes grew exponentially and increased over the course of the experiment, while the volume of peptide-treated tumors (TTs) only slightly increased. A decrease of 70-80% in tumor volume was obtained in the three tumor types upon peptide treatment (Fig. 6E).
After sacrificing the mice, tumors were excised and each half tumor was fixed with formaldehyde and used for hematoxylin-eosin (H&E) and IHC staining, while the second half was frozen and used for RNA isolation and qPCR. Control tumors and R-Tf-D-LP4-TTs were analyzed by qPCR using specific primers for cell proliferation factor Ki-67. The results showed a decrease of two-to threefold in Ki-67 mRNA levels in peptide-TTs (Fig. 6F). Ki-67 levels were also analyzed from paraffin-embedded sections, showing strong staining of nuclei in the three types of untreated tumors, while staining in the R-Tf-D-LP4-TTs was pronouncedly low (Fig. 6G). Quantitative analysis indicated that Ki-67-positive cell numbers were lower by 70-80% in the R-Tf-D-LP4-TTs, relative to the control tumors (Fig. 6H).
As the mode of action of the peptide involves apoptosis induction, we analyzed the presence of apoptotic cells in the R-Tf-D-LP4-TTs in situ using TUNEL staining (Fig. 7A-C). While no TUNEL-positive cells were apparent in control HBSS-TTs, the majority of the cells were TUNEL-positive in the R-Tf-D-LP4-TTs, with staining colocalizing with PI nuclear staining. The results clearly indicate that the peptide induced apoptotic cell death. Thus, the marked decrease in tumor size in the peptide-treated xenografts  can be attributed both to inhibition of cell proliferation (Fig. 6F-H) and to peptide-induced cell death ( Fig. 7A-C).
Furthermore, in A549-, MDA-MB-231-, and U-87MG-derived tumors treated with the R-Tf-D-LP4 peptide, p53 expression levels were highly elevated (over two-to fivefold) (Fig. 7D-F). Caspase-8, mediating the extrinsic apoptosis pathway, and caspase-3 were also overexpressed (two-to threefold) in R-Tf-D-LP4-TTs (Fig. 7D,E). Cyto c, a key player in mitochondria-mediated apoptosis, was also highly expressed (two-to fourfold) in the R-Tf-D-LP4-TTs (Fig. 7D,F). Similarly, expression levels of the proapoptotic protein SMAC/Diablo were much higher (two-to fivefold) in R-Tf-D-LP4-TTs (Fig. 7D,E). Similar results were obtained with other VDAC1-based peptides, with D(1-14)N-Ter-Antp-TTs and Tf-D-LP4-TTs showing elevated levels of apoptosis-related proteins, such as Cyto c, p53, and caspases (Shteinfer-Kuzmine et al., 2017), These results were further confirmed when tumor sections were IHC-stained with specific antibodies against the pro-apoptotic proteins p53, Cyto c, and AIF. In all cases, a massive increase in staining intensity was seen in R-Tf-D-LP4-TTs (Fig. 7G-I). The overexpression of pro-apoptotic proteins may explain the high level of apoptosis obtained in the R-Tf-D-LP4-TTs.

R-Tf-D-LP4 peptides alter the expression levels of metabolic enzymes and transporters in vivo
Next, we analyzed the effect of R-Tf-D-LP4 treatment on the metabolic features of residual tumors derived from A549, MDA-MB-231, and U-87MG cells using IHC and qPCR (Fig. 8). The R-Tf-D-LP4 peptide decreased the expression of metabolism-related proteins in A549-, MDA-MB-231-, and U-87MG-derived tumors, as analyzed by IHC and qPCR (Fig. 8). In the R-Tf-D-LP4-TTs, the expression levels of the glucose transporter (Glut-1) and of the glycolytic enzymes HK-I and glyceraldehyde dehydrogenase (GAPDH) were dramatically decreased, with qPCR analysis showing decreases of three-to fourfold. The expression levels of mitochondrial proteins associated with metabolism were also decreased. The level of VDAC1, a key protein in cancer cell energy and metabolism homeostasis, was also decreased, as were the levels of the Krebs cycle enzyme citrate synthase (CS), the electron transport chain protein complex IV, and ATP synthase subunit 5a (Fig. 8).
