Drosophila were raised under standard conditions in a temperature-controlled incubator with a 12 h:12 h light:dark cycle at 25 °C and 65% relative humidity, on food consisting of agar, cornmeal, molasses, propionic acid and yeast. Unless specifically indicated otherwise, experiments using adult flies were performed on animals aged ~ 2 days (1–3 d).

Cisd KO (Cisd^−/−) and UAS-Cisd were previously reported [32]. UAS-ref(2)P (p62) was generously provided by Prof. D. Walker [33], park²⁵ mutants [34], and UAS-mito-QC [8] lines have been previously described. Pink1^B9 mutants [35] were kindly provided by Prof. J. Chung (Seoul National University). The following strains were obtained from Bloomington Drosophila Stock Center (BDSC, RRID:SCR_006457): w¹¹¹⁸ (BDSC_6326), da-GAL4 (BDSC_55850), nSyb-GAL4 (BDSC_51635), Mef2-GAL4 (BDSC_27390), TH-GAL4 (BDSC_8848), UAS-lacZ (BDSC_8529), UAS-mito-GFP (BDSC_8843), UAS-hCISD1 (BDSC_77990), UAS-hCISD2 (BDSC_76845), UAS-Atg5 RNAi (BDSC_27551), UAS-Atg8a (BDSC_84981), UAS-GFP-mCherry-Atg8a (BDSC_37749) and UAS-mCherry-Atg8a (BDSC_37750, [36]). The following strains were obtained from Vienna Drosophila Research Centre (VDRC): UAS-lacZ RNAi GD (v51446), Cisd RNAi GD (v33925) and KK (v104501).

Where multiple UAS transgenes were present in experimental animals, an inert UAS transgene (e.g., UAS-lacZ, UAS-lacZ RNAi, UAS-fluorophore, etc.), appropriate for the experimental conditions, was always included to account for possible titration effects. Since X chromosome nondisjunction is present in multiple balanced Pink1^B9 mutant stocks, correct genotypes were determined by either combining paternal animals with X chromosome markers or by PCR-based genotyping of discarded tissue after dissection.

The following mouse antibodies were used: ATP5A (Abcam, ab14748, 1:10000), Ubiquitin (clone P4D1, Cell Signalling Technology, 3936, 1:1000), CISD2 (Proteintech, 66082-1-Ig, 1:1000), TOM20 (clone F-10, Santa Cruz, sc-17,764, 1:1000), HA (Cell Signaling, 2367, 1:1000), tyrosine hydroxylase (Immunostar, 22,491), Tubulin (Sigma, T9026, 1:5000). The following rabbit antibodies were used in this study: pS65-Ub (Cell Signaling Technologies, 62,802 S, 1:1000), GABARAP/Atg8a (Abcam, ab109364, 1:1000), ref(2)P/p62 (Abcam, ab178440, 1:1000), PINK1 (Cell Signaling Technology, 6946, 1:1000), CISD1 (Proteintech, 16006-1-AP, 1:1000), CISD2 (Proteintech, 13318-1-AP, 1:1000), FLAG (Cell Signaling, 2368 S, 1:1000), Tubulin (Abcam, ab18251, 1:5000). Rat anti-Calnexin (clone W17077C, BioLegend, 699,401).

The following secondary antibodies were used: Goat anti-mouse-HRP (Abcam, Ab6789-1, 1:10000), Goat anti-rabbit-HRP (Invitrogen, G21234, 1:10000), Goat anti-mouse-IRDye 680RD (Li-Cor, 926-68070), Goat anti-rabbit- IRDye 800CW (Li-Cor, 926-32211), Goat anti-mouse-AF405 (Invitrogen, A48255), Goat anti-mouse-AF488 (Invitrogen, A11001), Goat anti-mouse-AF594 (Invitrogen, A11005), Goat anti-rabbit-AF488 (Invitrogen, A11008) and Donkey anti-rat-AF488 (Invitrogen, A21208).

DMEM/F12, GlutaMAX Supplement (31,331,093, Gibco), DMEM, high glucose, GlutaMAX Supplement, pyruvate (31,966,047, Gibco), Fetal Bovine Serum (FBS) (10,270,106, Gibco), MEM Non-Essential Amino Acids Solution (100X) (11,140,050, Gibco), Penicillin-Streptomycin (10,000 U/mL) (15,140,122, Gibco), Opti-MEM (31,985,062, Gibco), Lipofectamine RNAiMAX (13,778,075, Invitrogen), Lipofectamine3000 (L3000008, Invitrogen), phosphate buffered saline (DPBS (1X), no calcium, no magnesium, 14190-169, Gibco), Formaldehyde (FA) (28,908, Fisher Scientific Ltd) and Pierce BCA Protein Assay Kit-500 mL (23,223, Life Technologies) were purchased from Thermo Fisher Scientific. The following reagents were obtained from Sigma: Dimethyl sulfoxide (DMSO) (276855-250ML), Rosiglitazone (R2408-50MG), Antimycin A (A8674-25MG). 5X siRNA Buffer (B-002000-UB-100, Horizon Discovery), mitoNEET Inhibitor NL-1 (475825-10MG, Calbiochem), cOmplete protease inhibitors (4,693,159,001, Roche), PhosSTOP phosphatase inhibitors (4,906,837,001, Roche), Bovine Serum Albumin (BSA, Fraction V, Cold-ethanol Precipitated, BP1605100, Fisher BioReagents), Oligomycin A (495455-10MG, Merck Life Science Limited), Chloroquine (C6628-25G, Merck Life Science Limited).

