Deficiency of parkin and PINK1 impairs age-dependent mitophagy in Drosophila

1Loss-of-function mutations in PARK2 and PINK1, which encode the cytosolic E3 ubiquitin ligase parkin and the mitochondrial ubiquitin kinase PINK1, respectively, are the most prevalent recessive causes of Parkinson’s disease (PD), an age-dependent neurodegenerative disorder. Parkin and PINK1 cooperate in the selective autophagic degradation of damaged mitochondria (mitophagy). Upon mitochondrial damage, PINK1 is stabilized on the outer mitochondrial membrane (OMM) and phosphorylates ubiquitin and parkin at their respective S65 residues. This activates parkin to catalyze ubiquitination of many OMM proteins that are then either degraded by the proteasome or act as signals for autophagic clearance of the mitochondrion. Parkin-mediated mitochondrial ubiquitination and mitophagy are counteracted by specific deubiquitinases. PINK1/parkin-mediated mitophagy can be triggered in a variety of neuronal and non-neuronal cells in vitro, typically by exposure to mitochondrial depolarizing agents. Along with other mitochondrial quality control mechanisms PINK1/parkin-mediated mitophagy contributes to the maintenance of a healthy mitochondrial network.

2Despite the wealth of mechanistic information on PINK1/parkin-mediated mitophagy in cultured cells, questions still surround the existence and physiological relevance of this pathway in vivo. Direct evidence for the occurrence of PINK1/parkin-mediated mitophagy in vivo is still scarce. Ubiquitin phosphorylated at S65 (pS65-Ub), a biomarker of PINK1 activity, accumulates in brains from elderly human subjects and from mice with a genetic defect in mitochondrial DNA proofreading, suggesting that PINK1-mediated mitophagy is induced by aging and accumulation of mitochondrial damage. However, a recent study reported that mitophagy occurs independently of PINK1 in the mouse. In Drosophila, deficiency of PINK1 or parkin causes reduced life span and severe flight muscle degeneration with accumulation of swollen mitochondria. In parkin mutant flies, half-lives of mitochondrial proteins are drastically prolonged, consistent with a role for parkin in mitophagy, but the effect of PINK1 on mitochondrial protein turnover is much more limited. Overall, it remains to be directly demonstrated whether PINK1 and parkin can target damaged mitochondria to lysosomes in vivo.

3Recently, the mt-Keima reporter was developed to quantitatively image mitophagy in vitro and in vivo. Here, we used live mt-Keima imaging and correlative light and electron microscopy (CLEM) to assess mitophagy in Drosophila in vivo and to determine the role of parkin and PINK1 in this pathway.

4The phenotype of parkin- and PINK1-deficient flies is especially striking in flight muscle. We therefore focused on the role of mitophagy in this tissue. We generated flies that express mt-Keima specifically in muscle (mef-2-Gal4). The mt-Keima reporter is a mitochondrially targeted form of Keima, a fluorescent protein that resists degradation by lysosomal proteases. The peak of the mt-Keima excitation spectrum shifts when mitochondria are delivered to acidic lysosomes, which allows dual-excitation ratiometric quantification of mitophagy (Figure 1A). Transmission electron microscopy (TEM) analysis showed that mt-Keima expression did not change the morphology of muscle mitochondria (Figure 1B). The mt-Keima signal extensively colocalized with the mitochondrial protein ATP synthase β, confirming that mt-Keima was properly targeted to mitochondria (Figure 1C). Interestingly, a small subset of mt-Keima structures had high 543 nm/458 nm ratio values, indicative of an acidic environment (Figure 1D). These high 543/458 ratio mt-Keima puncta colocalized with the lysosomal dye LysoTracker (Figure 1E). Thus, the high 543/458 ratio mt-Keima puncta probably represented mitochondria that had been delivered to lysosomes. The abundance of these puncta was very low in muscle cells of 1-week-old adult flies, but strongly rose with aging, showing an approximately tenfold increase by the age of 4 weeks (Figure 1F,G). To assess the role of autophagy in the biogenesis of the high 543/458 ratio mt-Keima puncta, we overexpressed a kinase-dead version of Atg1 (Atg1K38A), the homologue of mammalian ULK1. Atg1/ULK1 is needed in the early steps of autophagosome formation and is also involved in mitophagy. When overexpressed, kinase-dead Atg1 exerts dominant-negative effects. Overexpression of Atg1K38A suppressed the high levels of mitophagy observed in aged muscle cells (Figure 1F,G). The low residual level of high 543/458 ratio mt-Keima puncta that persisted after Atg1K38A overexpression, may result from Atg1-independent mechanisms of autophagy induction.

