PINK1 and Parkin Control Localized Translation of Respiratory Chain Component mRNAs on Mitochondria Outer Membrane

PINK1 Regulates the mRNA Localization and Protein Abundance of Certain nRCC Genes in Drosophila

To understand the OXPHOS defects in PINK1 mutant, we examined the expression of RCC proteins in NM tissues of control and PINK1^B9 mutant flies. Ten RCC (7 nRCC and 3 mtRCC) proteins were chosen for analysis. Levels of 4 nRCC proteins, including the 30 kD subunit of complex-I (C-I 30), OSCP subunit of C-V, core protein 2 of complex-III, and ND75 subunit of complex-I (Figure 1A, B; data not shown), were reduced, whereas other nRCC proteins tested were unaltered in PINK1 mutant (Figure S1A). Levels of 3 mtRCC proteins tested, cytochrome c oxidase subunit I (C-IV-s1) (Figure 1A, B), ATP synthase subunit 6, and cytochrome b (Figure S1A), were not changed. The levels of several nuclear-encoded, mito proteins, e.g., Porin (VDAC) and MCU, were also unaffected (Figure S1A). PINK1 appears to specifically regulate the expression of select nRCC proteins.

We next examined the mechanisms underlying the selective reduction of nRCC proteins in PINK1 mutant. Semi-quantitative RT-PCR and quantitative real-time RT -PCR (qRT-PCR) revealed that total mRNA levels for the affected proteins were largely unaffected (Figure 1A–C). To test whether localized translation might be involved, we analyzed nRCC mRNA distribution in subcellular fractions. In the cytosolic (cyto) fraction, nRCC mRNA levels were comparable between control and mutant (Figure 1A–C). Intriguingly, a fraction of the mRNAs for the affected nRCC subunits was found in Percoll gradient-purified mito (Figure 1A–C), the purity of which was verified by the absence of other membrane structures such as the ER and endosomes (Figure S1B). RNase A-treatment of purified mito degraded the nRCC mRNAs, but not mtRCC RNAs residing in the matrix (Figure S1C), suggesting that the nRCC mRNAs are MOM-bound. Fluorescent in situ hybridization (FISH) confirmed the mito localization of select nRCC mRNAs (Figure 1D, E; Figure S1D), and transmission EM analysis revealed the association of ribosomes with muscle MOM, suggesting ongoing translation on MOM (Figure S1E). Strikingly, correlating with the change of protein levels, levels of MOM-bound nRCC mRNAs, but not mtRCC mRNAs, were significantly reduced in NM tissues of PINK1 mutant (Figure 1A–C). In contrast, in the intestine, which was phenotypically normal in PINK1 mutant, the mRNA localization and protein expression of those nRCC genes were unaltered (Figure S1F). PINK1 thus regulates the mRNA localization and protein expression of select nRCC genes in a tissue-specific manner. Due to the importance of C-I to PD in general (Schapira, 2010), and C-I 30 to PINK1 function in particular (Wu et al., 2013), we focused on C-I 30 mRNA in later analyses.

PINK1 Regulates the mRNA Localization and Protein Expression of Select nRCC Genes in Human DNs

We next tested possible conservation of PINK1 function in nRCC mRNA regulation. In HEK293T cells, the nRCC mRNAs homologous to those regulated by PINK1 in flies were found in highly purified mito (Figure 2A, B; S2A). Mito localization of mammalian C-I 30 mRNA was further confirmed using the MS2-GFP/MS2-bs tagging system (Bertrand et al., 1998) (Fig. S2B, C). Moreover, in PINK1 siRNA-treated cells, levels of MOM-bound nRCC mRNAs were significantly reduced, whereas mtRCC RNA (C-IV s1) was unaffected (Figure 2A, S2D). Similar to PINK1 RNAi, expression of pathogenic PINK1(G309D) (Hoepken et al., 2007) greatly reduced levels of MOM-bound nRCC mRNAs, but not mtRCC RNA (Figure 2A, B), suggesting that PINK1(G309D) acts in a dominant-negative manner. Not all nRCC mRNAs or mRNAs for nuclear-encoded mito proteins were MOM-bound (Figure 2A, S2D). Correlating with the change in abundance of MOM-bound, but not total, mRNAs of select nRCC genes, their protein levels were also reduced when PINK1 was inhibited (Figure 2C). Thus, the function of PINK1 in nRCC mRNA localization and protein expression is conserved in human cells.

