The loss of LRPPRC function induces the mitochondrial unfolded protein response.

The inactivation of the LRPPRC gene, which has previously been associated with the neurodegenerative French Canadian Leigh Syndrome, results in a decrease in the production of mitochondria-encoded subunits of complex IV, thereby causing a reduction in complex IV activity. Previously we have shown that reducing complex IV activity triggers a compensatory and conserved mitochondrial hyperfusion response. We now demonstrate that LRPPRC knock-down in mammalian cells leads to an imbalance between mitochondria-encoded and nuclear-encoded subunits of complex IV and that this imbalance triggers the mitochondrial unfolded protein response (UPRmt). The inactivation of the LRPPRC-like gene mma-1 in C. elegans also induces UPRmt, which demonstrates that this response is conserved. Furthermore, we provide evidence that mitochondrial hyperfusion and UPRmt are coordinated but mediated by genetically distinct pathways. We propose that in the context of LRPPRC mma-1 knock-down, mitochondrial hyperfusion helps to transiently maintain mitochondrial ATP production while UPRmt participates in the restoration of mitochondrial proteostasis. Mitochondrial proteostasis is not only critical in pathophysiology but also during aging, as proteotoxic stress has been shown to increase with age. Therefore, we speculate that the coordination of these two mitochondrial stress responses plays a more global role in mitochondrial proteostasis.


INTRODUCTION
Mitochondria are essential eukaryotic organelles that participate in processes such as cellular energy production, cell signaling and apoptosis [1][2][3]. The production of cellular energy by mitochondria produces as a byproduct reactive oxygen species (ROS). While the effect of ROS on the accumulation of mutations in the mitochondrial genome is still under debate, ROS have been shown to promote protein oxidation and consequently misfolding and/or unfolding of these proteins inside mitochondria [4,5]. The resulting proteotoxic stress has been shown to increase with age and to participate in several age-related disorders, such as neurodegenerative diseases [6]. To maintain mitochondrial proteostasis and, hence, mitochondrial function, it is important that damaged proteins are Research Paper eliminated by mitochondrial proteases and that mitochondrial chaperones assist in the folding of nascent proteins. Another challenge that mitochondria face is the fact that the assembly of complexes of the Electron Transport Chain (ETC) in the inner mitochondrial membrane (IMM) requires a proper stoichiometric ratio of their subunits. Hence, misfolding of one subunit of an ETC complex results in the failure to assemble the entire complex. Consequently, the other subunits will accumulate in the mitochondrial matrix and thereby compromise mitochondrial proteostasis. Finally, ETC components are encoded by the mitochondrial and the nuclear genome. In order to maintain their respective stoichiometric ratios, the expression of mitochondrial and nuclear encoded proteins therefore has to be properly coordinated [7].
Several mitochondrial stress response pathways that maintain mitochondrial function have been described. One or more of these pathways are activated, depending on the extent to which mitochondrial proteostasis is compromised or mitochondria are damaged. For example, severely damaged mitochondria, which have irremediably lost their membrane potential, are eliminated from the functional mitochondrial network by mitochondrial-specific autophagy (mitophagy) (see for review [8]). The accumulation of unfolded or misfolded mitochondrial proteins activates the mitochondrial unfolded protein response (UPR mt ). This response leads to the increased production and import into mitochondria of chaperones (such as HSP60 and HSP70) and proteases (such as ClpP), which help misfolded proteins to fold properly or cause their degradation, respectively (see for review [9]). Finally, the mitochondrial hyperfusion response has recently been described in mammalian cells and in C. elegans. Various forms of stress, such as a decrease in complex IV activity, the inhibition of cytosolic protein synthesis or starvation, induce mitochondrial hyperfusion in order to maintain mitochondrial ATP production [10][11][12].
The human LRPPRC gene encodes a leucine-rich pentatricopeptide repeat containing protein that is imported into mitochondria and that is mutated in patients with French Canadian Leigh Syndrome, a neurodegenerative disorder associated with complex IV deficiency [13]. Inside the mitochondrial matrix, the LRPPRC protein is part of a ribonucleoprotein complex that post-transcriptionally controls the expression of specific mitochondrial mRNAs such as the mRNA coding for COX I, a component of complex IV [14,15]. We have previously shown that reducing LRPPRC function in mammalian tissue culture cells or reducing the function of the LRPPRC-like gene mma-1 (mma, mitochondrial morphology-abnormal) in C. elegans leads to a decrease in the level of COX I and, consequently, a decrease in complex IV activity [11]. This decrease in complex IV activity is compensated by an evolutionarily conserved mitochondrial hyperfusion response [11]. COX I is one of three mitochondriaencoded subunits of complex IV. The remaining 11 subunits in mammals and 6 subunits in C. elegans are nuclear-encoded [16,17]. Therefore, we hypothesized that the reduction of COX I protein level upon LRPPRC mma-1 knock-down might cause an imbalance between nuclear-and mitochondria-encoded subunits and, hence, trigger UPR mt . Here we report that in addition to triggering a mitochondrial hyperfusion response, LRPPRC mma-1 knock-down also triggers UPR mt in both mammalian cell cultures and C. elegans. We propose that these two responses act together to maintain and restore mitochondrial function, in response to decreased complex IV activity.

