Amyotrophic lateral sclerosis-associated mutant SOD1 inhibits anterograde axonal transport of mitochondria by reducing Miro1 levels

Abstract Defective axonal transport is an early neuropathological feature of amyotrophic lateral sclerosis (ALS). We have previously shown that ALS-associated mutations in Cu/Zn superoxide dismutase 1 (SOD1) impair axonal transport of mitochondria in motor neurons isolated from SOD1 G93A transgenic mice and in ALS mutant SOD1 transfected cortical neurons, but the underlying mechanisms remained unresolved. The outer mitochondrial membrane protein mitochondrial Rho GTPase 1 (Miro1) is a master regulator of mitochondrial axonal transport in response to cytosolic calcium (Ca2+) levels ([Ca2+]c) and mitochondrial damage. Ca2+ binding to Miro1 halts mitochondrial transport by modifying its interaction with kinesin-1 whereas mitochondrial damage induces Phosphatase and Tensin Homolog (PTEN)-induced Putative Kinase 1 (PINK1) and Parkin-dependent degradation of Miro1 and consequently stops transport. To identify the mechanism underlying impaired axonal transport of mitochondria in mutant SOD1-related ALS we investigated [Ca2+]c and Miro1 levels in ALS mutant SOD1 expressing neurons. We found that expression of ALS mutant SOD1 reduced the level of endogenous Miro1 but did not affect [Ca2+]c. ALS mutant SOD1 induced reductions in Miro1 levels were Parkin dependent. Moreover, both overexpression of Miro1 and ablation of PINK1 rescued the mitochondrial axonal transport deficit in ALS mutant SOD1-expressing cortical and motor neurons. Together these results provide evidence that ALS mutant SOD1 inhibits axonal transport of mitochondria by inducing PINK1/Parkin-dependent Miro1 degradation.


Introduction
Amyotrophic lateral sclerosis (ALS) is an adult onset neurodegenerative disorder, which is characterized by the selective death of motor neurons in the spinal cord, motor cortex and brain stem (1). Approximately 10% of all ALS cases are inherited and mutations in a number of genes have been associated with excitotoxicity, neuroinflammation, loss of protein homeostasis, mitochondrial dysfunction, and defective axonal transport (4)(5)(6)(7).
Neuronal function, integrity and survival are largely dependent on the correct distribution of proteins and organelles to their designated destinations. The architecture of neurons renders them particularly vulnerable to disruptions of transport processes and numerous lines of evidence suggest that defective axonal transport contributes to human neurodegenerative diseases including ALS (8). We and others have shown that impaired axonal transport is a very early event in both in vitro and in vivo models of ALS (5). In case of SOD1-related ALS we have shown that ALS mutant SOD1 selectively reduces anterograde transport of mitochondria in primary cortical and motor neuron cultures (9). Subsequent time-lapse recordings in single axons in the intact sciatic nerve of presymptomatic SOD1 G93A transgenic mice and rats revealed deficits in both anterograde and retrograde axonal transport of mitochondria in vivo (10)(11)(12).
Here we investigated the mechanism underlying impaired axonal transport of mitochondria in SOD1-related ALS. We report that ALS mutant SOD1 reduces anterograde mitochondrial transport by inducing PINK1/Parkin-dependent degradation of Miro1.

