Loss of Drosophila i-AAA protease, dYME1L, causes abnormal mitochondria and apoptotic degeneration

Abstract

Mitochondrial AAA (ATPases Associated with diverse cellular Activities) proteases i-AAA (intermembrane space-AAA) and m-AAA (matrix-AAA) are closely related and have major roles in inner membrane protein homeostasis. Mutations of m-AAA proteases are associated with neuromuscular disorders in humans. However, the role of i-AAA in metazoans is poorly understood. We generated a deletion affecting Drosophila i-AAA, dYME1L (dYME1L^del). Mutant flies exhibited premature aging, progressive locomotor deficiency and neurodegeneration that resemble some key features of m-AAA diseases. dYME1L^del flies displayed elevated mitochondrial unfolded protein stress and irregular cristae. Aged dYME1L^del flies had reduced complex I (NADH/ubiquinone oxidoreductase) activity, increased level of reactive oxygen species (ROS), severely disorganized mitochondrial membranes and increased apoptosis. Furthermore, inhibiting apoptosis by targeting dOmi (Drosophila Htra2/Omi) or DIAP1, or reducing ROS accumulation suppressed retinal degeneration. Our results suggest that i-AAA is essential for removing unfolded proteins and maintaining mitochondrial membrane architecture. Loss of i-AAA leads to the accumulation of oxidative damage and progressive deterioration of membrane integrity, which might contribute to apoptosis upon the release of proapoptotic molecules such as dOmi. Containing ROS level could be a potential strategy to manage mitochondrial AAA protease deficiency.

Main

Mitochondria dictate the survival and well being of the eukaryotic cells, but their unique genetic system and complex biophysical characteristics make for great challenges in maintaining organelle integrity and function.¹ One challenge is ensuring the proper assembly of the protein complexes carrying out mitochondrial functions. Most mitochondrial proteins are encoded by the nuclear genome and imported into the mitochondria after synthesis.² However, mitochondria also contain their own genome, which encodes the core components of the electron transport chain (ETC). The mitochondrion-encoded subunits of the ETC assemble on the inner mitochondrial membrane (IMM) with the nuclear-encoded ones. Unassembled polypeptides have to be removed to maintain the stoichiometry of the ETC complexes. Another challenge is the production of reactive oxygen species (ROS), the unavoidable by-products of electron transfer, which are generated mainly at complex I (NADH/ubiquinone oxidoreductase) and complex III (ubiquinol-cytochrome c oxidoreductase) in the ETC.³ Excessive ROS can damage proteins and impair mitochondrial functions.

An elaborate system of chaperones and proteases has evolved to ensure mitochondrial proteostasis.⁴ The proteases are located in different submitochondrial compartments and carry out critical steps of mitochondrial biogenesis and turnover, including processing, assembly and degradation of mitochondrial proteins. Mitochondrial proteases of the AAA class (ATPases Associated with diverse cellular Activities) are the main regulators of proteostasis on the IMM,⁵ which houses many important complexes including those of the ETC. The catalytic domains of AAA proteases face either the matrix (mitochondrial m-AAA proteases) or the intermembrane space (IMS) (mitochondrial i-AAA protease).⁶ Despite their different topologies, mitochondrial m-AAA proteases and i-AAA protease share highly conserved protein structures and catalytic mechanism, and even an overlapping substrate specificity.⁷ Mutations in the mitochondrial m-AAA proteases are responsible for neurodegenerative disorders including hereditary spastic paraplegia (HSP), spinocerebellar ataxia (SCA28) and spastic ataxia neuropathy syndrome.^{8, 9, 10} However, the degenerative mechanisms remain elusive,¹¹ and the presence of multiple mitochondrial m-AAA proteases with redundant functions in eukaryotes complicates their analysis in animal models. By contrast, only one mitochondrial i-AAA protease has been identified in eukaryotic genomes. It coordinates mitochondrial fusion and fission,¹² and couples the mitochondrial dynamics to oxidative phosphorylation.¹³ Knocking down mitochondrial i-AAA protease in cultured cells perturbed mitochondrial morphology and sensitized cells to oxidative stress and apoptotic stimuli.^{14, 15, 16} However, the pathophysiological consequences of i-AAA loss of function at the animal level have been largely unknown. Yet, the absence of gene redundancy makes mitochondrial i-AAA protease particularly suitable for genetic studies exploring the function of mitochondrial AAA proteases in animal models.

Drosophila melanogaster has been widely used to understand the biochemical processes underlying a variety of human diseases,¹⁷ including many mitochondrial disorders such as Parkinson’s disease.^{18, 19, 20} In these studies, some key phenotypes of mitochondrial diseases, such as impaired locomotor activities and neural and muscular degeneration, have been successfully recapitulated in Drosophila. Here we demonstrate that loss of mitochondrial i-AAA protease (dYME1L) in Drosophila melanogaster perturbs mitochondrial proteostasis, causes mitochondrial anomalies and triggers apoptotic degeneration in neurons and muscles.

Results

CG3499 encodes the Drosophila homolog of the human i-AAA protease YME1L

Based on sequence analysis, CG3499 is the Drosophila homolog of human i-AAA protease YME1L.²¹ It is predicted to have a single transmembrane domain followed by an ATPase domain and an M41 peptidase domain, both of which are highly conserved among mitochondrial AAA proteases, and with human YME1L. To determine whether the CG3499 protein localizes to the mitochondria, we built a CG3499-GFP fusion protein and coexpressed it with Tom20-mCherry, a fluorescent mitochondrial marker, in Drosophila Schneider 2 (S2) cells. Confocal microscopic analysis showed overlapping fluorescent signals for GFP and Tom20-mCherry (Figure 1a), confirming that CG3499 (dYME1L) is a mitochondrial protein.