LRPPRC siRNA leads to an imbalance between mitochondria-encoded and nuclear-encoded subunits of complex IV and triggers UPR mt
Inactivation of LRPPRC results in a decrease in the production of mitochondria-encoded subunits of complex IV [11,14,15]. We reasoned that this decrease may lead to an imbalance between mitochondriaencoded and nuclear-encoded subunits of this complex and thereby trigger UPR mt . To test this hypothesis, we inactivated LRPPRC in SH-SY5Y cells using small interfering RNA (siRNA) and quantified the level of COX I protein (a mitochondria-encoded subunit of complex IV) and the level of COX IV protein (a nuclear-encoded subunit of complex IV). As shown in Figure 1A, the level of COX I protein decreases after 3 days of LRPPRC siRNA (down to 50% of the level in control siRNA cells). In contrast, the level of COX IV protein remains stable, resulting in an imbalance between COX I and COX IV subunits with ~1.8 times more COX IV subunits than COX I subunits upon LRPPRC siRNA.
In order to test whether this imbalance triggers UPR mt , we measured the levels of the mitochondrial chaperones HSP60 and HSP70 as well as the mitochondrial protease ClpP. We observed that 3 days of LRPPRC siRNA triggers a ~3-fold increase in the level of endogenous HSP70 protein ( Figure 1B; n=3; p=0.0473 by one sample t-test), a ~1.3-fold increase in the level of endogenous HSP60 protein ( Figure 1B; n=3; p=0.178 by one sample t-test) and a ~2.3-fold increase in the level of the endogenous mitochondrial protease ClpP ( Figure 1B; n=3; p=0.0202 by one sample t-test). Similar results were observed using HEK293T cells ( Figure S1). Based on these results, we conclude that the imbalance between nuclear-and mitochondriaencoded subunits of complex IV caused by LRPPRC siRNA triggers UPR mt .

The transient activation of UPR mt by LRPPRC siRNA correlates with the restoration of the balance between complex IV subunits
In order to test the effect of the induction of UPR mt upon LRPPRC siRNA on the balance between nuclearand mitochondria-encoded complex IV subunits, we performed a time course experiment. The imbalance between COX I and COX IV is first observed after 3 days of LRPPRC siRNA (Figure 2A). This time point www.impactaging.com also corresponds to the highest level of induction of HSP70 and ClpP ( Figure 2B). After 5 days of LRPPRC siRNA, the level of COX IV protein decreases (down to 50% of the level in control siRNA cells), restoring the balance between COX I and COX IV subunits ( Figure  2A). This time point coincides with a decrease of the levels of HSP70 and ClpP back to the levels measured in control cells ( Figure 2B). A similar transient activation of UPR mt was observed in HEK293T cells ( Figure S1). Therefore, the transient activation of UPR mt correlates with the restoration of mitochondrial proteostasis in the context of LRPPRC siRNA. www.impactaging.com

UPR mt and mitochondrial hyperfusion are transient responses to LRPPRC siRNA that follow similar kinetics
We have previously shown that mitochondrial hyperfusion is a transient response, which peaks at 3 days of LRPPRC siRNA ( Figure S2; day 3: 50% of cells have hyperfused mitochondria, 32% have tubular mitochondria and 18% have fragmented mitochondria). After 4 or 5 days of inactivation, the population of cells that display hyperfused mitochondria decreases while the population of cells that display fragmented mitochondria increases ( Figure S2; day 5: 35% of cells have hyperfused mitochondria, 45% have tubular mitochondria and 28% have fragmented mitochondria). Similarly, we observed that the induction of UPR mt is at its maximum at 3 days of LRPPRC siRNA, in both SH-SY5Y cells and HEK293T cells ( Figure 2B, Figure S1). After 4 and 5 days of inactivation, the UPR mt decreases in both cell lines ( Figure 2B, Figure S1B). Hence, in mammalian cell culture, UPR mt and the mitochondrial hyperfusion response both are transient and follow similar kinetics.