ARPE-19 cells, stably expressing mCherry-GFP-Fis1(101–152) referred to as ARPE-19-MQC, were a kind gift of Dr Ian Ganley (University of Dundee), and hTERT-RPE1, hTERT-RPE1-YFP-PARKIN cells were generously donated by Prof. Jon D Lane (University of Bristol). Cells were cultured in Dulbecco’s modified Eagle’s medium DMEM/F12 (hTERT-RPE1, hTERT-RPE1-YFP-PARKIN, and ARPE-19-MQC) or DMEM (U2OS) supplemented with 10% FBS, 1% non-essential amino acids and 1% penicillin/streptomycin. For siRNA experiments, cells were reverse-transfected with 20 nM of non-targeting (NT1) or target-specific siRNA oligonucleotides (Dharmacon On-Target Plus or siGenome, ThermoFisher Scientific), using Lipofectamine RNAi-MAX (Invitrogen) according to manufacturer’s instructions for 72 h. Alternatively, cells were seeded and transfected for 48 h with pcDNA3.1-CISD1-FLAG or pcDNA3.1-CISD2-FLAG using Lipofectamine3000 (Invitrogen) as per manufacturer instructions. For drugs treatment experiments, cells were treated with 100 µM of rosiglitazone or NL-1 for 24 h.

siRNAs used in this manuscript were as follows: ON-TARGETplus Non-Targeting siRNA oligo #1 (D-001810-01-20, Horizon Discovery), ON-TARGETplus Human CISD1 siRNA SMARTpool (L-020954-01-0020, Horizon Discovery), ON-TARGETplus Human CISD2 siRNA SMARTpool (L-032593-02-0020, Horizon Discovery). ORFs from hCISD1 and hCISD2 were amplified from transgenic flies carrying hCISD1 (BDSC_77990) or hCISD2 (BDSC_76845). FLAG was introduced at the C-terminal during the PCR amplification. Gibson assembly was used to introduce these PCR products into a pcDNA3.1 cut with KpnI and NotI. The following primers were used for the PCR amplification:

hCISD1_KpnI_pCDNA31_Fwd: CTGGCTAGCGTTTAAACTTAAGCTTGGTACCATGAGTCTGACTTCCAGTTCCA.

For the analysis of cell culture lysates by immunoblot, cells were rinsed once with cold PBS and subsequently lysed with 100–200 µL cold RIPA buffer (50 mM Tris pH8, 180 mM NaCl, 1mM EDTA, 1% Triton-X100, 0.5% SDS), supplemented with cOmplete protease inhibitors and phosphatase inhibitors. Cell lysates were incubated on ice for 15-20 min before being harvested into an Eppendorf tube. In the case of Drosophila lysates, animals (5 to 20 per replicate) with the right genotype were selected under CO₂ and flash-frozen in liquid N₂, all direct comparing genotypes being harvested equally at the same time. 150–200 µL of ice-cold RIPA buffer supplemented with cOmplete protease inhibitors and phosphatase inhibitors was added to 2 mL tubes containing 1.4 mm ceramic beads (15,555,799, Fisherbrand) to which flies were added and lysed using a Minilys homogeniser (Bertin Instruments). Flies were shaken twice at maximum speed, 10 s followed by a brief incubation on ice, after which samples were shaken again at maximum speed, 10 s, making a total of three cycles. After the lysis, Drosophila samples were returned to ice for at least 40 min. Both cell culture and Drosophila lysates were then centrifuged 5 min at maximum speed (21,000 x g) at 4 °C. Supernatant was then transferred to a fresh Eppendorf tube and centrifuged. This step was repeated until samples were completely cleared. For experiments evaluating Cisd dimer bands (reducing vs. non-reducing conditions), RIPA buffer was additionally supplemented with fresh 20 mM iodoacetamide.

Protein content was determined by BCA assay as per manufacturer instructions. Samples were then diluted in Laemmli ×4 Sample Buffer (1,610,747, Bio-Rad) supplemented with 100 mM 2-Mercaptoethanol and analysed by SDS-PAGE using Mini-PROTEAN TGX Gels 4–20% (4,561,093, Bio-Rad) or similar. Gels were transferred onto precast nitrocellulose membranes (1,704,158, BioRad) using the BioRad Transblot Turbo transfer system, and membranes were immediately washed in distilled water and stained with ponceau (P7170-1 L, Sigma) Membranes were then blocked by incubation with 5% (w/v) semi-skimmed milk in TBS 1X containing 0.1% (v/v) Tween-20 (TBStw) for 1 h at least. Membranes were then incubated in a shaker at 4 °C overnight with appropriate primary antibodies in 5% milk-TBStw. 3 washes of 10 min with TBStw were performed to rinse the membranes and then were incubated for one hour in secondary antibodies made in 5% milk-TBStw. Membranes were then washed 3 times,10 min each, in TBStw. Membranes were developed either with ECL reagent (RPN2232, Amersham) using the Amersham Imager 680 (GE Healthcare), or alternatively with the LICOR secondary antibodies and imaged in an Odyssey LICOR imaging system. Image analysis was performed using of FIJI (Image J). For experiments evaluating Cisd dimer bands (reducing vs. non-reducing conditions), samples were diluted either in Laemmli x4 Sample Buffer supplemented with 100 mM 2-Mercaptoethanol (reducing) and/or in Laemmli x4 Sample Buffer alone (non-reducing conditions).

5 adult flies per biological replicate were homogenised in TRIzol and RNA extraction was carried out using Direct-zol RNA MiniPrep kit (Zymo Research, Z2050). DNA was removed with Turbo DNase free (Ambion, AM1907), and cDNA synthesis was performed using a Maxima First Strand kit (ThermoFisher, K1641) followed by a qRT-PCR using SYBR green (ThermoFisher, 4,309,155) with a CFX96 Touch Real-Time PCR Detection System. The relative transcript levels of each target gene were normalised against RpL32 and Tubulin (Tub84B) mRNA levels; quantification was performed using the comparative C_T method [37]. Primers used:

Tubulin - Fwd: TGGGCCCGTCTGGACCACAA; Rev: TCGCCGTCACCGGAGTCCAT [38],

Drosophila brains were dissected from 30-day-old flies and immunostained with anti-tyrosine hydroxylase as described previously [41]. Tyrosine hydroxylase-positive neurons were counted under blinded conditions. For immunostaining of adult flight muscles and larval brains, animals were dissected in PBS 1X and fixed in 4% FA/PBS for 30 min at room temperature, permeabilized in 0.3% Triton X-100 for 30 min, and blocked with 1% BSA in 0.3% Triton X-100 PBS for 1 h at RT. Tissues were then incubated on a shaker with the appropriate primary antibody: ATP5A antibody (Abcam, ab14748, 1:500), pUb (pS65-Ub, Cell Signalling Technologies, 62,802 S, 1:250) and/or ref(2)P/p62 (Abcam, ab178440, 1:1000), diluted in 1% BSA in 0.3% Triton X-100 PBS overnight at 4 °C. Next day, samples were rinsed 3 times 10 min with 0.3% Triton X-100 in PBS, and incubated with the appropriate fluorescent secondary antibodies overnight at 4 °C. The tissues were washed 3 times in 0.3% Triton X-100 in PBS followed by a last wash on PBS and mounted on slides using Prolong Diamond Antifade mounting medium (Thermo Fisher Scientific) and imaged next day. Tissues were imaged with a Zeiss LSM880 confocal microscope (Carl Zeiss MicroImaging) equipped with a Nikon Plan-Apochromat 63x/1.4 NA oil immersion objective. Images were analyses using FIJI (Image J).

Third instar larvae overexpressing UAS-mito-GFP with the pan-neuronal driver nSyb-GAL4 were dissected in PBS and fixed in 4% FA/PBS for 30 min at room temperature. The tissues were washed 3 times in PBS and mounted on slides using Prolong Diamond Antifade mounting medium (Thermo Fisher Scientific) and image next day in a Zeiss LSM880 confocal microscope (Carl Zeiss MicroImaging) equipped with Nikon Plan-Apochromat 63x/1.4 NA oil immersion objectives.

The repetitive iteration startle-induced negative geotaxis (RISING, or ‘climbing’) assay was performed using a counter-current apparatus. Experiments were performed using multiple small groups of flies of the indicated genotypes and ages. Briefly, flies were collected under minimal CO₂ anaesthesia 24 h before the assay. On the day of assay, flies are transferred to assay tubes and allowed to acclimatise in the behaviour room for ~ 1 h. Cohorts of ~ 20 flies (25 max.) were placed into the first chamber of the counter-current apparatus, tapped to the bottom and given 10 s to climb a 10 cm distance. Flies that climbed > 10 cm within this time were shifted to the neighbouring tube. This procedure was repeated five times, and the number of flies that remained in each chamber were counted. The weighted performance of several groups of flies for each genotype was normalised to the maximum possible score and expressed as the climbing index [34]. To age flies for climbing assays, they were transferred to fresh tubes every 2–3 days. For lifespan experiments, flies were grown under identical conditions at low density. Progeny were collected under very light (< 30 s) anaesthesia and kept in tubes of 10–25, transferred to fresh media every 2–3 days and the number of dead flies recorded. Percent survival was calculated at the end of the experiment after correcting for any accidental loss or escape using https://flies.shinyapps.io/Rflies/ (Luis Garcia).

For mitolysosome imaging in UAS-mito-QC, tissues were dissected in PBS, fixed in 4% FA/PBS at pH 7 for 30 min at room temperature and mounted. Third instar larval and adult brains were mounted in Prolong Diamond Antifade mounting medium (Thermo Fisher Scientific), while thoraces were mounted in VECTASHIELD Antifade Mounting Medium (H-1000, Vector Laboratories). Larval and adult brains were imaged using Andor Dragonfly spinning disk microscope (Oxford Instruments Group), equipped with a Nikon Plan-Apochromat 100x/1.45 NA oil immersion objective and iXon camera taking 10–13 μm z-stacks with 0.2 μm step size. Adult thoraces were imaged with a Zeiss LSM 880 confocal microscope (Carl Zeiss MicroImaging) equipped with Nikon Plan-Apochromat 63x/1.4 NA oil immersion objectives, taking 10 μm z-stacks with 1 μm step size. We find that the time elapsed from dissection-mounting and imaging is critical for the retrieval of red-only dots. Tissue specific optimisation was required. Larval brains were imaged early next day after dissection, while adult brains and thoraces were imaged briefly after (2–4 h later). Image analysis of larval brains was done employing Imaris (version 9.7.0) analysis software (BitPlane; RRID:SCR_007370) as previously described [8]. For adult Drosophila brains and thoraces, as well as for ARPE-19-MQC cells, mitolysosomes were analysed using the FIJI plug-in mito-QC Counter [42] as advised by authors. Analysis was performed in maximum projection images of Drosophila adult brain and ARPE-19-MQC cells, while Drosophila adult thoraces were analysed as per single plane.

The average number of mitolysosomes per cell was analysed per animal in larval and adult brain, as well as in ARPE-19-MQC cells, while mitolysosomes per area (µm) was calculated in Drosophila muscles. Data points in the quantification charts show the average mitolysosomes per biological replicate, where n ≥ 6 animals or n ≥ 60 cells for each condition.

Thoraces were dissected in 0.1 M cacodylate buffer and fixed in 1% glutaraldehyde and 4% paraformaldehyde in 0.1 M cacodylate buffer overnight at 4 °C. After washing with 0.1 M cacodylate buffer, the samples were treated with 1% OsO₄ for 1 h at 4 °C. After washing with ddH₂O, the samples were dehydrated in a gradient series using 30%, 50%, 70%, 95%, and 100% ethanol solutions at room temperature. The samples were then infiltrated using series epoxy resin at room temperature and mounted in pure resin. After polymerization at 60 °C for 2 days, 90 nm sections were prepared and observed with a Tecnai G2 Spirit TWIN transmission electron microscope (FEI, Hillsboro, OR, USA).

