Prevention of mitochondrial impairment by inhibition of protein phosphatase 1 activity in amyotrophic lateral sclerosis

Amyotrophic lateral sclerosis (ALS) is a fatal neurodegenerative disease caused by progressive loss of motor neurons (MNs) and subsequent muscle weakness. These pathological features are associated with numerous cellular changes, including alteration in mitochondrial morphology and function. However, the molecular mechanisms associating mitochondrial structure with ALS pathology are poorly understood. In this study, we found that Dynamin-related protein 1 (Drp1) was dephosphorylated in several ALS models, including those with SOD1 and TDP-43 mutations, and the dephosphorylation was mediated by the pathological induction of protein phosphatase 1 (PP1) activity in these models. Suppression of the PP1-Drp1 cascade effectively prevented ALS-related symptoms, including mitochondrial fragmentation, mitochondrial complex I impairment, axonal degeneration, and cell death, in primary neuronal culture models, iPSC-derived human MNs, and zebrafish models in vivo. These results suggest that modulation of PP1-Drp1 activity may be a therapeutic target for multiple pathological features of ALS.


Introduction
Amyotrophic lateral sclerosis (ALS) is a neurodegenerative disorder characterized by progressive and selective loss of motor neurons (MNs) 1,2 . The clinical symptoms of ALS include muscle weakness, dysphagia, and speech problems, and the disorder eventually leads to death from dyspnea within three to 5 years of diagnosis 3,4 . Most ALS cases are sporadic, and roughly 10% of them are familial forms caused by various genetic mutations in specific genes, including superoxide dismutase 1 (SOD1), transactive response DNA-binding protein-43 (TDP-43), and chromosome 9 open reading frame 72 (C9orf72) 5,6 . Mutations in these genes are also found in the sporadic form of ALS 1 . For example, hexanucleotide repeat expansion of C9orf72 was identified in a large proportion of both familial and sporadic forms of ALS patients, suggesting that somatic or de novo mutations in causative genes may also be responsible for sporadic ALS 7,8 . Although the causes of ALS may differ, most ALS cases exhibit common pathological features with biochemical/cellular hallmarks such as protein aggregations, abnormal Ca 2+ buffering 9 , glutamate excitotoxicity 10 , glial activation 11 , disruption of RNA metabolism 12 , and mitochondrial dysfunction 13,14 . Therefore, it has been postulated that multiple cellular changes are involved in the progression of ALS symptoms, making it difficult to develop effective treatments. Although riluzole and edaravone are the drugs for ALS currently approved by FDA, their efficacies are modest in clinical trials, and their pharmacological targets appear to be different [15][16][17] . It is, therefore, of great importance to develop novel treatments targeting several of the common pathological features responsible for the complex nature of ALS pathology.
Mitochondrial dysfunction is commonly found in both familial and sporadic ALS cases [18][19][20] , suggesting that mitochondria may serve as an important hub for ALS pathophysiology. For example, it has been reported that mitochondria were morphologically fragmented and functionally deteriorated from the pre-symptomatic period in patients and mouse models of ALS (SOD1 G93A) 21,22 . C9ORF72-associated poly(GR) compromises mitochondrial function and induces mitochondrial fragmentation, leading to DNA damage and an increase in cell death 23,24 . In sporadic ALS cases, mitochondrial morphological abnormalities have been detected in the anterior horn of the spinal cord 19 . Mitochondria are essential for cellular ATP production, and MNs require large amounts of energy from the mitochondria in order to function. Thus, it has been suggested that MNs are one of the most vulnerable types of cells to mitochondrial dysfunction [25][26][27] . On the other hand, the morphological changes in the mitochondria in ALS pathology may hint at additional links between mitochondria and MN function. The morphological dynamics of mitochondria are homeostatically controlled by the balance between continuous fission and fusion, and this is necessary for the maintenance of mitochondrial quality 28,29 . Mitochondrial dynamics are controlled by large GTPase dynamin-related proteins, including Dynamin-related protein 1 (Drp1), Mitofusin1/2 (MFN1/2), and Optic atrophy 1 (Opa1) 30 . Recently, it was reported that the expression of Drp1 was induced in the spinal cord of SOD1 G93A model mice 31 , and the suppression of Drp1 inhibited mutant SOD1-induced mitochondrial fragmentation and cell death in vitro 2 , suggesting that alteration of Drp1 activity may mediate at least some aspects of MN degeneration in ALS. However, the precise molecular mechanisms underlying the Drp1 function and ALS pathology are yet to be described.
In an attempt to identify novel therapeutic targets for ALS, we examined whether Drp1-dependent mitochondrial dynamics are an effective target for ALS therapy and explored the upstream mechanism modulating Drp1 activity in ALS models. To this end, we identified that Drp1 is pathologically dephosphorylated in SOD1 G93A mice, and the blockade of Drp1 activity prevents cell death induced by the expression of mutant SOD1 or TDP-43. Furthermore, we identified PP1 as a phosphatase responsible for the dephosphorylation of Drp1 and found that the inhibition of PP1 activity suppresses neurotoxicity in several ALS models. Based on our observations, we propose that pathological induction of PP1 plays a key role in the progression of cellular events associated with ALS, providing novel therapeutic targets for ALS.