Figure 1

Generating mitochondrial i-AAA protease mutant in Drosophila melanogaster. (a) Coexpression of CG3499-GFP and Tom20-mCherry in Drosophila S2 cells showing their colocalization to the mitochondria. Scale bar, 5 μm. (b) A schematic illustration of the genomic region of CG3499 and the CG3499-deletion allele (dYME1L^del). DNA gel electrophoresis shows a 3.6 kb fragment amplified from wild-type genomic DNA (wt) but a 1.6 kb fragment amplified from dYME1L^del genomic DNA (del)

To obtain a dYME1L mutant, we mobilized the P element inserted in the first intron of CG3499 (BL14902: KG01861). We recovered an imprecise excision with a 2 kb deletion (dYME1L^del), which removed the majority of the dYME1L coding region (Figure 1b). dYME1L^del flies were homozygous-viable, but male-sterile. Compared with the wild-type flies (w¹¹¹⁸), dYME1L^del flies had a markedly shortened lifespan (Figure 2c). The lifespan was even shorter for the homozygous mutant progeny of homozygous mothers (dYME1L^del/dYME1L^del) compared with that for the homozygous progeny of heterozygous mothers (dYME1L^del/CyO) (Supplementary Figure S1), suggesting a maternal effect consistent with the maternal inheritance of mitochondria. Although most maternal effects are manifested during embryonic development,^{22, 23, 24} this mitochondrial mutation has a cumulative effect that extends to adulthood.

Figure 2

Phenotypic characterization of dYME1L^del flies. (a) dYME1L^del flies show an age-dependent locomotor impairment in the climbing tests. Expressing dYME1L or human homolog YME1L restores the climbing ability in dYME1L^del flies. Each data point represents the average of three groups of 10 flies. (b) dYME1L^del flies show an age-dependent decrease in the resistance to mechanical stress, which can be rescued by dYME1L or YME1L expression. Each data point represents the average of 10 individual flies. Student's t-test is performed between the individual genetic manipulation and its corresponding control for (a and b): dYME1L^del versus w¹¹¹⁸. dYME1L^del, Da-gal4 versus Da-gal4. dYME1L^del, Da-gal4>dYME1L versus dYME1L^del, Da-gal4 and dYME1L^del, Da-gal4>YME1L versus dYME1L^del, Da-gal4. *P<0.01 (c) dYME1L^del flies are short-lived and their lifespan can be rescued by the expression of either dYME1L or YME1L transgene. Log-rank test is performed between the individual genetic manipulation and its corresponding control, as indicated within parentheses after the abbreviated genotype. *P<0.0001. (d) Representative electron microscopic images from different genotypes at 3 days (3d) and 4 weeks (4w) after eclosion showing retinal degeneration in dYME1L^del flies. The red arrows indicate dying photoreceptor cells featured with signs for apoptosis: large electron-dense area in the nucleus for condensed chromatin, cell blebs and shrinkage, cytoplasmic condensation and losing cell–cell contacts with neighboring cells. Scale bar, 1 μm. The genotypes of flies used are as follows and the number of flies used in (a) are indicated as follows (within parentheses): wild type – w¹¹¹⁸ (N=269); mutant – dYME1L^del (N=277); Da-gal4 – Da-gal4/+ (N=115); dYME1L^del, Da-gal4 – dYME1L^del; Da-gal4/+ (N=187); dYME1L^del, Da>dYME1L or dYME1L^del, dYME1L – dYME1L^del; Da-gal4/UAS-dYME1L-GFP (N=100); dYME1L^del, Da>YME1L or dYME1L^del, YME1L – dYME1L^del; Da-gal4/UAS-YME1L-GFP (N=79)

dYME1L^del causes neuronal degeneration and abnormal crista structure

Mitochondrial mutations afflict most severely the highly energy-demanding tissues such as the nervous system and the muscles. Thus, we examined whether dYME1L^del caused any neuronal or muscular deficiencies, using locomotor assays. Young dYME1L^del flies appeared similar to the age-matched wild-type flies in the climbing test and the bang sensitivity assay. However, they gradually lost their climbing ability (Figure 2a), and became increasingly susceptible to seizure following mechanical stress as they aged (Figure 2b).

The progressive loss of motor control and the increased sensitivity to mechanical stress suggested defects in the muscles, the nervous system or both. To assay whether dYME1L^del caused neurodegeneration, we examined the morphology of the adult eye, a widely used model for various human neurodegenerative diseases,²⁵ using electron microscopic imaging and optical neutralization techniques. dYME1L^del flies had the full complement of ommatidium components at a young age but gradually lost rhabdomeres with age (Figure 2d and Supplementary Figure S2d), indicating an age-dependent degeneration of photoreceptor neurons. Ubiquitous expression of the dYME1L cDNA in the mutant background rescued most dYME1L^del phenotypes, including shortened lifespan (Figure 2c), age-dependent locomotor impairments (Figures 2a and b) and neurodegeneration (Figure 2d and Supplementary Figure S2e). These results demonstrate that the dYME1L mutation is responsible for the observed neuronal, muscular and behavior defects in dYME1L^del flies. Of note, expressing human YME1L also rescued these phenotypes (Figure 2 and Supplementary Figure S2e), confirming that CG3499 is indeed the Drosophila ortholog of the human mitochondrial i-AAA protease.

We further examined mitochondrial ultrastructure in the indirect flight muscle (IFM) by transmission electron microscopy (Figure 3). Compared with the wild-type mitochondria (Figures 3a, b and e), the cristae of dYME1L^del mitochondria were misoriented (Figures 3c, d, f and g), giving rise to a ‘patchy’ appearance, and often loosely packed. Swollen mitochondria were also observed. In old dYME1L^del flies, cristae became more disorganized and there were many swollen mitochondria with much fewer cristae. In addition, old mutant flies had many electron-dense structures (Figures 3d and f) similar to the dense flocculant structures known to appear in mitochondria during aging and oxidative stress.²⁶

Figure 3

Mitochondrial anomalies in dYME1L^del flies. Electron micrographs of w¹¹¹⁸ (a, b and e) and dYME1L^del (c, d, f and g) IFM. w¹¹¹⁸mitochondria display a relatively uniform morphology with densely packed arrays of cristae at both 3 days (3d) (a) and 3 weeks (3w) (b and e). By contrast, dYME1L^del mitochondria display various morphological anomalies, including reduced crista density (arrow in c) and disorganized cristae (f and g), and, in aged flies, dense aggregates (white arrowheads in d and f), swelling and membrane disruption (black arrowheads in d and g). Scale bar for (a–d), 2 μm and for (e–j), 0.5 μm

Apoptosis underlies neurodegeneration in dYME1L^del flies

Mitochondrial dysfunction has been reported to trigger apoptosis in mammals.²⁷ However, mitochondrial involvement in apoptosis is still controversial in Drosophila.²⁸ The transmission electron microscopy of aged dYME1L^del flies displayed dying photoreceptor cells with apoptotic morphology, such as condensed nuclei, cell blebs and shrinkage, cytoplasmic condensation and losing cell–cell attachment (Figure 2d and Supplementary Figure S2a). Thus, we examined whether apoptotic cell death underlies the neuromuscular degeneration observed in dYME1L^del flies.

