Mitolysosome exocytosis, a mitophagy-independent mitochondrial quality control in flunarizine-induced parkinsonism-like symptoms.

Mitochondrial quality control plays an important role in maintaining mitochondrial homeostasis and function. Disruption of mitochondrial quality control degrades brain function. We found that flunarizine (FNZ), a drug whose chronic use causes parkinsonism, led to a parkinsonism-like motor dysfunction in mice. FNZ induced mitochondrial dysfunction and decreased mitochondrial mass specifically in the brain. FNZ decreased mitochondrial content in both neurons and astrocytes, without affecting the number of nigral dopaminergic neurons. In human neural progenitor cells, FNZ also induced mitochondrial depletion. Mechanistically, independent of ATG5- or RAB9-mediated mitophagy, mitochondria were engulfed by lysosomes, followed by a vesicle-associated membrane protein 2- and syntaxin-4-dependent extracellular secretion. A genome-wide CRISPR knockout screen identified genes required for FNZ-induced mitochondrial elimination. These results reveal not only a previously unidentified lysosome-associated exocytosis process of mitochondrial quality control that may participate in the FNZ-induced parkinsonism but also a drug-based method for generating mitochondria-depleted mammal cells.

INTRODUCTION

Mitochondrial quality control is essential for mitochondrial homeostasis and function (, ). Mitophagy mainly contributes to quality control by removal of damaged mitochondria (–). One of the most well-defined mechanisms underlying mitophagy is the PTEN-induced putative kinase protein 1 (PINK1)–PARK2–dependent pathway (), in which stabilized PINK1, through phosphorylation of ubiquitin or mitofusin 2, recruits PARK2, an E3 ligase, to damaged and depolarized mitochondria (, ). Damaged mitochondria can also be eliminated through other autophagy pathways, including BCL2 interacting protein 3-like (BNIP3L)– or FUN14 domain containing 1 (FUNDC1)–dependent mitophagy (, ), and Unc-51–like kinase 1 (ULK1)– and RAB9-dependent, autophagy-related 5 (ATG5)–independent noncanonical autophagy (, ). Alternatively, mitochondria can also be eliminated by PINK1-PARK2–dependent mitochondrial-derived vesicles (MDVs) (, ). Mitochondria may be sequestered in vacuoles and subsequently extruded from dying cells (, ). Mitochondria may also be secreted through a migrasome-mediated mitocytosis mechanism ().

Parkinsonism is a neurodegenerative syndrome that is pathologically characterized by the degeneration of multiple anatomical structures within the brain (). Evidence has pointed to a mitochondrial etiology of parkinsonism (). Mitochondrial quality control, in particular, is essential for normal cellular function and has been causally related to parkinsonism and other metabolic diseases (, ). Chronic use of flunarizine (FNZ), a piperazine derivative with calcium-antagonizing and anticonvulsant properties, often leads to parkinsonism (). FNZ is widely prescribed in continental Europe and China, but not in the United Kingdom or in the United States, for the treatment of vertigo of central and peripheral origins, migraine, epilepsy, and occlusive peripheral vascular diseases (, ). However, FNZ can produce parkinsonism as one of its side effects (), partially owing to its inhibition of dopamine receptors (). Furthermore, FNZ was reported to inhibit complexes I and II of the mitochondrial electron transport chain in tissue-derived submitochondrial particles (). The exact mechanisms of FNZ-induced parkinsonism remain unknown.

In this work, we aimed to clarify the mechanisms underlying FNZ-induced parkinsonism. FNZ treatment was found to cause mitochondria elimination both in vivo and in vitro. Further studies revealed that, during FNZ treatment, mitochondria entered lysosomes, followed by VAMP2- and STX4-dependent exocytosis. This mechanism of mitochondrial elimination was independent of ATG5- and RAB9-mediated mitophagy. We also report here an FNZ-based method for generating mitochondria-depleted mammal cells, without the need for genetic manipulation.

RESULTS

FNZ induced motor dysfunction and brain mitochondrial mass decrease in vivo

Given the link between FNZ and parkinsonism (), we asked whether FNZ could cause parkinsonism-like symptoms in mice. To answer this, we administrated FNZ to ICR mice by intraperitoneal injections at a daily dose of 30 mg/kg for 2 weeks (Fig. 1A). The body weight of the FNZ-treated mice decreased in the first week but rebounded in the second week to be indistinguishable from control animals (fig. S1A). Rotarod testing of motor functions revealed substantial performance deterioration in FNZ-treated mice, indicative of an impaired coordination and balance (Fig. 1B). FNZ-treated mice also performed poorly on the open-field test and Morris water maze test (Fig. 1, C to E, and fig. S1, B to H). To determine the possible effect of FNZ on dopaminergic neurons, we assessed the numbers of nigral dopaminergic neurons by stereological analysis and measured striatal dopamine concentrations by ultraperformance liquid chromatography (UPLC). The results showed that FNZ did not induce the death of dopaminergic neurons (fig. S2) but decreased the striatal dopamine concentrations (Fig. 1F). These results show that FNZ treatment produces motor and memory deficits in mice, similar to FNZ-induced parkinsonism.

Fig. 1.

FNZ induced motor dysfunction and reduced mitochondrial content in vivo.

(A) Experimental design for FNZ and PEG400 treatment in ICR mice. (B to E) Behavior assessment of male mice following FNZ or control PEG400 treatment for 2 weeks. FNZ-treated mice showed defects on the rotarod test (B) (n ≥ 7 mice) and open-field test (C to E) (n ≥ 10 mice). (F) The striatal dopamine concentrations were detected by UPLC assay in mice treated with FNZ for 2 weeks (n ≥ 5 mice). (G to I) PET/CT imaging and biodistribution analysis of 18F-FDG in brains of mice treated with FNZ or PEG400. Representative images from mice before (Baseline) and after treatment are shown, with values for maximum percentage injected dose indicated (G). Quantification of k1 value (H). Waterfall plot of the percentage change in k1, each bar representing an individual mouse (I) (n = 12 mice). (J and K) Western blot analysis of endogenous mitochondrial OMM proteins (TOM20 and VDAC), IMM protein (PHB1), and matrix protein (HSP60) in mouse brain homogenates (J). Quantification of protein levels (K) (n ≥ 4 mice). (B) to (E) and (H) show means ± SEM, and (F) and (K) show means ± SD; *P < 0.05, **P < 0.005, ***P < 0.001, in t test.

Parkinsonism has been reported to be associated with mitochondrial dysfunction (), which could be indicated by the rate of cerebral glucose metabolism. We used [18F]-fluoro-2-deoxyglucose (18F-FDG) positron emission tomography/computed tomography (PET/CT) scanning to dynamically assess glucose metabolism in the brains of FNZ-treated mice. 18F-FDG uptake kinetics were monitored by transfer from bloodstream to tissue (k1), transfer from tissue back to the bloodstream (k2), phosphorylation (k3), and the plasma clearance of 18F-FDG to tissue (ki). We observed that k1 was significantly elevated in the brains of FNZ-treated mice compared with the PEG400 (polyethylene glycol 400) control group (Fig. 1, G to I), while no significant difference was observed for k2, k3, and ki in FNZ-treated mice (fig. S3). This indicates an increase in glucose uptake in the brains of FNZ-treated mice. Next, we monitored mitochondrial mass in multiple tissues, including brain, muscle, liver, kidney, heart, and spleen. FNZ treatment led to significant reductions in outer mitochondrial membrane (OMM) proteins (TOM20 and VDAC), inner mitochondrial membrane (IMM) protein (PHB1), and matrix protein (HSP60) in brain, but not in other tissues (Fig. 1, J and K, and fig. S4). Hence, FNZ specifically decreases mitochondrial mass in brain.