To validate these findings in disease -relevant neurons, we used induced DNs (iDNs) trans-differentiated from patient skin fibroblasts (Caiazzo et al., 2011). Expression of DN-related markers and DN-like transcriptomes were confirmed in iDNs (Figure S2E–G). The functionality of iDNs was validated by rescue of locomotor deficit in 6-OHDA-lesioned rats upon transplantation (Figure S2H). Comparison of PINK1(G309D) patient iDNs with control iDNs revealed disease-specific abnormal mito distribution and morphology in neuronal processes (Figure 2D, E), and reduced ATP levels (Figure 2F). Importantly, reduced mito-association of select nRCC mRNAs and reduced protein levels were observed in patient iDNs (Figure 2G, H). Patient fibroblasts, however, did not show significant change in mito distribution, morphology (Figure 2D, E), ATP production (Figure 2F), or nRCC protein expression (Figure 2G, H), indicating that the PINK1(G309D) mutation exerts DN-specific effects on mito function.

PINK1 Functionally Interacts with Pum and the TOM/TIM Complex to Regulate the Localization and Translation of nRCC mRNAs

The tight correlation of levels of select nRCC proteins with their mRNAs in the mito fraction, but not their mRNAs in the cyto fraction which constitute the majority of total mRNAs (Figure S3A) and were unchanged in PINK1 mutant (Figure 1C), suggests that the MOM-bound nRCC mRNAs are translationally active, whereas those in the cyto are repressed. Indeed, we observed cytoplasmic ribosomes associated with MOM (Figure S1E), and detected cytoplasmic translation factors in highly purified mito from fly tissues or human cells (Figure S3B). Moreover, polysome analysis of cyto and mito fractions indicated that PINK1-regulated nRCC mRNAs are translationally more engaged in the latter fraction (Figure S3C), a notion supported by in vitro translation assays (Figure S3D). Consistently, PINK1-regulated nRCC proteins are present almost exclusively in the mito fraction, whereas ND42 and Tom20, which are synthesized in the cyto and subsequently imported into mito, are present in both the cyto and mito fractions (Figure S3E).

To probe the molecular mechanism of translational control of nRCC mRNAs, we tested the role of the translational repressor Pum (Quenault et al., 2011). Pum is present in both the cyto and mito fractions in fly tissues (Figure 3A) and mammalian cells (Figure S3F). The mito localization of Pum appeared to be regulated byPINK1, as the a mount of mito-bound Pum, but not total Pum, was reduced in PINK1 mutant tissues (Figure 3B) or PINK1(G309D)-transfected HEK293 cells (Figure S3G). To test whether the mito localization of Pum depends on its binding to mRNA, we treated purified mito with RNase A and observed significantly reduced mito-association of Pum (Figure S3H). Moreover, an RNA binding-deficient form of Pum (Friend et al., 2012) was compromised in mito localization (Figure S3I), suggesting that Pum may be recruited to MOM via binding to mRNAs.

We next tested the role of Pum in nRCC mRNA translation. Pum RNAi moderately increased mito-bound nRCC mRNAs and their protein levels in fly muscle tissues (Figure 3C, 3D, S3J). Similar effect was observed in HEK293 cells (Figure S3K). Importantly, Pum RNAi effectively restored levels of MOM-bound nRCC mRNAs and their proteins in PINK1 RNAi (Figure 3E) or PINK1 mutant (Figure S3L) muscle tissues.

Since all nRCC proteins require passage through the TOM/TIM complex to enter mito, we asked whether the MOM-targeting and translation of nRCC mRNAs is linked to nascent nRCC protein import. For this purpose, we knocked down components of the TOM/TIM complexes. Inhibition of Tom40 or Tim8 strongly exacerbated PINK1 mutant effects on the MOM-targeting and translation of nRCC mRNAs (Figure 3E, S3L). Independent RNAi lines for each gene produced similar effects (Figure S3N). The MOM-targeting and translation of nRCC mRNAs thus appears to be linked to protein import during RCC biogenesis.