As cancer cells undergo metabolic reprograming essential for cancer cell survival and growth (Vander Heiden et al., 2009), these findings point to a reversal of the metabolic reprograming of the cancer cells, originated from different cancer types, in response to R-Tf-D-LP4 peptide action.
3.6. The R-Tf-D-LP4 peptide decreased the expression levels of stem cells markers Cancer stem cell (CSC) populations have been identified in nearly all human malignancies, including those in the brain, pancreas, ovary, colon, and liver, as well as leukemia (Wang et al., 2011). CSCs are involved in tumor growth and disease progression and are thought to be highly resistant to a number of chemotherapeutic and radiation therapeutic strategies (Bao et al., 2006). We have previously demonstrated that glioblastoma cancer stem cells were eliminated by Tf-D-LP4 tumor treatment. Thus, we asked whether the R-Tf-D-LP4 peptide affects the expression of stem cells markers specific to each cancer type (Fig. 9).
As tumor spheroid formation (clonogenicity) is a well-known cancer stem cell feature and is considered as the gold standard in vitro method for evaluating the presence of CSCs (Valent et al., 2012;Weiswald et al., 2015), we performed this assay using the three cell lines used in the in vivo study. Using stem cell-specific medium (Pollard et al., 2009), we tested the effect of R-Tf-D-LP4 cell treatment on spheroid/aggregate formation and found that the peptide attenuated their formation, in comparison with peptide-untreated lung and breast cancer and GBM cell lines (Fig. 9B,D,F).

Discussion
Many malignant cells arise from the multistep process of tumorigenesis, involving accumulation of inherited or acquired genetic abnormalities, some of which drive malignant cells to reprogram energy production and afford protection from apoptosis, a hallmark of the majority of cancer types. Newly developed targeted therapies for cancer, including immunotherapy and anti-angiogenesis strategies, have shown that 1/3 of tumors respond to immunotherapy, and even less to angiogenesis inhibition, with patients showing varied responses, relapse, and resistance (Arbiser et al., 2017). Given these findings, and considering the heterogeneity and complexity of malignant tumors, a pandrug that acts regardless of the origin of the cancer and the mutations involved and which targets major cancer hallmarks, such as reprogramed metabolism and apoptosis, is required. VDAC1, as mitochondrial gatekeeper, functions in energy production, Ca 2+ homeostasis, oxidative stress, and apoptosis and as a hub protein interacting with over 150 proteins, including anti-apoptotic proteins , 2017a. Hence, targeting the interactions of VDAC1 with anti-apoptotic proteins (Bcl-2, Bcl-xL) or proteins supporting the metabolic requirements of cancer cells (HK-I, HK-II) is a promising strategy for the development of anticancer agents.
We have developed several VDAC1-based peptides that act as a 'decoy' to trap VDAC1-interacting proteins, thus competing with VDAC1 for interaction with Bcl-2, Bcl-xL, and HK and consequently interrupting their anti-apoptotic activities Arbel et al., 2012). The VDAC1-based peptides are discussed below with respect to the mode of cell-killing action, specificity to cancer cells, effectiveness in cancer mouse models, and advantages as potential anticancer drugs.  defective cancer cell lines (e.g., U-87MG and MCF-7 or MDA-MB-231 and HTB-66), and sensitivity to the peptides was independent of tumor origin (adenocarcinoma or carcinoma).