Materials and methods
Animal model and treatment SOD1 G93A transgenic mice were maintained as a hemizygous line by mating with wild type C57BL/6J mice.
Immunohistochemistry SOD1 G93A mice of each pathological stage were perfused in 4% paraformaldehyde, and the lumbar regions of the spinal cord were dissected. The tissues were post-fixed overnight at 4°C. The fixed tissues were soaked in 30% sucrose/1× PBS for cryo-protection. The lumbar regions of the spinal cord were cut serially into sections 40 μm thick on a cryostat, and these were transferred into 50% glycerol in PBS. The sections were blocked in 1xPBS containing 3% BSA and 0.2% Triton-X100 for 30 min at room temperature. After blocking, tissues were incubated with primary antibody overnight at 4°C and washed three times in 1× PBS. The primary antibodies used were phospho-DRP1 S616 (Cell Signaling, Cat. 3455), DRP1 (BD, Cat. 611113), and ChAT (Millipore, Cat. AB144P). The tissues were incubated with fluorescence-labeled secondary antibody and Hoechst33342 (10 μg/ml; Sigma, St. Louis, MO) for 30 min at room temperature and washed three times in 1× PBS. They were then mounted on slides and observed using a fluorescence or confocal microscope (Zeiss, Goettingen, Germany).

PP1 activity assay
PP1 activity was measured by the ProFluor Ser/Thr PPase assay kit (Promega, Cat. V1260). This assay is based on the dephosphorylation of a phosphorylated bisamide rhodamine 110 peptide substrate. In order to carry out a PP1 activity assay of the lumbar spinal cord, the tissue was sonicated in a lysis buffer (5 M NaCl, 1 M Tris-Cl, pH 7.5, 0.5 M EDTA, 0.5 M EGTA, 0.2 M Na 3 VO 4 , 0.1 M PMSF) and centrifuged at 10,000 rpm for 10 min. Following centrifugation, the supernatant was divided into two tubes and treated with either 0.2 nM (for PP2A inhibition) or 1 μM okadaic acid (for PP1 inhibition). The level of PP1 activity of each sample was measured based on the absorbance at 488 nm of nonphosphorylated S/T PPase R110 substrate and then normalized by protein levels (BCA assay). Relative PP1 activity was calculated as the value of phosphatase activity level (treated with 0.2 nM okadaic acid) minus the value of background phosphatase activity (treated with 1 μM okadaic acid) 34 .

Assay of NADH dehydrogenase and complex I activity of mitochondria
In order to carry out a histological assay of NADH dehydrogenase, the lumbar spinal cord was snap-frozen in pre-chilled isopentane and sectioned using a cryostat. Phosphate buffer (0.343 g NaH 2 PO 4 in 100 ml H 2 O) and NBT (250 mg nitro blue tetrazolium in 100 ml H 2 O) were prepared. The sections were incubated in a solution containing 5 mg NADH, 1 ml saline, 1 ml phosphate buffer, and 4 ml NBT with 2 ml distilled water. After incubation for 30 min at room temperature in humidity chamer, the sections were rinsed in distilled water and observed using a microscope. The mitochondrial complex I activity of the lumbar spinal cord and cultured neurons was measured using a microplate assay kit (Abcam, Cat. Ab109721) according to the manufacturer's instruction 35 .