To confirm the existence of apoptosis, we used the cleaved caspase-3-specific antibody to stain the thoracic muscles for the activated Drice,^{29, 30, 31} the Drosophila homolog of mammalian caspase-3, which resides in the cytosol as an inactive zymogen (procaspase-3) and translocates to the nucleus after proteolytic activation.³² The cleaved caspase-3 staining was observed either associated or within >80% of the muscle nuclei of aged dYME1L^del flies, but could hardly be detected in either young dYME1L^del flies or wild-type flies (Figures 4a and b and Supplementary Figure S3a). Consistent with the positive cleaved caspase-3 staining, we also detected TUNEL-labeled nuclei in the muscle of aged dYME1L^del flies but not in the age-matched wild-type control (Supplementary Figure S3b). Meanwhile, we detected apoptosis in the photoreceptor cells of aged dYME1L^del flies using both cleaved caspase-3 antibody and cleaved Dcp-1 antibody ³³ (Supplementary Figure S2b). These results demonstrate that the muscle and the retina of aged dYME1L^del flies undergo apoptosis.

Figure 4

Muscle and photoreceptor cells of aged dYME1L^del flies undergo apoptotic cell death. (a) Confocal microscopic sections of IFM preparations from 4-week-old w¹¹¹⁸ and dYME1L^del flies stained for cleaved caspase-3 (α-CC3, green), a marker of apoptotic cell death. Preparations are counterstained with DAPI (4',6-diamidino-2-phenylindole) to reveal nuclei (red) and phalloidin to reveal F-actin (blue). Scale bar, 10 μm. (b) Quantification of cleaved caspase-3-positive nuclei in wild-type and mutant IFM of aged flies (N=5). Eighty percent of nuclei display cleaved caspase-3 staining in aged mutant flies. Compared with wild-type flies, the number of apoptotic nuclei in dYME1L^del is significantly increased. Overexpressing ROS-scavenger proteins, such as SOD1 and SOD2, reduces cleaved caspase-3 staining, compared with the age-matched dYME1L^del, Da-gal4 flies. Student's t-test is performed between the individual genetic manipulation and its corresponding control: dYME1L^del versus w¹¹¹⁸. dYME1L^del, Da-gal4 versus Da-gal4. dYME1L^del, Da-gal4, SOD1 versus dYME1L^del, Da-gal4 and dYME1L^del, Da-gal4, SOD2 versus dYME1L^del, Da-gal4. *P<0.05, **P<0.01 and ***P<0.001. (c) The model of the Drosophila apoptosis cascade used to inform the genetic suppression experiments in (d) and (e). (d) Restoring or suppressing photoreceptor cell loss in dYME1L^del flies by genetic manipulation. The number of photoreceptors per ommatidium (N>100) is assessed in 4-week-old flies (N=5) using the optical neutralization assay. Compared with the age-matched wild type, dYME1L^del flies show eye degeneration as evident by assessing ommatidium integrity. Overexpressing dYME1L, YME1L, P35, DIAP1, SOD1 and SOD2 or knocking down dOmi, dArk and Dronc rescues or suppresses the retinal degeneration in the mutant background. Student's t-test is performed. Analysis between wild type and mutant or gal4 only in the mutant background is indicated above the corresponding columns. Analysis between the individual genetic background and the corresponding genetic manipulation is indicated within the parentheses. **P<0.01 and ***P<0.001. (e) Age-dependent locomotor impairment in dYME1L^del flies is suppressed either by knocking down dArk, Dronc and dOmi or overexpressing DIAP1. Student's t-test is performed. *P<0.05 and ***P<0.001. Genotypes in (b) are as follows: wild type – w¹¹¹⁸; mutant – dYME1L^del. Da-gal4-Da-gal4/+. dYME1L^del; Da-gal4 – dYME1L^del; Da-gal4/+. dYME1L^del; SOD1 – dYME1L^del, UAS-SOD1/dYME1L^del; Da-gal4/+. dYME1L^del; SOD2 – dYME1L^del, UAS-SOD2/dYME1L^del; Da-gal4/+. In (d) and (e), wt-Canton S. GMR-P35-gmr-P35; dYME1L^del. GMR-DIAP1-dYME1L^del; gmr-diap1. Da-gal4-dYME1L^del; Da-gal4/+. dYME1L^del, Da>dYME1L-dYME1L^del; Da-gal4/UAS-dYME1L-GFP. dYME1L^del, Da>YME1L-dYME1L^del; Da-gal4/UAS-YME1L-GFP. dYME1L^del, Da>SOD1-dYME1L^del, UAS-SOD1/dYME1L^del; Da-gal4/+. dYME1L^del, Da>SOD2-dYME1L^del, UAS-SOD2/dYME1L^del; Da-gal4/+. dYME1L^del, Da>dArk^IR-dYME1L^del; Da-gal4/UAS-dArk IR. dYME1L^del, Da>Dronc^IR-dYME1L^del; Da-gal4/UAS-Dronc IR. dYME1L^del, Da>dOmi^IR-dYME1L^del; Da-gal4/UAS-dOmi IR

To further explore whether apoptosis has causative roles in the degeneration, we overexpressed P35,^{34, 35, 36} a baculovirus inhibitor for effector caspases, in the eye under the control of an eye-specific gmr promoter (gmr-P35). The overexpression of P35 suppressed the loss of photoreceptors in the ommatidia of dYME1L^del flies (Figure 4d and Supplementary Figure S2e). To test the involvement of the caspase cascade in the degeneration in dYME1L^del flies, we either overexpressed or knocked down the relevant genes (Figure 4c), and assessed the integrity of their ommatidia by the optical neutralization assay (Figure 4d and Supplementary Figures S2c and e). The effective dArk and Dronc knockdown suppressed the retinal degeneration, confirming that the dArk/Dronc cascade is indeed involved. In addition, both DIAP1 overexpression and the effective dOmi (Drosophila Htra2/Omi) knockdown also suppressed the retinal degeneration. Similarly, the age-dependent locomotor impairment in dYME1L^del flies was suppressed by blocking apoptosis upon DIAP1 overexpression or dArk, Dronc and dOmi knockdown (Figure 4e). Collectively, these results indicate that the mitochondrial i-AAA protease mutation leads to degeneration through dArk/Dronc-mediated apoptotic cascade.³⁷