Adult muscle from GFP-mCherry-Atg8a reporter Drosophila were dissected in PBS, fixed with 4% FA/PBS at pH 7 for 30 min at room temperature, and mounted in Prolong Diamond Antifade mounting medium (Thermo Fisher Scientific). Samples were imaged the day after. Confocal images, acquired with a Zeiss LSM 880 microscope (Carl Zeiss MicroImaging) equipped with Nikon Plan-Apochromat 63x/1.4 NA oil immersion objectives, taking 10 μm z-stacks with 1 μm step size, were processed using FIJI plug-in mito-QC Counter [42]. The quantification of autolysosomes was performed in a plane-by-plane basis similarly to mitolysosome analysis in muscles. Data points in the quantification charts show average autolysosomes per area (µm) for individual animals, where n ≥ 6 animals for each condition.

ARPE-19-mito-QC-FIS1 (Ian Ganley, University of Dundee) cells were used to assess mitophagy in a human cell model. After 48 h knockdown with siRNA oligos, cells were seeded onto an Ibidi dish (IB-81,156, Ibidi, Thistle Scientific Ltd), and at 72 h post-knock down cells were imaged live using a spinning disk microscope. Generated images were processed using the FIJI plug-in mito-QC Counter [42] as previously described. For treatments with rosiglitazone and NL-1: cells were seeded onto an Ibidi dish, next day were treated with 100 µL rosiglitazone or NL-1 for 24 h and imaged and analysed as described above.

All steps were performed on ice or at 4 °C. 20–30 flies were prepared either fresh or after flash-freezing in liquid nitrogen, with all direct comparisons performed with flies that were prepared in the same manner. Flies were transferred into a Dounce homogeniser containing 400 µL of mitochondrial isolation buffer (225 mM mannitol, 75 mM sucrose, 5 mM HEPES, 0.1 mM EGTA, pH 7.4) containing cOmplete protease inhibitors (Roche) and PhosSTOP phosphatase inhibitors (Roche), and manually homogenised with 15 strokes of a pestle. The homogenate was transferred to an Eppendorf tube, after which further 400 µL of mitochondrial isolation buffer was added to the homogeniser and the sample was homogenised with a further 5 strokes. The homogenates were pooled and centrifuged at 1500 g at 4 °C for 6 min before being filtered through a 70 μm nylon cell strainer (352,350, Falcon). The sample was then centrifuged at 7000 g at 4 °C for 6 min and the resulting mitochondrial pellet was washed with isolation buffer once and finally resuspended into 150–200 µL RIPA buffer. The protein content was determined by BCA assay (Thermo Pierce).

For the assessment of ubiquitinated Cisd in mitochondrial fractions, USP2 deubiquitinating assay was performed as previously described [43]. 30 µg protein per mitochondrial fraction was treated with the pan-specific deubiquitinase USP2 (E-506, BostonBiochem). The USP2 enzyme was diluted in buffer (50 mM Tris–pH 7.5, 50 mM NaCl, 10 mM DTT) and then added to the subcellular fractions to a final USP2 concentration of 1 µM. The mixture was incubated for 45 min at 37 °C prior to analysis by immunoblotting.

Parental crosses were set on food containing 1 mM rosiglitazone diluted 1:1000 or DMSO as control. Resultant progeny of 0–3 days old were anaesthetised with CO₂ and the correct genotype was selected for further investigation.

0–3 days old flies were anaesthetised with CO₂ and thorax indentations and abnormally extended wing posture were assessed by light microscopy imaging using a Nikon SMZ stereo zoom microscope fitted with 1x Apo lens. The number of animals displaying each phenotype was counted in a binary manner (present-absent) respective to the total.

Charts report data as mean ± 95% CI for climbing behaviour or mean ± SEM for all other data sets. The number of biological replicates (N) are shown in individual graphs as individual data points or numerical value when N > 15. For statistical analyses of lifespan experiments, a log rank (Mantel-Cox) test was used. For climbing assay and qRT-PCR analyses, a Kruskal-Wallis nonparametric test with Dunn’s post-hoc correction for multiple comparisons was used. Where multiple groups were compared, statistical significance was calculated by one-way ANOVA with Bonferroni, Sidak’s or Geisser-Greenhouse post-test correction for multiple samples, as recommended in Prism. When only two groups were compared a Welch’s t test or Mann-Whitney test was used. All the samples were collected and processed simultaneously and therefore no randomization was appropriate. Unless otherwise indicated, image analysis was done in blind conditions. Statistical analyses were performed using GraphPad Prism 9 software (GraphPad Prism, RRID:SCR_002798). Statistical significance was represented in all cases as follows: *P < 0.05, **P < 0.01, ***P < 0.001 and ****P < 0.0001.

Since Cisd and its homologues have been identified as parkin substrates, we hypothesised that Cisd abundance could be increased in Drosophila parkin mutants. Examining the steady-state levels of Cisd, a ~ 14 kDa protein, by immunoblotting we found Cisd levels were increased in aged parkin null (park^–/–) and Pink1 null mutants (Pink1^–/–) (Fig. 1A, B). Mammalian CISD1 and CISD2 have been shown to form homodimers with high stringency [16, 44]. We also observed Cisd dimers in our immunoblots even under standard reducing conditions (Fig. 1A). In non-reducing conditions, Cisd is almost exclusively found as a dimer (Fig. S1A). We confirmed that the upper band was indeed a dimerised form of Cisd and not a ubiquitinated form as mitochondrial fractions treated with the promiscuous deubiquitinase USP2 retain the dimer band (Fig. S1B). We also noted that Cisd levels increased with age in wild-type (WT) flies (Fig. 1A, C), as previously described [32], with both monomer and particularly the dimer increasing in abundance (Fig. S1A). Age-dependent increase in Cisd was particularly notable in fly head lysates (Fig. S1C, D).