Measurement of mitochondrial membrane potential (ΔΨ m ) and velocity
Cultured cortical neurons were incubated with 500 nM TMRM (tetramethylrhodamine, methyl ester; Invitrogen, Cat. T668) for 30 min at 37°C. The signal of TMRM was imaged using a Carl Zeiss Observer Z1 with x630 magnification, and its intensity was measured using ImageJ.

Statistical analysis
All samples or animals were assigned randomly, and the experiment was performed in a blinded fashion. The data were tested for differences between multiple groups using one-way ANOVA followed by Scheffe's or Tukey's multiple comparisons tests for post hoc comparisons. Differences between the means of two groups were tested using the two-sided Student's t-test.

Ethical statements
All animal maintenance and experimental procedures were approved by members of the Laboratory Animal Research Center at Korea University College of Medicine. The use of hiPSCs was approved by the Institutional Review Board and Ethics Committee at the University of California, San Francisco (UCSF).

Changes in mitochondrial fission proteins in ALS models
We first explored whether molecules responsible for the morphological dynamics of mitochondria are altered in SOD1 G93A (G93A) ALS model mice. Although the expression level of Drp1 protein remained unaltered, the phosphorylation level of Drp1 S616 was significantly reduced at the onset period (when the mice were 100 days old) (Fig. 1a). The expression level of Fis1, a receptor for Drp1, also markedly increased during this period (Fig. 1a). Immunohistochemical labeling of the lumbar spinal cord with a phosphorylation-specific antibody against Drp1 S616 demonstrated that the reduction of Drp1 S616 phosphorylation mainly occurred in the MNs of G93A mice (Fig. 1b). Next, we overexpressed G93A or TDP-43 Q331K (Q331K) associated with ALS in primary neurons and found that these pathogenic mutants caused dephosphorylation of Drp1 (Fig. 1c). Collectively, these results suggest that pathogenic stimuli associated with ALS cause dephosphorylation of Drp1.

Prevention of Drp1 signaling reduces ALS-related neurodegeneration
It has been reported that phosphorylation of Drp1 S616 is controlled by cdk5, which prevents the oligomerization and subsequent mitochondrial localization of Drp1 in neurons 36 . Dephosphorylation of Drp1 S616 and the induction of Fis1 expression appear to promote Drp1-dependent mitochondrial fission, which is associated with neurodegeneration. In keeping with this, we found that the overexpression of mutant G93A in primary cortical neurons promoted mitochondrial fragmentation. However, suppression of Drp1 activation by transfection with shDrp1 or shFis1, and treatment with a chemical inhibitor (Mdivi-1), consistently prevented the mitochondrial fragmentation induced by ALS-related mutations, suggesting that Drp1-Fis1 signaling is necessary for ALS-related mitochondrial fragmentation ( Fig. 2a and Supplementary  Fig. 1a). Next, we tested whether a blockade of Drp1-Fis1 would prevent neurodegeneration. Following transfection with ALS-related mutant genes (SOD1 G93A and TDP-43 Q331K), neuronal death was enhanced, as assessed by activated caspase-3 immunoreactivity (Fig. 2b-d) and pyknotic nuclei (Supplementary Fig. 1b-d). Blocking Drp1 activation effectively prevented this cell death. Finally, we tested the effects of a blockade of Drp1 activity in vivo using an ALS zebrafish system (Fig. 2e). Degeneration of motor axons induced by transient overexpression of SOD1 G93A or TDP-43 Q331K was readily detectable in the models, but the blockade of Drp1 activity through treatment with Mdivi-1 effectively prevented this axonal degeneration. These results indicate that Drp1 activation contributes to ALS-related neurodegeneration in vivo.