Elevated ROS level contributes to neurodegeneration in dYME1L^del flies

To determine whether the dYME1L mutation had any impact on the ETC, we analyzed the amount of various subunits of the ETC complexes by western blot (Figure 5a and Supplementary Figures S4a and b). Except for complex I, all the tested complexes were present in comparable amounts in wild-type flies and dYME1L^del flies, regardless of their ages. By contrast, the amount of complex I, as monitored by NDUFS3, one of its subunits, was markedly reduced in aged dYME1L^del flies compared with the age-matched wild-type flies, although it was normal in young dYME1L^del flies. Consistent with this observation, the activity of complex I (Figure 5b), but not the other complexes (Supplementary Figures S4c and d), was also reduced in the aged mutant flies.

Figure 5

Complex I deficiency and elevated ROS in aged dYME1L^del flies. (a) Western blot analysis on whole-fly lysates reveals an age-dependent reduction in NDUFS3, a subunit of complex I in dYME1L^del flies. Subunits of complex IV (Cox4) and complex V (ATP5α) do not show such a pattern. (b) The complex I activity decreases with age in dYME1L^del flies. The isolated mitochondria from whole flies are tested. The activity is normalized to the young wild-type flies. CI, complex I. **P<0.01 in Student’s t-test. (c) m-Aconitase activity decreases with age in dYME1L^del flies. The activity is normalized to the age-matched wild-type flies. **P<0.01 in Student’s t-test. (d) Expression of ROS-scavenger proteins extends the lifespan of dYME1L^del flies. (e) NAC treatment also extends the lifespan of dYME1L^del flies (N=149) compared with the untreated control (N=91). The NAC dose used here, 1 μM, does not lead to lifespan extension in wild-type flies (Supplementary Figure S4e). Log-rank analysis is performed for (d) and (e). *P<0.0001. (f) The climbing assay shows ROS-scavenger proteins significantly improve the locomotor activities of dYME1L^del flies. *P<0.05 and **P<0.01 in Student's t-test. The genotypes in (a)–(c) are as follows. wt – w¹¹¹⁸; del – dYME1L^del. Young: 3 days old; old: 3 weeks old. The genotypes in (d) and (f) and the number of flies used in (d) (within parentheses) are as follows. Da-Da-gal4/+. dYME1L^del, Da-dYME1L^del; Da-gal4/+ (N=115); dYME1L^del, SOD1-dYME1L^del, UAS-SOD1/dYME1L^del; Da-gal4/+ (N=348); and dYME1L^del, SOD2-dYME1L^del, UAS-SOD2/dYME1L^del; Da-gal4/+ (N=308)

Complex I is a major site of ROS production in the respiratory chain.³ To test whether the impaired complex I leads to elevated ROS level in dYME1L^del flies, we used a colorimetric aconitase assay to indicate the status of oxidative damages in dYME1L^del flies (Figure 5c). Mitochondrial aconitase activity was indistinguishable between young wild-type flies and dYME1L^del flies. However, the aconitase activity was 20% lower in old dYME1L^del flies compared with that in the age-matched controls, indicating a higher amount of ROS in the mutant. To determine whether elevated ROS level contributes to the mutant phenotypes, we used the GAL4-UAS binary system to express ubiquitously the ROS-scavenging enzymes SOD1 and SOD2 in dYME1L^del flies under Da-gal4 driver. Expression of these enzymes reduced the fraction of apoptotic nuclei, as indicated by the positive cleaved caspase-3 staining in the muscle of aged dYME1L^del flies (Figure 4b and Supplementary Figure S2c). Moreover, overexpression of SOD1 or SOD2 extended the lifespan of dYME1L^del flies moderately (Figure 5d), improved their locomotor activities (Figure 5f) and completely suppressed the photoreceptor neuron degeneration in the eye at the age of 4 weeks (Figure 4d and Supplementary Figure S3e). We also treated flies with N-acetyl cysteine (NAC), an antioxidant increasing the pool of reduced glutathione.³⁸ As NAC itself can extend lifespan,³⁹ we first titrated its dosage in wild-type flies (Supplementary Figure S4e). At a dose where no significant increase of lifespan was observed in wild type, the lifespan of dYME1L^del flies was slightly extended (Figure 5e). Taken together, these results demonstrate that mitochondrial i-AAA protease deficiency impairs complex I activity and leads to an increased level of ROS, which contributes to the degenerative phenotypes.

dYME1L^del flies display elevated mitochondrial unfolded protein stress and are more sensitive to environmental and genetic stresses

Considering the proposed functions of i-AAA in mitochondrial proteostasis, we assessed the amount of HSP60, a marker for the unfolded protein response in mitochondria (UPR^mt).^{40, 41} Western blot analysis showed that HSP60 was more abundant in dYME1L^del flies compared with that in wild-type flies throughout adulthood (Figure 6a), supporting the idea that mitochondrial i-AAA protease deficiency disrupts mitochondrial protein homeostasis, leading to an elevated mitochondrial unfolded protein stress.