Fig. 1

Cisd accumulation disrupts mitochondria affecting locomotion and lifespan. A Representative immunoblot of whole fly lysates of the indicated genotypes from 2- and 30-day-old flies probed for Cisd (CISD2, Proteintech, 13318-1-AP) and Tubulin as loading control. B, C Quantification of Cisd monomer or dimer levels in 2- and 30-day-old flies analysed in A. D Confocal microscopy of flight muscle from 2-day-old wild-type control (WT ctrl) or Cisd overexpressing (OE) flies driven by da-GAL4, immunostained for ATP5A mitochondrial marker. E Climbing assay of 2- and 20-day-old WT and ubiquitous Cisd OE flies via the da-GAL4 driver. F Lifespan of WT and ubiquitous Cisd OE flies. N > 130 animals. G Confocal microscopy of neuronal soma from control (WT ctrl) or Cisd overexpressing (OE) larvae co-expressing the mito-GFP mitochondrial marker via the nSyb-GAL4 driver. H Climbing assay of 2- and 20-day-old WT and neuronal Cisd OE flies via the nSyb-GAL4 driver. I Lifespan of WT and neuronal Cisd OE flies. N > 90 animals. Statistical analyses: B RM one-way ANOVA with Geisser-Greenhouse correction; C paired t-test; E, H Mann-Whitney non-parametric test. *P < 0.05; **P < 0.01; ****P < 0.0001. Scale bars = 10 μm

As Cisd levels increased with age, and the Pink1-parkin pathway acts to restrict its steady-state levels, we reasoned that elevated Cisd levels may be a key contributor to age-related organismal decline. Thus, we induced transgenic overexpression of Cisd (Fig. S1E), with different tissue-specific drivers, and assessed mitochondrial integrity and organismal fitness and survival. Cisd overexpression via the moderate, ubiquitous driver da-GAL4 caused gross disruption to mitochondrial morphology, resulting in abnormally enlarged mitochondria in flight muscles (Fig. 1D), as previously reported [32]. It also severely impacted locomotor function and lifespan (Fig. 1E, F). Pan-neuronal overexpression of Cisd via nSyb-GAL4 also severely affected mitochondrial morphology and locomotion, and dramatically shortened lifespan (Fig. 1G–I). Notably, overexpression restricted just to DA expressing cells via TH-GAL4 also caused an age-related decline in motor ability (Fig. S1F), likely reflecting that DA neurons represent only part of a modulatory circuit for reactive locomotion [45], while muscle directed overexpression only caused a modest impact on locomotion and did not significantly affect lifespan (Fig. S1G, H).

Together, these data indicate that Cisd levels accumulate with normal ageing, as well as in Pink1 and parkin mutants, and that Cisd accumulation (especially neuronal accumulation) is sufficient to substantially impact organismal vitality.

Both CISD1 and CISD2 have been reported as substrates of Parkin in proteomic studies [14–16]. Evaluating the dynamics of their degradation during PINK1/Parkin-mediated mitophagy, we treated retinal pigment epithelial (RPE1) cells (with endogenous Parkin) or cells overexpressing YFP-Parkin (RPE1-Parkin OE) with antimycin A and oligomycin to induce mitophagy and assessed CISD1 and 2 levels over time (Fig. 2A). Monitoring PINK1-mediated phospho-ubiquitin (pUb) showed rapid induction of mitophagy that steadily increased over time with a concomitant gradual decrease in the OMM protein TOM20 (Fig. 2A). As expected, pUb deposition, its subsequent turnover, and TOM20 loss was accelerated in Parkin-overexpressing cells. In these conditions, CISD1 was rapidly degraded during mitophagy similar to TOM20, however, we found that CISD2 levels were minimally affected (Fig. 2A). Co-localisation studies analysing the distribution of FLAG-tagged CISD1 or 2 (Fig. 2B) showed that while the majority of CISD1 is localised to mitochondria (TOM20 staining), CISD2 is predominantly localised to the ER (Calnexin staining). These observations are consistent with other reports [22, 24] and align with the degradation of mitochondrial CISD1 while ER-localised CISD2 is largely spared during induced mitophagy.

Fig. 2

Drosophila Cisd is functionally more similar to CISD1. A Immunoblots of protein lysates from RPE1 cells ± YFP-Parkin overexpression (OE) treated with antimycin A (4 µM) and oligomycin (10 µM) for the indicated time to induce mitophagy, probed for mitophagy marker pUb, and degradation of TOM20 and CISD1/2 (CISD1, Proteintech, 16006-1-AP; and CISD2, Proteintech, 13318-1-AP), alongside respective loading control total Ub and Tubulin. Blot is representative of 3 replicate experiments. B Confocal micrographs of U2OS cells transfected with human CISD1-FLAG or CISD2-FLAG, counter-stained with antibodies against TOM20 (mitochondria) or Calnexin (ER). C Confocal micrographs of Drosophila larval neurons expressing transgenic mito-GFP and WT control, human CISD1-HA or CISD2-HA driven by nSyb-GAL4. D Immunoblots of protein lysates of 2- and 20-day-old whole flies expressing the indicated transgenes via da-GAL4 versus WT control, probed for pUb and CISD1/2 with (CISD2, Proteintech, 13318-1-AP). Scale bars = 10 μm

Drosophila encode a single CISD orthologue with substantial amino acid sequence homology to human CISD1 and CISD2, with marginally higher homology to CISD2 than CISD1 (CISD2; 45% identity, 66% similarity vs. CISD1; 42% identity, 65% similarity). Thus, we sought to determine whether Drosophila Cisd may function more similarly to one or other of the human orthologues. Ectopically expressing HA-tagged human CISD1 and CISD2 in Drosophila neurons, we found that expression of CISD1 induced the same mitochondrial hyperfused phenotype (Fig. 2C) as seen with overexpression of Cisd (Fig. 1G), while CISD2 expression had minimal effect (Fig. 2C). We previously noted that overexpression of Drosophila Cisd triggers mitochondrial pUb accumulation (discussed further below); this was mirrored by expression of CISD1 but not CISD2 (Fig. 2D). These results strongly support that Drosophila Cisd is functionally more related to CISD1 than CISD2.