PP1 is involved in Drp1 dephosphorylation in G93A mice
Because dephosphorylation of Drp1 was found to be one of the events associated with enhanced mitochondrial fission and neurodegeneration in the ALS models, we explored the upstream phosphatase(s) responsible for this dephosphorylation, under the assumption that upstream phosphatase activation may cause the Drp1-dependent mitochondrial defects found in ALS models (Fig. 3). The blockade of PP1 by either of two selective inhibitors, cantharidin or inhibitor-2 (I-2), promoted the phosphorylation of Drp1 S616 in a dose-dependent manner (Fig. 3a). Furthermore, the dephosphorylation of Drp1 was completely blocked by I-2 treatment in cortical neurons through G93A overexpression (Fig. 3b). These observations suggest that the activation of PP1 is responsible for the Drp1 dephosphorylation associated with ALS pathology. Moreover, we found that the lumbar spinal cord of G93A mice exhibited significantly higher PP1 activity in the symptomatic period than in the pre-symptomatic period, and the enhanced PP1 activity was maintained until the end stage (Fig. 3c). It has been reported that PP1 activity is regulated by its phosphorylation status, and dephosphorylation is required for its activation 37,38 . Western blot analysis showed that PP1α was activated in the pre-symptomatic period of ALS (Fig.  3d) and the phosphorylation of PP1 was decreased in the ChAT+ neurons of G93A mouse at pre-and postsymptomatic period, as well as in mSOD1 or TDP43 expressing cortical neurons ( Supplementary Fig. 2). Collectively, these results demonstrated that pathologic induction of PP1 activity may cause Drp1 activation and subsequent induction of cell death in ALS models.

PP1 inhibition rescues neurons from neurodegeneration
To test whether PP1 inhibition is sufficient for the prevention of Drp1-dependent mitochondrial fission in the ALS model, we first examined whether PP1 inhibition  prevented ALS-related mutant-dependent mitochondrial fragmentation. PP1 has three catalytic isoforms, α, β and γ, and I-2 can inhibit the action of all of them. Therefore, in order to explore their importance, the effects of specific shRNA against each isoform on mitochondrial fission were tested. Interestingly, the suppression of PP1α, γ, and I-2, but not PP1β, prevented the excessive fission of mitochondria in G93A and Q331K-overexpressed primary cortical neurons (Fig. 4a, b and Supplementary Fig.  3). Next, we tested whether suppression of PP1 prevented SOD1 G93A and TDP-43 Q331K-induced cell death. PP1 suppression by I-2 and shPP1γ reduced the number of active caspase-3-positive or pyknotic cells in the ALSrelated mutant gene-overexpressed neurons (Fig. 4c, d).
On the other hand, we found that the suppression of PP1α by shRNA significantly enhanced cell death. Yet the suppression of PP1 β and γ by shRNA significantly reduced the number of pyknotic cells in SOD1 G93A-or TDP-43 Q331K-overexpressed neurons (Fig. 4d). These results indicate that different PP1 isoforms may mediate different outcomes in ALS pathology via different substrate specificities. Furthermore, we also found that I-2 prevented the ALS-related mutant-induced retardation of axonal growth in zebrafish models (Fig. 4e, f). Expression of SOD1 G93A and/or TDP-43 Q331K reduced axonal growth, whereas I-2, shPP1α, and shPP1γ each restored axonal growth to a level equal to the control. Interestingly, this growth-restoring effect was not observed in Mdivi-1 treatments, suggesting that retardation of axonal growth was not entirely dependent on Drp1 activation and that PP1 may have additional targets for the progression of ALS-related axonal degeneration/growth retardation.

Blockade of PP1 restores mitochondrial respiratory functions
Because the mitochondrial bioenergetics is expected to influence axonal growth, we tested whether PP1 activity is also associated with the alteration of mitochondrial respiratory complex activities (Fig. 5). It has been reported that complex I activity is primarily affected in G93A mice 39 , and we confirmed that NADH dehydrogenase, which represents mitochondrial complex I activity, was substantially reduced in the lumbar spinal cord of G93A mice, compared to WT (Fig. 5a). Next, we tested whether a blockade of PP1 or Drp1 activity rescued the impairment of mitochondrial function induced by the overexpression of SOD1 G93A or TDP-43 Q331K, which rapidly reduced complex I activity and mitochondrial membrane potential (MMP; Fig. 5b, c) [40][41][42] . However, I-2 effectively prevented the reduction of mitochondrial complex I activity and MMP in the mutant-overexpressed neurons, while Mdivi-1 treatment failed to rescue these functional impairments of the mitochondria. These results suggest that pathological PP1 activity impaired mitochondrial respiration via a Drp1-independent route as well as through Drp1-mediated mitochondrial impairments.