Inactivation of the LRPPRC-like gene mma-1 in C. elegans also induces UPR mt
In order to address whether the induction of UPR mt in response to LRPPRC siRNA is conserved, we inactivated the LRPPRC-like gene mma-1 in C. elegans. We used the transcriptional reporters P hsp-6 GFP, P hsp-60 GFP and P hsp-4 GFP to monitor the expression of the genes encoding the mitochondrial HSP70 chaperone (hsp-6), the mitochondrial HSP60 chaperone (hsp-60) or the endoplasmic reticulum (ER) chaperone BiP (hsp-4), respectively [18]. We reduced mma-1 function using RNA-mediated interference (RNAi) by feeding [19]. Specifically, we grew L4 larvae of the reporter strains on RNAi plates seeded with an Escherichia coli strain expressing double-strand RNA (dsRNA) of mma-1. The plates were seeded with mma-1(RNAi) bacteria diluted 1:5 (v/v) with mock(RNAi) bacteria; hereafter referred to as 'mma-1(RNAi) 1:5 dil.'. The expression of the reporters was analyzed four days later in the F1 generation. Using these conditions, the level of MMA-1 protein is reduced by ~40% (see below, Figure S4). As shown in Figure 3, the P hsp-6 GFP reporter is strongly up-regulated in response to mma-1(RNAi) (on average ~200-fold compared to mock(RNAi); Figure 3A; n=5; p=0.037 by one sample t-test). P hsp-60 GFP expression was also increased in mma-1(RNAi) animals compared to mock(RNAi) animals, albeit to a lesser extent ( Figure  3B; n=9; p=0.011 by one sample t-test). Finally, mma-1 knock-down did not affect the expression of the ER chaperone hsp-4 ( Figure 3C; n=5; p=0.36 by one sample t-test). We also tested the effect of reducing the activity of the C. elegans homolog of paraplegin spg-7, which has previously been shown to induce UPR mt [18]. Using the RNAi conditions described above, the inactivation of spg-7 induces a strong up-regulation of the P hsp-6 GFP reporter and a weak up-regulation of the P hsp-60 GFP reporter but has no effect on the expression of the P hsp-4 GFP reporter ( Figure 3).
In order to confirm the effect of mma-1(RNAi) on the expression of hsp-6 and hsp-60 at the levels of endogenous HSP-6 and HSP-60 protein, we used a mouse monoclonal anti-HSP-60 antibody [20] and generated rabbit polyclonal anti-HSP-6 antibodies. We confirmed that the anti-HSP-6 antibodies specifically recognize HSP-6 ( Figure S3). Using these antibodies we found that the inactivation of mma-1 by RNAi leads to a ~1.7-fold increase in the level of endogenous HSP-6 protein ( Figure 3D; n=8; p=0.024 by one sample t-test).
In addition, we found that mma-1(RNAi) induces a ~1.4fold increase in the level of endogenous HSP-60 protein ( Figure 3E; n=10; p=0.022 by one sample t-test). The discrepancy between the level of induction for endogenous proteins (~1.4-fold and ~1.7-fold for HSP-60 and HSP-6, respectively) and for transcriptional reporters (~20-fold and ~200-fold for P hsp-60 GFP and P hsp-6 GFP, respectively) is most likely due to the copy number of the P hsp-6 GFP and P hsp-60 GFP transgenes integrated in the genome of the reporter strains; however, we cannot exclude that this discrepancy is also caused by post-transcriptional regulations of hsp-6 and/or hsp-60 expression. Based on these results we conclude that mma-1(RNAi) induces UPR mt and the transcriptional up-regulation of the hsp-6 and hsp-60 genes, which results in a significant increase in the level of endogenous HSP-6 and HSP-60 protein.