PINK1 has been shown to regulate mito fission/fusion dynamics (Yang et al., 2008) and motility (Liu et al., 2012). However, loss- or gain-of-function of genes involved in fission/fusion (Drp1, Marf) or motility (Miro) had no obvious effect on the levels of MOM-bound nRCC mRNAs, although for some unknown reason the protein levels of C-V (OSCP) and C-I 30 were reduced in Drp1-DN or Miro overexpression (OE) conditions (Figure S3O). Thus, defects in mito dynamics or motility cannot explain the effect of PINK1 on select nRCC mRNA localization or translation, and PINK1 appears to play a specific role in this process.

Altered Pum and TOM/TIM Activities Modulate PINK1 Mutant Effects on NM Tissue Function and Integrity in Drosophila

We next assessed the effects of Pum and Tom/Tim on PINK1 function at the organismal level. While Pum RNAi had little effect on NM tissue function or integrity on its own, it effectively rescued the wing posture, jumping ability, ATP level, and DN survival defects of PINK1 ^B9 mutant (Figure 3F) or PINK1 RNAi flies (Figure S3M). In addition, Pum RNAi rescued the abnormal mito morphology caused by PINK1 inactivation (Figure 3G, H). Consistent with Pum performing a conserved function, a portion of mammalian Pum-1 was localized to mito (Figure S3F), and Pum-1 RNAi rescued the reduced nRCC protein expression and ATP production in PINK1(G309D) iDNs (Figure 3I, J).

In contrast to the rescuing effect of Pum RNAi, Tim8 or Tom40 RNAi exacerbated PINK1 mutant phenotypes in muscle (Figure 3F, G, S3M) and DNs (Figure 3H). Moreover, Tim8 or Tom 40 RNAi, but not Pum RNAi, caused mild mito aggregation and DN loss (Figure S4A, B), emphasizing the importance of TOM/TIM to DN health. To further probe the function of the TOM/TIM complex, we inhibited additional components. RNAi of Tom7 and Tom20, but not Tim10 or Tim13, enhanced PINK1 mutant phenotypes (Figure S4C). Since the knockdown efficiencies of the various RNAi transgenes were similar (Figure S4D), their differential genetic interactions with PINK1 suggest that components of the TOM/TIM complex do not function equivalently in the PINK1 pathway. We further probed the genetic epistasis between PINK1 and TOM in gain-of-function studies. PINK1 mutant phenotypes were specifically rescued by Tom20 OE (Figure S4E); reciprocally, PINK1 OE rescued Tom20 RNAi effects (Figure S4F). These results, together with their mutual enhancement of mutant phenotypes shown later, suggest that PINK1 and Tom20 may act at the same level, e.g., by working together in a complex, rather than at different steps of a linear pathway.

We also examined Pum OE effects. While strong Pum OE with the muscle-specific Mhc-Gal4 driver was lethal, weak or medium Pum OE lines produced viable progenies with muscle defects (Figure 4A, S4G). Using the medium Pum OE line (Pum -m), we found that its phenotypic effects correlated well with impaired MOM-targeting and translation of nRCC mRNAs (Figure 4B, C). In DNs, abnormal mito aggregation and neuronalloss (Figure 4D) similar to that seen in PINK1mutant were observed in TH-Gal4>Pum-m flies. PINK1 OE failed to rescue Pum-m OE effects (Figure S4H). This result, together with the rescue of PINK1 mutant phenotypes by Pum RNAi (Figure 3), suggests that Pum may act downstream of PINK1.