Figure 6

dYME1L^del flies are more vulnerable compared with wild-type flies to genetic and environmental stresses. (a) Western blot analysis on whole-fly lysates showing that the UPR^mt, as monitored by the HSP60 levels, is upregulated throughout the adulthood in dYME1L^del flies (1–4 weeks old after eclosion). (b) Western blot analysis on whole-fly lysates showing an elevated level of phospho-AKT (P-AKT) and, to a greater extent, of phospho-eIF2α (P-eIF2α), in 3-day-old dYME1L^del flies compared with the age-matched w¹¹¹⁸. (c) Western blot analysis on whole-fly lysates exhibiting the increase of ATG8a-II form in 3-day-old dYME1L^del flies compared with the age-matched w¹¹¹⁸. The ATG8a-I form is also increased. (d) Western blot analysis on whole-fly lysates displaying elevated level of conjugated ubiquitin in 3-day-old dYME1L^del flies compared with the age-matched w¹¹¹⁸. (e) Overexpressing the endogenous mitochondrial protein Cox4 in the dYME1L^del background leads to complete lethality (*). The ectopic expression is driven by tub-gal4. (f) Paraquat treatment leads to quicker mortality in dYME1L^del compared with that in wild-type flies. Log-rank analysis is performed. *P<0.0001; n.s., not significant. wt in (b)–(e) and (f) are w¹¹¹⁸ and dYME1L^del/+, respectively. del: dYME1L^del

We also examined the levels of eIF2α and AKT phosphorylation that are considered as the downstream signal transducers of the matrix and IMS unfolded protein stresses,^{42, 43} respectively. We found that both eIF2α and AKT were hyperphosphorylated in dYME1L^del flies (Figure 6b), further confirming the elevated UPR^mt. However, eIF2α phosphorylation is also an indicator of ER unfolded protein response (UPR^ER). Indeed, BiP/GRP78, the master regulator of UPR^ER transduction pathway,^{44, 45, 46} was upregulated in dYME1L^del flies throughout their adulthood (Supplementary Figure S5a). In addition, the general cellular protein turnover mechanisms, autophagy and ubiquitin–proteasome system, were both activated in the mutant flies (Figures 6c and d).

Elevated mitochondrial unfolded protein stress might render young dYME1L^del flies more susceptible compared with wild-type flies to protein stoichiometric disruption. To test this idea, we further challenged the mitochondrial protein homeostasis by overexpressing an endogenous mitochondrial protein, cytochrome c oxidase subunit IV (Cox4). Although overexpression of this protein had no obvious detrimental effects on wild-type flies, it led to complete mortality in dYME1L^del flies (Figure 6e). To test whether dYME1L^del flies were also more sensitive, compared with wild type, to oxidative stress, we treated them with paraquat, a known inducer of oxidative damages.⁴⁷ Paraquat treatment led to premature mortality in both wild-type flies and dYME1L^del flies, but dYME1L^del flies were much more sensitive to paraquat compared with wild-type flies (Figure 6f): it took only 3 days after paraquat treatment for half of the mutant flies to die. More than 50% of the wild-type flies still survived 10 days after paraquat administration, but almost all of the dYME1L^del flies had died by that time. Taken together, these results suggest that the mitochondrial i-AAA protease mutation perturbs proteostasis inside mitochondria, and sensitizes them to genetic and environmental stresses.

Discussion

To understand the function of mitochondrial i-AAA protease in mitochondrial proteostasis and the degenerative mechanisms associated with mitochondrial AAA protease mutations, we generated a mitochondrial i-AAA protease mutant in Drosophila, thus creating the first YME1L-deficient animal model. We observed reduced complex I activities and consequently increased ROS levels in aged dYME1L mutant flies, which is consistent with studies on mammalian cultured cell lines.¹⁶ Importantly, deletion of dYME1L caused an increase in mitochondrial unfolded protein stress and rendered mutant flies more susceptible to genetic and environmental stresses. Furthermore, it led to impaired locomotor activity and neural and muscular degeneration. These phenotypes are very similar to some key symptoms of human degenerative diseases associated with mutations in mitochondrial m-AAA proteases.⁴⁸ Notably, ectopic expression of human mitochondrial i-AAA protease, YME1L, rescued the degenerative phenotypes in the mutant flies, suggesting conserved physiological functions of mitochondrial i-AAA protease in metazoans.

In accordance with the intimate relationship between mitochondria and apoptosis in mammals, many mitochondrial mutations cause apoptotic myopathies and neuropathies.^{49, 50} Several common neurodegenerative diseases such as Parkinson’s disease and Huntington’s disease are also associated with mitochondrion-dependent apoptosis.⁵¹ However, mitochondrial involvement in apoptosis in Drosophila remains uncertain.^{52, 53} Particularly, the cytochrome c involvement in the apoptosome activation is highly debatable.²⁸ Instead of being a direct ‘activator’, mitochondria are considered as the docking sites to receive death signals and release proapoptotic molecules. In Drosophila, the Apaf-1-related protein dArk is associated with and constitutively activates the initiator caspase Dronc, which in turn activates the effector caspase Drice. dArk/Dronc cascade is involved in the apoptosis of aged dYME1L^del flies as knockdown of either of them suppresses the degenerative phenotypes. DIAP1 inhibits both Dronc and Drice and thereby suppresses the apoptosis. DIAP1 can be degraded by the transcriptional upregulation of H99 cluster (RHG) genes, namely reaper, hid and grim.³⁷ However, we did not detect any significant transcriptional upregulation of these genes in aged dYME1L^del flies (Supplementary Figure S6). DIAP1 can also be degraded by dOmi, a mitochondrial serine protease in the IMS, when released into the cytoplasm.⁵⁴ Actually dOmi knockdown suppresses the degenerative phenotypes, namely retinal degeneration and locomotor impairment, in aged dYME1L^del flies (Figures 4d and e and Supplementary Figure S2e).

In addition, we found the dOmi upregulation at both transcriptional level and protein level (Supplementary Figures S5b and c), and the ruptured mitochondria in the electron microscopic images of the IFM from dYME1L^del flies (Figures 3d and g). Given that ectopic overexpression of dOmi in Drosophila developing eyes causes cell death,^{20, 54} these results suggest that dOmi might be released through mitochondrial breakages in dYME1L^del flies to trigger the apoptotic cascade (Supplementary Figure S7). Of note, dOmi protein level was also markedly increased in the old flies, regardless of their genotypes (Supplementary Figure S5c). The lack of cell death in old wild-type files argues that, although necessary, the increase of the endogenous dOmi alone might not be sufficient to trigger apoptosis. In yeast, the mitochondrial i-AAA protease is also required for the turnover of Ups1 and Ups2, which regulate the accumulation of cardiolipin (CL) and phosphatidylethanolamine, respectively.⁵⁵ Thus, impaired phospholipid metabolism might be another inroad leading to the membrane remodeling, rupture and eventually apoptosis in mitochondrial i-AAA protease mutant flies (Supplementary Figure S7). Our results not only support that mitochondrial disruption per se could trigger apoptosis in Drosophila but also suggest that apoptotic neuronal death might be a common degenerative mechanism associated with mitochondrial AAA protease mutations.