mma-1(RNAi)-induced up-regulation of hsp-6 transcription is dependent on ATFS-1 and HAF-1
In C. elegans, UPR mt is dependent on the peptide exporter HAF-1 as well as the bZip transcription factor ATFS-1 [21,22]. Consistent with previous studies [22], we confirmed that the inactivation of atfs-1 by RNAi suppresses the up-regulation of the P hsp-6 GFP reporter induced by spg-7(RNAi) (Figure 3F; n=7; p=0.0001 by one sample t-test). Furthermore, we found that mma-1(RNAi)-induced UPR mt is suppressed by atfs-1(RNAi) as well ( Figure 3F; n=10; p=0.0001 by one sample ttest). To rule out that the effects observed are due to differences in RNAi efficiency, we quantified the knock-down of mma-1 using anti-MMA-1 antibodies [11]. As shown in Figure S4A, the efficiency of mma-1(RNAi) in atfs-1(RNAi) animals is similar to that in wild-type animals (reduction by ~40% on average). Hence, the mma-1(RNAi)-induced up-regulation of hsp-6 transcription is dependent on ATFS-1. In contrast, the haf-1 loss-of-function mutation ok705 only partially suppresses the up-regulation of P hsp-6 GFP in response to mma-1(RNAi) (Figure 3G; n=5; p=0.0217 by one sample t-test). We observed a similar result for spg-7(RNAi)-induced P hsp-6 GFP up-regulation ( Figure 3G; n=6; p=0.017 by one sample t-test). We confirmed that the effects observed were not due to differences in the efficiency of mma-1(RNAi) ( Figure S4B). Therefore, mma-1(RNAi) induces UPR mt in a manner that is dependent on the bZip transcription factor ATFS-1 and that is partially dependent on the peptide exporter HAF-1.

Strong depletion of MMA-1 protein leads to both mitochondrial hyperfusion and UPR mt induction whereas mild depletion of MMA-1 protein only leads to UPR mt
When growing animals on 'non-diluted' mma-1(RNAi) plates, we achieved on average an ~80% reduction in the level of MMA-1 protein ( Figure 4A; mma-1(RNAi)). Under these conditions, as previously reported, animals www.impactaging.com exhibit hyperfused mitochondria in body wall muscle cells (76%; Figure 4A; n=17; two independent experiments) [11]. These animals also show a stronger UPR mt as reflected by the up-regulation of the P hsp- 6 GFP reporter (on average ~600-fold compared to mock(RNAi) animals; Figure 4A). When reducing the level of MMA-1 protein by only 40% (mma-1(RNAi) 1:5 dil.), the P hsp-6 GFP reporter is still up-regulated (on average 200-fold compared to mock(RNAi) animals; Figure 4A); however, these animals do not exhibit mitochondrial hyperfusion (6%; Figure 4A; n=16; two independent experiments). Hence, mild depletion of MMA-1 protein leads to the activation of the UPR mt pathway and strong depletion of MMA-1 protein additionally leads to the activation of the mitochondrial hyperfusion response pathway. www.impactaging.com

Figure 5. Mitochondrial hyperfusion and UPR mt act together to protect mitochondrial function. (A) Complex
IV of the ETC is composed of mitochondria-encoded and nuclear-encoded subunits. Upon LRPPRC mma-1 RNAi, the level of mitochondria-encoded subunits decreases, leading to a reduction in complex IV activity, which triggers the mitochondrial hyperfusion response and an accumulation of unassembled nuclear-encoded subunits, which triggers UPR mt . (B) While mitochondrial hyperfusion helps to maintain ATP level in response to a reduction in complex IV activity [11], UPR mt helps to restore the balance between mitochondria-encoded and nuclear-encoded subunits of this complex.
Mitochondrial hyperfusion can only compensate for reduced complex IV activity for a limited time. Prolonged inactivation of LRPPRC mma-1 eventually leads to mitochondrial fragmentation and drop in ATP level as previously shown [11].
www.impactaging.com background is slightly reduced; however, the difference is not statistically significant (p=0.22 by Student t-test). Based on these results we conclude that mma-1(RNAi)induced mitochondrial hyperfusion is independent of ATFS-1 and HAF-1.

ATFS-1-dependent UPR mt is not essential for viability in response to mma-1(RNAi)
We previously showed that mma-1(RNAi) causes synthetic embryonic lethality and synthetic midlarval arrest in the fzo-1(tm1133) loss-of-function mutant background, indicating that mitochondrial hyperfusion is essential for viability in response to mma-1(RNAi) [11].
To test whether UPR mt is also essential for viability in response to mma-1(RNAi), we inactivated mma-1 in animals carrying a loss-of-function mutation in the atfs-1 gene (atfs-1(tm4525)) [22]. We found that mma-1(RNAi) does not cause any obvious embryonic lethality or midlarval arrest in atfs-1(tm4525) mutant animals ( Figure  S5A). We also performed the experiment using double RNAi (afts-1(RNAi)+mma-1(RNAi)) and did not observe any obvious synthetic lethality or arrest ( Figure S5B). This indicates that in response to mma-1(RNAi), mitochondrial hyperfusion is essential for viability while ATFS-1-dependent UPR mt is not.