The results on cell death induced by the N-Ter-Antp and Antp-LP4 peptides point to three levels of sensitivity, as reflected by EC 50 values (Table 1). The highly sensitive group (EC 50 < 0.9-4.5 lM, 90-min incubation) includes hematological malignancies (e.g., myelogenous leukemia, histiocytic lymphoma, CLL, acute T-cell leukemia) and the prostate carcinoma cell lines (PC-3 and DU-145). The secondmost sensitive group (EC 50 , 6.5-10 lM) includes A549 and H358 lung cancer cells. The least sensitive group, with EC 50 values 20-to 30-fold higher than for the highly sensitive group, includes the PANC1 pancreatic cancer cell line and noncancerous PWR-1E and T-Rex-937 cells and PBCs and PBMCs. The different sensitivities of the various cancer cell types to the peptides could result from differences in antiapoptotic protein expression levels. For example, the most peptide-sensitive cells are those expressing mutated p53 and very high levels of HK-II (Tables 1  and 3).
The N-Ter-Antp, Antp-LP4, and Tf-LP4 peptides were further optimized (Figs 3 and S1, and Table 2) with respect to sequence length, D vs. L amino acid conformation, cell-penetrating peptide (CPP) type, and C-or N-terminal position. The N-Ter peptide was active only when Antp was used as the CPP but not Tf, a sequence recognized by the TfR. Moreover, deletion of the GXXXG motif (amino acids 21-26), shown to be involved in protein-protein interaction (Cymer et al., 2012), resulted in a nonactive peptide. The N-Ter-Antp peptide was active even after shortening the N terminus of the peptide by up to 18 of the 26 amino acids comprising the peptide (Fig. S1C).
As expected for a protease-protected peptide, D-Antp-LP4 showed the highest efficacy in killing prostate cancer cells, with an EC 50 in the submicromolar range, 8-to 30-fold less than that of the L-amino acidbased version (Fig. 2).
The better effectiveness of Tf-LP4 and its retro-version R-Tf-D-LP4 in inducing cell death than Antp-LP4 peptides (Table 2) may result from TfR being highly expressed in cancer cells (Daniels et al., 2012) (Fig. 3B). As R-Tf-D-LP4 was more soluble and stable than Tf-D-LP4 (Fig. S3C-E), and inhibited HK activity (Fig. S3A, B), it was selected for in vivo studies.
The findings that the VDAC1-based N-Ter-Antp, Antp-LP4, Tf-LP4, and, particularly, the protease-protected D-version of the peptides (Figs 1-3) efficiently killed such varied cancer cells, regardless of the cell mutation status, provide the opportunity for developing new therapies for different cancers.

Modes of action and advantages of VDAC1based peptides
Our results with cells in culture and in tumors suggest that the VDAC1-based peptides have a triple mode of action, namely affecting cell energy production and metabolism, preventing the protective effect of antiapoptotic proteins, and inducing apoptosis (Figs 4, 5 and S2). Cancer cells possess reprogramed metabolism, with the high use of glucose as an energy source via glycolysis and involving elevated expression of genes encoding glucose transporters and glycolytic enzymes, as HK, being a feature common to most solid tumors (Hanahan and Weinberg, 2011;Kaelin and Thompson, 2010;Koppenol et al., 2011).
Here, we demonstrated that our peptides (Antp-LP4, Tf-D-LP4, and R-Tf-D-LP4) interacted with purified HK-I and HK-II ( Fig. 4A-C), detaching bound HK-I (Fig. 4D,E). Thus, these 'decoy' peptides impaired the HK-VDAC1 interaction, leading to HK detachment and affecting the energy balance of high energy-demanding cancer cells. By detaching HK that catalyzes the first step of glycolysis, glycolysis is inhibited. Consequently, the TCA cycle is slowed as less acetyl-CoA is produced from pyruvate. As mitochondrial ATP is the preferential substrate for bound HK, detachment of mitochondria-bound HK decreased cellular ATP levels (Fig. 4F); this precedes a decrease in cell viability (Machida et al., 2006). Moreover, as the interaction of VDAC1 with HK mediates coupling between OXPHOS and glycolysis, enhanced cholesterol synthesis, decreased ROS production, and protection against apoptosis, disrupting this interaction should have multiple effects on cancer cells , 2017a. Accordingly, detachment of HK from VDAC1 represents a novel therapeutic strategy to impair cancer metabolism and augment apoptosis (Krasnov et al., 2013;Shoshan-Barmatz et al., 2015).