Blockade of PP1 activity alleviates ALS pathology in vivo and in a human iPSC model
Next, we tested whether PP1 inhibition can effectively prevent ALS-related neurodegeneration in a zebrafish model. PP1 inhibition by treatment with I-2 or okadaic acid (OA) prevented the axonal defects induced by SOD1 G93A or TDP-43 Q331K mutations (Fig. 6a, b). Interestingly, I-2 restored the motor axonal length of mutantexpressing MNs to a level equivalent to the control, which was significantly superior to the effects of axonal elongation by the Mdivi-1 treatment (Fig. 6c). Finally, we also tested whether these preventive effects of I-2 could be observed in a human iPSC-derived MN model in vitro. First, we differentiated multiple iPS cell lines from controls and TDP43 mutants (mTDP-43;A90V, Q343R, M337V) 43 into spinal MNs as described 21 . PP1 suppression induced by I-2 treatment over 2 weeks inhibited mitochondrial fragmentation, although in the periods between administrations of I-2, the mTDP group exhibited noticeable mitochondrial shortening (Fig. 6a). As anticipated, the number of active caspase-3-positive neurons and pyknotic cells was lower in the I-2-treated mTDP group, suggesting that inhibition of PP1 by I-2 suppresses neuronal toxicity in several ALS-associated models. The prevention of pathological PP1 activity could, therefore, be a viable therapeutic target.