Parkinsonism is thought to result from nigrostriatal dopaminergic dysfunction. We then measured mitochondrial content in the nigral dopaminergic neurons and found that mitochondrial content decreased in cell bodies and decreased more in neurites (Fig. 2, A to F). By detecting mitochondrial area in nondopaminergic neurons and adjacent glial fibrillary acidic protein (GFAP)–positive astrocytes in the substantia nigra (SN), we observed a pattern of mitochondrial content similar to that in dopaminergic neurons (Fig. 2, G to I, and fig. S5, A to C). These results indicate that FNZ could broadly deplete mitochondria in brain rather than specifically eliminating mitochondria in dopaminergic neurons.

Fig. 2.

Mitochondrial content decreased in dopaminergic neurons and astrocytes of mice treated with FNZ.

(A to F) Immunostaining was used to detect mitochondria (TOM20) in the dopaminergic neurons (TH) of mice treated with FNZ for 2 weeks. Representative images of mitochondria in neurite (A) and cell body (D) are shown. Quantification of relative mitochondrial area in neurites and cell bodies is shown in (B) and (C) (80 cell bodies from four mice) and (E) and (F) (≥60 neurites from four mice), respectively. (G to I) Immunostaining was used to detect mitochondria (TOM20) in the SN GFAP-positive astrocytes. Representative confocal microscopy images are shown in (G), and quantification of relative mitochondrial area in astrocytes is shown in (H) and (I) (60 cells from four mice). Scale bars, 10 μm. All graphs show means ± SD; **P < 0.005 and ***P < 0.001, in t test.

FNZ triggered mitochondrial elimination in vitro

To further investigate the effect of FNZ on mitochondria in brain, we treated human induced pluripotent stem cell (iPSC)–derived neurons and human primary astrocytes with FNZ for 3 days, found a substantial depletion of mitochondria in both neurons and glia, and observed almost complete loss in neurites as shown by TOM20 staining (fig. S5, D to H). We also labeled mitochondria of human iPSC-derived neural progenitor cells (NPCs) with green fluorescent protein (GFP) targeted to the mitochondrial matrix (mito-GFP). FNZ treatment caused aggregation of mitochondria after 1 day. Unexpectedly, mitochondrial mass decreased sharply, and nearly all mitochondria were depleted at day 3 of treatment (Fig. 3A). Similar results were obtained using MitoTracker Deep Red FM staining instead of mito-GFP (Fig. 3B). Mitochondrial DNA (mtDNA) copy number was also greatly decreased after FNZ treatment (Fig. 3C). Using immunofluorescence and immunoblot analyses, we observed that mitochondrial proteins of the OMM, IMM, and matrix disappeared after FNZ treatment (Fig. 3, D to F). Cellular metabolism analysis using Seahorse Flux Analyzer found greatly reduced basal and maximal mitochondrial respiration rates in FNZ-treated cells (Fig. 3, G to I). At the same time, glycolysis was increased (fig. S6), indicating a compensation in adenosine triphosphate (ATP) production during mitochondrial elimination. Last, we confirmed that FNZ also caused mitochondrial depletion in other cell types including human fibroblasts, mouse embryonic fibroblasts (MEFs), and Hela cells (fig. S7).

Fig. 3.

FNZ induced mitochondrial elimination in vitro.

(A) Representative microscopy images of mito-GFP–expressing NPCs treated with FNZ for 3 days. (B) Representative images showing MitoTracker Deep Red staining of NPCs with or without FNZ treatment for 3 days. (C) Quantification of relative mtDNA copy number (MT-ND1, MT-ND4, and MT-CO1) in NPCs with or without FNZ treatment for 3 days (n = 5 independent experiments). (D) Immunostaining of mitochondrial OMM protein (TOM20) and IMM protein (MT-CO1) in mito-GFP–expressing NPCs treated with FNZ for 3 days. (E and F) Western blot analysis (E) and quantification (F) of mitochondrial OMM proteins (VDAC and TOM20), IMM proteins (PHB1 and ATP5A), mtDNA-encoded IMM proteins (MT-ATP8 and MT-CO1), and matrix proteins (HSP60 and HSP70) in NPCs with or without FNZ treatment for 3 days (n ≥ 2 independent experiments). (G to I) Oligomycin, FCCP, and antimycin A and rotenone were added sequentially in NPCs treated with or without FNZ for 3 days (G). Mitochondrial ATP production capacity (H) and maximal respiration (I) (n = 2 independent experiments). Scale bars, 10 μm. All graphs show means ± SD; *P < 0.05, **P < 0.005, and ***P < 0.001, in t test.

We then investigated whether FNZ could eliminate other cellular organelles. We labeled mitochondria with mito-GFP, endoplasmic reticulum (ER) with ER-DsRed, lysosomes with LysoTracker Deep Red, or Golgi apparatus with DsRed-Golgi and found that, in contrast to mitochondria, the structures of the other membranous organelles (i.e., ER, lysosomes, and Golgi apparatus) remained intact in FNZ-treated cells (fig. S8). This indicated that FNZ specifically depletes mitochondria.

Besides FNZ, parkinsonism can be triggered by many other drugs such as typical antipsychotics [chlorpromazine (CPZ) and haloperidol (HP)], gastrointestinal prokinetic drugs [domperidone (DOM)], neurotoxins [1-methyl-4-phenyl-1,2,3,6-tetrahydropyridine (MPTP)], and calcium channel blockers [cinnarizine (CNZ)] (, ). We then asked whether these parkinsonism-induced drugs could also induce mitochondrial elimination. Only FNZ and CNZ, a piperazine derivative similar to FNZ, induced mitochondrial elimination (fig. S9). Thus, FNZ may trigger parkinsonism via a unique mechanism, the elimination of mitochondria.

Mitochondrial elimination by FNZ is distinct from canonical and noncanonical mitophagy

Given that LC3-mediated mitophagy plays a predominant role in mitochondrial degradation (), we interrogated whether FNZ-induced mitochondrial elimination depends on this pathway. We labeled NPCs with mito-DsRed and LC3B-GFP and used bafilomycin A1 (BAF) to block lysosomal fusion and acidification. The mammalian target of rapamycin (mTOR) inhibitor rapamycin was used as a positive control to promote LC3-dependent mitophagy (). The colocalization ratio of mito-DsRed and LC3B significantly increased under BAF treatment in the rapamycin group. Unexpectedly, no significant differences were observed in the FNZ group (Fig. 4, A and B). This finding led us to further examine the dependence of mitochondrial elimination on the canonical autophagy factor ATG5 and mitophagy receptors PINK1 and FUNDC1. We observed that knockdown of ATG5, PINK1, or FUNDC1 in NPCs failed to abrogate FNZ-induced mitochondrial elimination (Fig. 4, C and D, and fig. S10, A to C). Neither did knockout (KO) of ATG5, PARK2, or FUNDC1 in MEFs inhibit FNZ-induced mitochondrial elimination (Fig. 4, E and F). Our data demonstrate that ATG5-related canonical mitophagy is not involved in the FNZ-induced mitochondrial elimination. The observed mitochondrial elimination in PARK2 KO cells (Fig. 4, E and F) likely also excludes a major role for the MDV pathway, which is dependent on PARK2 (). Mitochondrial elimination could also occur via ULK1-RAB9–dependent, noncanonical mitophagy (, ). Knockdown of RAB9 and ULK1 by short hairpin RNA (shRNA) also failed to block the mitochondrial elimination (Fig. 4, G and H, and fig. S10, D and E), ruling out a role for the ULK1-RAB9–dependent pathway. Last, we determined the effect of 3-methyladenine (3MA), an autophagy inhibitor, on FNZ-induced mitochondrial elimination. As expected, 3MA had no effect (Fig. 4, I and J). Together, these results exclude canonical and noncanonical mitophagy as mechanisms of mitochondrial elimination by FNZ.