We recently described that Pink1-mediated pUb can be readily detected in Drosophila by immunoblotting or immunohistochemistry [43], and noted that pUb accumulates on dysfunctional mitochondria in parkin mutants due to blocked autophagic flux (Fig. 3A, B). Here, we found that Cisd overexpression is also sufficient to cause a similar accumulation of pUb by immunoblotting (Fig. 3A). Immunohistochemistry showed a concomitant accumulation of pUb surrounding selected mitochondria in flight muscle (Fig. 3B). The build-up of pUb is consistent with Cisd accumulation blocking efficient mitochondrial degradation and mirrors blocked mitophagy due to the loss of parkin. To verify these observations using an orthogonal approach we analysed mitophagy flux using the mito-QC mitophagy reporter flies [8]. This revealed that Cisd overexpression caused a significant reduction in mitolysosome number in larval and adult neurons (Fig. 3C–F). Thus, Cisd overexpression is sufficient to block mitophagy flux.

Fig. 3

Cisd overexpression blocks mitophagy flux. A Immunoblot analysis of whole fly lysates from 2-day-old flies of the indicated genotypes, analysed for pUb, and Tubulin or total protein levels as loading controls. Blot is representative of 3 replicate experiments. Cisd overexpression was driven by da-GAL4. B Confocal microscopy analysis of adult Drosophila flight muscle from 2-day-old flies of WT control, parkin mutant and Cisd overexpression driven by da-GAL4 immunostained for mitochondria (ATP5A) and pUb. C–F Confocal analysis of mitophagy reporter mito-QC (OMM-localised tandem RFP-GFP) of WT or Cisd overexpression driven by nSyb-GAL4, in larval (C, D) or adult (E, F) neurons with ‘red-only’ mitolysosomes shown. D, F Number of mitolysosomes quantified shown in C and E. Data points indicate individual animals analysed. Statistical analysis: unpaired t-test; **P < 0.01; ****P < 0.0001. Scale bars = 10 μm

We postulated that the mitophagy block caused by Cisd accumulation may be due to defective autophagy. To investigate this, we immunostained adult flight muscle overexpressing Cisd for the autophagy receptor p62 (also called ref(2)P in Drosophila) alongside the autophagosome reporter, mCherry-Atg8a (Drosophila homologue of LC3). This showed that p62- and Atg8a-positive puncta accumulate close to mitochondria, forming large aggregates (Fig. 4A). Analysing these structures by electron microscopy confirmed that indeed numerous autophagosomes accumulate around abnormally enlarged and vesiculated mitochondria in Cisd overexpressing tissue (Fig. 4B).

Fig. 4

Cisd overexpression causes autophagosome accumulation and prevents autophagy. A Confocal micrographs of WT control versus Cisd overexpressing adult flight muscle via Mef2-GAL4, immunostained for p62 alongside imaging mCherry-Atg8a autophagosome reporter. B Electron micrographs of flight muscle as in A, showing multiple autophagic vesicles (inset) in proximity to disrupted mitochondria (arrowheads). C Larval neurons of WT control versus Cisd overexpressing or Atg5 knockdown (via nSyb-GAL4 driver) animals immunostained for p62 alongside ATP5a (mitochondria) and DAPI. D Immunoblot analysis of protein lysates from whole flies overexpressing Cisd (via da-GAL4) or WT controls. Blots were probed with antibodies against p62, Atg8a (LC3), Cisd (CISD2, Proteintech, 13318-1-AP) and Tubulin. Quantification of replicate blots is shown in Fig. S2A, B. E Quantification of the number of autolysosomes shown in F. Data points indicate individual animals analysed. Statistical analysis: unpaired t-test; ***P < 0.001. F Confocal microscopy analysis of adult flight muscle WT control versus Cisd overexpressing animals co-expressing the autophagy flux reporter GFP-mCherry-Atg8a driven by Mef2-GAL4. G, H Locomotor climbing assay of 2-day-old adult flies expressing the indicated transgenes (via da-GAL4). I Immunoblot analysis of equivalent samples analysed in G and H, probed for autophagy markers (p62 and Atg8a), Cisd and Tubulin. Quantification of replicate blots is shown in Fig. S2C. Statistical analyses: Kruskal-Wallis non-parametric test with Dunn’s post-hoc correction. ***P < 0.001; ****P < 0.0001. Scale bars = 10 μm for light microscopy, or indicated on image for EM

Immunostaining larval brains also revealed an accumulation of p62 puncta upon Cisd overexpression, similar to that seen in autophagy-deficient Atg5 RNAi flies (Fig. 4C). Likewise, immunoblot analysis of Cisd overexpressing flies showed increased levels of p62 and accumulation of the lipidated form of Drosophila LC3 (Atg8a-II) (Fig. 4D and Fig. S2A, B), indicative of blockage in autophagy flux. Consistent with these observations, the autophagy flux reporter, GFP-mCherry-Atg8a, also showed that autolysosome number was significantly decreased in Cisd overexpressing muscle (Fig. 4E, F). These data all indicate that maturation of autophagosomes to autolysosomes is impaired.

Finally, complementing these observations with organism-scale genetic manipulations, we found that overexpression of p62 or Atg8a was sufficient to significantly rescue the climbing deficit caused by Cisd overexpression (Fig. 4G, H). Interestingly, immunoblot analysis of these animals revealed that upregulation of either p62 or Atg8a significantly reduced the levels of overexpressed Cisd (Fig. 4I and Fig. S2C). The reasons for this are currently unclear and are likely due to increased autophagic flux [33], but certainly correlate well with the phenotypic rescue.

Taken together, these data suggest that the Cisd-related block in mitophagy is caused by an impairment in autophagosome engulfment of defective mitochondria, which ultimately results in autophagosome accumulation and a blockage of general autophagy flux.

Since Cisd accumulation blocks mitophagy, we reasoned that reducing Cisd levels could have the opposite effect and facilitate mitophagy. To approach this, we first assessed available loss-of-function reagents for Cisd. A null mutant, Cisd^–/–, had significant impact on organismal and mitochondrial phenotypes, conferring shorter lifespan, reduced climbing ability and fragmented mitochondria (Fig. S3A–D), consistent with a previous report [32]. Two independent transgenic Cisd RNAi constructs (termed GD and KK), which substantially reduced the Cisd protein levels by 99% and 96%, respectively (Fig. S3A), produced similar but generally milder phenotypes (Fig. S3B–D). Hence, for reasons of greater versatility (i.e., tissue-specific targeting) the RNAi lines were used for subsequent studies and the KK line was preferentially used.

