Autophagy inhibition rescues structural and functional defects caused by the loss of mitochondrial chaperone Hsc70-5 in Drosophila

ABSTRACT We investigated in larval and adult Drosophila models whether loss of the mitochondrial chaperone Hsc70-5 is sufficient to cause pathological alterations commonly observed in Parkinson disease. At affected larval neuromuscular junctions, no effects on terminal size, bouton size or number, synapse size, or number were observed, suggesting that we studied an early stage of pathogenesis. At this stage, we noted a loss of synaptic vesicle proteins and active zone components, delayed synapse maturation, reduced evoked and spontaneous excitatory junctional potentials, increased synaptic fatigue, and cytoskeleton rearrangements. The adult model displayed ATP depletion, altered body posture, and susceptibility to heat-induced paralysis. Adult phenotypes could be suppressed by knockdown of dj-1β, Lrrk, DCTN2-p50, DCTN1-p150, Atg1, Atg101, Atg5, Atg7, and Atg12. The knockdown of components of the macroautophagy/autophagy machinery or overexpression of human HSPA9 broadly rescued larval and adult phenotypes, while disease-associated HSPA9 variants did not. Overexpression of Pink1 or promotion of autophagy exacerbated defects. Abbreviations: AEL: after egg laying; AZ: active zone; brp: bruchpilot; Csp: cysteine string protein; dlg: discs large; eEJPs: evoked excitatory junctional potentials; GluR: glutamate receptor; H2O2: hydrogen peroxide; mEJP: miniature excitatory junctional potentials; MT: microtubule; NMJ: neuromuscular junction; PD: Parkinson disease; Pink1: PTEN-induced putative kinase 1; PSD: postsynaptic density; SSR: subsynaptic reticulum; SV: synaptic vesicle; VGlut: vesicular glutamate transporter.


Introduction
Parkinson disease (PD), the most prevalent movement disorder and the second most prevalent neurodegenerative disease, is characterized by resting tremor, stiffness, and slowness of movement. Analysis of genetic and environmental factors contributing to PD suggests that impairments in mitochondrial function, lysosomal degradation pathways, and synaptic transmission are of central importance to pathogenesis and progression of PD [1][2][3][4]. Due to their complex morphology and high-energy demands, neurons in the adult brain are particularly susceptible to dysregulation of mitochondrial quality control systems [5]. These quality control systems operate at different levels protecting cells and tissues from dysfunctional mitochondria. While a significant proportion of PD cases are sporadic, mutations in at least 11 genes have been implicated in monogenic typical or atypical forms of parkinsonism [3], providing crucial insights into the cellular and molecular pathways involved in PD. In all cases, the detailed molecular mechanisms leading to disease development have not been fully elucidated. However, a number of PD-associated genes have been implicated in mitochondrial function, altered mitochondrial dynamics, or the accumulation of dysfunctional mitochondria, which are characteristic features of PD [6][7][8][9][10][11]. For example, the PD-associated genes Pink1 (PTEN-induced kinase 1) and park are important regulators of mitochondrial quality control and mitophagy, a form of macroautophagy [6][7][8]. Mitochondrial dysfunctions have also been implicated in various other neurodegenerative diseases such as Alzheimer disease where mitochondria are key targets of both Aβ42 and tau toxicity [12].
We have previously addressed the importance of mitochondrial quality control and selective vulnerability of dopaminergic neurons by generating a model for loss of function of Drosophila Hsc70-5 [13]. Enhanced mitophagy is identified as an early pathological feature caused by knockdown in a presymptomatic model of loss of Hsc70-5 function [13] prior to the emergence of locomotion defects. Consistent with the upregulation of mitophagy upon loss of Hsc70-5 function in a Drosophila, human fibroblasts obtained from a carrier of the PD-associated HSPA9 A476T variant exhibit increased colocalization of mitochondria with autophagosomes.
In the current study, we investigated in detail the consequences of neuronal loss of Hsc70-5 function in vivo and examine epistatic interactions in this genetic background.