ALS mutant SOD1 impairs axonal transport of mitochondria after lentiviral delivery in primary rat motor neurons
We have shown previously that ALS mutant SOD1 disrupts axonal transport of mitochondria in transfected primary embryonic rat cortical neurons and in embryonic SOD1 G93A transgenic mouse motor neurons (9). To test if this defect could be recapitulated by expressing ALS mutant SOD1 in primary cultures of embryonic rat motor neurons we delivered EGFP, EGFP-SOD1 wild type (WT), A4V, G37R or G93A to DIV3 rat motor neurons by lentiviral transduction and analyzed axonal transport of MitoTracker Red CMXRos-labelled mitochondria from timelapse recordings and corresponding kymographs as described by us before (Fig. 1A, a) (9,28,29). As a positive control, we quantified axonal transport in rat cortical neurons co-transfected with EGFP, EGFP-SOD1 WT or G93A and DsRed2mito (Fig. 1B, a) (9).
In EGFP and EGFP-SOD1 WT transduced motor neurons approximately 50% of mitochondria were motile and anterograde and retrograde transport was balanced (Fig. 1A, b and c). Lentiviral delivery of EGFP-SOD1 A4V, G37R or G93A significantly reduced anterograde transport of mitochondria while retrograde transport remained unchanged (Fig. 1A, b); as a result, the balance of transport shifted toward net retrograde transport (Fig. 1A, c).
In EGFP and EGFP-SOD1 WT-expressing cortical neurons approximately 45% of mitochondria were motile, with near equal amounts of mitochondria moving in anterograde and retrograde directions (Fig. 1B, b). EGFP-SOD1 G93A caused a significant reduction in the number of anterograde mitochondria but did not affect the number of retrograde mitochondria (Fig. 1B,  b). As a result, there was a significant shift toward net retrograde mitochondrial transport (Fig. 1B, c). Transfection of ALS mutant SOD1 did not affect axon length (Supplementary Material, Fig. S1).
Hence lentiviral delivery of ALS mutant SOD1 in embryonic rat motor neurons caused impairment of axonal transport of mitochondria similar to that reported in embryonic SOD1 G93A transgenic mouse motor neurons and transfected cortical neurons (9). Taken together these data also indicate that for the purpose of studying the effect of ALS mutant SOD1 on axonal transport of mitochondria, the method of delivery of SOD1, transfection or lentiviral transduction, and the neuronal cell type, cortical or motor neurons, are interchangeable.

ALS mutant SOD1 does not affect cytosolic Ca 2þ levels
We reported previously that expression of the ALS type-8 (ALS8)-associated vesicle-associated membrane proteinassociated protein B (VAPB) mutant VAPBP56S in rat cortical neurons impairs anterograde transport of mitochondria similar to ALS mutant SOD1 (29). Mechanistically, overexpression of VAPBP56S increased cytosolic Ca 2þ levels ([Ca 2þ ] c ), which caused a reduction in the amounts of tubulin but not kinesin-1 that were associated with Miro1 and as a consequence transport was arrested (29,30).
To investigate whether elevated [Ca 2þ ] c also played a role in ALS mutant SOD1-induced defective axonal transport of mitochondria, we examined resting [Ca 2þ ] c in cortical neurons expressing EGFP, EGFP-SOD1 WT or G93A by Fura2 ratio imaging.

ALS mutant SOD1 reduces Miro1 levels in a Parkindependent fashion
It is well established that ALS mutant SOD1 accumulates in mitochondria and causes mitochondrial damage (7), and mitochondrial dysfunction has been linked to Miro1 degradation and impairment of axonal transport of mitochondria (23). Furthermore, Miro1 levels have been shown to be decreased in the spinal cord of transgenic mice expressing ALS mutant SOD1 G93A (31). To investigate if reduced levels of Miro1 may be at the basis of the axonal transport defect in SOD1-related ALS we first explored if expression of ALS mutant SOD1 could affect because of a selective block of anterograde (Ab, Bb-Anterograde), but not retrograde (Ab, Bb-Retrograde) transport. As a consequence, SOD1 G93A disturbed the balance of transport to promote net retrograde movement (Ac, Bc). Results are shown as mean 6 SEM, statistical significance was determined by one-way ANOVA followed by Fisher's LSD test, ns, not significant, * P < 0.05, **** P