Fig. 4.

FNZ-induced mitochondrial elimination was independent of canonical or noncanonical mitophagy.

(A and B) Representative images (A) and quantification of the ratio of LC3-positive mitochondria (B) in NPCs expressing mito-DsRed and LC3B-GFP treated with FNZ or rapamycin for 12 hours in the presence or not with BAF (n ≥ 29 cells from three independent experiments). (C and D) Representative images (C) and quantification of cells without mitochondria (D) in mito-GFP–expressing NPCs transduced with ATG5, PINK1, FUNDC1, or control TRC and treated with FNZ for 3 days (n ≥ 3 independent experiments). (E and F) Representative images (E) and quantification of cells without mitochondria (F) in wild-type, ATG5−/−, PARK2−/−, and FUNDC1−/− MEFs transfected with mito-GFP and treated with FNZ for 3 days (n = 3 independent experiments). (G and H) Representative images (G) and quantification of cells without mitochondria (H) in mito-GFP–expressing NPCs transduced with TRC, RAB9, or ULK1 shRNA and treated with FNZ for 3 days (n ≥ 3 independent experiments). (I and J) Representative images (I) and quantification of cells without mitochondria (J) in mito-GFP–expressing NPCs treated with FNZ and 3MA for 3 days (n = 3 independent experiments). Scale bars, 10 μm. All graphs show means ± SD; *P < 0.05 and ***P < 0.001, in t test.

FNZ-induced mitochondrial elimination through a lysosome-associated exocytosis mechanism

To test the role of lysosomes in FNZ-induced mitochondrial elimination, we treated cells with chloroquine (CLQ), an inhibitor of lysosomal proteases, which blocked the elimination of mitochondria during FNZ treatment (Fig. 5A). Confocal profile analysis and stimulation emission depletion (STED) three-dimensional (3D) reconstruction of cells expressing mito-DsRed and LAMP1-GFP (lysosomal marker) revealed that most mitochondria were located in lysosomes after FNZ treatment (Fig. 5, B and C, and fig. S11A). We also used mito-DsRed and OMP25-mCherry to label mitochondrial matrix and OMM, respectively, and found that mitochondria were engulfed by lysosomes intact, as evidenced by the colocalization of OMM and matrix in lysosomes (fig. S11, B and C). To further investigate the engulfment of mitochondria by lysosomes, we used mito-mKeima, which emits different-colored signals at acidic and neutral pH (). The ratio of 543/458 nm increased significantly after FNZ treatment (fig. S11, D and E), indicating mitochondria presence within acidic lysosomes. Furthermore, transmission electron microscopy (TEM) showed the presence of mono-membrane engulfed mitochondria during FNZ treatment (Fig. 5D). These findings demonstrate that lysosomes are involved in mitochondrial elimination during FNZ treatment. We named this mitochondrial structures engulfed by lysosomes as mitolysosome. CLQ did not affect the ratio of mitolysosome to total mitochondria (fig. S11, F and G), suggesting that it interferes at a downstream step of elimination.

Fig. 5.

Mitochondrial elimination through a lysosome-dependent exocytosis mechanism.

(A) Representative images and quantification of cells without mitochondria in NPCs treated with FNZ and CLQ for 3 days. (B) Representative images and profile analysis (white lines) of mito-DsRed– and LAMP1-GFP–expressing NPCs treated with FNZ for 24 hours. (C) Imaris 3D reconstruction of mitochondria and lysosomes [scale bars, 3 and 1 μm (right)], and quantification of the ratio of lysosomal mitochondria (n = 10 cells from two independent experiments). (D) Representative TEM images of NPCs during FNZ treatment. Arrows indicate the engulfment of mitochondria by lysosomes. (E) Representative images of mito-DsRed–expressing NPCs treated with FNZ for 24 hours and stained with FM1-43FX and LysoTracker. (F) Immunoblot analysis and quantification of MT-ATP8, TOM20, H3, and actin in cell lysates and culture supernatants (n ≥ 2 independent experiments). (G and H) Representative images (G) and quantification of cells without mitochondria (H) in NPCs transduced with indicated shRNA and treated with FNZ for 3 days (n = 4 independent experiments). (I and J) Immunoblot analysis (I) and quantification (J) of TOM20 and actin in cell lysates and culture supernatants of NPCs transduced with indicated shRNA and treated with FNZ for 2 days (n = 2 independent experiments). Scale bars, 10 μm in (A), (B), (E), and (G) and 500 nm in (D). All graphs show means ± SD; *P < 0.05, **P < 0.005, and ***P < 0.001, in t test (A and C) and one-way analysis of variance (ANOVA) test (F, H, and J).

We then asked whether FNZ could trigger mitolysosome exocytosis. To this end, we labeled plasma membrane of NPCs with FM1-43FX, a membrane-impermeable dye that incorporates into the plasma membrane lipids (). Normal NPCs displayed a polygonal shape, whereas FNZ-treated cells were shrunken and presented multiple globular structures (fig. S12A). We observed that the FM1-43FX–labeled plasma membrane encircled areas containing mito-DsRed and LysoTracker Deep Red, a lysosome marker (Fig. 5E), suggestive of mitolysosome exocytosis via membrane blebbing. To further confirm mitolysosome exocytosis, we performed polymerase chain reaction (PCR) to investigate whether the culture supernatant of FNZ-treated NPCs contained mtDNA. We were able to amplify mitochondrial genes (but not nuclear genes) from the culture supernatant (fig. S12B). Furthermore, we observed an increase in the amount of the mitochondrial proteins MT-ATP8 and TOM20 as well as actin in culture supernatants from FNZ-treated cells. In contrast, extracellular levels of H3 (a nuclear marker as a control) were not increased (Fig. 5F). Consistent with these results, actin was observed in the membrane blebs that encircled mitolysosome in FNZ-treated NPCs by phalloidin staining (fig. S12C). Fluorescence-activated cell sorting (FACS) also showed an increased amount of extracellular mitochondria after FNZ treatment (fig. S12, D and E). Hence, FNZ treatment triggered lysosome-associated exocytosis of mitochondria.

mtDNA can be released by neutrophil extracellular traps (NETs) upon OMM protein TOM20 mobilization to the cell surface (). However, we failed to detect TOM20 on the cell surface of nonpermeabilized cells (fig. S13), ruling out NETs as a mechanism. We reasoned that lysosome-associated secretion could participate in mitolysosome exocytosis. Lysosomal secretion is controlled by SNARE-dependent membrane fusion (), a process inhibited by high calcium concentrations (). We thus increased intracellular calcium concentration by addition of extracellular CaCl2 and ionomycin, a calcium ionophore, ensuring rapid equilibration of transmembrane calcium ions. We showed that calcium concentrations at higher than 1.6 μM effectively blocked mitochondrial elimination by FNZ treatment (fig. S14), implicating a calcium-dependent process such as SNARE-mediated fusion, as supported by transcriptome analysis (fig. S15 and data S1 and S2). Vesicle v-SNARE and plasma membrane t-SNARE, VAMP2, and STX4, respectively, are key to lysosome cell membrane fusion and secretion (, ). Immunofluorescence microscopy analysis showed that VAMP2 puncta were located on lysosomes and also on the cell surface enclosing mitochondria after FNZ treatment (fig. S16, A and B), indicating a role for VAMP2 in mitolysosome exocytosis. Furthermore, knockdown of VAMP2 or STX4 by shRNA blocked FNZ-induced mitochondrial elimination (Fig. 5, G and H, and fig. S16, C and D), as well as mitolysosome exocytosis (Fig. 5, I and J), but not lysosomal engulfment of mitochondria (fig. S16, E and F). Together, FNZ-induced mitochondrial depletion occurs through a lysosome-associated, VAMP2-STX4–dependent process.