DISCUSSION
Depending on the type and the severity of the stress to which mitochondria are exposed, different mitochondrial stress responses (such as UPR mt , mitochondrial hyperfusion or mitophagy) can be activated in order to protect mitochondrial function [23]. Here, we show that reducing LRPPRC or mma-1 function leads to an evolutionary conserved coactivation of UPR mt and mitochondrial hyperfusion. Depletion of LRPPRC or MMA-1 protein results in a decrease in the production of mitochondria-encoded subunits of complex IV and thereby causes a reduction of complex IV activity [11,14,15]. Depletion of LRPPRC or MMA-1 protein also leads to an imbalance between nuclear-encoded and mitochondria-encoded subunits of this complex. We show that the activation of the UPR mt occurs when this balance and, hence mitochondrial proteostasis, is disrupted (Figure 2; day 3 of LRPPRC siRNA). This finding suggests that the accumulation of unassembled complex IV subunits is most likely the signal that triggers UPR mt . Furthermore, our data indicate that UPR mt is transiently activated until mitochondrial proteostasis is restored (Figure 2; day 5 of LRPPRC siRNA). We, therefore, propose that the transient activation of UPR mt participates in the restoration of mitochondrial proteostasis by either proteolytic degradation of unassembled subunits by ClpP or assisted folding of nascent subunits by HSP70.
In addition, the response may also involve the downregulation of the genes encoding nuclear-encoded subunits of complex IV.
Whereas UPR mt and mitochondrial hyperfusion in response to LRPPRC mma-1 RNAi follow similar kinetics, genetic experiments in C. elegans indicate that the pathways that mediate these two responses are distinct. Specifically, LRPPRC mma-1 RNAi-induced UPR mt is dependent on the previously described HAF-1, ATFS-1 pathway, whereas LRPPRC mma-1 RNAiinduced mitochondrial hyperfusion is not. This notion is furthermore supported by the fact that in C. elegans, a mild depletion of MMA-1 protein activates only the UPR mt pathway whereas a strong depletion of MMA-1 activates both pathways. The pathway through which a reduction in LRPPRC mma-1 function induces mitochondrial hyperfusion remains to be elucidated. Similarly, the mechanism through which mitochondrial hyperfusion and UPR mt are coordinated is currently unknown.
We propose that the mitochondrial hyperfusion response and UPR mt act together to maintain and restore mitochondrial function ( Figure 5). We have previously shown that mitochondrial hyperfusion helps to maintain cellular ATP levels despite a reduction in complex IV activity caused by LRPPRC mma-1 RNAi [11]. We now propose that this transient compensation enables the cell to restore the balance between mitochondria-encoded and nuclear-encoded subunits of complex IV. However, as previously shown, prolonged inactivation of LRPPRC mma-1 eventually leads to mitochondrial fragmentation, a drop in ATP level and the loss of cellular and organismal functions [11]. The two mitochondrial stress response pathways we unraveled are therefore likely to be important for the maintenance of mitochondrial function in response to moderate and short-term changes in mitochondrial proteostasis. Interestingly, UPR mt as well as an 'elongation' of mitochondria are also induced in response to the infection of C. elegans by Pseudomonas aeruginosa [24]; however, how these two stress responses are induced in this context remains to be determined. In contrast, UPR mt and mitochondrial hyperfusion may not protect against more drastic and long-term changes in mitochondrial proteostasis, such as in C. elegans in which mma-1 has been chronically inactivated or in French Canadian Leigh Syndrome patients carrying mutations in the LRPPRC gene. In cell lines derived from these patients, mitochondria are fragmented [15], which indicates that the mitochondrial hyperfusion response has already failed. Whether UPR mt is activated in these cell lines is currently unknown. In case it is www.impactaging.com activated, it will be interesting to determine whether this pathway contributes to the pathophysiology of this neurodegenerative disorder, which has not been completely elucidated yet.
We previously showed that the mitochondrial hyperfusion response is essential for viability in response to mma-1(RNAi) [11]. In contrast, we present evidence that ATFS-1-dependent UPR mt is not essential for viability upon inactivation of mma-1. Furthermore, we show that ATFS-1-dependent UPR mt is also not essential for viability in response to spg-7(RNAi). One possible explanation for this finding is the existence of an ATFS-1-independent UPR mt pathway. Indeed, Haynes and co-workers have shown that while spg-7(RNAi) induces the up-regulation of 685 genes, 294 of these genes are still up-regulated in atfs-1(tm4525) mutant animals [22]. Hence, in the context of mma-1 inactivation, the ATFS-1-independent UPR mt in combination with the mitochondrial hyperfusion response may be sufficient for the completion of embryonic and larval development.
The effect of the activation of UPR mt on aging remains controversial. While Houtkooper et al reported an extension of lifespan in response to the activation of UPR mt [7], Bennett et al observed no clear correlation between UPR mt and lifespan [25]. We and others have previously shown in C. elegans and in rat that mitochondrial fragmentation increases with age and that this fragmentation is accompanied with a loss of mitochondrial volume [26,27]. Interestingly, a recent study reports that fibroblasts isolated from long-lived human individuals exhibit hyperfused mitochondria [28]. While the genetic determinants resulting in this change in mitochondrial morphology remain unknown, the mitochondrial hyperfusion in the cells of these individuals has been shown to increase mitochondrial mass by preventing mitophagy [28]. Consequently, while the activity of their individual mitochondria decreases with age, the overall mitochondrial function of their cells is maintained, implicating a beneficial role of mitochondrial hyperfusion in aging [28]. In conclusion, a better understanding of UPR mt and the mitochondrial hyperfusion response has not only the potential to increase our knowledge of how cells respond to mitochondrial stress under physiological and pathophysiogical conditions (i.e. French Canadian Leigh Syndrome) but also during aging.

