A VDAC1-Derived N-Terminal Peptide Inhibits Mutant SOD1-VDAC1 Interactions and Toxicity in the SOD1 Model of ALS

Mutations in superoxide dismutase (SOD1) are the second most common cause of familial amyotrophic lateral sclerosis (ALS), a fatal neurodegenerative disease caused by the death of motor neurons in the brain and spinal cord. SOD1 neurotoxicity has been attributed to aberrant accumulation of misfolded SOD1, which in its soluble form binds to intracellular organelles, such as mitochondria and ER, disrupting their functions. Here, we demonstrate that mutant SOD1 binds specifically to the N-terminal domain of the voltage-dependent anion channel (VDAC1), an outer mitochondrial membrane protein controlling cell energy, metabolic and survival pathways. Mutant SOD1G93A and SOD1G85R, but not wild type SOD1, directly interact with VDAC1 and reduce its channel conductance. No such interaction with N-terminal-truncated VDAC1 occurs. Moreover, a VDAC1-derived N-terminal peptide inhibited mutant SOD1-induced toxicity. Incubation of motor neuron-like NSC-34 cells expressing mutant SOD1 or mouse embryonic stem cell-derived motor neurons with different VDAC1 N-terminal peptides resulted in enhanced cell survival. Taken together, our results establish a direct link between mutant SOD1 toxicity and the VDAC1 N-terminal domain and suggest that VDAC1 N-terminal peptides targeting mutant SOD1 provide potential new therapeutic strategies for ALS.


INTRODUCTION
Amyotrophic lateral sclerosis (ALS) is a progressive and fatal neurodegenerative disease caused by the death of upper and lower motor neurons in the brain and spinal cord (Cleveland and Rothstein, 2001). The age of onset is typically between 50 and 60 years, followed by progressive paralysis and death 2-5 years after diagnosis (Mulder et al., 1986;Dorst et al., 2019). Most cases of ALS are sporadic and lack any apparent genetic linkage, although in 10% of cases, the disease is inherited in a dominant manner. About a fifth of these familial cases have been attributed to mutations in the gene encoding cytoplasmic Cu/Zn superoxide dismutase (SOD1) (Rosen et al., 1993).
To date, more than 180 different human SOD1 mutations have been identified throughout the length of the SOD1 protein that are directly linked to familial ALS (fALS), 1 including active dismutase mutants, such as SOD1 G93A and SOD1 G37R , and inactive dismutase mutants in which the mutation affects the metal-binding region, such as SOD1 G85R and SOD1 H46R (Abu-Hamad et al., 2017). The latter group of mutants are more unstable than are the former. Moreover, wild type human SOD1 (SOD1 WT ) can become misfolded and toxic, thus sharing an aberrant conformation with SOD1 mutants, when oxidatively modified (Tiwari et al., 2009;Bosco et al., 2010;Guareschi et al., 2012;Lim and Song, 2016;Medinas et al., 2018;Xu et al., 2018). The exact mechanism which drives motor neuron degeneration and disease progression remains unknown, although multiple hypotheses have been proposed to explain mutant SOD1-dependent toxic effects (Ilieva et al., 2009). These include ER stress, oxidative stress, glutamate-mediated excitotoxicity and mutant SOD1 misfolding and aggregationinduced pathology. Indeed, aggregation of misfolded SOD1 proteins is a common pathological observation among subjects with different SOD1 mutations and is, therefore, believed to be central to ALS pathogenesis (Bruijn et al., 1998;Wang et al., 2005;Prudencio et al., 2009;Abu-Hamad et al., 2017;Shvil et al., 2018).
Although predominantly a cytosolic protein, SOD1 is also found localized in other cellular compartments, including mitochondria. Both in mouse and rat models of ALS and post-mortem tissue samples from ALS patients, mutant SOD1 was found in fractions enriched for mitochondria derived only from affected but not unaffected tissues (Mattiazzi et al., 2002;Liu et al., 2004;Vijayvergiya et al., 2005;Bergemalm et al., 2006;Deng et al., 2006;Vande Velde et al., 2008). Moreover, a clear temporal correlation between disease progression and mitochondrial association was shown for different SOD1 mutants in rodent models (Liu et al., 2004). In addition, we have recently reported a clear inverse correlation between mutant SOD1 mitochondrial association in motor neuron-like NSC-34 cells and disease duration in patients carrying mutations in SOD1 (Abu-Hamad et al., 2017). 1 http://alsod.iop.kcl.ac.uk/als/ Highly purified floated mitochondria coupled with protease accessibility has demonstrated deposition of mutant SOD1 on the cytoplasmic-facing surface of spinal cord mitochondria (Liu et al., 2004;Vande Velde et al., 2008). Sensitivity to proteolysis and immunoprecipitation with specific antibodies for misfolded SOD1 further demonstrated that misfolded species of SOD1 are associated with the outer mitochondrial membrane of the spinal cord (Vande Velde et al., 2008). In addition, mutant SOD1 was proposed to interact with other components of the outer mitochondrial membrane, including Bcl-2 (Pedrini et al., 2010) and the protein import machinery (Li et al., 2010), thus affecting the corresponding functions. Importantly, this was seen only for spinal cord mitochondria but not for mitochondria isolated from unaffected tissues Li et al., 2010). More specifically, direct binding of misfolded SOD1 to the voltage-dependent anion channel-1 (VDAC1) was previously shown, causing reduction of VDAC1 conductance and channel instability, leading to inhibition of VDAC1 transport of adenine nucleotides across the outer mitochondrial membrane Magri et al., 2016). VDAC1, also known as the mitochondrial porin, is located at the outer mitochondrial membrane, where it assumes a crucial position controlling the metabolic cross-talk between the mitochondria and the rest of the cell, thus regulating the metabolic and energetic functions of mitochondria. VDAC1 is also a central player in mitochondria-mediated apoptosis and has been implicated in apoptotic-related functions, given its role as the target for pro-and anti-apoptotic Bcl2family of proteins (Shimizu et al., 1999;Arbel and Shoshan-Barmatz, 2010) and due to its function in the release of apoptotic proteins from the mitochondrial inter membrane space (Tajeddine et al., 2008;Abu-Hamad et al., 2009). VDAC1, the main VDAC isoform, is composed of 19 transmembrane β-strands forming a membrane-embedded β-barrel and a flexible amphipathic 26-residue-long N-terminal domain lying inside the pore but able to translocate from within the pore to the channel surface (Geula et al., 2012). This mobility is important for controlling channel gating but also for interactions with pro-and anti-apoptotic proteins (Abu-Hamad et al., 2009;Arbel and Shoshan-Barmatz, 2010;Shoshan-Barmatz et al., 2010Arbel et al., 2012;Geula et al., 2012). Importantly, cells expressing an N-terminally truncated form of VDAC1 are resistant to apoptosis (Abu-Hamad et al., 2009). These findings suggest that the VDAC1 N-terminal domain is required for interaction with VDAC1-associated proteins and apoptosis.
Here, we demonstrate the direct interaction of VDAC1 with mutant SOD1 and show that this interaction requires the VDAC1 N-terminal domain. Moreover, SOD1-mediated toxicity was prevented by synthetic VDAC1-N-terminal peptides. Finally, we show that a VDAC1 N-terminal peptide enhanced the survival of mutant SOD1 G93A motor neuron-like NSC-34 cells and mutant SOD1 G93A mouse embryonic stem cellderived motor neurons. These findings point to VDAC1 N-terminal peptides as offering possible novel therapeutic strategies for ALS.