A number of agents, including 2-deoxyglucose (2-DG), 3-bromopyruvate (3-BP), an alkylating reagent, and lonidamine, have been used to inhibit HK activity and disrupt glycolysis (Krasnov et al., 2013). In addition, the antifungal agents clotrimazole and bifonazole were shown to disrupt the HK-VDAC1 complex (Goldin et al., 2008;Penso and Beitner, 1998). However, these agents are not very potent, not targeted to cancer cells, and they are toxic to certain normal tissues where glucose is the main energy source, such as brain, retinae, and testes (Krasnov et al., 2013). In contrast, the Tf-D-LP4 and R-Tf-D-LP4 peptides are targeted to cancer cells overexpressing TfR and thus specifically disrupt the HK-VDAC1 interaction in cancer cells and affect overall cellular bioenergetics, as well as activating apoptosis (Figs 4, 5 and S2).
The R-Tf-D-LP4 but not the Tf-D-LP4 peptide inhibits HK activity (Fig. S3). As HK is a cellular protein, its inhibition by the R-Tf-D-LP4 peptide takes place in cells overexpressing TfR, such as cancer cells (Daniels et al., 2012). Thus, R-Tf-D-LP4 targets HK preferably in cancer cells.
As expected from VDAC1-based peptides that interact with HK, Bcl-2, and Bcl-xL and preventing their anti-apoptotic activities (Abu-Hamad et al., 2008;Shoshan-Barmatz et al., 2009;Zaid et al., 2005), cell treated with the peptide underwent massive apoptosis, both in culture and in tumors (Figs 5 and 7, and S2). The peptide induced mitochondria-mediated apoptosis, as reflected in the release of Cyto c, and the presence of apoptotic cells. As found for apoptosis inducers, such as STS, As 2 O 3 , and selenite (Weisthal et al., 2014), the peptide also disrupted [Ca 2+ ]i homeostasis, increasing [Ca 2+ ]i levels (Fig. 4). Furthermore, in tumors treated with the peptide, massive cell death was observed using TUNEL staining, and the levels of pro-apoptotic proteins, Cyto c, p53, AIF, and caspases were highly elevated (Fig. 7). Hence, the dramatic effect of the peptide on tumor growth can be attributed to its action on two important cancer hallmarks, namely cancer cell energy status and induction of apoptotic cell death.
4.3. VDAC1-based peptides similarly influence tumors derived from three different cancers by reprograming cell metabolism, inhibiting tumor growth, inducing cell death, and eliminating CSCs In a recent study (Shteinfer-Kuzmine et al., 2017), we demonstrated that in U-87MG cell-derived tumors, VDAC1-based peptides perturbed cell energy homeostasis, inhibited tumor growth, induced apoptosis, and eliminated CSCs. Here, we demonstrate that the R-Tf-D-LP4 peptide similarly affects tumors derived from the U-87MG glioblastoma, A549 lung cancer, and the MDA-MB-132-triple-negative breast cancer cell lines. Indeed, in three different tumors, the R-Tf-D-LP4 peptide similarly inhibited tumor growth (~80%) by inhibiting cell proliferation, as revealed by decreased Ki-67 staining (Fig. 6F-H). This inhibition may result from the peptide impairing cancer cell energy and metabolism homeostasis (Fig. 8). Strikingly, the expression levels of energy-related enzymes, such as certain enzymes associated with glycolysis (HK-I, Glut-1, and GAPDH), Krebs cycle activity [citrate synthase (CS)], and oxidative phosphorylation (complex IVc and ATP synthase 5a subunits), were downregulated, as revealed by IHC and qPCR (Fig. 8). These results suggest that in the residual peptide-TTs, the reprogramed metabolism of the cancer cell was reversed upon interruption the interaction of VDAC1 with energy production-associated enzymes, such as HK.