Miro1 levels in HEK293 cells and primary cortical neuron cultures.
We co-transfected HEK293 cells with myc-tagged Miro1 (Myc-Miro1) and EGFP-SOD1 WT, A4V, G93A, G37R or G85R and quantified Myc-Miro1 levels on immunoblots. Compared to SOD1 WT all four ALS-associated SOD1 mutants caused a decrease in Myc-Miro1 levels of approximately 20% (Fig. 3A). We next determined if ALS mutant SOD1 affected the amount of endogenous Miro1 in rat cortical neurons transduced with EGFP-SOD1 WT or G93A. In these samples, endogenous Miro1 was readily detected on immunoblots as a doublet with a molecular weight of approximately 80 kDa as described before (32). Compared to SOD1 WT, ALS mutant SOD1 G93A significantly reduced the level of endogenous Miro1 in cortical neurons indicating that the ALS mutant SOD1associated impairment of mitochondrial axonal transport may be due to reduced Miro1 levels (Fig. 3B).
To further investigate the involvement of the PINK1/Parkin pathway in ALS mutant SOD1-induced reductions in Miro1 levels we took advantage of HeLa cells that lack endogenous Parkin (33,34). We co-transfected HeLa cells with EGFP-SOD1 WT or ALS mutant SOD1 G93A and myc-Miro1 and determined Miro1 levels. ALS mutant SOD1 G93A did not affect Miro1 levels in native HeLa cells lacking Parkin (Fig. 3C). To confirm that this lack of effect of SOD1 G93A was indeed due to Parkin deficiency we restored Parkin expression by cotransfection of YFP-Parkin. It has been shown previously that expression of YFP-Parkin in HeLa cells restores the PINK1/ Parkin pathway (35). In presence of YFP-Parkin ALS mutant SOD1 G93A caused a marked decrease in Miro1 compared to SOD1 WT (Fig. 3C).

Expression of Miro1 rescues the effects of ALS mutant SOD1 on mitochondrial motility
To further investigate the role of Miro1 in the axonal transport defects observed in ALS mutant SOD1-expressing neurons we enquired if expressing wild type Miro1 or a Ca 2þ insensitive mutant of Miro1 in which the EF hands were disrupted (Miro1 E208K/ E328K ) could rescue the effect of ALS mutant SOD1 on mitochondrial transport. Expression of either wild type Miro1 or Miro1 E208K/E328K should rescue the effect of ALS mutant SOD1 on mitochondrial transport if the defect was caused by reduced levels of Miro1, but only Miro1 E208K/E328K should rescue if the defect was Ca 2þ dependent. Indeed, we have previously shown that in agreement with a Ca 2þ dependent mechanism, expression of Miro1 E208K/E328K but not wild type Miro1 rescued defective mitochondrial axonal transport in VAPBP56S-expressing cortical neurons (29).
We co-expressed Miro1 or Miro1 E208K/E328K with EGFP-SOD1 WT or G93A in rat primary cortical neurons and quantified transport of mitochondria. In SOD1 WT-expressing neurons Miro1 E208K/E328K but not Miro1 lowered the number of retrograde mitochondria (Fig. 4A, b). As a result, mitochondrial transport was skewed toward net anterograde transport in SOD1 WT þ Miro1 E208K/E328K expressing neurons (Fig. 4A, c). In these neurons, the number of anterograde mitochondria was slightly elevated but not to a significant extent. These data are consistent with the previously proposed model in which binding of Ca 2þ to the EF hands of Miro1 enables retrograde transport by inactivation of kinesin-1-mediated anterograde transport (24,25). Compared to SOD1 WT, transfection of SOD1 G93A alone again reduced anterograde transport of mitochondria and consequently shifted the balance of transport toward net retrograde transport (Fig. 4A, b and c). Both wild type Miro1 and Miro1 E208K/E328K fully rescued anterograde axonal transport (Fig. 4A, b) and restored the balance of transport in SOD1 G93Aexpressing neurons (Fig. 4A, c).
The rescue by wild type Miro1 was further confirmed in motor neurons by quantification of axonal transport of mitochondria in rat motor neurons co-transduced with wild type Miro1 and EGFP-SOD1 WT or G93A (Fig. 4B). Expression of SOD1 G93A in absence of wild type Miro1 reduced anterograde axonal transport of mitochondria compared to SOD1 WT and, as was the case in cortical neurons, co-expression of wild type Miro1 fully rescued the effect of SOD1 G93A on mitochondrial transport (Fig. 4B, b and c). Together these data confirm that ALS mutant SOD1 does not act via Ca 2þ but by reducing Miro1 levels to affect anterograde transport of mitochondria.