Analysing mitophagic flux using the mito-QC reporter, we observed a significant increase in mitophagy with age both in muscle (Fig. 5A, B) and neurons (Fig. S3E, F). Interestingly, Cisd knockdown was able to significantly increase mitophagy specifically in muscle tissue of older flies (Fig. 5A, B). No induction was observed in young flies or in aged neurons (Fig. S3E, F). Of note, induction of mitophagy by loss of Cisd did not appear to occur via the upregulation of Pink1-mediated pUb as levels were indistinguishable from wild type (Fig. S3G). Moreover, we verified that the effects we observe upon Cisd OE/RNAi manipulation were specific to Cisd and not due to indirect transcriptional effects on Pink1 or parkin expression (Fig. S2D, E).

Fig. 5

Loss of Cisd promotes mitophagy flux. A, B Confocal microscopy analysis of mitophagy reporter mito-QC in flight muscle from WT control and Cisd knockdown flies driven by Mef2-GAL4 of the indicated ages. B Quantification of the number of mitolysosomes shown in A. Data points indicated individual animals analysed. Statistical analysis: unpaired t-test; *P < 0.05; **P < 0.01. C Immunoblot analysis of ARPE-19 cells expressing mito-QC, shown in D, with non-targeting siRNAs (Ctrl) or targeting CISD1, CISD2 or both. D Confocal microscopy analysis of mitophagy using mito-QC in cells shown in C. E Quantification of the number of mitolysosomes shown in D. Data points indicate replicate experiments. Statistical analysis: one-way ANOVA with Dunnett’s post-hoc correction; *P < 0.05; **P < 0.01. Scale bars = 10 μm

Seeking to validate the effect of Cisd loss of function in human cells, we knocked down CISD1 and/or CISD2 in a mitophagy reporter cell line (ARPE-19-mito-QC) (Fig. 5C) and assessed the formation of mitolysosomes. In agreement with the in vivo data, we found that siRNA-mediated depletion of CISD1, CISD2 or both fragmented mitochondria and significantly increased basal mitophagy (Fig. 5D, E). Together, these data indicate that reducing CISD levels can increase mitophagy in Drosophila and human cells.

We have established that Cisd is a key Parkin substrate that accumulates in Pink1/parkin mutants and its pathologic build-up induces mitochondrial disruption and neurodegenerative phenotypes. Therefore, we reasoned that reducing Cisd levels in Pink1/parkin mutants could ameliorate the pathological phenotypes. Pink1/parkin mutants exhibit an array of well characterised neurodegenerative phenotypes, including locomotor deficits, degeneration of flight muscle accompanied by gross mitochondrial disruption, progressive dopaminergic (DA) neuron loss, and shortened lifespan [34, 35, 46]. Ubiquitous knockdown of Cisd in parkin or Pink1 mutants significantly suppressed their climbing defects (Fig. 6A, B) and disruption of flight muscle mitochondria (Fig. 6C, D). The two RNAi lines suppressed equally well, so for simplicity subsequent experiments were conducted with the KK RNAi line. Cisd knockdown also significantly suppressed the degeneration of DA neurons (Fig. 6E, F) and shortened lifespan (Fig. 6G, H) in both parkin and Pink1 mutants. Interestingly, muscle-specific depletion of Cisd was sufficient to partially rescue the locomotor phenotype of parkin mutants while pan-neuronal knockdown did not (Fig. S4A, B). Overall, these data demonstrate that reduction in Cisd levels is sufficient to significantly reduce the pathology associated with loss of Pink1 or parkin. Supporting the genetic interaction between Cisd and Pink1/parkin, it was notable that Cisd overexpression, while viable in a wild-type background, caused complete lethality in Pink1/parkin mutants (Fig. S4C). Thus, there exists a strong genetic interaction between Cisd and Pink1/parkin implicating a strong functional interaction.

Fig. 6

Cisd knockdown ameliorates Pink1/parkin mutant phenotypes. A, B Climbing analysis of WT control versus (A) parkin or (B) Pink1 mutants with control or Cisd RNAi. GD and KK indicate independent RNAi lines driven by da-GAL4. See also Fig. S 3B. N is shown inside bars. C, D Confocal microscopy of flight muscle from (young) wild-type control (WT ctrl) or (C) parkin or (D) Pink1 mutants expressing control or Cisd RNAi stained for the mitochondrial marker ATP5a. E Quantification of dopaminergic (DA) neurons shown in F. F 30-day-old WT or parkin or Pink1 mutants expressing control or Cisd RNAi, immunostained for tyrosine hydroxylase. G, H Lifespan analysis of WT versus (G) parkin or (H) Pink1 mutants with control or Cisd RNAi. Statistical analyses: (A, B) Kruskal-Wallis non-parametric test with Dunn’s post-hoc correction, (E) one-way ANOVA with Sidak’s post-hoc correction; (G, H) Log rank (Mantel-Cox) test. *P < 0.05; **P < 0.01; ***P < 0.001; ****P < 0.0001. Scale bars = 10 μm

Having shown that loss of Cisd promotes mitophagy in wild-type flies and human cells, we assessed whether reducing Cisd increased mitophagy in the Pink1/parkin mutants. As expected, we found that mitochondria from Pink1 and parkin mutants accumulate high levels of p62 and Atg8a-II (Fig. 7A, B), consistent with a block in mitochondrial turnover, which also occurs upon Cisd overexpression (Figs. 4D and 7A and B). In this context, ubiquitous Cisd knockdown prevented the aberrant build-up of p62 and Atg8a-II in the Pink1/parkin mutants (Fig. 7A, B). Moreover, Cisd knockdown also substantially reduced the dramatic build-up of pUb that occurs in parkin mutants (Fig. 7C), consistent with a de-repression of the blocked autophagic flux, complementing the p62 analysis. This increased mitophagy flux is not sufficient to fully eliminate pUb build-up, back to WT levels, indicating that additional mechanisms are required. Immunostaining thoraces of these flies showed similar depositions of p62 aggregates in Pink1 and parkin mutants, which were also substantially reduced upon Cisd knockdown (Fig. 7D). The reduction in p62 build-up correlated with a notable improvement in mitochondrial morphology in flight muscle, labelled with ATP5a (Fig. 7D). Investigating mitophagy flux directly using the mito-QC reporter, we found that, indeed, reducing Cisd increased mitophagy in parkin or Pink1 mutant muscle (Fig. 7E–H), which likewise correlated with improved mitochondrial morphology and general tissue health.