GeCKO screen to identify genes necessary for mitochondrial elimination

To identify genes necessary for mitochondrial elimination, we generated a GeCKO (genome-wide CRISPR KO) library of NPCs, as previously described (), and performed a genome-scale, loss-of-function, genetic screen (Fig. 6A). The NPC-GeCKO library cells were subjected to FNZ treatment and FACS-sorted for cells in which mitochondria were preserved (mito+) (fig. S17A). Comparative analysis of single-guide RNA (sgRNA) representation between the NPC-GeCKO library and the sorted mito+ group revealed robust enrichment of specific sgRNAs (fig. S17B). These genes were further filtered according to their enrichment (>2-fold) and P < 0.05. Of the 20,834 genes tested, the KO screen identified an enrichment of 671 genes in the mito+ group, which were further analyzed for protein location and using gene ontology (GO) (fig. S17C and data S3). Fourteen highly enriched candidates, representing various biological processes including mitochondria/lysosome membrane function, macromolecule catabolic process, ATP metabolic process, and secretion by cell, were selected for the generation of individual CRISPR KO cell lines for further validation (Fig. 6B, fig. S17D, and table S1). These KOs were treated with FNZ for 3 days. A significant reduction in mitochondrial elimination was observed for all these KOs as compared to NT1 control cells, demonstrating that the selected genes are essential for FNZ-induced mitochondrial elimination (Fig. 6C).

Fig. 6.

GeCKO screen to identify genes necessary for FNZ-induced mitochondrial elimination.

(A) Schematic of GeCKO screen. (B) Identification of top candidate genes using the RNAi Gene Enrichment Ranking (RIGER) P value analysis. The functional categories of the candidate genes, manually curated, are shown in the bottom panel. (C) Validation of the candidate genes. Quantification of the fraction of cells without mitochondria in NPCs expressing Cas9 and an sgRNA against candidate genes (bottom, n = 3 independent experiments). (D) Identification of genes related to mitochondrial entry into lysosomes. Quantification of the fraction of mitochondria in lysosomes in NPCs expressing mito-GFP, Cas9, and indicated sgRNA treated with FNZ for 1 day and stained with LysoTracker (bottom, n = 30 cells from three independent experiments). (E and F) Immunoblot analysis of TOM20 and actin in cell lysates and culture supernatants in indicated KO NPCs treated with FNZ for 2 days (E). Quantification of protein levels is shown in (F) (n = 2 independent experiments). (G) Graphic scheme of mitochondrial elimination induced by FNZ. Essential genes identified in the GeCKO screen are shown in the table below. Scale bars, 10 μm. All graphs show means ± SD; *P < 0.05, **P < 0.005, and ***P < 0.001, in one-way ANOVA test.

To better understand the mechanisms by which the identified genes could contribute to mitochondrial elimination, we checked mitochondrial engulfment by lysosomes and mitolysosome exocytosis in the KO cell lines. We observed a significant reduction in mitochondrial entering into lysosomes for BAX, GBAS, UBR7, BACE1, FBXO30, ATP5A1, HKDC1, NDUFS4, and PTPRN KOs as compared to NT1 control cells (Fig. 6D). We further quantified the amount of mitolysosome in the culture supernatant for the 14 KOs treated with FNZ for 2 days. We observed a marked reduction of TOM20 for BAX, GBAS, UBR7, P2RX7, BACE1, FBXO30, NDUFS4, SYT5, PTPRN, LGI3, FEM1C, and HKDC1 KOs (Fig. 6, E and F, and fig. S17, E and F). KO of P2RX7, FEM1C, LGI3, and SYT5 blocked mitolysosome secretion but had no effect on mitochondrial entering into lysosomes, suggesting a specific role for these genes in mitolysosome exocytosis. No substantial defect was observed for FNBP1 KO in terms of both mitochondrial entering into lysosomes and mitolysosome exocytosis, indicating that other unknown pathways involving FNBP1 may also contribute to the FNZ-induced mitochondrial elimination.

DISCUSSION

Drug-induced parkinsonism (DIP) is the second most common cause of parkinsonism following Parkinson’s disease (). Chronic use of FNZ often leads to parkinsonism. Here, we describe parkinsonism-like symptoms induced by FNZ in mice. The major symptoms of the Parkinson’s disease are caused by the loss of dopaminergic neurons in the SN (). Our data (fig. S2) and a previous report () both show that FNZ did not induce the loss of dopaminergic neurons. However, mitochondrial deficits were found in the cell bodies and especially in the neurites of dopaminergic neurons, and dopamine concentrations were also reduced in the striatum, implying that the dysfunction of midbrain dopaminergic neurons could be triggered by FNZ. Further studies on other aspects of dopamine metabolism and dopamine neuron function are needed to examine the dysfunction of dopamine neurons. We also found mitochondrial deficits in nondopaminergic neurons and astrocytes, suggesting that broader defects in brain may occur after FNZ treatment. Considering the nondegeneration of dopamine neurons and the broader mitochondrial deficits in brain, the FNZ-treated mice may not serve as a canonical model of parkinsonism.

We report that FNZ induces mitochondrial elimination through a lysosome-associated exocytosis mechanism (Fig. 6G). FNZ induced a decrease in mitochondrial content in dopaminergic neurons of mice without affecting the number of nigral dopaminergic neurons. Other group recently reported that the disruption of mitochondrial complex I function in mice dopaminergic neurons by conditional Ndufs2-KO could compromise dopamine release, with the survival of dopaminergic neurons enabled by a Warburg-like shift in metabolism in the early stage (). Similar metabolic shifts were also observed in our study. In addition, dopamine D2 receptor blockage by FNZ has been reported (, ). We also found the mitochondrial content in nondopaminergic neurons and astrocytes was decreased upon FNZ treatment. It is reasonable to speculate that the broad mitochondrial depletions might contribute to FNZ-induced parkinsonism.

FNZ at a concentration higher than 15 μM could obviously deplete mitochondria (fig. S9), suggesting that the depletion effects appear only when FNZ reaches a certain concentration in vivo. FNZ, as a lipophilic drug, can rapidly penetrate the blood-brain barrier, can easily accumulate in the brain, and has a relatively long circulatory half-life (), with concentrations 10 times higher than in plasma (). We speculate that other organs are not affected because the concentration of FNZ is not high enough in these organs. We also found the FNZ can be neutralized by the compositions of serum, i.e., bovine serum albumin, and a higher concentration is needed for mitochondrial depletion in serum medium (fig. S18). In addition, different compositions of tissue fluid may also lead to the observed different effects.