Detection of early synaptic changes due to loss of Hsc70-5 function
To further investigate structural changes in response to loss of Hsc70-5 function, we stained for synaptic vesicles (SVs) and active zone (Az) markers. To this aim, we performed confocal imaging on fixed larval NMJs. Analysis and quantification revealed that SV proteins Csp (cysteine-string protein, Figure 1C) and VGlut (vesicular glutamate transporter, Figure 1D) were reduced at NMJs of symptomatic larvae compared to control larvae. A similar reduction was observed for the central organizing component of Azs, brp (bruchpilot, Figure  1E). We observed no significant differences in presymptomatic larvae for SV proteins or Az components compared to control ( Figure 1C-E).
We proceeded to investigate synapse maturation by analyzing subsynaptic reticulum (SSR) and postsynaptic density (PSD) maturation. The SSR, a complex system of membrane tubules and lamellae, is formed by the invagination of postsynaptic membranes following presynaptic innervation. Thus, an increased percentage of terminal synaptic boutons devoid of SSR can be used as a marker for delayed synapse terminal maturation [18,19]. To examine SSR folding and morphogenesis, we labeled NMJs with antibodies against dlg and the neuronal marker hrp. dlg reliably surrounded terminal boutons in all genotypes examined, suggesting that Hsc70-5 knockdown did not affect SSR morphogenesis ( Figure 1F). Furthermore, quantification of the number, intensity, and size of glutamate receptor cluster per synapse revealed no postsynaptic phenotypes ( Figure 1G).
We next assessed defects in presynaptic maturation by quantifying the percentage of PSDs unopposed by Azs. To this aim, we analyzed the fraction of PSDs apposed by Az component brp that reliably localizes to mature synapses. Az apposed to newly forming PSDs usually recruit detectable brp puncta within 2-4 h of their formation and more than 95% of all glutamatergic synapses are apposed by brp [20]. A higher percentage of brp-negative synapses is indicative of neurodegenerative or neurodevelopmental defects [21]. elav>Hsc70-5 KK100233 larvae displayed defects in the apposition of PSDs and Azs. Twice as many unapposed, putatively immature synapses were detected ( Figure 1G). This defect might be caused by impairments in presynaptic maturation or stabilization.
Defects in microtubule (MT) cytoskeletal organization have been associated with PD models [22,23] in particular, and with loss of synaptic mitochondria in general [24,25]. The Drosophila homolog of the LRRK2 (leucine-rich repeat kinase 2), linked to familial and sporadic PD, controls MT stability at Drosophila NMJs by suppressing futsch function [23]. Additionally, a decreased number of MT loops, which is indicative of stable MT, has been observed [24]. In this study, fewer MT loops were detected in terminal boutons at the NMJs of elav>Hsc70-5 KK100233 larvae (Figure 2A and S3). In addition to defective MT morphology, we observed a reduction in the percentage of boutons connected to the stable MT network (Fig. S3). This impaired long-range intracellular transport. Next, we employed the synaptic footprint assay and examined neuronal membranes by labeling hrp. Synaptic footprints are biomarkers for late-stage synapse retraction [26], while hrp inhomogeneity has been associated with early-stage synapse disassembly [25]. We visualized the presynaptic compartment by labeling for the membrane marker hrp and the SV marker VGlut. Neither synaptic footprints nor hrp inhomogeneity was detected in response to Hsc70-5 knockdown ( Figure 2B) suggesting that elav>Hsc70-

Functional consequences of loss of Hsc70-5 function
Next, we investigated whether the aforementioned morphological changes affect synaptic function. To examine whether symptomatic elav>Hsc70-5 KK100233 larvae have impaired synaptic transmission, we recorded evoked excitatory junctional potentials (eEJPs) by using current-clamp recording at NMJ from muscle 6 in segment A5. The amplitude of eEJPs in elav>Hsc70-5 KK100233 was reduced compared to control ( Figure 3A). We also recorded spontaneous events ( Figure  3B and Figure 3C) and the response of nerve terminals during and following high-frequency stimulations ( Figure 3D-F). We noted a decrease in the amplitude of miniature excitatory junctional potential (mEJP) and an increase in their frequency in elav>Hsc70-5 KK100233 larvae compared to control. The ratio of eEJP to mEJP or quantal content was reduced in mutants compared to control ( Figure 3C). 10 Hz stimulation at mutant terminals caused no changes in EJP amplitude during a 10 Hz stimulation paradigm ( Figure 3D and Figure 3E). However, a drastic time-dependent increase in failure in mutant compared to control was observed ( Figure 3D and Figure 3F).