Knockdown of PINK1 rescues the effects of ALS mutant SOD1 on mitochondrial motility
The above data strongly suggested that ALS mutant SOD1 impairs axonal transport of mitochondria by triggering PINK1/ Parkin-dependent degradation of Miro1. To further verify this mechanism, we disrupted the PINK1/Parkin pathway by ablation of PINK1 expression using miRNA in cortical and motor neurons expressing EGFP-SOD1 WT or G93A and analyzed axonal transport of mitochondria. As control, we used a nontargeting control miRNA. The efficiency of the PINK1 miRNAmediated knockdown was verified by qRT-PCR; on average 70% knockdown was achieved (Supplementary Material, Fig. S3).
In non-targeting control miRNA þ SOD1 WT expressing cortical and motor neurons approximately half of mitochondria were motile and these were near equally divided into anterograde and retrograde mitochondria ( Fig. 5A and B, a and  b). Control miRNA þ SOD1 G93A-expressing cortical and motor neurons exhibited impaired anterograde transport of mitochondria ( Fig. 5A and B, a and b) and this caused a shift toward net retrograde transport ( Fig. 5A and B, c). Thus, control miRNA did not affect transport per se and did not alter SOD1 G93Aassociated reductions in anterograde transport (compare with Fig. 1). In contrast, whereas PINK1 miRNA did not affect axonal transport of mitochondria in SOD1 WT-expressing cortical or motor neurons, it fully restored anterograde transport in SOD1 G93A-expressing neurons to control levels ( Fig. 5A and B, b) and rebalanced transport (Fig. 5A and B, c). Thus, in agreement with a PINK1/Parkin-dependent reduction in Miro1 levels as the underlying cause of the ALS mutant SOD1-associated defect in mitochondrial trafficking, miRNA-mediated knockdown of PINK1 rescued mitochondrial axonal transport in SOD1 G93Aexpressing cortical and motor neurons.