Unlike the calcium channel blocker FNZ or CNZ, several other tested parkinsonism-inducing drugs such as antipsychotics, gastrointestinal motility drugs, and antiepileptic drugs () failed to show similar effects on mitochondria. It may thus be argued that the identified mitochondrial exocytosis pathway could be specific to FNZ (and to CNZ). However, as in the present study, we were not able to screen all of the parkinsonism-inducing drugs, an essential role for this mitochondrial quality control pathway in DIP may not be excluded, for other untested drugs. On the other hand, mitochondrial exocytosis seems to be a quite universal process occurring in not only pathological but also physiological settings. For instance, neurons and astrocytes can release mitochondria for its recycling and transferring in mice (). In addition, defective mitochondria were reported to be eliminated from cardiomyocytes through exophers (). Mitochondria were also reported to be extruded from dying cells (, ). Different mechanisms, such as CD38-mediated transfer, LC3-dependent exophers, and vacuole-mediated extrusion, were found to mediate the mitochondrial exocytosis in the aforementioned settings. It is tempting to extrapolate that the lysosome-associated mitolysosome exocytosis mechanism identified in this study could also play a role in mitochondrial quality control under basal conditions.

The mechanisms of how FNZ could trigger the mitolysosome exocytosis remain unknown. Several scenarios may explain the drug-induced mitochondrial clearance. First, FNZ, as an antagonist of T type–specific calcium channels, may trigger mitochondrial exocytosis by changing the calcium influx. However, sequestration of calcium did not recapitulate the mitochondrial exocytosis, in contrast to FNZ (fig. S19, A and B). Therefore, we consider that it is less likely that the possible inhibition of calcium channels by FNZ is responsible for triggering the mitochondrial clearance. Second, FNZ could interfere with mitochondrial respiration through inhibition of complexes I and II of the electron transport chain (). Furthermore, FNZ was also reported to compromise mitochondrial function (e.g., collapsing the membrane potential) in Jurkat cells (). These findings were consistent with our observations that FNZ collapsed mitochondrial membrane potential in NPCs (fig. S19, C and D), suggesting severely damaged mitochondrial function. This may trigger the mitochondrial clearance. In addition, possible perturbation of membrane properties (e.g., influence of the gel to liquid-crystalline transition temperature of phosphatidylserines and increasing membrane fluidity by intercalating into cholesterol-rich domains of the target membrane) by FNZ, a lipophilic drug (, ), may further facilitate mitochondrial entering into lysosomes. BAX, a mitochondrial OMM protein that was identified in the present study to be essential for mitochondrial entry into lysosomes, could also be translocated to the lysosome membranes (). Hence, it seems also likely that BAX could participate in the fusion of mitochondria and lysosomes during FNZ treatment.

In addition to the implications for DIP, this work also presents an FNZ-dependent method for total depletion of mitochondria without exogenous expression of genes. The mitochondria-free cells generated this way could survive for nearly 1 month, although they exhibited defective proliferation and differentiation (fig. S20). Mitochondria-free cells may be useful to elucidate previously unidentified functions of mitochondria or could be exploited to replace mutated mtDNA. mtDNA mutations are known to cause debilitating syndromes (), the severity of which is dependent on specific gene mutations and the ratio of mutant to wild-type mtDNA (i.e., heteroplasmy) (). Mitochondrial replacement therapy using pronuclear transfer was previously shown to offer exciting prospects in early human oocytes and embryos (). Alternatively, spontaneously intercellular exchange of mitochondria in certain cells was also reported to partly rescue the function of mtDNA mutated cells such as astrocytes (). However, the abovementioned technologies cannot completely replace mutant mtDNA, in contrast to the FNZ-based method, which allows the depletion of almost all mitochondria and mtDNA. The proposed method is valuable for further exploitation to facilitate the depletion of mutated mtDNA of patients, followed by replacement with normal mitochondria.

MATERIALS AND METHODS

Plasmids

The plasmid pDsRed2-mito (632421, Clontech) was used to mark mitochondria, with a mitochondrial targeting sequence from COX8A (cytochrome c oxidase subunit 8A) fused to the fluorescence protein DsRed2. mito-DsRed was subcloned into a lentiviral expression vector—pLenti. pLenti-mito-GFP was constructed by replacing DsRed2 with GFP. LAMP1 fused with GFP was cloned into pMXs vector as a lysosome tracker. DsRed-Golgi was subcloned from pDsRed-Monomer-Golgi (632480, Clontech) into pMXs vector as a Golgi apparatus marker. LC3-GFP, mt-mKeima, and YFP-PRKN were described in our previous report (). shRNA for targeted genes was constructed into the pLKO.1 vector (8453, Addgene). Target sequences of shRNA for a variety of genes are shown in table S2.

Reagents

FNZ (S2030, Selleck) was dissolved in dimethyl sulfoxide (DMSO) at a concentration of 10 mM as a stock solution for the culture cell. The solutions were further diluted with culture cell medium to obtain 15 μM. For in vivo studies, FNZ was dissolved in 50% PEG400 and 50% saline at a concentration of 6 mg/ml and daily injected at a dose of 30 mg/kg. CHIR99021 (S1263, Selleck), dorsomorphin (S7840, Selleck), SB431542 (S1067, Selleck), ionomycin (I24222, Thermo Fisher Scientific), CNZ (S4727, Selleck), CPZ (S5749, Selleck), HP (S1920, Selleck), DOM (S2461, Selleck), MPTP (S4732, Selleck), BAF (C2501, Gene Operation), and rapamycin (HY-10,219, MedChemExpress) were dissolved in DMSO. CLQ (C6628, Sigma-Aldrich) was dissolved in distilled water. 3MA (orb181186, Biorbyt) was dissolved in ethanol. The solutions were further diluted with culture cell medium.

Mouse studies

Animals and FNZ administration

All housing, breeding, and other animal procedures were approved by the Guangzhou Institutes of Biomedicine and Health (GIBH) Ethical Committee and in accordance with GIBH Institutional Animal Care and Use Committee (IACUC) guidelines. Young adult male ICR mice (Charles River Laboratories) (8 to 10 weeks of age) were maintained under standard laboratory conditions (12-hour light/dark cycle, room temperature 21° ± 1°C) with free access to water and food and were adapted to the laboratory conditions for at least 1 week. FNZ was administered intraperitoneally at a daily dose of 30 mg/kg for 1 or 2 weeks. Control groups received 50% PEG400 treatment at the same volume. Animals were evaluated daily for body weight, and motor function in rotarod, open field, and Morris water maze was recorded weekly before the injection.

Rotarod

A rotarod test was performed on the rotarod apparatus (UgoBasile) to assess motor learning, coordination, and balance. Each mouse was given a training session (four 5-min trials, 5 min apart) to acclimate them to the rotarod apparatus. During the test period (1 hour later), each mouse was placed on the rotarod with an increasing speed, from 4 to 60 rpm in 6 min. The latency to fall off the rotarod within this time period was recorded. Each mouse received two consecutive trials, and the mean latency to fall was calculated.

Open field

An open-field test was performed to assess exploratory activity and anxiety. A square open field (100 cm × 100 cm × 40 cm) was divided into a center area (60 cm × 60 cm) and four side areas (20 cm × 60 cm). Mice were individually placed in the center and allowed to freely explore for 10 min while the trial was videotaped. Subsequent video scoring was completed by an EthoVision-computerized image analyzing system (Noldus Information Technology) on total distance traveled and percentage of time spent in the center and frequency of cross-central area.