Peptides
The peptides used in this study (listed in Table 1) were synthesized by GL Biochem (Shanghai, China). The peptides were first dissolved in DMSO as a 40 mM solution and then diluted 20-fold in the appropriate buffer. Peptide concentrations were determined as described previously (Shteinfer-Kuzmine et al., 2018). The final concentration of DMSO in control and peptidecontaining samples was ≤0.5%.

Protein Purification
Recombinant hSOD1 wt , hSOD1 G93A , and hSOD1 G85R were expressed in sf-9 cells and purified by hydrophobic interaction chromatography using phenyl-Sepharose 6 Fast Flow high sub (Amersham Biosciences), followed by ion exchange chromatography using a HiTrap Q-Sepharose anion exchange column (Amersham Biosciences), as described previously (Hayward et al., 2002).

VDAC1 Purification
VDAC1 was purified from rat liver mitochondria using celite:hydroxyapatite and CMC chromatography, as previously described (Gincel et al., 2001). DNA sequences encoding full-length murine VDAC1 and N-terminally truncated VDAC1 ( N-VDAC1) lacking residues 1-26 were cloned into the pET21a vector (Novagen) using the NheI/XhoI sites. Escherichia coli BL21(DE3) cells were transformed with plasmid pET21a harboring the VDAC1 or N-VDAC1 genes. Protein expression was induced for 3 h using 1 mM isopropyl-β-Dthiogalactopyranoside (IPTG; Sigma). Proteins were purified on agarose-packed nickel-nitrilotriacetic acid resin (Ni-NTA; Qiagen) in the presence of 8 M urea. Refolding of the eluted protein was performed essentially as described previously (Hiller et al., 2008). The refolded protein was further purified as above for mitochondrial VDAC1.