Discussion
Defective axonal transport of mitochondria and other cargoes is one of the earliest neuropathological features observed in models of ALS which has led to the suggestion that it may play a causative role in disease (5). Mutations in essential axonal transport components such as cytoplasmic dynein, kinesin-1 and tubulin that have been shown to cause ALS and other motor neuron disorders, have reinforced this idea. We and others have shown that axonal transport of mitochondria is impaired in in vitro and in vivo models of mutant SOD1-related ALS but the underlying causes of defective transport remained unclear (9)(10)(11)(12)36,37).
Here we show that impaired axonal transport of mitochondria in mutant SOD1-expressing neurons correlates with reduced levels of Miro1 (Fig. 3). Degradation of Miro1 has been shown to be an early event in PINK1/Parkin-dependent mitophagy. In mitophagy, mitochondrial damage causes stabilization of PINK1 on the outer mitochondrial membrane where PINK1 phosphorylates ubiquitin which in turn drives recruitment of Parkin, itself a PINK1 substrate, to the mitochondria (38). PINK1 also phosphorylates Miro1 which targets Miro1 for degradation in a Parkin-dependent manner (23). As a consequence, molecular motors are prevented from attaching to mitochondria and damaged mitochondria are immobilized. Consistent with PINK1-dependent reduction of Miro1 levels perturbing transport of mitochondria, expression of Miro1 and ablation of PINK1 rescued axonal transport in ALS mutant SOD1-expressing cortical and motor neurons (Figs 4 and 5). Furthermore, Miro1 degradation assays directly confirmed the Parkin dependent degradation of Miro1 (Fig. 3). It is well-established that ALS mutant SOD1 accumulates in mitochondria where it interacts with voltage-dependent anion channel 1 (VDAC1) and Bcl-2, and causes mitochondrial dysfunction (39)(40)(41)(42)(43)(44)(45). Thus, our data suggest that ALS mutant SOD1-induced mitochondrial damage activates the PINK1/Parkin pathway and as a result halts mitochondrial transport. In line with such a model it has been shown that Miro1-expression is decreased in the spinal cord of SOD1 G93A transgenic mice (31).
We previously reported that mutant VAPBP56S deregulates Ca 2þ homeostasis leading to increased [Ca 2þ ] c that impairs mitochondrial transport by binding to the EF hand motifs of Miro1 (29 Fig. S2) and consistent with a Ca 2þ -independent mechanism both wild type and Ca 2þ insensitive mutant Miro1 were able to restore axonal transport of mitochondria in ALS mutant SOD1 expressing neurons (Fig. 4). Thus, two distinct genetic causes of ALS, mutant VAPB and mutant SOD1, that impair axonal transport of mitochondria converge on Miro1 to halt transport, albeit by different mechanisms.
Impaired axonal transport of mitochondria has also been observed in models of TDP-43, FUS, SIGMAR1 and C9orf72-related ALS (29,36,(50)(51)(52)(53).  impaired axonal transport of mitochondria are in these cases, but mitochondrial damage and Ca 2þ mishandling are welldocumented in all of ALS (7,57). Both ALS mutant TDP-43 and FUS have been shown to accumulate in mitochondria and disrupt mitochondrial function (58,59). C9orf72 GGGGCC repeat expansion-associated glycine/arginine (GR) dipeptide repeat protein (DPR) has been shown to bind to mitochondrial ribosomes and compromise mitochondrial function (60). Mitochondria appear swollen and vacuolated in motor neurons of ALS patients (54). TDP-43, FUS, and SIGMAR1 have all been linked to Ca 2þ mishandling, probably due to disruption of ER-mitochondria contact sites (53,(61)(62)(63)(64)(65), and evidence suggests altered Ca 2þ in the motor terminals of ALS patients (66). In agreement with mitochondrial damage-associated reductions in Miro1 causing axonal transport defects in non-SOD1 ALS, decreased levels of Miro1 have been reported in TDP-43 M337V transgenic mice and in ALS patients (31), and, in the case of FUS, a genetic interaction with PINK1/Parkin has been described in Drosophila (52). Thus, several ALS-associated insults may converge on axonal transport via Miro1 and trigger motor neuron demise.
Defective axonal transport of mitochondria leading to depletion of mitochondria from axons is one of the earliest phenomena observed in ALS models and has been proposed to play an important causal role in disease (5). However, increasing the motility of mitochondria in SOD1 G93A transgenic mice did not slow disease in these animals (67). Our data now suggest that mitochondrial damage is the upstream cause of defective axonal transport of mitochondria in mutant SOD1-associated ALS. Hence it may not be surprising that stimulating anterograde transport of damaged mitochondria did not alter the course of disease in SOD1 G93A transgenic mice. Indeed, our findings indicate that, at least in the case of ALS mutant SOD1, the relative increase in retrograde transport of mitochondria at the cost of anterograde transport represents increased clearance of damaged mitochondria via mitophagy. If this is also the case in Quantitative analysis of mitochondrial transport shows that expression of ALS mutant SOD1 significantly impairs overall motility of mitochondria (Ab, Bb-Motile) because of a selective block of anterograde (Ab, Bb-Anterograde), but not retrograde transport (Ab, Bb-Retrograde). As a consequence, SOD1 G93A disturbed the balance of transport to inhibit anterograde and promote retrograde movement (Ac, Bc). Ablation of PINK1 expression fully rescued impaired transport of mitochondria (Ab, Bb) and rebalanced anterograde and retrograde transport (Ac, Bc). Results are shown as mean 6 SEM, statistical significance was determined by one-way ANOVA followed by Fisher's LSD test, ns, not significant, * P < 0.05, *** P < 0.001, **** P non-SOD1 ALS is not yet clear but these results suggest that successful therapeutic strategies will probably need to target both mitochondrial dysfunction and transport simultaneously.
A rat PINK1-targeting miRNA expression construct was designed using Invitrogen's RNAi Designer (www.invitrogen.com/ rnai) and cloned into the pcDNA6.2 TM -GW/EmGFP-miR expression vector according to the manufacturer's instructions (BLOCK-iT TM Pol II miR RNAi expression vector kit, Invitrogen). Sequences for the top strands were 5 0 -TGC TGT GTC CTA TCA GAT AAT CCT CCG TTT TGG CCA CTG ACT GAC GGA GGA TTC TGA TAG GAC A (PINK1) and 5 0 -TGC TGA AAT GTA CTG CGC GTG GAG ACG TTT TGG CCA CTG ACT GAC GTC TCC ACG CAG TAC ATT T (negative control; provided with the kit). In order to exchange the EmGFP tag for a ECFP tag, ECFP was amplified using primers 5 0 -TTT AAA ACC ATG GTG AGC AAG GGC GAG GAG (fw) and 5 0 -TTT AAA CGA TCT TAC TTG TAC AGC TCG TCC ATG CC (rev) from pECFP (Clontech). After sub-cloning into pCR V R -Blunt II-TOPO (Life Technologies), ECFP was cloned into pcDNA6.2 TM -GW/EmGFP using DraI sites (Thermo Scientific).