Morris water maze

The apparatus was a circular pool (140 cm diameter) filled with water. Tests were performed at 22°C. The pool was painted with nontoxic black caramel. A 12-cm-diameter transparent platform was placed 1 cm below the water surface at a fixed position. Mice were taken to the behavior room and trained on five consecutive days (four trials per day). The starting point changed after each trial of a daily training session. Each trial lasted 60 s or until the mice found the platform. If the platform was not located during the time period, the mice were directed to the platform. After each trial, the mice were placed on the platform for 10 s. On day 6, upon completion of the training phase, the platform was removed for the probe trial. The duration of probe trial was 60 s. All parameters were recorded semiautomatically by a video tracking system (EthoVision).

18F-FDG small-animal PET imaging and image analysis

All PET images were acquired with a high-resolution small-animal PET/CT (Siemens Inveon). Before 18F-FDG scanning, mice were fasted for 8 hours, followed by inducing anesthesia with 5% isoflurane, and maintained with 1.5% isoflurane. Dynamic PET scans were acquired at 0 to 60 min after intravenous injection of 5.55 MBq 18F-FDG (150 μCi). At the end of the PET scan, a CT scan was conducted to register with 18F-FDG images. PET images in a size of 128 × 128 matrix (pixel size: 0.7764 mm × 0.7764 mm × 0.7963 mm; 16 frames: 6 × 10 s, 4 × 60 s, 1 × 300 s, and 5 × 600 s) were reconstructed by using 3D ordered-subset expectation maximization (OSEM3D) algorithms.

PET images and CT images were co-registered by PMOD software (version 3.902, PMOD Technologies Ltd.) (). PMOD kinetic modeling tool (PKIN) was used to perform kinetic analysis. The inferior vena cava was chosen to extract 18F-FDG image–derived input functions due to its less partial volume effect and visible on the PET images (). The inferior vena cava was clearly identified on the first two frames of the dynamic imaging in the coronal plane. The volume of interest (VOI) of inferior vena cava was constructed by drawing region of interest (ROI) in over four successive axial slices. A second VOI was drawn in the brain. The time activity curve (TAC) of the inferior vena cava and brain was calculated according to their VOI. A two-tissue, three-compartment model was chosen to fit the TAC of mouse brain (). Briefly, the Patlak plot defines a constant (K) that incorporates the forward (k1) and reverse (k2) rate constants from plasma to tissue and the phosphorylation (k3) constant. The formula is expressed as Ki = (k1 × k2)/(k2 + k3). The dephosphorylation constant (k4) is assumed to be 0.

Determination of striatal dopamine concentrations

Striatal dopamine concentrations were measured using the Waters ACQUITY UPLC H-Class chromatographic system coupled with 2465 Electrochemical Detector. Striata samples were weighted and homogenized in ice-cold 0.1 M HClO4 (0.1 g of original wet tissue/ml) containing 0.1% l-cysteine and dihydroxybenzylamine (DHBA) (0.3 ng/μl) as internal standard. Homogenates were centrifuged at 15,000g for 20 min at 4°C. Supernatants were collected, and 4 μl was injected onto an ACQUITY UPLC BEH C18 1.7 μm 2.1 × 50 mm column. The mobile phase consisted of 3 mM sodium 1-heptanesulfonate, 100 mM sodium acetate, 85 mM citric acid, and 0.2 mM EDTA (final pH 3.7). The column and detection temperature were set at 35°C, and the detection potential was +0.80 V (versus Ag/AgCl reference). The flow rate was set at 0.40 ml/min, and typical retention times were as follows: dopamine, 2.694 min; DHBA, 1.591 min. Dopamine contents were determined by comparing sample peak areas with those recorded from standards run on the same day.

Stereological cell counting and densitometric analysis

Mice were perfused and fixed. The brains were postfixed using 4% paraformaldehyde (PFA) (wj0012, Genesion), cryoprotected in 30% sucrose, and sectioned into 40-μm coronal sections. Coronal sections were sliced throughout the brain, including the SN and striatum. The slices were treated with a 1:10,000 dilution of rabbit polyclonal anti-tyrosine hydroxylase (TH) (AB152, Millipore) followed by biotinylated secondary antibody and VECTASTAIN Elite ABC Reagent (PK-6101, Vector Laboratories). Positive immunostaining was visualized using 3,3′-diaminobenzidine (DAB) followed by a reaction with hydrogen peroxide (DAB kit, SK-4100, Vector Laboratories). Stained sections were mounted onto slides. Every fourth section was analyzed for stereological cell counting. The total numbers of TH-stained neurons within the SN region were counted with the aid of the Optical Fractionator tool in Stereo Investigator software (MicroBrightfield). Striatal TH-positive fiber density was measured by densitometry at three coronal levels (+1.0, 0.0, and −1.0) relative to bregma. Data are represented as the mean of the three levels.

Immunofluorescence

Every selected section of encompassing the midbrain per mouse was preincubated with 5% donkey serum for 1 hour at room temperature before incubation with the primary antibodies overnight at 4°C. After three washes, the sections were then incubated with the secondary antibodies at room temperature for 1 hour. Last, the sections were visualized under a confocal microscope.

For cell slides, cells were fixed with 4% PFA for 30 min at room temperature, washed in phosphate-buffered saline (PBS), and then incubated with primary antibodies in PBS containing 10% goat serum and 0.3% Triton X-100 (T8200, Solarbio). Then, cells were washed three times in PBS, incubated with secondary antibodies for 1 hour at room temperature, and washed three times. Cells were then stained with 4′,6-diamidino-2-phenylindole (D9542, Sigma-Aldrich) and mounted on glass slides with fluorescent mounting medium (36307ES08, Yeasen). Fluorescence images were acquired on a Zeiss 710 NLO laser scanning confocal microscope system. The detailed information of primary antibodies used in this study was shown in table S3.

Protein sample harvesting

At days 7 and 14 after intraperitoneal injection, the mice were deeply anesthetized. Brain, muscle, liver, kidney, heart, and spleen were rapidly excised and placed in cold PBS to remove excess blood. Upon isolation, each tissue was collected in a single 1.5-ml microcentrifuge tube and snap-frozen in liquid N2. Tissue samples were stored at −80°C until further processing. To make protein extracts, all tissues were weighed and defrosted on wet ice in fivefold mass excess of freshly prepared, ice-cold radioimmunoprecipitation assay (RIPA) lysis buffer (P0013K, Beyotime). Phenylmethylsulfonyl fluoride (PMSF) (ST506, Beyotime) and protease inhibitor cocktails (B14002, Bimake) were added immediately. Tissue lysates were obtained using a tissue homogenizer [PBI shredder SG3 (Oroboros)]. An ultrasonic crusher was used to ensure complete crushing of proteins before clarification by centrifugation at 14,000 rpm for 10 min at 4°C. Supernatants were carefully harvested for downstream biochemical analyses.

For obtaining cell samples, cells were lysed in RIPA buffer containing both PMSF and protease inhibitor cocktails. An ultrasonic crusher was used to ensure complete crushing of proteins. In some experiments, supernatant of FNZ-treated cultures, harvested 24 hours after fresh medium replacement, was spun at 500g for 5 min and passed through 0.45-μm filters to remove dead cells. The resultant supernatant fractions were resolved directly in loading buffer.