Downregulation of autophagy was protective
ATP and proper mitochondrial function are important to maintain synaptic transmission at increased temperatures [29]. Mutants with compromised mitochondrial function paralyze faster compared to controls. We thus performed a candidate-based screen to identify modifiers of Hsc70-5 loss of function ( Figure  5A). Downregulation of synaptic proteins involved in neurotransmission and endocytic machinery enhanced heat-stress induced paralysis observed upon loss of Hsc70-5. Knockdown of Parkinson disease-associated dj-1βand Lrrk, as well as components of retrograde transport machinery, were suppressors of Hsc70-5 knockdown-induced paralysis.
We identified five members of the autophagy machinery Atg1, Atg5, Atg7, Atg12, and Atg101 as suppressors of Hsc70-5 loss of function induced paralysis ( Figure 5A). To verify the RNAi efficiency of modifiers isolated in the screen, we tested the Atg1, Atg5, Atg7, Atg12, and Atg101 RNAi strains for their ability to inhibit starvation-induced autophagy ( Figure 5B). Overexpression of RNAi against Atg1, Atg5, Atg7, Atg12, and Atg101 in the larval fat body using a mosaic system [34,35] for genetic analysis in cells expressing GFP completely suppressed starvation-induced autophagy ( Figure 5B). After validation of these constructs, we tested if knockdown of autophagy-related genes could ameliorate phenotype caused by loss of Hsc70-5. We used UAS-RNAi strains against lmd and w (white) with Hsc70-5 KK100233 (Hsc70-5 KK100233 + w-RNAi and Hsc70-5 KK100233 + lmd-RNAi) as controls to compare larvae bearing Hsc70-5 KK100233 in combination with a second UAS site expressing RNAi against autophagic components. Panneuronal expression of RNAi strains targeting components of the autophagic machinery, Atg1, Atg5, Atg7, Atg12, and Atg101 in the elav>Hsc70-5 KK100233 background restored the righting reflex to control levels ( Figure 5C). Next, we extended the analysis to flies. As previously demonstrated, 4-d-old elav>Hsc70-5 KK100233 ,tub-GAL80 ts symptomatic flies displayed severe defects in climbing and wing posture ( Figure 5D and Figure 5E) [14]. Knockdown of autophagic components reversed impairments in locomotion, wing posture and ATP levels ( Figure 5D-F) induced by Hsc70-5 knockdown in young flies. Knockdown of autophagy genes alone did not reveal any significant differences compared to controls ( Figure 5C-F). These results suggested that inhibiting autophagy was sufficient to rescue Hsc70-5 knockdown associated impairments in larval and adult symptomatic models.