Discussion
Recent large-scale genetic analyses have revealed that genes mediating many different biological cascades are associated with ALS, suggesting that many different pathways should merge into a common 'hub' cascade to exhibit common symptoms such as MN death, axonal degeneration, and mitochondrial defects 44,45 . In this study, we identified that dephosphorylation of Drp1 is mediated by pathological activation of PP1 induced by SOD1 G93A or TDP-43 Q331K overexpression in cellular models, and in mouse models with the G93A mutation. Blocking the PP1-Drp1 cascades effectively prevented mitochondrial defects and subsequent neurodegeneration in various ALS models. The gain-of-function mutation of SOD1 is proposed to impair protein misfolding-mediated clearance pathways, including ubiquitin-dependent proteolysis, protein aggregations, and autophagy, in addition to causing mitochondrial dysfunction [46][47][48] . On the other hand, TDP-43 is an RNA-binding protein that may trigger defects in mRNA processing, with multiple consequences including defective nucleocytoplasmic transport 49 , mitochondrial deficits, and abnormalities in mitophagy 50,51 . Given that these different causative mutations commonly compromise mitochondrial function, which can be prevented by the blockade of PP1-Drp1 cascades, these form an important common biological hub for ALS pathology, and may be important therapeutic targets against the multiple pathological features of ALS. Further analysis with different ALS models caused by other types of mutation is necessary to expand upon this intriguing notion (Fig. 7).
Mitochondrial defects have been considered important mediators in the progression of multiple neurodegenerative diseases, including ALS. In ALS models, the mitochondria are hyper-fragmented, and their axonal transport, ATP synthesis, and mitophagic removal are greatly affected 28,52,53 . Drp1 has been proposed as a key player in this mitochondria-related ALS pathology because several causative gene mutations associated with ALS promote Drp1 activity 36,[54][55][56][57] . The induction of Drp1 activity is apt to promote cell death signaling via multiple mediating processes 39,58 . Thus, the inhibition of Drp1 led to the exhibition of therapeutic effects in the G93A mouse model 59 . Our current observations further confirmed this notion in SOD1 mutations and extended it to include ALS models with the TDP-43 mutation. Notably, we observed the dephosphorylation of Drp1 S616 sites in several ALS models, and Drp1 S616 has been widely considered as an activating phosphorylation site 30 . However, we previously found that phosphorylation of Drp1 S616 by cyclindependent kinase 5 (CDK5) inhibited the translocation of Drp1 to the mitochondria in neurons, resulting in reduced mitochondrial fission 36,60 . Given that CDK5 is specifically expressed in post-mitotic neurons, this cascade appeared to be specific to neurons at that particular stage. In this respect, the dephosphorylation of Drp1 S616 in the neurons of ALS models appears to promote Drp1 activation. We found that a blockade of dephosphorylation prevented Drp1-related ALS symptoms, which supports our current hypothesis.
While Drp1 inhibition proved to be clearly beneficial, there were nonetheless several reasons for us to seek upstream targets, such as PP1, for ALS therapy. Previously, we observed that a blockade of Drp1 activity in normally developing chick MNs resulted in an increase in programmed cell death and the impairment of axonal growth 61 . Considering that mitochondrial fusion and fission are essential events for cellular homeostasis 28,30,53 , there is a narrow range in which mitochondrial dynamics must be maintained for optimal survival. In concurrence with our previous observations in developing MNs, we also found that the suppression of Drp1 activation had little benefit for axonal growth. It should also be noted that Drp1 performs many functions in addition to mitochondrial fission. Drp1 can interact with many different proteins localized at different subcellular locations, thereby being involved in many biological processes such as peroxisomal fragmentation, vesicle endocytosis/recycling, MMP changes, and cytoskeletal remodeling 29,62 . Indeed, restoration of Drp1 activity to a normal physiological level in the long-term may be difficult to achieve, meaning that alternative robust and druggable targets for ALS therapy are highly desirable.
Here we identified PP1 as an upstream phosphatase responsible for the dephosphorylation of Drp1 and tested whether regulation of PP1 activity is beneficial for the suppression of ALS-related pathological consequences. PP1 is a serine/threonine phosphatase and exists as a cellular holoenzyme composed of catalytic and regulatory subunits 38,63,64 . It is required for many cellular functions, including cell division, apoptosis, cytoskeletal organization, protein synthesis, and metabolism 38 . Moreover, it has been reported that PP1 activity is associated with several neurodegenerative diseases. For instance, the dephosphorylation of mutant leucine-rich repeat kinase 2 (LRRK2) by PP1 is increased in Parkinson's disease 65 . Previous studies reported that, in Alzheimer's disease, tau phosphorylation was induced by GSK3β, which is activated by PP1 66,67 , and I-2 prevented amyloid-β oligomer (AβO)-induced defects in brain-derived neurotrophic factor (BDNF)-containing vesicle transport 67 . In this study, we found that PP1 activity was increased in the SOD1 G93A ALS mouse model before the onset of the disease, and the blockade of PP1 activity prevented mitochondrial defects. Importantly, the PP1 blockade prevented axonal degeneration and restored the complex I activity of the mitochondria in the ALS model, results that were not achieved by the inhibition of Drp1 in our model. We found that the different PP1 isoforms may mediate different consequences of ALS-related pathologies because selective inhibition of each isoform resulted in different outcomes with respect to mitochondrial fragmentation, axonal growth, and cell survival. Das et al. reported that Sephin1, a novel inhibitor of PP1 and PPP1R15α (GADD34) binding, prevented the aggregation of mutant SOD1 G93A protein and MN death in an ALS model, improving motor deficiency in SOD1 G93A mice 68 . They proposed that Sephin1 inhibited the phosphorylation of eukaryotic initiation factor 2 (eIF2α) without affecting PP1 catalytic activity. On the other hand, we observed that the inhibition of PP1 activity improved ALS-related pathologies and failed to observe any dephosphorylation of eIF2 α induced by I-2 treatment (Supplementary Fig. 4). Therefore, PP1 may affect the propagation of ALS-related pathologies in catalytic activity-dependent or -independent manners.
While PP1 appears to affect ALS pathology via Drp1 dephosphorylation, it must also have other substrates. For instance, mitochondrial complex I, which is composed of 44 subunits 69 , is a pacemaker of the mitochondrial respiratory system 70,71 . Thus, impairment of complex I is often reported in ALS models 1,2 . Complex I deficiency represents severe defects within the mitochondrial system and induces ALS via impairment of mitochondrial metabolism [72][73][74] . Some subunits in complex I, which are closely linked to its activity, are regulated by (de)phosphorylation 63,[75][76][77] . Therefore, although we cannot completely rule out the possibility that the pathological effect of PP1 activation is also mediated by Drp1, we favor the idea that pathological induction of PP1 activity may impair other cellular processes, including mitochondrial complex activity, independently of Drp1, via dephosphorylation of its target protein(s). Unbiased identification of PP1 targets and the further assessment of the precise molecular targets of PP1 should be addressed in the future to clarify this and related issues. Fig. 7 Hypothetical schematic diagram: abnormal promotion of PP1 activity in the MNs of ALS and dephosphorylation of the S616 site of Drp1, resulting in the induction of mitochondrial fragmentation. Also, abnormal PP1 activity decreases the mitochondrial complex I activity, leading to axonal degeneration. Thus, the blockade of PP1 activity reduces the ALS-induced mitochondrial fission and improves mitochondrial functions such as the complex I activity of the mitochondrial electron transport chain and membrane potential, resulting in improvement of axonal impairment. Therefore, inhibition of excessive PP1 activity may promote MN survival by improving mitochondrial function.