VDAC1 Channel Reconstitution, Recording and Analysis
The reconstitution of recombinant WT or N-VDAC1 into a planar lipid bilayer (PLB) prepared from soybean asolectin, and subsequent single channel current recordings and data analysis were carried out as described previously (Gincel et al., 2001). Currents were recorded under voltage-clamp conditions before and 5 min after the addition of 40 µg of recombinant hSOD1 wt , hSOD1 G93A , or hSOD1 G85R to the cis compartment The VDAC1-based peptides used in this study with the amino acid sequence number, molecular mass, calculated molar extinction coefficient and purity are presented. The cell-penetrating sequence is underlined. The molar extinction coefficient was calculated based on amino acid composition using the following link: http://www.biomol. net/en/tools/proteinextinction.htm.
using a Bilayer Clamp BC-535B amplifier (Warner Instrument, Hamden, CT, United States). Current amplitude histograms were prepared using AxoGraph X software. Relative conductance was determined as the average steady-state conductance at a given voltage normalized to the conductance at 10 mV, the maximal conductance. Relative conductance-voltage plots were prepared using Microsoft Excel software.
Cell Treatment With VDAC1-Based Peptides, Cell Death and XTT Analyses Apoptotic cell death was also analyzed using Acridine Orange (AcOr)/ethidium bromide (EtBr) staining (McGahon et al., 1995). Cells in 24-well plates were washed with 200 µl PBS and 10 µl of a solution containing 100 mg/ml AcOr and 100 mg/ml EtBr in PBS was added. The cells were then visualized by fluorescence microscopy (ZOE fluorescence cell imager, Bio-Rad), images were recorded and cells at early and late apoptotic stages were counted.

Immunostaining
For immunostaining, SH-SY5Y cells (4.5 × 10 4 ) were grown on sterilized coverslips in 24-well plates. 30 h post-transfection, cells were fixed using 4% paraformaldehyde (PFA; diluted in PBS) for 15 min, and then washed 3 times with PBS (5 min each wash). Cells were then permeabilized with 0.3% Triton X-100 in PBS for 5 min followed by washing with PBS. Cells were then blocked for 1 h with blocking buffer (1% BSA free fatty acids diluted in PBS). Anti-VDAC1 polyclonal antibody (ab15895) and mouse anti misfolded SOD1 (B8H10, Medimabs) were incubated at room temperature for 1-2 h in a buffer of 1% BSA free fatty acids and 0.3% Triton-X100 in PBS. Following incubation with primary antibodies, cells were washed with PBS and incubated with fluorescent conjugated secondary Alexa Flour 488 anti-rabbit and Alexa flour 647 anti-mouse antibodies. The coverslips were carefully dried and mounted on slides using Immumount (Immumount TM , Thermo). After overnight drying, images were acquired on an Olympus IX81 confocal microscope.

Mouse Embryonic Stem Cell (mESC) Cultures
Mouse embryonic stem cell lines harboring human mutant SOD1 (SOD1 G93A ) were a kind gift from Dr. Kevin Eggan (Harvard Stem Cell Institute). This cell line carries green fluorescent protein (GFP) under the control of the promoter for the motor neuron (MN)-specific transcription factor HB9 (HB9:GFP cells) (Di Giorgio et al., 2007). We used the SOD1 G93A mESC line to derive GFP + MNs.
For an in vitro differentiation assay, EBs were enzymatically dissociated after 7 days in culture with TrypLE Express (Gibco) and seeded on 0.01% poly-l-ornithine (Sigma) precoated coverslips followed by laminin (10 µg/ml, Sigma). Cells (5 × 10 4 ) were seeded on coverslip in 24-well plates with ADFNB cell medium supplemented with of Ciliary neurotrophic factor (10 ng/ml, CNTF) and Glial cell-derived neurotrophic factor (GDNF; Miltenyi Biotec). During plating, 1-20N-Ter-Antp VDAC1-peptide was added to the cultures. Half of the medium was replaced with fresh medium every second day and at the indicated time points. The cultures were fixed in 4% paraformaldehyde in PBS.