i-AAA is required for maintaining stoichiometry of all the ETC complexes encoded by both the nuclear and the mitochondrial genomes. Why dYME1L^del specifically impairs complex I in aged flies remains to be explored. Probably, complex I is the largest ETC complex and consists of more than 40 different subunits, which may render it particularly vulnerable. Indeed many mitochondrial mutations perturbing mitochondrial proteostasis compromise complex I.^{41, 56} Flies carrying the Drosophila Sicily mutation that disrupts complex I assembly also display reduced complex I activity, elevated ROS level and retinal degeneration.⁵⁶ However, there is no detectable apoptosis in the Sicily mutant, suggesting the existence of caspase-independent cell death and the insufficiency for ROS alone to elicit apoptosis. We found that the expression of ROS-scavenger proteins suppressed the retinal degeneration and extended the lifespan of dYME1L^del flies. Further, we showed in proof of principle that the antioxidant NAC was also able to extend the lifespan of dYME1L^del flies. Thus, ROS may act synergistically with other mitochondrial deficiencies to trigger or promote apoptosis in dYME1L^del flies (Supplementary Figure S7). More importantly, our work resonates with a recent study showing that NAC treatment rescued mitochondrial defects in cultured AFG3L2-depleted mouse neurons,⁵⁷ and suggests that manipulating ROS level could be a potential disease-curbing strategy.

In mammals, OPA1, a substrate of i-AAA, controls crista remodeling and mediates apoptotic neurodegeneration.^{58, 59} However, the S2 cleavage site of i-AAA is not present in Drosophila OPA1.⁶⁰ Ectopic overexpression of another mitochondrial protease, Rhomboid-7, could process dOPA1.⁶¹ Our western blot analysis revealed similar patterns of dOPA1 isoforms in wild-type and mutant flies (Supplementary Figure S5d), further substantiating that dYME1L is unlikely to be required for OPA1 processing. On the other hand, although the eIF2α phosphorylation attenuated protein translation, all the OPA1 isoforms appeared to be more abundant in the dYME1L^del background compared with that in the wild-type background, suggesting a role of i-AAA protease in removing excessive IMS proteins. Nonetheless, it is still possible that other substrates of i-AAA could regulate crista remodeling in Drosophila as OPA1 in mammals.

In dYME1L^del flies, sustained UPR^ER and general protein turnover mechanisms are observed. Thus, it would be interesting to dissect how a mitochondrial stress triggers UPR^ER and understand the pathophysiological significance of their interplay.

Materials and Methods

Molecular biology

Drosophila genes CG3499-PB (dyme1l-PB), CG7654 (tom20), CG10664 (cox4) and CG8479-PA (dOpa1-PA) were cloned from embryonic cDNA library. The human mitochondrial i-AAA protease subunit gene, yme1l, was cloned from cDNA clone MGC: 3533 IMAGE: 3637576 (NIH-MGC Project, Gaithurberg, MD, USA). Tags, such as GFP, mCherry and myc, were fused at the C terminus of the individual mitochondrial proteins. The genes were cloned into pAW and pPW vectors.

Cell culture

Drosophila S2 cells were cultured in Schneider’s Drosophila medium (Invitrogen, Carlsbad, CA, USA) supplemented with 10% FBS. Cellfectin II (Invitrogen) was used as the vehicle to deliver plasmids into cells. Before confocal fluorescence images were taken, cells were seeded in concanavalin A-pretreated coverglass chambers.

Drosophila genetics

Fly stocks and all genetic crosses were maintained at 25 °C on standard cornmeal agar media under a 12:12 light-dark cycle. The fly stocks, BL14902 (P-element insertion in CG3499 gene region), BL5782 (gmr-diap1), BL5774 (gmr-P35), BL28544 (dOmi RNAi) and BL32963 (Dronc RNAi), were obtained from Bloomington Stock Center (Bloomington, IN, USA). dArk RNAi line was generously provided by Pascal Meier (London, UK) through NIG and MITILS.⁶² UAS-SOD1 and UAS-SOD2 were kindly provided by John Phillips (Guelph, Ontario, CA, USA). Gal4 driver lines, tub-gal4 and Da-gal4, were kept in the lab. Transgenic flies, UAS-dYME1L-GFP, UAS-YME1L-GFP, UAS-Cox4-GFP and UAS-dOPA1-myc, were made by BestGene embryonic injection service. w¹¹¹⁸ was used as a wild-type control, and experiments were carried out on male flies, unless otherwise indicated. Fly age was counted as days per weeks after eclosion.

We generated CG3499-deletion line by imprecise excision of the P-element insertion line BL14902, y¹;CG3499^KG01861/CyO;ry⁵⁰⁶. P-element excision was screened in the CG3499 gene region by PCR analysis on the extracted genomic DNA using two primers that amplify the CG3499 gene region (left primer: 5′-CGATAGGGACTCTTTGTCTATCCT-3′; right primer: 5′-ACATAGGTTGCAGTAGTCTGGTCAC-3′). In wild-type and in precise excision lines, these primers amplify a 3.6 kb fragment. Smaller PCR products indicate imprecise excisions. The PCR products were sequenced to identify imprecise excisions removing large portions of the CG3499 gene region (left primer: 5′-ACTGTACGGAGAGGCAGTTTAGAT-3′; right primer: 5′-GGTTCGATTAACGGAGTAGTAGTGC-3′). One imprecise excision line removing 2 kb CG3499-genomic region including the majority of the exons was recovered and named dYME1L^del.

Immunohistochemistry

Drosophila tissues were dissected in 4% PFA PBS and permeablized in 4% PFA PBST (0.1% Triton X-100 in PBS) for 30 min. The samples were washed with PBST and blocked in 5% FBS, 0.5% BSA PBST (blocking buffer) for 1 h and then incubated with primary antibodies overnight at 4 °C. Phalloidin (Alexa Fluor; Life Technologies, Eugene, OR, USA; 1 : 40) was used to highlight F-actin, whereas the cleaved caspase-3 antibody (9661; Cell Signaling; 1 : 400) or the cleaved Dcp-1 antibody (9578; Cell Signaling; 1 : 100) was used to detect apoptosis. After extensive washing with the blocking buffer, the samples were treated with secondary antibodies (Alexa Fluor; Life Technologies; 1 : 200) in the blocking buffer for 45 min. After extensive washes, samples were placed in VECTASHIELD Mounting Medium with DAPI (Vector Lab, Berlingame, CA, USA) before confocal imaging. For the TUNEL assay, Click-iT TUNEL Alexa Fluor Imaging Assay Kit (Life Technologies) was used following the manufacturer’s protocol.