MATERIALS AND METHODS
General C. elegans methods and strains. C. elegans strains were cultured as previously described [29]. Bristol N2 was used as the wild-type strain. Mutations used in this study were described by Riddle and coworkers [30] except: (LG IV) haf-1(ok705) (OMRF Knockout Group); (LG V) atfs-1(tm4525) (National BioResource Project). A summary of the transgenic lines is provided in Table S1.
C. elegans RNA interference. RNAi by feeding was performed as previously described [11] with the following modifications. RNAi plates containing 1 mM IPTG were inoculated with 50 μl of 0.5 OD 600nm /ml of concentrated mma-1(RNAi) or spg-7(RNAi) bacterial clones from the Ahringer library [31]  culture medium (25 mM glucose) was added. For the analysis of mitochondrial morphology, cells were kept in normal culture medium until harvested. For the analysis of UPR mt , the normal culture medium was replaced with medium containing only 5 mM glucose 24 hours before analysis.
Protein analysis. Rabbit polyclonal anti-HSP-6 antibodies were generated using the antigenic peptide C 472 QEAKTAEEPKKEQN (cysteine plus HSP-6 Cterminal sequence) by Thermo Fisher Scientific, as previously described [33] and used at 1:5000 for Western analysis. Worm protein extracts were subjected to SDS/PAGE and Western. To detect HSP-60, GFP and α-Tubulin, we used a mouse monoclonal antibody developed by Nonet and co-workers and available at DHSB [20] (1:2000), rabbit polyclonal antibodies from Abcam ab290 (1:2000) and a mouse monoclonal antibody from Abcam ab7291 (1:2000), respectively. Images were quantified using the ChemiDoc XRS+ System (Bio-Rad). The data presented are ratios relative to the mock(RNAi). Only samples for which mma-1 knock-down was at least 20% were used for the statistically analysis. Statistical analysis was performed by using one sample t-test. Normality of the data was assessed by Shapiro-Wilk Normality test.
Analysis of mitochondrial morphology in SH-SY5Y cells. Mitochondrial morphology was analyzed as previously described with the following modifications [11]. Cells grown on glass coverslips were fixed with PBS 3.7% PFA at 20°C for 10 minutes and permeabilized with PBS 0.2% Triton X-100 for 10 minutes at 20°C. The coverslips were blocked with PBS 5% BSA for 1 hour at 4°C and incubated overnight at 4°C in PBS 5% BSA with polyclonal anti-TOM20 antibodies (1:2000, Santa Cruz Biotechnology). After three washes with PBS, the coverslips were incubated with an Alexa555 goat anti-rabbit antibody (1:2000, Molecular Probes) in PBS 5% BSA for 2 hours at 20°C. After three washes with PBS, three washes with PBS 0.2% Tween and three washes with PBS, the glass coverslips were mounted on glass slides with MOWIOL 4-88 containing DAPI (Sigma). At least 300 cells per coverslip were counted in a blinded manner. Results are based on three independent experiments.