Determination of Neurite Outgrowth, Intersections Between Neurites and Survival
Stereological estimation of neurite lengths to evaluate neurite outgrowth in cultured cells was carried out as described previously (Ronn et al., 2000). The total neuritic length per cell was estimated by counting the number of GFP + soma and neurite intersections with test lines of an unbiased counting frame superimposed on images of cell cultures obtained using a 20× objective (NA 0.75) of a Nikon Eclipse E800 epifluorescence microscope equipped with a Nikon DXM1200F CCD camera. The absolute length, L, of neurites per cell was subsequently estimated from the number of neurite intersections, I, per cell by means of the equation L = (πd/2)I describing the relationship between the number of neurite intersections and the vertical distance, d, between the test lines used.

Statistical Analyses
Statistical comparisons between groups were performed by a two-tailed unpaired Student's t-test. The mean ± SEM of results obtained from at least three independent experiments are presented. The significance of differences was calculated by a two-tailed Student's t-test and is reported as * * p < 0.01. Statistical comparisons between conditions in the mESC MN assays was performed by one-way ANOVA followed by Dunnett's Multiple Comparison test against the control condition ( Figure 6B) or Tukey's Multiple Comparison test (Figures 6C,E). The confidence interval was stated at the 95% confidence level, placing statistical significance at p < 0.05. GraphPad Prism 6 was used for plotting data and statistical analysis.

VDAC1 N-Terminal Domain and a VDAC1-Derived Peptide Specifically Interact With Mutant but Not Wild Type SOD1
The interaction of purified WT and the SOD1 mutants SOD1 G93A and SOD G85R with purified VDAC1 (Figure 1A) was assayed by MST (Figures 1B-E). MST measures any variation in the thermal movement of a fluorescently labeled binding partner. The subsequent fluorescence depletion in a heated spot of the protein solution is measured as a function of increasing interacting partner concentration, with dissociation constants (K D ) values being derived from the depletion curves (Wienken et al., 2010; Figure 1D). Fluorescently labeled VDAC1 incubated with increasing concentrations of WT or mutant SOD1 (0-100 µM) showed that mutant SOD1 G93A and SOD1 G85R but not SOD1 WT bound to VDAC1 ( Figure 1B).
Next, to identify the binding site for mutant SOD1 in VDAC1, we took advantage of different VDAC1-based peptides which we have developed and previously tested (Arzoine et al., 2008;Arbel and Shoshan-Barmatz, 2010). As the N-terminal domain of VDAC1 has been the shown to interact with several proteins, such as hexokinase, Bcl-2 and Bcl-xL (Arzoine et al., 2008;Arbel and Shoshan-Barmatz, 2010;Arbel et al., 2012), we tested whether a synthetic N-terminal peptide interacts with SOD1. Accordingly, fluorescently labeled mutant SOD1 G93A or SOD1 G85R protein was incubated with increasing concentrations of the synthetic VDAC1 N-terminal peptide and changes in fluorescence were monitored ( Figure 1C). By plotting the percentage change of normalized fluorescence ( F Norm %) as a function of peptide concentration, a fitted curve yielded dissociation constants (K D ) for the three versions of SOD1 ( Figure 1D). The results showed that the VDAC1 N-terminal peptide bound both mutant SOD1 G93A and SOD1 G85R , but not to SOD1 WT (Figures 1C,D), indicating that these mutants interact specifically with the N-terminal region.
The specificity of the N-terminal peptide to mutant SOD1 was demonstrated by testing the binding of another VDAC1derived peptide, LP3. This peptide, representing the sequence of a VDAC1 loop facing the cytosol (Arzoine et al., 2008), showed significantly lower binding to mutant SOD1 than did the (1-26)-N-terminal peptide ( Figure 1E). These results show that VDAC1 and the N-terminal VDAC1 peptide specifically interact with mutant SOD1 G93A and SOD1 G85R .