Confocal microscopic image acquisition

Images were acquired using an UltraVIEW VoX spinning disk confocal imaging system (Perkin-Elmer, Waltham, MA, USA) on the AxioObserver.Z1 inverted microscope (Zeiss, Thornwood, NY, USA) equipped with a × 63 objective lens (numerical aperture: 1.4; immersion medium: oil) and cameras (Hamamatsu C9100-13 and C10600-10B (ORCA-R2, Bridgewater, NJ, USA)). Live cells with transient transfection were imaged in Schneider’s Drosophila medium supplemented with 10% FBS. Fixed tissues were imaged in VECTASHIELD Mounting Medium with DAPI. Imaging processing was performed using Volocity (Perkin-Elmer). Brightness and contrast were adjusted but no nonlinear adjustment (gamma settings) was allowed.

Longevity assay

Male flies were collected and maintained at 25 °C with a transfer for fresh food every 2–3 days. The number of dead flies was recorded on a daily basis. The survival rate was calculated by the percentage of total flies surviving.

Behavior assays

To avoid the impact of CO₂ anesthesia, flies were grouped into individual vials (10 per vial, 3 vials per age for the climbing assay) or were individually placed into a vial (10 flies per age for the bang sensitivity test) one day before the test. In the climbing test, the 10 flies were transferred into a long glass tube by tapping down from the original vials. The tube was gently tapped down on the bench top to send the flies to the bottom of the tube. The flies were allowed to climb up the tube and the time that took them to reach a predefined height was measured. If it took more than 60 s for the flies to reach the desired height, the time was arbitrarily scored 60 s. Three groups of 10 flies for each individual genotype at a desired age (three biological replicates) were tested. In the bang sensitivity test, flies were transferred into an empty vial individually by tapping down from the original vials. The vial was then vortexed (maximal setting) for 10 s to paralyze the fly temporarily. The time for each fly to recover (right itself and resume normal behavior) was recorded. Any trial in which the fly remained stunned for more than 60 s was terminated and recorded as 60 s. Ten flies were individually tested for each genotype and age. The age-matched w¹¹¹⁸, da-gal4/+ and dYME1L^del; da-gal4/+ flies were used as control for dYME1L^del flies, dYME1L^del; da -gal4/+ flies and dYME1L^del; da-gal4/UAS construct flies, respectively. Experiments on the flies at the same age after eclosion were performed at the same time, to minimize variability contributed by environmental factors.

Viability assay

The crosses between predefined genotypes were performed and the expected ratio of different genotypes in the progeny was calculated. Seven days later, the parental flies were removed. When the progeny eclosed, the number of individual genotypes were counted for 10 days. The real ratio was calculated. The viability was expressed as the percentage of the real ratio to the expected ratio. Three individual experiments were set up for each cross.

Immunobloting

Whole flies were collected and homogenized in RIPA buffer supplemented with protease inhibitors, Complete Mini Protease inhibitor cocktail tablets (Roche, Penzberg, Germany), and phosphatase inhibitors, PhosSTOP EASYpack (Roche). Total protein was assayed by the BCA method (Thermo Scientific, Rockford, IL, USA). Equal amount of total protein was subjected to Tris-glycine SDS-PAGE and subsequently transferred to PVDF membrane. The membrane was blocked by 5% non-fat milk TBST (0.1% Tween-20 in TBS) or SuperBlock blocking buffer (Thermo Scientific). Specific antibodies were used to detect NDUFS3 (17D95; Abcam, Cambridge, MA, USA), Cox4 (ab16056; Abcam), ATP5α (15H4C4; Abcam), actin (C4; Millipore, Billerica, MA, USA), HSP60 (D307; Cell Signaling), phospho-Drosophila AKT (Ser505) (4054; Cell Signaling), AKT (9272; Cell Signaling), eIF2α (ab26197; Abcam), phospho-eIF2α (ab32157; Abcam), β-tubulin (E7; Hybridoma Bank, Iowa City, IA, USA), ubiquitin (3933; Cell Signaling), BiP (PAB2462; Abnova, Iowa City, IA, USA), dOmi (Emad Alnemri, Philadelphia, PA, USA)⁶³ and ATG8a (Gabor Juhasz, Budapest, Hungary).⁶⁴ We used myc antibody (9E10; Hybridoma Bank) to detect the overexpression of myc-tagged dOPA1 (dOPA1-myc). HRP-conjugated anti-mouse Ig and anti-rabbit IgG secondary antibodies were used (GE Healthcare, Pittsburgh, PA, USA). SuperSignal West Pico or Dura Chemiluminescent Substrate (Thermo Scientific) was used as a developer and ImageQuant LAS4000 (GE) was used for imaging. Densitometry quantification was performed using the 1D gel analysis function of ImageQuant TL software (GE).

Mitochondrial isolation and complex activity assay

Mitochondria were isolated following a published protocol.⁵⁶ Briefly, whole flies were homogenized in mitochondrial isolation buffer (5 mM HEPES, 210 mM d-mannitol, 70 mM sucrose and 1 mM EGTA). The lysate was subjected to centrifugation at 3000 × g for 10 min at 4 °C. The supernatant was decanted to a new microcentrifuge tube and centrifuged at 8000 × g for 20 min at 4 °C. The supernatant (cytoplasmic fraction) was discarded and the pellet was resuspended and washed with mitochondrial isolation buffer. The resuspension was again subjected to centrifugation at 8000 × g for 20 min at 4 °C. The supernatant was discarded and the mitochondrion-enriched pellet was suspended in the assay buffer (250 mM sucrose, 2 mM EDTA and 100 mM Tris-HCl, pH 7.4).