Cell culture and neural differentiation

All cultures were maintained in a humidified incubator containing 5% CO2 at 37°C. Normal human 90DiPSCs as reported previously () were cultured in mTesR1 medium (85850, STEMCELL Technologies). Colonies, following 0.5 mM EDTA (E116428, Aladdin) treatment, were cultured on Matrigel-coated (356234, Becton Dickinson) plates. Neurons were differentiated from 90D iPSCs. Briefly, 0.5 mM EDTA–dissociated iPSCs were split 1:4 on Matrigel-coated plates. On the following day, the iPSC medium was replaced with the N2B27 medium, i.e.,1:1 mixture of Dulbecco’s modified Eagle’s medium/F12 (SH30023-018, Hyclone) and neurobasal medium (21103-049, Gibco) supplemented with 0.5× N2 (17502048, Gibco), 0.5× B-27 (17504044, Gibco), 1× GlutaMAX (35050-061, Gibco),1× nonessential amino acid (11140-050, Gibco), and 1× penicillin/streptomycin (SV30010, Gibco). CHIR99021 (3 μM), dorsomorphin (2 μM), and SB431542 (2 μM) were added in the medium. The culture medium was changed every other day. 90D iPSCs were maintained under the above condition for 7 days, followed by dissociation with dispase (1 mg/ml) (17105041, Gibco) and passaged at 1:6 in the medium described above. Then, 1 μM CHIR99021, 2 μM dorsomorphin, and 2 μM SB431542 were added for another 7 days for differentiation into NPCs. The NPCs were expanded with N2B27 medium containing epidermal growth factor (EGF) (20 ng/ml) (10605-HNAE-1, Sino Biological) and basic fibroblast growth factor (bFGF) (20 ng/ml) (10014-HNAE-1, Sino Biological) and passaged with accutase (A6964, Sigma-Aldrich). For neuron differentiation, NPCs were plated onto Matrigel-coated plates and cultured in N2B27 medium supplemented with brain-derived neurotrophic factor (BDNF) (10 ng/ml) (4004, BioVision) and 1 μM adenosine 3′,5′-monophosphate (cAMP) (A6885, Sigma-Aldrich).

For primary astrocyte (1800, ScienCell), astrocytes were cultured on poly-l-lysine–coated (P4707, Sigma-Aldrich) plates in Astrocyte Medium (1801, ScienCell). Then, cells were passaged to Matrigel-coated plates in N2B27 medium for FNZ treatment.

Lentivirus production

Human embryonic kidney (HEK) 293T cells were plated onto 10-cm dishes and cotransfected with target plasmids and packaging vectors [PMD2.G (12259, Addgene) and PSPAX2 (12260, Addgene) for pLenti vector, and pCL (10045P, Imgenex) for pMXs vector] using polyethylenimine-based transfection. The viruses were filtered with 0.45-μm filters (SLHV033RB, Millipore) and centrifuged at 50,000g for 2.5 hours. The precipitation was used to infect cells.

FNZ treatment of NPCs in vitro

Normal human iPSC-derived NPCs were infected with mito-GFP/mito-DsRed virus, and then passaged and seeded in NPC medium on six-well plates, with 8 × 106 cells per well to ensure approximately 80% surface coverage at the time of the experiment. After overnight incubation, cells were treated with 15 μM FNZ for 3 days, followed by microscopy detection or fixation using 4% PFA. Cells with less than three mitochondrial clusters were defined as cells without mitochondria, and the fraction of cells without mitochondria was calculated as a percentage of total cells as previously reported ().

Western blot

Total proteins, electrophoresed on 12 to 15% polyacrylamide gel containing sodium dodecyl sulfate, were immediately transferred onto the polyvinylidene difluoride (PVDF) membranes (ISEQ00010, Millipore). Membranes were blocked in 5% milk, followed by incubation for 1 hour at room temperature or overnight at 4°C with primary antibodies. Then, membranes were washed in TBST (tris-buffered saline–Tween 20), incubated with secondary antibodies for 1 hour at room temperature, and washed in TBST. Enhanced chemiluminescence (ECL) substrate solution (WBKLS0500, Millipore) was added for chemiluminescence, the intensity of which was quantitated using a chemiluminescence imager (SmartChemil). The gray-scale values of protein bands were measured using the ImageJ (National Institutes of Health) software.

Quantitative reverse transcription PCR and PCR

Total RNA was extracted with TRIzol (15596026, Invitrogen), and 2 μg of RNA was used to generate complementary DNA. For mitochondrial genomic level detection, total DNA was extracted from cells after 3 days of FNZ treatment using the Universal Genomic DNA Kit (DP304, TIANGEN) and diluted to a concentration of 3 ng/ml. Quantitative PCR (qPCR) experiments were performed with SYBR Green Master Mix (RR420L,Takara) using the Bio-Rad CF96 Real-Time PCR System. Thermal condition was 95°C for 10 min, followed by 45 cycles of 95°C for 15 s, 60°C for 15 s, and 72°C for 15 s. The expression was normalized using glyceraldehyde-3-phosphate dehydrogenase (GAPDH) as an internal reference. For PCR of culture supernatant, the supernatant of the FNZ-treated culture, harvested 30 min after fresh medium replacement, was spun at 500g for 5 min and passed through 0.45-μm filters to remove dead cells. The thermal cycling was an initial 95°C for 5 min, followed by 29 cycles of 95°C for 30 s, 60°C for 30 s, and 72°C for 30 s. The primers used are listed in tables S4 and S5.

OCR and ECAR measurements using seahorse cellular flux assays

Seahorse plates were pretreated by coating with poly-d-lysine (0.1 mg/ml) and Matrigel. NPCs treated with FNZ for 3 days were passaged and seeded in NPC medium onto the pretreated Seahorse plates with 1 × 106 cells per XF24 well to ensure ~90% surface coverage at the time of the experiment. For extracellular acidification rate (ECAR) analysis, medium was exchanged for glycolysis stress medium [i.e., XF Base Medium (102353, Agilent) supplemented with 2 mM glutamine] at 1 hour before the assay. The cultures were then incubated in a non-CO2 incubator at 37°C to equilibrate. Substrates and selective inhibitors were injected during the measurement to achieve final concentrations of glucose at 2.5 mM, oligomycin at 1 μM, and 2-deoxy-d-glucose (2-DG) at 50 mM, according to the manufacturer’s instructions. For oxygen consumption rate (OCR) analysis, medium was exchanged for mitochondrial stress medium (i.e., XF Base Medium supplemented with 2 mM glutamine and 10 mM glucose) at 1 hour before the assay. The cultures were similarly equilibrated. Substrates and selective inhibitors were injected to achieve final concentrations of oligomycin at 1.0 μM, carbonyl cyanide p-trifluoromethoxyphenylhydrazone (FCCP) at 0.5 μM, and rotenone/antimycin A at 0.5 μM, according to the manufacturer’s instructions. The OCR and ECAR values were further normalized to the number of cells present in each well. The baselines of OCR and ECAR were defined as the average values. Changes in OCR and ECAR in response to substrates and inhibitors were defined as the maximal change after the chemical addition as compared with the baseline.

FACS analysis

Standard FACS analysis was performed with BD Fortessa. Extracellular mitochondria were analyzed using an adapted method as previously reported (). Briefly, mito-DsRed–labeled NPCs were treated with FNZ for 3 days (the culture medium was replaced freshly at day 2). The culture supernatant, harvested at day 3, was spun at 500g for 5 min and passed through 0.45-μm filters to remove dead cells. The resultant supernatant was used for FACS analysis. Cell surface staining of TOM20 (ab78547, Abcam) was performed as described previously (). Briefly, NPCs were treated with FNZ for 1 day, followed by cell collection (without fixation) and staining with a TOM20 antibody and secondary antibody. Gating was performed on the basis of an antibody-free control.