Cell culture, plasmid transfection, lentiviral transduction
Cortical neurons from E18 rat embryos were isolated and cultured as described before (9,28,70). 375 000 cells were seeded onto 18 mm cover slips coated with poly-L-lysine. Cells were cultured for 5-7 d and then transfected with Lipofectamine LTX (Life Technologies; 0.5 ml/mg of DNA). Transduction with lentiviruses was performed on DIV5 using an MOI of 5.

SDS-PAGE and immunoblot
Samples were separated on poly-acrylamide gels and transferred to Protran nitrocellulose membranes (0.45 lm, GE Healthcare) using a Pierce G2 Fast Blotter (Thermo Scientific) or Bio-Rad TransBlot cell. Membranes were blocked for 1 h at room temperature in TBS containing 5% milk powder/0.1% Tween-20 (TBST-M) and probed with primary antibodies in TBST-M for either 1 h at room temperature or 16 h at 4 C. After washing with TBS/0.1% Tween-20 (TBST), membranes were incubated with horseradish peroxidase (HRP)-conjugated or alkaline phosphatase (AP)-coupled, species-specific secondary antibodies. Following incubation with SuperSignal West Pico Chemiluminescence Substrate (Thermo Scientific) or AttoPhos V R AP Fluorescent Substrate (Promega) signals were detected on Hyperfilm ECL (GE Healthcare) or using a G: Box imager (Syngene). Films were scanned on a CanoScan LiDE 60 photo scanner and signals were quantified using the Fiji distribution (http://fiji.sc/) (72) of the ImageJ image processing software [National Institutes of Health (NIH), http://rsb.info.nih.gov/ij/] (73).
Live microscopy of mitochondrial axonal transport was performed using an Axiovert200 microscope (Carl Zeiss) equipped with a Polychrome IV monochromator (Till Photonics), an EGFP/ DsRed filter set (Chroma Technology Corp.), a 40Â EC Plan-Neofluar 1.3 N.A. objective (Zeiss), a high speed HF110A emission filter wheel controlled by a ProScan III Controller (Prior Scientific) and a Hamamatsu C9100-12 EMCCD camera (Hamamatsu Photonics). The microscope setup was controlled using MicroManager 1.4.21 (74). 48 h post-transfection or 72 h post-transduction, coverslips were transferred to a heated observation chamber (SA-20LZ, Warner Instruments) mounted on the stage of the microscope. In order to visualize mitochondria, motor neurons were stained with MitoTracker V R Red CMXRos (66 nM, Life Technologies) beforehand. In addition to the heated chamber, an objective heater (IntraCell) was used to maintain the cells at 37 C. Mitochondrial movements were recorded for 5 min with 3 s time-lapse interval in Neurobasal medium (Life Technologies) supplemented with 5 mM HEPES pH 7.0 and 2 mM GlutaMAX TM -I. Axonal transport was analyzed using ImageJ as described previously (9,28,29). Briefly, the overall transport of mitochondria was quantified from kymographs by calculating the distance between the position of individual mitochondria at the start and end of time-lapse recordings and dividing by the time elapsed. This yielded an average transport velocity for each mitochondrion that includes anterograde and retrograde movements and stationary periods. Mitochondria were classified as motile when their velocity exceeded 0.1 mm/s or as stationary when their velocity was 0.1 mm/s. Mitochondria were classified as anterograde or retrograde according to their predominant direction of travel.