Complex I activity was assayed in a spectrophotometer at 340 nm in a reaction mixture of 25 mM potassium phosphate, pH 7.5, 0.2 mM NADH and 1.7 mM potassium ferricyanide. Complex II plus III (succinate-cytochrome c oxidoreductase) activity was assayed at 550 nm in a reaction mixture of 25 mM potassium phosphate, pH 7.5, 2 mM potassium cyanide, 80 μM cytochrome c and 5 mM succinate.⁶⁵ Complex IV (cytochrome c oxidase, COX) activity was assayed in 10 mM potassium phosphate and 25 μM reduced cytochrome c.⁶⁶ The enzymatic kinetics was measured with the SWIFT II Wavescan software (GE healthcare) and the slope (velocity) was recorded. All the activities were normalized to total protein assayed. Then, all the samples were further normalized to the samples of the young wild-type flies (3 days old).

Aconitase assay

Mitochondrial aconitase activity of the whole flies was assayed with a Colorimetric Kit (Sigma, St. Louis, MO, USA) according to the manufacturer’s instructions (MAK051; Sigma). Biological triplicates were assayed per genotype per age. All the activities were normalized to total protein assayed. All samples were then normalized to the age-matched wild-type flies.

Analysis of retinal degeneration

Male flies of predefined genotypes were reared at 25 °C under a 12:12 light-dark cycle. Flies were killed at different ages (5 per group) and the heads were cut off before counting the mean number of rhabdomeres per ommatidium using the optical neutralization technique.⁶⁷ White-eyed flies were crossed with Canton S to make the genotypes to test in a red-eyed background. Data for each individual fly was based on examination of more than 100 ommatidia.

Transmission electron microscopy

Tissues were dissected out in fixative (2.5% glutaraldehyde, 1% PFA, 0.12 M sodium cacodylate buffer, pH 7.4, with the addition of 0.1% dishwashing detergent). Then, they were collected into microporous specimen capsules and fixed in the fixation buffer. The samples were held in a fixative without a detergent overnight at 4 °C. The muscles were postfixed in 1% OsO₄, en bloc stained with 1% uranyl acetate, dehydrated in an ethanol series, embedded in Embed 812 resin (Electron Microscopy Sciences, Hatfield, PA, USA) and cured at 60 °C for 2 days. Sections were stained with uranyl acetate and lead citrate and examined with a JEM 1400 electron microscope (JEOL, Pleasanton, CA, USA) equipped with an AMT XR-111 digital camera (Advanced Microscopy Techniques, Wobrun, MA, USA).

Paraquat resistance test

We used dYME1L^del flies as mutant flies and dYME1L^del/+ flies as the wild-type (control) flies. Flies (100 flies per group) were starved for 6 h in vials supplemented with only filter paper soaked with 5% sucrose solution. Later, the flies were transferred to vials with only filter paper soaked with 10 mM paraquat in 5% sucrose solution or filter paper soaked with 5% sucrose solution overnight. Then, flies were transferred to fresh vials. Survival was determined on a daily basis for the subsequent 10 days.

NAC treatment

To test whether NAC could improve the lifespan of dYME1L^del flies, we first titrated the dosage on wild-type flies. A drop of NAC solution (50 μl) with different concentrations (0, 1, 10 and 100 μM) was dropped on the fresh food a day before these vials were used to replace the old vials. Longevity assay was performed on dYME1L^del flies treated with 1 μM NAC.

Transcriptional level analysis

Total RNA from Drosophila tissues was isolated using TRIzol reagent (Ambion, Carlsbad, CA, USA). The isolated RNA was treated with RQ1 RNase-free DNase (Promega, Madison, WI, USA) at 37 °C for 15 min. Then, the treated RNA was cleaned up using RNeasy Mini Kit (Qiagen, Valencia, CA, USA). RevertAid First Strand cDNA Synthesis Kit (Thermo Scientific) was used for first-strand synthesis. One-step quantitative real-time PCR was performed on 7900HT Fast Real-Time PCR system by using SYBR GreenER qPCR SuperMix (Life Technologies).

dOmi (CG8464) was detected by dOmi set (forward primer dOmi-RT-L: 5′-TCGGCACGGACAAGATTACG-3′; reverse primer dOmi-RT-R: 5′-ACGGCTTCAATGGTGTTGGG-3′). dArk (CG6829) was detected by dArk set (forward primer dArk-RT-L: 5′-TTTGAAGAAGCCCATAAACAAGTG-3′; reverse primer dArk-RT-R: 5′-TGGAAACGACCGTGGTTAGTAA-3′). Dronc (CG8091) was detected by Dronc set (forward primer NC-RT-L: 5′-TACCGAGTGTTTCGTAATGG-3′; reverse primer NC-RT-R: 5′-GGAAATGGTCCTTGATCTTC-3′). Reaper (CG4319) was detected by Reaper set (forward primer Reaper-RT-L: 5′-AGCAGAAGGAGCAGCAGATC-3′; reverse primer Reaper-RT-R: 5′-ATGGCTTGCGATATTTGCCG-3′). Hid (CG5123) was detected by Hid set (forward primer Hid-RT-L: 5′-TCATCTTCGTCCTCCGCATC; reverse primer Hid-RT-R: 5′-CGTGAAATTGCAAGAGGGGC). Grim (CG4345) was detected by Grim set (forward primer Grim-RT-L: 5′-CCAAGGACAACTGCAACAGC-3′; reverse primer Grim-RT-R: 5′-CGCTGCTGATCTCGAAGGAT-3′). β-actin (forward primer actin-RT-L: 5′-GCCCATCTACGAGGGTTATGC-3′; reverse primer actin-RT-R: 5′-CAAATCGCGACCAGCCAG-3′) or rp49 (forward primer rp49-RT-L: 5′-GCAAGCCCAAGGGTATCGA-3′; reverse primer rp49-RT-R: 5′-TAACCGATGTTGGGCATCAG-3′) was set up as the internal control. Expression levels of the individual genes were normalized to the level of the internal control.

Statistics

The data was shown as the mean±S.D. of at least three independent experiments unless specified differently. Significance of differences was calculated by the two-tailed Student’s t-test using the T Test function provided by Microsoft Excel. For the survival and lifespan experiments, log-rank analysis was performed by virtue of the built-in function of survival in GraphPad Prism 6 (La Jolla, CA, USA).