5To demonstrate the occurrence of mitophagy in Drosophila indirect flight muscle more definitively, we resorted to CLEM. We first performed live mt-Keima imaging of muscle cells to identify regions of interest that contained both ‘acidic’ and ‘neutral pH’ mt-Keima structures (Figure 2A). After fixation, we burned laser marks around the regions of interest using near-infrared branding (NIRB) to be able to re-identify these regions after processing for TEM (Figure 2B–D). Interestingly, ‘neutral pH’ mt-Keima structures colocalized with typical mitochondria at the TEM level, while ‘acidic’ mt-Keima puncta showed remarkable overlay with smaller organelles with the characteristic electron-dense appearance of lysosomes (Figure 2E–N). Many of the mt-Keima-positive lysosomes contained densely packed concentric membranes surrounding a dense core, giving them the appearance of multilamellar bodies (Figure 2N). Multilamellar bodies are lysosomal organelles that are found in various cell types in physiological conditions but also accumulate in lysosomal storage diseases.

6We then crossed the mt-Keima-expressing flies with PINK1 loss-of-function mutant flies (PINK1B9) and parkin RNAi flies. The low level of mitophagy in 1-week-old adult flies was not affected by deficiency of PINK1 or parkin (Figure 3A,B). However, the age-dependent increase in mitophagy in 3- and 4-week-old control flies was completely abolished in the PINK1B9 and parkin RNAi flies (Figure 3A,B).

7The DUBs USP15 and USP30 oppose parkin-mediated mitochondrial ubiquitination and mitophagy in cultured human cells. Interestingly, the yeast homologue of USP15 (Ubp12) deubiquitinates the MOM protein Fzo1, the yeast homologue of the mammalian parkin substrates mitofusin 1 and 2, pointing to a conserved role for USP15 in deubiquitination at the MOM. Knockdown of the Drosophila homologs of USP15 (CG8334, hereafter called dUSP15) and USP30 (CG3016, hereafter dUSP30) largely rescues the mitochondrial defects of parkin-deficient fly muscle in vivo. To assess the effects of dUSP15 and dUSP30 on PINK1/parkin-mediated mitophagy in vivo, we knocked down dUSP15 and dUSP30 in mt-Keima-expressing flies using RNAi. Levels of dUSP15 mRNA in dUSP15 RNAi lines were 57.6 ± 2.5% of control levels (n = 6), and dUSP30 mRNA levels in dUSP30 RNAi lines were 16.4 ± 5.8% of controls (n = 6). Knockdown of dUSP15 and, to a lesser extent, dUSP30 rescued the mitophagy defect of two different parkin RNAi fly lines (Figure 3C,D). The mitophagy defect of PINK1B9 flies was partially rescued by dUSP15 knockdown, but unaffected by dUSP30 knockdown (Figure 3C,D).

8Dopaminergic neurons of parkin mutant Drosophila also accumulate abnormal mitochondria. We therefore expressed mt-Keima in dopaminergic neurons (TH-GAL4). As in muscle cells, mitophagy increased with aging (Figure 4A,B). Loss of parkin suppressed mitophagy in 4-week-old flies (Figure 4A,B), and this deficit was rescued by dUSP15 knockdown (Figure 4C,D).

9Our data show that mitophagy occurs in Drosophila flight muscle and dopaminergic neurons in vivo, even in the absence of exogenous mitochondrial toxins. Mitophagy in these cells rises with aging, and this age-dependent increase is abrogated by PINK1 or parkin deficiency. Thus, PINK1 and parkin appear to be particularly important for mitophagy during aging, which may explain why loss of these proteins in patients causes an age-dependent neurodegenerative disease rather than a congenital or developmental disorder. Consistent with this, pS65-Ub, a readout of PINK1 activity, is almost undetectable in postmortem brains from neurologically normal, young human subjects, but accumulates in brains from elderly individuals.