PINK1-TOM20 Interaction Promotes the MOM-Targeting and Translation of nRCC mRNAs

We sought to elucidate the molecular basis of the genetic interactions between TOM and PINK1, especially between Tom20 and PINK1. Tom20 is a key component of the protein import system that serves as a receptor for precursor proteins containing mito targeting sequences (MTS). Recently, PINK1 was found to associate with the TOM complex in depolarized mito in mammalian cells (Lazarou et al., 2012), although its physiological role was unclear. We hypothesized that Tom20 helps anchor translating nRCC mRNAs and associated ribosomes/RNPs to the MOM, presumably through binding to MTS of nascent nRCC proteins during co-translational import, and that PINK1 may facilitate this process. To test this model, we analyzed Tom20 immunoprecipitate (IP) prepared from fly muscle (Figure 4E) and HEK293T cells (Figure 4F). To stabilize endogenous PINK1, we briefly subjected flies or cultured cells to mild CCCP treatment (Narendra et al., 2010). We found that PINK1 and Pum, but not Parkin, associated with Tom20. The association of PINK1 or Pum with Tom20 was dramatically reduced by RNase A treatment (Figure 4E, F), whereas Tom20-Tom40 interaction was not affected (Figure 4F). The RNA-dependency of the above interactions prompted us to test the involvement of nRCC mRNAs. RNA-IP of fly muscle (Figure S4I) or HEK293 cells (Figure S4J) showed that the levels of nRCC mRNAs associated with Tom20 were reduced when PINK1 was inhibited. Supporting a critical role of Tom20 -PINK1 interaction in the localization and translation of nRCC mRNAs, Tom20 RNAi inhibited the MOM-targeting and translation of nRCC mRNAs (Figure 4G, H). Moreover, Tom20 RNAi enhanced the reduced MOM-targeting and translation of nRCC mRNAs (Figure 4I), mito aggregation (Figure S4L, M), and muscle degeneration (Figure S4C) effects ofPINK1 mutation, whereas Tom20 OE had opposite effects (Figure S4E, S4K).

RNA-dependent Association of PINK1 with eIF4G and Other Components of the Translation Initiation Complex

We next probed the mechanism by which PINK1 regulates nRCC mRNA translation. We found that PINK1 associated with mRNA 5′ cap structure, as detected by m⁷-GTP-sepharose chromatography (Figure 5A). Importantly, the G309D mutation impaired such association (Figure 5B). The association of PINK1 with m⁷-GTP-sepharose was RNA-dependent, whereas m⁷-GTP-sepharose binding by eIF4E, a translation initiation complex (TIC) component that directly binds to 5′ cap, was RNA-independent (Figure 5A), suggesting that PINK1 may associate with mRNA 5′ cap indirectly through the TIC. Indeed, endogenous PINK1 exhibited RNA-dependent associations with TIC components eIF4A and eIF4G, which occurred in the mito and cyto fractions (Figure 5C). It is worth pointing out that the association between PINK1 and eIF4G was not eliminated by RNase A treatment (Figure 5C), and PINK1 did not associate with eIF4E, suggesting certain degree of specificity and directness of PINK1-eIF4G interaction. Human genetics studies recently implicated eIF4G as a candidate FPD gene (Chartier-Harlin et al., 2011). We tested whether the PINK1-eIF4G association might be altered in disease condition and found that the G309D mutation attenuated PINK1-eIF4G interaction (Figure 5D). PINK1 also associated RNA-dependently with polyA binding protein (PABP), a protein known to interact with the TIC to promote translation (Figure 5E). Together, these data support a role of PINK1 in promoting translation through RNA-dependent and possibly RNA-independent associations with the TIC.

To further demonstrate the role of PINK1 in regulating the translation of mito-bound nRCC mRNAs, we made a translational reporter in which the synthesis of a luciferase-C-I 30 fusion protein is under the control of the 5′ and 3′ UTRs of C-I 30 mRNA. Combining in vitro transcribed reporter mRNA with purified mito from PINK1-WT transfected HEK293T cells led to stimulated translation of the reporter (Figure S5A). Using the same assay, we found that purified mito from PINK1 or parkin mutant fly tissues or PINK1−/− knockout mice tissues translated the reporter mRNA at a lower efficiency than WT mito (Figure S5B–D). Collectively, these data support a role of PINK1 in stimulating the translation of MOM-bound C-I 30 mRNA.