As demonstrated, VDAC1-based peptides activate apoptosis in cells in culture (Figs 5 and S2) and in tumors ( Fig. 7A-C). R-Tf-D-LP4 induced massive apoptotic cell death in the three types of tumors tested. The intense apoptosis revealed by TUNEL staining, reflecting DNA degradation, can be explained by the multiple actions of the peptide, specifically antagonizing the anti-apoptotic effect of anti-apoptotic proteins, and inducing overexpression of key proteins in apoptosis, including caspases 3 and 8, p53, Cyto c, AIF, and SMAC/Diablo (Fig. 7D-I).
Finally, a large number of anticancer chemotherapeutic agents exert their therapeutic actions by inducing apoptosis of malignant cells (Costantini et al., 2000;Hail, 2005;Kaufmann and Vaux, 2003), mainly by activating the Cyto c/caspase-9 pathway. However, they do not act on CSCs, which are resistant to chemo-and radiotherapies (Chen et al., 2012;Visvader and Lindeman, 2012). As shown in our previous study of GBM Shteinfer-Kuzmine et al., 2017), the R-Tf-D-LP4 peptide eliminated CSCs in tumors derived from glioblastoma, lung, and breast cancers (Fig. 9).
The effect of the peptide on CSCs was also reflected in the inhibition of spheroid/aggregate formation (Fig. 9). Cell spheroid/tumorsphere-forming capacity was used for evaluating the presence of CSCs (Valent et al., 2012;Weiswald et al., 2015). This effect may result from the peptide inhibiting CSC proliferation or/and inducing cell death.
The results reported here show that tumors derived from three different cell lines, despite carrying different mutations, responded similarly to the peptide. MDA-MB-231 (but not U-87MG or A549) cells express mutated p53 and do not express the estrogen receptor (ER). At the same time, A549 cells express low levels of the pro-apoptotic protein Bax and mutated KRAS, LKB1, and CDNKA, while U-87MG cells express mutated PTEN. Thus, the anticancer activity of VDAC1-based peptides is achieved regardless of cancer origin and mutational status and, most importantly, also involves the targeting of CSCs, known to be resistant to chemo-and radiotherapies.

Conclusions
To conclude, cell-penetrating VDAC1-based peptides were developed. The selected peptides possess multiple effects on cancer cells, including disruption of cell energy and metabolism homeostasis, and interference with the activity of anti-apoptotic proteins, leading to apoptotic cell death with perceived specificity toward the cancerous cells.
In mouse models for GBM, and lung and breast cancers, the R-Tf-D-LP4 peptide simultaneously attacked several cancer hallmarks, causing impairment of energy and metabolic homeostasis, including reduction in metabolic enzymes expression, inhibition of tumor growth, induction of apoptosis, and overexpression of pro-apoptotic proteins. These manifold effects explain the prompt effect of the peptide on cancer cells. Moreover, the mode of action of the R-Tf-D-LP4 peptide explains why it is not directed to a specific cancer type but rather acts on a variety of cancers, regardless of the mutations involved or the survival mechanisms acquired. These features of the proteaseprotected, cell-penetrating peptides point to their great therapeutic potential in various cancers, including those chemoresistant cancers for which traditional therapies are ineffective. Zhang CL, Huang T, Wu BL, He WX and Liu D (2017) Stem cells in cancer therapy: opportunities and challenges. Oncotarget 8, 75756-75766.

Supporting information
Additional Supporting Information may be found online in the supporting information tab for this article: Fig. S1. Cell death-inducing activity of various VDAC1-based peptides  Table S1. Amino acid sequences, MS/MS data, and analytical data for the peptides used in this study Table S2. Antibodies used in this study