10A recent study reported that PINK1 is dispensable for mitophagy in the mouse in vivo. This conclusion was based on imaging using the mito-QC reporter, which differs from the mt-Keima probe used in our study. mito-QC is targeted to the OMM, whereas mt-Keima is localized to the mitochondrial matrix. Upon mitochondrial damage, activated parkin ubiquitinates a wide variety of OMM proteins, many of which are then extracted from the OMM and degraded by the proteasome before engulfment of the mitochondrion into an autophagosome. The mito-QC probe consists of a tandem mCherry-GFP tag fused to a large portion of the OMM protein FIS1, which is itself a substrate of parkin. If mito-QC is ubiquitinated by parkin and extracted from the OMM, this would prevent mito-QC delivery to lysosomes and reduce its sensitivity as a reporter for PINK1/parkin-mediated mitophagy. An alternative explanation for the discrepant results could be that mice may not accumulate sufficient mitochondrial damage in their lifespan to activate PINK1/parkin-mediated mitophagy. The abundance of pS65-Ub in cerebral cortex is very low in wild-type mice, but is substantially higher in mice expressing a proof-reading-deficient version of mitochondrial DNA polymerase γ. This suggests that a ‘second hit’ in addition to aging may be required to induce PINK1/parkin-mediated mitophagy in mice and may explain why PINK1 or parkin deficiency by itself does not cause a degenerative phenotype in this species.

11In a recent Drosophila study Lee et al. were unable to detect mitophagy in flight muscle and found no evidence for a major role for PINK1 or parkin in mitophagy in other fly tissues. The discrepancy with our correlative mt-Keima and TEM imaging findings may be due to the fact that Lee et al. mostly relied on imaging with mito-QC, a probe that may be less sensitive for parkin-mediated forms of mitophagy, as discussed above. Also, Lee et al. restricted their analysis of flight muscle to 2-day-old or younger flies, and may thus have missed the PINK1/parkin-dependent rise in mitophagy that occurs at later ages.

12Recent publications proposed a distinction between basal and induced mitophagy. According to this terminology, the physiological mitophagy observed in our study could be labeled as basal. However, this may be confusing, because the term ‘basal’ suggests a relatively stationary background phenomenon, whereas our data show that ‘basal’ mitophagy in the fly is strongly induced by normal aging.

13We did not detect a significant reduction in mitophagy in 1-week-old PINK1B9 and parkin RNAi flies, at a time point when these flies already display mitochondrial abnormalities. This suggests that the mitochondrial changes in 1-week-old PINK1B9 and parkin RNAi flies are caused by loss of aspects of PINK1 and parkin function that are unrelated to mitophagy. For example, parkin also regulates Ca2+ transfer from ER to mitochondria. PINK1 promotes mitochondrial complex I activity through phosphorylation of the complex I subunit NdufA10 and is involved in crista junction remodeling via phosphorylation of dMIC60. Alternatively, the early mitochondrial abnormalities in PINK1- and parkin-deficient flies may be due to more subtle reductions in mitophagy that are not detectable with mt-Keima imaging.

14Genetic manipulation in Drosophila is relatively straightforward. As illustrated by our dUSP15 and dUSP30 knockdown experiments, this novel mt-Keima fly model is a convenient tool to determine the impact of individual genes on PINK1/parkin-mediated mitophagy in vivo. This model will greatly facilitate the identification of targets for modulation of a pathway with growing relevance for neurodegenerative diseases.

15All Drosophila melanogaster stocks and experimental crosses were kept on standard corn meal and molasses food at room temperature. The mt-Keima construct (mt/mKeima/pIND(SP1)) was a gift from Dr. A. Miyawaki (RIKEN Brain Science Institute, Japan). The mt-Keima cDNA was cloned into the NotI and Xba1 sites of pUAS-attB and inserted in integration site VK20 after in-house injection. CG8334 transgenic UAS-RNAi (18981) and parkin KK UAS-RNAi (107919, named parkin2 RNAi in this paper) lines were obtained from the Vienna Stock Center (VDRC) and CG3016 transgenic UAS-RNAi line (3016 R-2) from NIG-Fly Stock Center. Parkin (37509, named parkin1 RNAi in this paper) and control (luciferase, 31603) TRiP UAS-RNAi lines, mef-2-GAL4, TH-GAL4, PINK1B9 and Atg1K38A were obtained from Bloomington stock center (Indiana, USA). To quantify parkin, CG8334 and CG3016 mRNA levels under control of mef-2-GAL4, RNA was isolated from adult thoraces and real-time RT-PCR was performed as previously described using primers 5’-CCAGCAATGTCACCATCAAAG-3’ and 5’-GCGTGTCCACTCAGTCTG-3’ for parkin, 5’-GGAGTGACGCATCTTGAG-3’ and 5’-TTCTTTGGTATGGGTGGACTG-3’ for CG8334 and 5’ TACGCCATAGCAATCTGGGG-3’ and 5’- CTCGTGTATCTGCTGGCGTT-3’ for CG3016. The data were normalized using RP-49, a ribosomal gene. Real-time PCR showed that parkin mRNA levels in parkin1 RNAi flies were 49,7 ± 2,3% of control levels (n = 5). Parkin mRNA levels in parkin2 RNAi flies were determined previously.