Autophagy induction in symptomatic flies exacerbated Hsc70-5 knockdown induced defects.
Loss of Hsc70-5 function has been associated with increased mitophagy in Drosophila and human fibroblasts [14]. Consistent with a protective role for autophagy, PRKN overexpression is sufficient to rescue alterations in mitochondrial morphology and increased apoptosis [36] caused by HSPA9 silencing in HeLa cells and the dopaminergic SH-SY5Y cell line. To address whether promoting autophagy was detrimental or protective in vivo, we at first modulated autophagic flux genetically. Quantification of larval locomotion revealed that panneuronal overexpression of Atg1 at 25°C was sufficient to cause a sluggish righting reflex. Concomitant overexpression of elav>Hsc70-5 KK100233 did not further exacerbate this phenotype ( Figure 6A). Using the Gal4/Gal80 system, we investigated the effect of Atg1 overexpression on Hsc70-5 knockdown related phenotypes in flies by performing the longevity assay ( Figure 6B). Knockdown of Hsc70-5 caused a significant reduction in median and maximum life expectancy compared to control flies. Atg1 overexpression in the control background enhanced median but not maximum survival. Notably, concomitant overexpression of Atg1 with Hsc70-5 KK100233 caused a reduction in both median and maximum lifespan compared to Hsc70-5 KK100233 alone ( Figure 6B). Atg1 overexpression in the elav>Hsc70-5 KK100233 ,tub-GAL80 ts background exacerbated the loss of Hsc70-5-induced climbing impairment ( Figure 6C); however, a slight improvement was observed in wing phenotypes ( Figure 6D). Atg1 overexpression alone did not impair adult locomotion of 4-d-old flies examined using the climbing ability nor induce abnormal wing phenotypes ( Figure 6C and Figure 6D). These results suggested that overexpression of Atg1 did not modify Hsc70-5 knockdown induced defects at the larval stage but reduced lifespan and climbing ability in flies.
Rapamycin has been used to induce autophagy in Drosophila [37]. Feeding rapamycin was sufficient to enhance autophagy as demonstrated by increased accumulation of autophagy markers Atg8a and relative levels of Atg8a-II, in addition to reduced levels of ref(2)P, homolog of mammalian SQSTM1/ p62 ( Figure 6E and Figure 6F). Hsc70-5 knockdown caused the same effects in fly heads ( Figure 6E and Figure 6F). We investigated whether supplementing rapamycin could further enhance autophagy caused by Hsc70-5 knockdown. Quantification of western blots showed a further increase in the accumulation of total Atg8a and relative levels of Atg8a-II with concomitant reduction of ref(2)P ( Figure 6E and Figure  6F). Rapamycin treatment prolonged lifespan in control flies but caused a reduction in elav>Hsc70-5 KK100233 ,tub-GAL80 ts flies ( Figure 6G). Moreover, rapamycin treatment exacerbated the climbing impairment and wing phenotypes in symptomatic flies ( Figure 6H and Figure 6I). PINK1 has been implicated in familial PD. It regulates the degradation of old and dysfunctional mitochondria in health and upon exposure to the mitochondrial uncoupler CCCP [6][7][8]38,39]. Hence, we examined if Pink1 overexpression could modulate the loss of Hsc70-5 phenotypes. Pink1 overexpression was sufficient to induce a sluggish righting reflect at the larval stage ( Figure 6J). In elav>Hsc70-5 KK100233 larvae it rescued locomotion defects ( Figure  6J). Pink1 overexpression in the elav>Hsc70-5 KK100233 ,tub-GAL80 ts genetic background reduced lifespan ( Figure 6K) and exacerbated the climbing impairment and wing phenotypes in symptomatic adult flies ( Figure 6L and Figure 6M).

Atg1 knockdown rescued synaptic mitochondrial loss and synaptic defects observed in symptomatic larvae.
Atg1 is required for the initiation of autophagosome formation [40]. We thus characterized the cellular changes caused by modulation of Atg1 expression in the elav>Hsc70-5 KK100233 background. Atg1 knockdown reversed the loss of mitochondria observed at the NMJs of elav>Hsc70-5 KK100233 larvae ( Figure 7A) and rescued the decrease in mitochondria area fraction, mitochondrial number, size, and altered morphology at the NMJ in elav>Hsc70-5 KK100233 larvae ( Figure 7B).
NMJs of early symptomatic elav>Hsc70-5 KK100233 larvae were characterized by reduced presence of SV marker proteins and Az components, impaired synapse maturation, and alterations in the MT cytoskeleton. We next assessed MT abundance as a marker for synapse stability. Atg1 knockdown restored usual MT abundance in elav>Hsc70-5 KK100233 larvae ( Figure 7C). Furthermore, Atg1 knockdown in elav>Hsc70-5 KK100233 larvae alleviated impairments in VGlut ( Figure 8A and Figure 8B) and brp abundance ( Figure 8C and Figure 8D), and synapse maturation impairment ( Figure 8E and Figure 8F). Overexpression of Atg1 failed to rescue defects in MT abundance ( Figure 7C), SV proteins abundance ( Figure  8A and Figure 8B), and synaptic maturation defects caused by loss of Hsc70-5 at the NMJ ( Figure 8E and Figure 8F). These findings indicated that restoring mitochondrial mass by Atg1 knockdown could rescue loss of Hsc70-5 induced synaptic defects.