Mutant but Not Wild Type SOD1 Interacts With VDAC1 and Reduces Its Channel Activity
To test whether mutant hSOD1 G93A and hSOD1 G85R binding to VDAC1 affects VDAC1 function, as well as the requirement of the N-terminal domain for such binding, full length and N-terminally truncated VDAC1 were expressed in E. coli, purified ( Figure 2M) and reconstituted into a PLB as described previously (Gincel et al., 2001).
Single-channel conductance under voltage-clamp conditions was measured as a function of time, reflecting ions passing through the channel in response to an applied voltage gradient. Current-time traces recorded at −10 or 10 mV from purified recombinant full length VDAC1 showed a stable full open state that was maintained for extended periods. Addition of hSOD1 WT , even at the highest concentration (60 µg/ml), had no effect on the current (Figures 2A,B). VDAC1 showed a bellshaped relative conductance curve as function of the voltage, with hSOD1 WT having no effect on VDAC1 channel conductance at all tested voltages, i.e., −60 to +60 mV ( Figure 2C). This was also revealed in the current amplitude histograms (Figure 2D), which showed a single channel conductance of 32 pA, at 10 mV.
In contrast to hSOD1 WT , both mutant hSOD1 G93A and hSOD1 G85R reduced the channel conductance of bilayerreconstituted VDAC1 at all tested voltages and decreased the current amplitude histograms (Figures 2E-L). hSOD1 G93A was found to be more effective in reducing VDAC1 conductance than was hSOD1 G85R when added at the same concentration, resulting in 57 and 40% inhibition of channel conductance, respectively ( Table 2).
As reported previously (Prezma et al., 2013;Shteinfer-Kuzmine et al., 2018), VDAC1 lacking the N-terminal 26 residues ( N-VDAC1) showed no voltage-dependent gating, with hSOD1 WT having no effect on channel conductance at all voltages tested or on the channel current amplitude histograms (Figures 3A-D). Moreover, and in contrast to what was observed upon SOD1 mutants interaction with VDAC1 (see Figures 2E-L), hSOD1 G93A or hSOD1 G85R had no effect on N-VDAC1 conductance at all tested voltages (Figures 3E-L). These results thus suggest that VDAC1 N-terminal is important for binding of hSOD1 G93A and hSOD1 G85R to VDAC1.
The summary of the effects of SOD1 WT , hSOD1 G93A , and hSOD1 G85R on the channel conductance of VDAC1 and N-VDAC1 is presented in Table 2.

Cell-Penetrating VDAC1 N-Terminal Peptides Inhibit Cell Death of NSC-34 Cells as Induced by Mutant SOD1 G93A
In our previous study (Shteinfer-Kuzmine et al., 2018), several novel VDAC1 N-terminal-derived peptides were designed and tested for their ability to induce cell death in cancer cells. Here, we sought to determine whether any of these peptides does not induce cell death yet can interact with misfolded mutated SOD1 and protect against SOD1 G93A -mediated cell death in neuronal cultures. Accordingly, we assessed the effects of synthetic cell-penetrating VDAC1-N-terminal-derived peptides in inducing cell death in NSC-34 cells, a mouse motor neuronlike hybrid cell line, and in the U-87MG and A549 cancer cell lines (Figures 4A-C). We considered five cell-penetrating peptides derived from the VDAC1-N-terminal domain in these studies: (1) (Table 1). FIGURE 2 | Mutant SOD1 but not wild type SOD1 interacts with VDAC1 and inhibits channel conductance. (A,B) Recombinant full length VDAC1 purified from E. coli was reconstituted into a PLB and channel currents through VDAC1, in response to a voltage step from 0 to 10 mV (A) or to -10 mV (B), before and 15-20 min after addition of 40 µg/ml (final concentration) of SOD1 were recorded. (C) VDAC1 relative conductance as a function of voltage in a 60 to -60 mV step before (•) and after addition of SOD1 WT (•). Relative conductance (conductance/maximal conductance) was determined as the average steady-state conductance at a given voltage normalized to the conductance at 10 mV, considered the maximal conductance.  3.65 ± 0.14 3.92 ± 0.08 Results are the conductance (1 M NaCl, pH 7.4, at 10 mV), calculated from experiments as in Figures 2, 3 presented as mean ± SE (n = 3).
peptides, yielded peptides that could not induce cell death (Figures 4A-C).
To test in vitro whether the different shortened versions of the VDAC1 N-terminal peptide could prevent mutant SOD1 toxicity in neurons, NSC-34 cells were transfected to express SOD1 WT or mutant SOD1 G93A protein and then treated with or without the above indicated VDAC1 N-terminal peptides. Cell viability was quantified using the XTT assay and apoptosis using acridine orange and ethidium bromide staining. Whereas SOD1 WT had not effect on cell viability, expressing SOD1 G93A reduced cell viability by 25-30%. The presence of VDAC1 N-terminal peptides reduced the toxic effect of SOD1 G93A in a concentration-dependent manner (Figures 4D-G). Importantly, this effect was observed using the three non-cell death-inducing VDAC1-derived N-terminal peptides, i.e., the (1-20)-N-Ter-Antp, (5-20)-N-Ter-Antp and (10-20)-N-Ter-Antp peptides.
In order to show that the rescue effect of the N-terminal peptide is not specific for mutant SOD1 G93A , but a general effect, we have expressed two different dismutase active mutants SOD1 G93A or SOD1 G37R to induce cell death, increasing it from 17% in the control plasmid-transfected cells to 67 and 70%, in cells expressing mutant SOD1 G93A or SOD1 G37R , respectively. Incubation with the (10-20)-N-Ter-Antp peptide, decreased SOD1 G93A and SOD1 G37R -mediated cell death by 68 and 81%, respectively ( Figure 4H).
In addition, to test whether the N-terminal VDAC1-based peptide could promote misfolded SOD1 detachment from the mitochondria and more specifically from VDAC1, SH-SY5Y neuronal cells were transfected to express two different mutated SOD1 G93A or SOD1 G37R proteins and then treated with or without the VDAC1 (10-20)-N-Ter-Antp peptide. Colocalization of misfolded SOD1 with VDAC1 was determined by immunostaining analysis using an anti-VDAC1 antibody that does not target the N-terminus region of the protein, and the B8H10 antibody specifically recognizing misfolded SOD1. In cells transfected with SOD1 G93A or SOD1 G37R , VDAC1 staining is punctuated as expected for mitochondrial localization, while misfolded SOD1 shows both punctuated staining but mostly diffused, indicating that part of the misfolded SOD1 protein is mitochondria bound (Figures 5A,C). Yet, a clear co-localization of misfolded SOD1 with VDAC1 can be observed (Figures 5A,C). However, this co-localization was greatly eliminated in cells subjected to treatment with (10-20)-N-Ter-Antp peptide (Figures 5B,D).