Fig. 7

Loss of Cisd rescues Pink1 and parkin mutant degenerative phenotypes. A Immunoblot of mitochondrial protein lysates from Cisd knockdown driven by da-GAL4 in WT versus Pink1 and parkin mutant backgrounds, alongside respective controls, probed for p62 and Atg8a (LC3) levels with ATP5a loading control. B Relative p62 levels quantified from blots in A, normalised to WT control. Statistical comparisons are against WT control unless indicated. C Immunoblots of whole fly lysates of genotypes as in A, probed for pUb, Ub, Cisd (CISD2, Proteintech, 13318-1-AP) and Tubulin as loading control. D Confocal micrographs of adult flight muscle from Cisd knockdown driven by da-GAL4 in WT versus Pink1 and parkin mutant backgrounds, alongside respective controls, immunostained for APT5a and p62. E Confocal microscopy analysis of mitophagy reporter mito-QC in flight muscle from Cisd knockdown driven by Mef2-GAL4 in 2-day-old WT and parkin mutant backgrounds, alongside WT control. F Quantification of the number of mitolysosomes shown in E. G Confocal microscopy analysis of mitophagy reporter mito-QC in flight muscle from Cisd knockdown driven by Mef2-GAL4 in 2-day-old WT and Pink1 mutant backgrounds, alongside WT control. H Quantification of the number of mitolysosomes shown in G. Data points indicate individual animals analysed. Statistical analysis: one-way ANOVA with Sidak’s post-hoc correction; *P < 0.05, **P < 0.01; ****P < 0.0001. Scale bars = 10 μm

Taken together, these data demonstrate that Cisd-depletion induces upregulation of mitophagy independently of Pink1 and parkin, and rescues the abnormal accumulation of pUb, p62 and Atg8a back towards basal levels, preventing neurodegenerative phenotypes in Pink1 and parkin mutant flies. Consistent with these results, the Cisd knockdown-induced increase in mitophagy observed in 30-day-old wild-type animals (Fig. 5A, B) was also parkin-independent (Fig. S4D, E). Interestingly, in this experiment mitolysosomes were largely absent in parkin mutant muscle, however, this result should be interpreted with caution since it is well documented that flight muscle in parkin mutants undergoes apoptosis in young adults [34, 35, 47].

Given that genetic reduction of Cisd levels robustly rescued Pink1/parkin phenotypes, we considered the potential for CISD1/2 small-molecule inhibitors to have similar beneficial effects. Several compounds, including rosiglitazone, pioglitazone and NL1, have been reported to potently inhibit CISD1 and CISD2 [48]. Following our findings that CISD1/2 knockdown was sufficient to induce mitophagy, we tested the effects of two structurally distinct compounds, rosiglitazone and NL1, for their potential to induce mitophagy. Exposing mitophagy reporter cells to rosiglitazone and NL1 for 24 h, we found that both compounds intensely induced mitophagy as well as mitochondrial fragmentation (Fig. 8A, B), mirroring the effects of genetically reducing CISD1/2.

Fig. 8

CISD inhibitors induce mitophagy and rescue rescues Pink1 and parkin mutant phenotypes. A Confocal microscopy analysis of WT ARPE-19 cells expressing mito-QC to visualise mitolysosomes (shown separately) treated with 100 µM rosiglitazone (Rosi), NL1 or vehicle. B Quantification of the number of mitolysosomes per cell of conditions shown in A. Data points indicate replicate experiments. Statistical analysis: RM one-way ANOVA with Geisser-Greenhouse correction; *P < 0.05; ****P < 0.0001. C Analysis of Climbing, (D) thoracic indentations and drooped-wing phenotype, and (E) mitochondrial morphology in flight muscle of Pink1 and parkin mutants alongside WT control flies, treated with 1 mM rosiglitazone (Rosi) or vehicle. Statistical analysis: Chi-squared test. ***P < 0.001; ****P < 0.0001. Scale bars = 10 μm. F Immunoblot analysis of protein lysates from whole flies upon treatment with 1 mM rosiglitazone (Rosi) or vehicle. Samples were homogenised under reducing conditions and blots were probed for Cisd (CISD2, Proteintech, 13318-1-AP) and Tubulin

These results motivated us to test the therapeutic potential of CISD inhibitors in vivo. The preceding results showed that rosiglitazone had an overall greater and more consistent effect on mitophagy induction than NL1, hence, we focussed on rosiglitazone for in vivo analysis. To this end, we treated Pink1 and parkin mutants with rosiglitazone-dosed food and examined their phenotypes. Pink1/parkin mutants treated with rosiglitazone showed a modest but significant improvement in their climbing ability (Fig. 8C). There was also a general modest improvement of thoracic indentations and abnormal wing posture (Fig. 8D), caused by the degeneration of underlying flight muscle. Furthermore, rosiglitazone treatment visibly improved mitochondrial morphology of Pink1/parkin flight muscle (Fig. 8E). Although dimer levels were modestly reduced, we did not observe major effects on Cisd stability upon rosiglitazone treatment in whole fly lysates (Fig. 8F). Together, these results support the potential benefit of CISD inhibitors in promoting mitophagy and alleviating degenerative phenotypes in these PD models.