Ca 2þ imaging
Resting [Ca 2þ ] c levels in transfected cortical neurons were determined by Fura2 ratio imaging as described previously (29). First, cells were stained for 20 min at 37 C with 5 lM Fura2-AM (Life Technologies) in external solution (ES; 145 mM NaCl, 2 mM KCl, 5 mM NaHCO 3 , 1 mM MgCl 2 , 2.5 mM CaCl 2 , 10 mM glucose, 10 mM HEPES pH 7.0). After destaining by incubation for another 20 min in ES only, Fura2 340 nm/Fura2 380 nm image pairs were recorded for 400 s with 1 s time-lapse interval using the same microscopy setup as described above, but using WinFluor V3.7.4 software (written by John Dempster, University of Strathclyde; http://spider.science.strath.ac.uk/sipbs/software_imaging.htm). Neurons were perfused continuously with ES (2 ml/min; Gilson Minipuls Evolution). To ensure that only viable neurons were taken into account, a transient Ca 2þ influx was invoked by depolarization with 50 mM KCl for 1 min. Signals were recorded and analyzed using the WinFluor software.

Quantitative real time PCR
RNA from cultured cells was isolated according to the manufacturer's instructions using TRIzol V R Reagent (Life Technologies). After resuspension in nuclease-free water, 0.5-2 lg RNA were treated with DNase I [NEB; DNase I was inactivated by addition of 2.5 mM EDTA (final concentration) and incubation for 10 min at 75 C] and reverse transcribed using SuperScript V R III Reverse Transcriptase and Oligo(dT) primers (Thermo Scientific). 100 ng of template was subjected to a quantitative real time PCR utilising Brilliant III Ultra-Fast 2Â SYBR V R Green QPCR master mix (Agilent Technologies) or HOT Firepol EvaGreen PCR Mix Plus (Solis Biodyne) using an Mx3000p qPCR System (Agilent Technologies) or a BioRad CFX96 Real-Time System (C1000 Touch Thermal Cycler; BioRad). PINK1 primers were as follows: rnPINK1 fw: 5 0 -TGT CAG GAG ATC CAG GCA ATT, rnPINK1 rev: 5 0 -CTT CAT ACA CAG CGG CAT TGCA, GAPDH or RPL19 was used as endogenous control: rnGAPDH fw: 5 0 -TGA AGG GTG GGG CCA AAGG, rnGAPDH rev: 5 0 -GGT CAT GAG CCC TTC CAT GA, rnRPL19 fw: 5 0 -CTC GAT GCC GGA AGA ACA CC -3 0 , rnRPL19 rev: 5'-GAG CGT TGG CAG TAC CCT T -3 0 .

Statistical analysis
Calculations and statistical analysis were performed using Excel (Microsoft Corporation) and GraphPad Prism software (GraphPad Software).

Supplementary Material
Supplementary Material is available at HMG online.