PINK1 Binds to nRCC mRNAs

We sought to elucidate the molecular basis of the RNA-dependency of PINK1 association with TIC. RNA-IP analysis revealed PINK1 association with nRCC mRNAs (Figure 5F). To test whether PINK1 directly binds mRNAs, we performed cross-linking and immunoprecipitation (CLIP) assays (Ule et al., 2005). Endogenous PINK1 directly bound to C-I 30 and C-V α mRNAs but not Mfn2 mRNA (Figure 5G). Moreover, compared to PINK1-WT, PINK1(G309D) bound to C-I 30 mRNA poorly (Figure 5H), supporting the importance of nRCC mRNA-binding to PINK1 function. CLIP assays performed on mito fraction showed that PINK1 binds to MOM-localized C-I 30 mRNA (Figure 5I).

Pum also exhibited C-I 30 mRNA -binding activity in CLIP assays (Figure 5J). In HEK293 cells transfected with PINK1 -WT, mRNA binding by Pum was significantly weakened. In contrast, nRCC mRNA binding by Pum was enhanced in PINK1(G309D) transfected cells or PINK1 RNAi cells (Figure 5J, K). These data suggest that PINK1 negatively impacts the nRCC mRNA-binding activity of Pum. We tested whether this effect might be mediated by a PINK1/Pum physical interaction. Our repeated co-IP assays were unable to detect such interaction (Figure S5E).

PINK1 Promotes Translational Derepression and Activation of nRCC mRNAs on MOM

Data presented earlier suggest that the MOM-localized nRCC mRNAs are translationally repressed in the cyto before reaching MOM, where they are reactivated by PINK1. To dissect the molecular mechanism, we generated a construct expressing C-I 30 mRNA tagged with MS2 binding sites, allowing purification of C-I 30 RNPs using GST-MS2. To specifically monitor events occurring on MOM-bound C-I 30 mRNAs, we used mito for RNP purification. Analysis of mito C-I 30 RNPs purified from PINK1-WT or PINK1-G309D transfected cells showed that PINK1-WT increased the association of eIF4G and PABP with C-I 30 mRNA (Figure 6A). Intriguingly, a number of translational repressors, including Dcp1, POP2, Pum-1, hnRNP-F, and hnRNP-H, were enriched in the C-I 30 RNPs purified from PINK1-G309D transfected cells (Figure 6A). In the case of hnRNP-F, we observed a modified form in PINK1-WT transfected cells. These results suggest that PINK1 may translationally derepress nRCC mRNAs on MOM by displacing and/or modifying translation repressors and further stimulate translation by promoting activator association.

We next tested the in vivo relevance of the above findings. The fly homologues of hnRNP-F/H, known as Glo (Kalifa et al., 2006), and Dcp1 were found to associate with C-I30 mRNA in muscle tissues (Figure 6B). Moreover, in PINK1 mutant more C-I 30 mRNA was bound by Glo and Dcp1 in the mito fraction (Figure 6C). Binding of repressors to MOM -bound C-I 30mRNA was also observed in HEK293 cells (Figure S6A), and increased repressor binding to nRCC mRNAs was found in PINK1−/− mouse midbrain tissues (Figure S6B). Considering that Dcp1 is a core component of P –bodies (Franks and Lykke-Andersen, 2008), cytoplasmic foci involved in mRNA turnover or repression, we tested association of other P-body components with C-I 30 mRNA. GW182, a conserved key component of P-body, also bound to C-I 30 mRNA (Figure 6B, C), and GW182-positive foci were found residing in the immediate vicinity of mito in PINK1 mutant but not wild type muscle tissues (Figure 6D), suggesting that nRCC mRNAs might be targeted to P-bodies in PINK1 mutant condition. We further tested the functional relationships between PINK1 and the translational repressors. Knockdown of the translation repressors effectively rescued PINK1 loss-of-function phenotypes, whereas eIF4G knockdown had opposite effect (Figure 6E–J). The phenotypic effects of these genetic interactions correlated well with effects on C-I 30 protein expression (Figure S6C). Thus, both the positive and negative regulators associated with C-I 30 mRNA are important effectors in the PINK1 pathway of nRCC mRNA regulation.