A VDAC1 N-Terminal Peptide Enhances the Survival of Mouse Embryonic Stem Cell-Derived Motor Neurons Expressing SOD1 G93A
The effects of increasing concentrations of the VDAC1-based (1-20)-N-Ter-Antp peptide on three characteristics of SOD1 G93Aexpressing ESC-derived motor neurons were analyzed in vitro. Specifically, we assessed neurite outgrowth, MN survival shortly after induction of final differentiation and survival of maturing SOD1 G93A -expressing MNs. To follow neuronal development and remodeling of neuronal extensions, we determined neurite outgrowth in cell cultures by tracing neurites and their branches using the stereological procedure, as described previously (Ronn et al., 2000). SOD1 G93A -expressing MNs grown in relatively pure cultures die rapidly over the 1st week (Aggarwal et al., 2017). Therefore, the first two assays were performed over the first 2 days of culture. This period is characterized by small to minimal loss of MNs, with the exception of those cells that fail to attach or which die immediately. The third assay was performed after the onset of cell death, when around 50% of cells had died.
Images showing the effects of the N-terminal peptide on neurite outgrowth of mESC-derived SOD1 G93A -expressing MNs (Figures 6A,D), as well as quantification of neurite outgrowth (Figure 6B), are presented. Cells were incubated with the indicated concentrations of the peptide and analyzed 24 h later. The results showed that adding the peptide at a 10 µM concentration significantly (p < 0.05) improved neurite outgrowth of SOD1 G93A -expressing MNs (Figures 6A,B). The effect of the peptide on the survival of mESC-derived SOD1 G93A -expressing MNs was analyzed 24 h after plating. The results showed that when added at concentrations of 5 or 10 µM, the peptide significantly (p < 0.05) improved MN density (Figures 6A,C). The reduced effect of the peptide at higher concentrations (25 µM) may result from other non-specific interactions. Moreover, the VDAC1 N-terminal peptide extended survival of the mESC-derived SOD1 G93Aexpressing mature MNs (Figures 6D,E). When five cultures were incubated with the indicated peptide concentration for 96 h, it was seen that the peptide significantly (p < 0.05) improved MN survival when administered at a 10 µM concentration (Figures 6D,E). Thus, the peptide improved survival of SOD1 G93A -expressing MNs by twofold.