16Flies were dissected in HL3 buffer. Indirect flight muscle fibers and complete brains were immobilized on a glass slide in low gelling temperature agarose and analyzed using a Leica TCS SP5 II confocal microscope equipped with a 63x objective lens (HC PL APO 63x/1.4 CS2), a multi-argon laser (458, 476, 488 nm) and a He/Ne laser (543 nm). Mt-Keima was imaged in two channels via two sequential excitations (458 nm, green; 543 nm, red) and using a 600 to 695 nm emission range. Images from random microscopic fields were captured and analyzed by an investigator blinded to genotype and age. For muscle analysis, at least 7 z-stacks with 0.2 µm slice thickness were taken per fly. For dopaminergic neuron analysis, at least 10 PPL1 neurons were analyzed per fly. Ratio (543/458) images were created using the Ratio Plus plugin in ImageJ. High (543/458) ratio areas were segmented and quantified with the Analyze Particles plugin in ImageJ. The total mitochondrial area was quantified with the Analyze Particles plugin by calculating the area of the total emission at 543 nm excitation. The parameter (high [543/458] ratio area/total mitochondrial area) was used as an index of mitophagy, as described. LysoTracker (DND-26, Thermo Fisher) was imaged using a 488 nm excitation and a 495–550 nm emission filter.

17Indirect flight muscle was dissected in HL3, live imaged for mt-Keima and fixed in 4% paraformaldehyde for 1 hr. Samples were washed three times for 10 min in PBS and placed in blocking solution (5% normal donkey serum in PBS, 0.1% Triton X-100) for 30 min, after which the samples were incubated overnight at 4°C with antibody against ATP synthase β (1:500; ab14730, Abcam) in PBS, 2% normal donkey serum, 0.1% Triton X-100. After washing, samples were incubated for 2 hr with Alexa 488-conjugated secondary antibody in PBS, 2% normal donkey serum, 0.1% Triton X-100. Samples were washed and mounted on a glass slide in Vectashield mounting medium (Vector Laboratories). Alexa 488 fluorescence was imaged using a 488 nm excitation and a 495–550 nm emission filter.

18CLEM studies using near-infrared branding (NIRB) were performed as described previously. Tissues were first imaged live in low gelling temperature agarose (3% in 0.1 M phosphate buffer, pH 7.4 [PB]), and fixed overnight in 0.5% glutaraldehyde, 4% PFA in 0.1 M PB, pH 7.4 at 4°C. Samples were rinsed in 0.1 M PB after which fixed samples were imaged using the Leica SP5 confocal microscope to check whether fixing affected overall tissue morphology. The bleaching function of the ZEN2010 software (Zeiss) was used to perform NIRB on a LSM 780 inverted confocal microscope. Branding marks were introduced to the tissue with a Mai Tai DeepSee two-photon laser (Spectra-Physics) at 800 nm and 40% maximal power output. Z-stacks of the region of interest were acquired before and after NIRB. Processed samples were post-fixed in 2% glutaraldehyde, 4% paraformaldehyde and 0.2% picric acid in 0.1 M PB at 4°C overnight or until further processing. Next, samples were osmicated for 1 hr in 2% OsO4, 1.5% potassium ferrocyanide in PB and subsequently stained for 30 min in 0.2% tannic acid, followed by overnight incubation in 0.5% uranyl acetate in 25% methanol. Next day, samples were stained en bloc with lead aspartate and dehydrated in an ascending series of ethanol solutions followed by flat embedding in Agar 100 (Laborimpex; Agar Scientific). Flat-embedded sections were mounted on aluminum pin stubs (Gatan) with conductive epoxy (Circuit Works). A Zeiss Sigma Variable pressure SBF-SEM with 3View technology (Gatan) was used to approach the region of interest. Imaging was done at 1.3 kV with a pixel size of 20 nm and sections of 200 nm. Regions of interest could be located based on the branding marks and muscle fiber morphology. When the region of interest was reached, 70 nm thick, serial ultrathin sections were cut using a Reichardt Ultracut E ultramicrotome. All sections were collected as ribbons of 4–5 sections on triple slot grids (Ted Pella). Images were taken on a JEOL JEM-1400 Transmission Electron Microscope operated at 80 kV.

19Values and error bars represent mean ± SEM, and n refers to the number of biological replicates. Significance of differences between conditions was analyzed with one-way ANOVA and post-hoc Holm-Sidak test (SigmaStat 3.5, Systat). No outliers were excluded.