Microscopy

Fluorescence microscopy

Cells were infected with virus for expressing YFP-PARK2, GFP-LC3B, LAMP1-GFP (), or mitochondria-localization fluorescent proteins and were passaged to an appropriate density before FNZ addition. Images and time series were acquired with a Zeiss confocal microscope. For mito-mKeima, dual-excitation (458/543) ratiometric imaging of mito-mKeima was used to monitor the delivery of mitochondria to lysosomes as described previously (). For FM1-43FX (F35355, Invitrogen) staining, cells were exposed to FM1-43FX (100 ng/ml) for 10 s at room temperature, washed in PBS, and then imaged. For autophagy inhibition experiments, 10 μM CLQ or 1 mM 3MA was added simultaneously with FNZ for 3 days, followed by fixation and imaging. For analyzing the colocalization of mito-DsRed and LC3-GFP, FNZ or 100 nM rapamycin was added for 12 hours, and then cells were treated with 100 nM BAF for 4 hours before fixation to prevent autophagosome-lysosome fusion. For ionomycin treatment, cells were treated for 3 days with FNZ, 0 to 3.2 μM CaCl2, and 1 μM ionomycin (a calcium ionophore allowing rapid equilibration of transmembrane calcium concentrations). The relative mitochondrial area and ratio of LC3-positive/lysosomal mitochondria to total mitochondria were analyzed by ImageJ. Briefly, for relative mitochondrial area analysis, mitochondria were analyzed in the periphery of cells by first creating a binary mask of the cell morphometric channel (TH, TUJ, GFAP, or increased brightness) and using this channel to select one cell, and total cell area was measured. Then, images were thresholded to select mitochondria. From the thresholded fluorescence, binary images were generated, and the mitochondrial area was measured. The relative mitochondrial area was generated by dividing the mitochondrial area by this total cell area. For ratio of GFP-positive mitochondria to total mitochondria, a binary mask of the mitochondrial channel mito-DsRed was created and the total mitochondrial area was quantified. To select LC3-positive/lysosomal mitochondria puncta for further analysis, LC3-GFP or LAMP1-GFP fluorescence was thresholded. The thresholding value was determined as the average threshold value required for selecting GFP puncta. The thresholded image was converted to a binary image, and the lysosomal mitochondrial puncta area was determined. Then, the ratio of GFP-positive mitochondria puncta area to total mitochondrial area was quantified as the ratio of LC3-positive/lysosomal mitochondria to total mitochondria.

TEM analysis

FNZ-treated cells were fixed in 1% osmium tetroxide in Sörensen’s phosphate buffer for 1 hour, washed three times, and dehydrated in increasing concentrations of ethanol (25, 50, 75, and 96% for 2 × 10 min, respectively, and 100% for 2 × 15 min). Before embedding, the slices were placed in 100% acetone for 2 × 20 min and then overnight in a mixture of acetone and epon resin polybed 812 (1:1; Polysciences). The specimen was transferred to pure resin for at least 4 hours before embedding in new pure resin and polymerized at 60°C for 48 hours. Then, embedded specimen was sectioned in an ultratome (Super Nova) at 50 nm and mounted on slot copper grids previously covered with a thin film of Pioloform. Grids were stained in 4% uranyl acetate for 30 min at 40°C and in 0.5% lead citrate for 2 min at room temperature, followed by observation with a Tecnai G2 Spirit electron microscope (Nanoscience Initiative).

STED superresolution microscope imaging

NPCs expressing mito-DsRed and LAMP1-GFP were treated with FNZ for 24 hours, followed by fixation in 4% PFA for 30 min at room temperature. A Leica TCS SP8 STED 3× system (Leica) with two continuous-wave lasers at 592 and 660 nm for the depletion was used for STED imaging. An HyD detector and a STED WHITE objective (100×/1.40 OIL or 93×/1.30 GLYC) were used. The 3D STED images were acquired with excitation at 488 nm (WLL), emission in the range of 500 to 550 nm, depletion at 660 nm, excitation at 568 nm (WLL), emission in the range of 580 to 640 nm, depletion at 592 nm, gated detection at tg = 3 ns, and Z stacks of 4 μm. In general, images were recorded with a pixel resolution of 22.7 nm × 22.7 nm or higher, with a scan speed of 100 Hz, with a directional mode, with a line average of 3, and with a pinhole size of 1 Airy unit. The 3D images were further processed using Imaris software (Oxford Instruments).

The volume of total and lysosomal mitochondria was quantified according to the Imaris’s instruction. Briefly, surface style was used to create the surface of GFP and Red, followed by selection of the region of a cell and acquisition of the volume of Red (i.e., total mitochondrial volume). Mark was used to create a Red channel from GFP surface, followed by creation of the surface of a new Red channel by the same threshold. The volume of the new Red surface was the volume of lysosomal mitochondria.

RNA sequencing

Total RNA was extracted with TRIzol. Libraries were prepared using an Illumina TruSeq RNA Sample Prep kit (FC-122-1001, Illumina) following the manufacturer’s instruction. The RNA sequencing (RNA-seq) experiment was performed at Annoroad Gene Technology Co. Ltd. (Beijing, China). Paired-end reads were split into forward and reverse reads (RNA-seq_NPC_FNZ_Day0-Day5: https://doi.org/10.5061/dryad.rn8pk0pc4). Data were analyzed with an RSEM software. The GO analyses were performed using DAVID 6.7 database. The P values represent the modified Fisher’s exact corrected EASE score.

NPC-GeCKO library generation and FNZ-based screen

Mito-GFP–expressing NPCs were infected with the lentiGuide-Puro two-vector system for Cas9 and sgRNA delivery as previously described (). Briefly, Cas9-NPCs cells were generated via lentivirus-mediated transduction of the Cas9 gene, followed by selection with blasticidin (10 μg/ml). Cas9-NPCs were further transduced with lentivirus particles containing a human sgRNA library at a multiplicity of infection (MOI) of 0.3 to achieve no greater than 1 sgRNA per cell, followed by selection with both puromycin (1 μg/ml) and blasticidin (10 μg/ml) for more than 9 days. Then, 5 × 106 cells were collected for determining initial sgRNA abundance. Infected cells, expanded to 5 × 107 cells, were treated with FNZ for 3 days, followed by FACS sorting of GFP-positive cells. Genomic DNA was isolated from all samples, and the sgRNA sequences were amplified by PCR and sequenced on HiSeq 2500 (GeCKO_NPC_mito − GFP_FNZ_CT + MITO_1_2; https://doi.org/10.5061/dryad.bcc2fqzdc). MAGeCK was used to score genes under each condition relative to the preswap reference samples. Two infection replicates and screen were performed.

Gene-specific CRISPR-Cas9 sgRNA infection

CRISPR KO NPCs were generated using the lentiCRISPR v2 system (52961, Addgene). Briefly, annealed oligonucleotides were cloned into the lentiCRISPR v2 vector between Bsm BI sites. Oligonucleotide sequences for the gene-specific sgRNAs are listed in table S6. Polyclonal CRISPR KO NPCs were generated via lentivirus transduction of the gene-specific sgRNA-containing vector, followed by selection with puromycin (1 μg/ml) for 14 days. Vector control cells were generated as described above, using the NT1 sgRNA as a control. Genomic DNA was extracted from the KO cells, followed by PCR amplification of the targeted region using gene-specific primers specified in table S4. PCR products were cloned into pTA2 (TAK-201, Takara) as per the manufacturer’s recommendation, and ≥10 clones per KO were sequenced by Sanger sequencing.