DISCUSSION
The results of the present study contribute to the better understanding of mutant SOD1-mediated mitochondrial dysfunction and cellular toxicity with relevance to ALS pathogenesis. Our results demonstrate that SOD1 mutants bind FIGURE 3 | The N-terminal domain of VDAC1 is required for mutant SOD1 interaction with VDAC1 and inhibition of channel conductance. (A,B) Currents passing through bilayer-reconstituted recombinant N-terminally truncated VDAC1 ( N-VDAC1) were recorded in response to voltage step from 0 to 10 mV (A) or -10 mV (B) before and 15-20 min after the addition of 40 µg/ml (final concentration) of SOD1 WT . (C) Relative conductance of N-VDAC1 as a function of voltage in a step from 60 to -60 mV before ( ) and after addition of SOD1 WT ( ). Relative conductance (conductance/maximal conductance) was determined as the average steady-state conductance at a given voltage normalized to the conductance at 10 mV, taken as the maximal conductance. directly and selectively to the N-terminal domain of VDAC1. Both dismutase-active SOD1 G93A and dismutase-inactive SOD G85R mutants, but not wild type SOD1, bind to the VDAC1 N-terminal region. Moreover, we demonstrated that different versions of an N-terminal peptide suppressed mutant SOD1 toxicity in motor neuron-like NSC-34 cells expressing mutant The rescuing effects of the VDAC1 N-terminal-derived peptides are shown as a percentage of cell death. The significance of quantitative analysis of triplicates of different biological repeats (n = 3) was performed by Student's t-test; * * P < 0.01. (H) SH-SY5Y cells (4.5 × 10 4 cells/well in 24-well plates) were transfected with an empty plasmid or a plasmid encoding for mutant SOD1 G93A or SOD1 G37R . Twenty-four hours post-transfection, the cells were incubated for 5 h with (10-20)N-Ter-Antp peptide (20 µM) and then analyzed for apoptosis using acridine orange and ethidium bromide staining, as described previously (McGahon et al., 1995). Fluorescence microscopy images were analyzed and about 100 to 300 cells were counted for each treatment in representative microscopic fields. The significance of quantitative analysis of triplicates of different biological repeats (n = 3) was performed by one-way Anova; * * P < 0.01.

Mutant SOD1 Interacts With VDAC1 to Mediate Mitochondrial Dysfunction
The direct interaction of mutant SOD1 with VDAC1 was previously determined by immunoprecipitation using anti-SOD1, anti-VDAC1 and anti-misfolded SOD1 antibodies together with mutant SOD1 from rat spinal cord tissues or using purified proteins in a lipid bilayer system Magri et al., 2016). Now, we extended these findings and showed using both MST and VDAC1 channel conductance that the dismutase-active SOD1 G93A mutant or the dismutase-inactive SOD1 G85R mutant but not SOD1 wt specifically interact with VDAC1.
Binding of misfolded SOD1 species to mitochondrial membranes was shown to disrupt transport of metabolites required for oxidative phosphorylation, reduce membrane potential, and the activity of electron transport chain complexes (Liu et al., 2004), and to the generation of reactive oxygen (ROS) or nitrogen species with damaging effects on respiratory chain complexes (Martin et al., 2007). All these effects can be produced by interactions of mutant SOD1 with VDAC1, which mediates the transport of metabolites, ions (including Ca 2+ ) and ROS .