Parkin Cooperates with PINK1 to Promote hnRNP-F/Glo Ubiquitination and nRCC mRNA Translation

Parkin was previously shown to act downstream of PINK1 (Clark et al., 2006; Park et al., 2006; Yang et al., 2006). We tested the role of Parkin in translational control of nRCC mRNAs. Parkin associated with Pum -1 in HEK293 cells (Figure 7A). This association raised the possibility that Parkin may mediate the effect of PINK1 in recruiting Pum to mito. Contrary to this expectation, mito-bound Pum was increased in parkin mutant (Figure S7A), suggesting that not only Parkin is not required for the MOM localization of Pum, it may act at a later step to displace Pum from MOM-localized nRCC mRNAs to relieve translation al repression. Supporting a role for Parkin in translational regulation, we found that transfected (Figure 7B) and endogenous Parkin (Figure 7C) associated with 5′ cap in an RNA-dependent fashion, that pathogenic mutations disrupted this association (Figure 7D), that Parkin associated with TIC component PABP (Figure 7E), and that Parkin stimulated the translation of the C -I 30 reporter (Figure S5D). As in PINK1 mutant, the levels of MOM-bound nRCC mRNAs and their proteins were reduced in parkin mutant (Figure 7F, G). However, unlike in PINK1 case, parkin mutant phenotypes were not rescued by Pum RNAi (Figure S7B). Moreover, unlikePINK1 -OE, Parkin OE was able to rescue Pum OE effects on nRCC mRNA translation and tissue integrity (Figure 7H, S7C), suggesting that Parkin may function downstream of Pum.

We next searched for potential target(s) of Parkin mediating its effect on nRCC mRNA translation. A number of observations led us to focus on Glo. First, Glo RNAi effectively rescued PINK1 mutant (Figure 6E–J), whereas Glo OE had opposite effect (Figure S7D, E). Second, in PINK1 and parkin mutants, more nRCC mRNAs bound to Glo in the mito fraction (Figure 7I). Third, Glo physically associated with Parkin in fly tissues (Figure S7F). Fourth, we observed modification of hnRNP-F by PINK1-WT in C-I 30 RNP (Figure 6A). We tested whether Parkin might ubiquitinate Glo. A Ub-positive, higher molecular weight species consistent with mono-ubiquitinated Glo was detected in Glo-IP prepared from wild type but not parkin or PINK1 mutant flies, suggesting that PINK1 and Parkin regulate Glo ubiquitination (Figure 7J). Interestingly, in PINK1 mutant expressing a phospho-mimetic, but not a non-phosphorylatable forms of Parkin with the PINK1 phosphorylation site mutated (Shiba-Fukushima et al., 2014), Glo ubiquitination was restored (Figure S7G), indicating that PINK1 acts through Parkin to regulate Glo ubiquitination. Ubiquitination of hnRNP-F by PINK1/Parkin was also observed in mammalian cells, in which the WT (Parkin-WT) or phospho-mimetic (Parkin-S65E), but not E3-dead (Parkin-C431S) or non-phosphorylatable (Parkin-S65A) forms of Parkin, induced hnRNP-F mono-ubiquitination (Figure 7K). These data support that Glo/hnRNP-F is a direct substrate mediating the effect of PINK1/Parkin on nRCC mRNA translation. Glo or hnRNP-F protein level was not obviously changed by altered PINK1 or Parkin activities (Figure 7I–K), consistent with mono-ubiquitination generally not affecting protein stability. We hypothesize that ubiquitination of Glo alters the structure of the repressor complexes and their binding to nRCC mRNAs. Supporting this notion, repressor binding to C-I 30 mRNA was increased in parkin mutant, but reduced in PINK1 mutant upon Parkin OE (Figure S7H, I).

Drosophila Genetics

Fly culture and crosses were performed according to standard procedures and were raised at indicated temperatures. The UAS-Tom20, UAS-Tom40 and UAS-Tom70 transgenic flies were generated by BestGene Inc. Other fly stocks were obtained from VDRC or Bloomington Drosophila stock centers or from other investigators.

RT-PCR, qRT-PCR, CLIP Assays, and Purification of Mito-Associated mRNAs

For RT-PCR analysis, total RNAs were extracted from cell cultures, fly tissues, or adult mouse midbrain (P27) using an RNeasy® Mini kit, and one-step RT-PCR was performed using an RT-PCR kit. For qRT-PCR analysis, purified RNA were mixed with primers and Maxima SYBR Green/ROX qPCR Master Mix (Thermo Scientific) and PCR products detected by StepOnePlus real-time PCR system (Applied Biosystems). CLIP assay was performed under conditions modified from a published method (Ule et al., 2005). Intact mito from in vitro cell cultures, fly heads, fly thoraces, fly intestinal, and mouse midbrain (P7) tissues were purified at described previously (Kristian et al., 2006) and mito-associated mRNAs were extracted as described above.