VDAC1 Interacts With Mutant SOD1 Through the VDAC1 N-Terminal Domain
We identified the binding sites mediating VDAC1-mutant SOD1 interactions by showing that SOD1 G93A and SOD1 G85R do not bind the VDAC1 N-terminally truncated protein, but interact with the VDAC1-derived N-terminal peptide, suggesting that mutant SOD1 interacts with the VDAC1 N-terminal domain to mediate its cell toxic effects. This finding is not surprising as the N-terminal domain of VDAC1 was shown to be the interaction site for many proteins (Arzoine et al., 2008;Abu-Hamad et al., 2009;Arbel and Shoshan-Barmatz, 2010;Arbel et al., 2012;Geula et al., 2012) and to possess an ATP-binding site (Yehezkel et al., 2007). The first eight amino acids of the VDAC1-N-terminal domain are hydrophobic in nature, thus providing a natural possible site of contact with misfolded SOD1. Wild type recombinant SOD1 remains soluble, whereas mutations in the SOD1 protein leads to exposure of certain hydrophobic residues normally buried with the protein core. These structural changes lead to misfolding and aggregation of mutant SOD1 proteins via exposure of hydrophobic residues, such that they interact with intracellular membranes, such as mitochondria, ER and others (Israelson et al., 2015).
The interaction of misfolded SOD1 with VDAC1 is further suggested by the co-localization of two different SOD1 mutants with VDAC1 at the mitochondria ( Figure 5). Furthermore, the N-terminal-derived peptide decreasing the extent of this co-localization, points to VDAC1-N-terminus as the misfolded SOD1 interaction site. Surprisingly, we have noticed a tendency for a nuclear localization of VDAC1 in cells accumulating misfolded SOD1. This phenomenon is not clear and should be further investigated.
The VDAC1 N-terminal region is proposed to move within the channel pore (Hiller and Wagner, 2009) and to translocate from the internal pore to the channel surface (Geula et al., 2012), allowing it to interact with cytosolic proteins. The multiple glycine residues ( 21 GlyTyrGlyPheGly 25 ) following this domain represent a GXXXG motif that connects the N-terminal domain to β-strand 1, and confer the flexibility required for N-terminal domain translocation out of the channel pore (Geula et al., 2012). The GXXXG motif has been shown to be involved in dimerization in proteins such as glycophorin A (Gerber and Shai, 2001), human carbonic anhydrase (Whittington et al., 2001), yeast ATP synthase (Saddar and Stuart, 2005), carnitine palmitoyltransferase (Jenei et al., 2009), and others. In VDAC1, this motif is not required for VDAC1 dimerization but it might be involved in interaction with VDAC1-associated proteins (Geula et al., 2012). Interestingly, SOD1 contains three GXXXG (residues 11-16, 32-36, and 36-41) and two GXXXXG motifs (residues 51-56 and 56-61), but their importance for the interaction of VDAC1 with mutant SOD1 proteins is unknown.
In a PLB, reconstituted VDAC1 but not the N-terminally truncated protein bound a misfolded SOD1 mutant but not wild type SOD1, leading to reduced VDAC1 conductance. It is very likely that the exposure of residues in mutant SOD1 proteins that are normally hidden leads to increased interactions between SOD1 mutants and the VDAC1 N-terminal domain. Such association would reduce the overall transit of ions through the VDAC1 pore, thereby leading to the observed reduction in VDAC1 conductance. This interaction is expected to also reduce ATP, ADP, metabolite and ROS transport, in turn leading to an inhibition of cell growth and induction of mitochondrial dysfunction and cell toxicity (Figure 7). FIGURE 7 | Mutant SOD1 binds to VDAC1 and inhibits VDAC1 activities, with addition of a VDAC1 N-terminal peptide preventing such inhibition. Schematic model showing mutant SOD1 binding to VDAC1, with the N-terminal peptide serving as a decoy. (A) Mutant SOD1 is proposed to bind to the VDAC1 N-terminal domain and inhibit VDAC1 conductance, thereby suppressing both influx and efflux of different mitochondrial metabolites and ions, including Ca 2+ , and ROS. This reduction in metabolite flux results in reduced energy production and increased oxidative stress, leading to mitochondrial dysfunction and cell death. (B) VDAC1 N-terminal-derived peptides bind mutant SOD1 and prevent its association with VDAC1, thereby preventing mitochondria dysfunction. The N-terminal peptide thus provides a new therapeutic approach for inhibiting mutant SOD1 toxicity in ALS. OMM and IMM indicate outer and inner mitochondrial membrane, respectively while IMS indicates, the intermembrane space. We have shown here that different versions of the N-terminal peptide are able to suppress the toxicity of motor neuron-like NSC-34 cells expressing mutant SOD1 G93A in a concentrationdependent manner. For these experiments, we used short versions of the VDAC1 N-terminal domain peptide which lack the GXXXG motif, as this domain is required for N-terminal peptide-induced cell death (Figures 4A-C) but not for the interaction with mutant SOD1 (Figures 4D-F).
Finally, experiments with mESC-derived SOD1 G93Aexpressing MNs showed that the (1-20)-N-Ter-Antp VDAC1-based peptide significantly improved neurite outgrowth and the survival of SOD1 G93A -expressing MNs in a concentration-dependent manner (with an optimal peptide concentration of 10 µM). This not only corroborates the insight gained from studies on NSC-34 cells but also demonstrates that the VDAC1-based peptide acts as a neuroprotectant against misfolded SOD1-mediated MN toxicity. Further, these results underline the importance of a direct misfolded SOD1-VDAC1 interaction for the appearance of misfolded SOD1 toxicity. It remains to be shown whether the VDAC1-based (1-20)-N-Ter-Antp peptide also has the same effect on fully mature human motor neurons. Indeed, the results justify more thorough in vitro and in vivo analysis of the therapeutic potential of the VDAC1-based (1-20)-N-Ter-Antp peptide in treating ALS. These include improving peptide stability, such as by introducing amino acids in the D-conformation and carrying out toxicological studies, as performed with another VDAC1based peptide specifically inducing cell death of cancer cells (Shteinfer-Kuzmine et al., 2018).
In summary, we have shown here that SOD1 mutants interact with VDAC1 through the VDAC1 N-terminal domain to exert their inhibitory effect on VDAC1 channel conductance. Moreover, we have shown that VDAC1 N-terminal-derived peptides specifically bind mutant SOD1 and inhibit mutant SOD1-induced toxicity in motor neuron-like NSC-34 cells expressing mutant SOD1 or in mouse embryonic stem cellderived motor neurons. Thus, we suggest that this VDAC1-based peptide represents a new strategy for interfering with mutant SOD1-mediated cell toxicity.