Immunohistochemistry, RNA FISH, and TEM analysis

Immunohistochemical analyses of mito morphology of adult fly brains and muscle tissues were essentially done as described (Wu et al., 2013). For immunohistochemical analysis of mitochondrial morphology of adult fly brains and muscle tissues, male flies at around 5 days of age and raised at 29 degree were used. For analysis of DA neuron number in the various genetic backgrounds, male flies at around 14 days of age and raised at 29 degree were used. For RNA FISH in fly muscle, a protocol modified from a previously published method was used. Digoxigenin-labeled riboprobes for C -I 30, Oscp and Marf were generated using a PCR -based method. For TEM analysis, adult male flies at around 5 days of age and raised at 29°C were used. Samples were further sectioned and prepared for TEM analysis by the Cell Sciences Imaging Facility of Stanford University School of Medicine.

DNA Cloning, IP, RNP purification, m⁷-GTP Sepharose Affinity Chromatography, Western Blotting, RNAi, and Translational Reporter Assays

DNA cloning, IP, Western blotting, RNAi, and luciferase-based translational reporter assays were performed essentially as described (Gehrke et al., 2010). For RNP purification, cells were transfected with MS2-GST, C-I 30-MS2-bs, and PINK1 constructs. Sixty hours post-transfection, cells were UV-crosslinked, mito purified, and mito extracts subjected to GST-pull down and WB analyses. For IP and m⁷-GTP sepharose affinity chromatography, extracts made from HEK293T cells, fly tissues prepared from 5 day-old flies raised at 29 degree, or mouse midbrain (P7) tissue were prepared in lysis buffer [50 mM Tris-HCl, pH7.4, 150 mM NaCl, 5 mM EDTA, 10% glycerol, 1% Triton X-100, 0.5 mM DTT, 60 mM β-glycerolphosphate, 1 mM sodium vanadate, 20 mM NaF, and Complete protease inhibitor cocktail (Roche)] and subjected to IP using the indicated antibodies, or pull-down using m⁷-GTP sepharose beads at 4°C for 4 hours. For WB analysis of fly samples, male flies at around 5days of age and raised at 29 degree were used.

iDN Culture

Fibroblasts were seeded at 50% confluence onto matrigel-coated wells and infected with concentrated lentiviral particles expressing human TFs. Lentiviral particles were mixed with MEF media containing 8 μg/ml polybrene. 24 hours post-infection the medium was replaced with MEF medium containing 2 μg/ml doxycycline to activate TF expression. Medium was gradually changed to N3 medium (DMEM/F-12 medium containing 25 μg/ml insulin, 50 μg/ml transferrin, 30 nM sodium selenite, 20 nM progesterone, 100 nM putrescine, 2 μg/ml doxycycline) and cells were cultured for 2–3 weeks before immunohistochemical and biochemical analyses.

Behavioral Assays and ATP Measurement

Wing posture, flight and jumping ability assays on adult flies, and measurements of ATP contents in thoracic muscle were performed as described previously (Wu et al., 2013). For all wing posture assays, male flies at 1-day and 13~15-day old and aged in 29°C were used, except for the Mhc-Gal4>Pum-m OE flies, which were assayed at day 1 and day 7, as noted in Figure 4A. For the jumping and flight ability assays, male flies at 3–5 days of age raised at 29°C were used. For each genotype at least three independent experiments were performed for these assays. For ATP measurement in mammalian cells, in vitro culture d cells trypsinized from 3.7 cm ² culture plates were collected and transferred into a new 1.5 ml reaction tube, centrifuged at 1,000 g at 4°C, and washed once with ice cold 1x PBS before ATP measurement.

Statistical Analysis

Statistical significance of all data were evaluated by unpaired Student’s t-tests. Error bars represent standard error of the mean (SEM).

See Extended Experimental Procedures for additional details.