Autophagy suppression was protective against oxidative stress
Next, we tested whether knockdown of Hsc70-5 enhances vulnerability to oxidative stress. We treated young flies with hydrogen peroxide (H 2 O 2 ). H 2 O 2 induces generalized oxidative stress which is less specific than the effects of the mitochondrial complex inhibitors like rotenone or paraquat [41]. dj-1β and Lrrk are crucial for protection against oxidative stress [41,42]. H 2 O 2 has also been shown to reduce the lifespan of flies that display a reduced function of dj-1β and Lrrk. elav>Hsc70-5 KK100233 ,tub-GAL80 ts flies were more vulnerable to H 2 O 2 treatment than control flies ( Figure 9A). While Atg1 knockdown caused a minor reduction of lifespan upon exposure to oxidative stress compared to control, it proved to be effective in restoring the diminished stress resistance in the elav>Hsc70-5 KK100233 ,tub-GAL80 ts background ( Figure  9A). We thus concluded that suppression of autophagy in the elav>Hsc70-5 KK100233 ,tub-GAL80 ts background might be beneficial for survival under oxidative stress.

Autophagy suppression improved healthspan but did not extend lifespan
To investigate the long-term impact of Atg1 knockdown, we examined the lifespan of flies under normal conditions ( Figure 9B). Atg1 knockdown led to a reduction of the lifespan in elav>Hsc70-5 KK100233 ,tub-GAL80 ts flies. As a minor decrease in lifespan had been observed, we thought to assess the effect of Atg1 knockdown at various pathological stages. Atg1 knockdown was beneficial in 4-d-old symptomatic flies and improved locomotion. In 10-d-old flies, concomitant knockdown resulted in impairment in climbing using a less challenging climbing assay ( Figure 9C). We referred to these 10-d-old flies as late-symptomatic flies (Table S1). We concluded that reduced autophagy flux was beneficial for improving locomotion ability in Hsc70-5 reduced background in young flies, but impaired climbing ability and lifespan in elav>Hsc70-5 KK100233 ,tub-GAL80 ts flies in the long run.

Investigating the impact of rare Hsc70-5 mutations in vivo.
Three mutations have been identified in HSPA9 in a congenital disease termed epiphyseal, vertebral, ear, nose, plus associated finding syndrome or EVEN-PLUS [28]. Biallelic mutations are identified in three individuals from two families, including the previously reported R126W variant [17]. Functional studies investigating the impact of variant HSPA9 P509S reveals a lower ATP catalysis rate and reduced chaperoning activity compared to WT HSPA9 [17]. Next, morphological studies in human fibroblasts from a heterozygous carrier of variant HSPA9 A476T shows mitochondrial defects compared to WT HSPA9 homozygous sibling [17]. Another study investigates the effects of human variant R126W, and P509S in yeast by generating analogous substitutions in the Saccharomyces cerevisiae ortholog of Hsc70-5, the SSC1 gene [16]. The substrate-binding domain mutation SSC1 A453T is shown to cause mitochondrial dysfunction, enhanced ROS levels, and reduces ability to prevent aggregate formation of unfolded substrates [16].
In this study, we used a Drosophila model of loss of Hsc70-5 function to test the functional relevance of rare HSPA9 mutations HSPA9 R126W , HSPA9 A476T, andHSPA9 P509S in vivo. Overexpression of human HSPA9 and Drosophila Hsc70-5, in a genetic background with reduced Hsc70-5, unlike mutant variants HSPA9 R126W , HSPA9 A476T andHSPA9 P509S , were able to rescue the loss of synaptic mitochondria, locomotion defects, abnormal wing phenotype, and reduced ATP levels. This study provided evidence that the investigated amino acid replacements in the ATPase and substrate-binding domain impaired HSPA9 function. In vitro overexpression studies have suggested no difference in overexpression levels or localization of the mutant variants [17] and future studies should address this in in vivo setting.

Hsc70-5 knockdown induced synaptic defects.
Neuronal mitochondria have been implicated in diverse functions including spine formation [43], synaptic plasticity [44], and axonal branching [45]. Loss of Miro dramatically excludes mitochondria from distal compartments and causes a gradual timedependent reduction in EJP amplitude following 10 Hz stimulation, although no changes are observed in EJP or mEJP amplitude under baseline conditions [24]. Similar findings are reported in Drp1 mutants where dramatic synaptic mitochondria reduction is observed [46].
Loss of neuronal mitochondria as a consequence of Hsc70-5 silencing caused a reduction in EJP and mEJP amplitudes and quantal content ( Figure 3A-C). There was an increase in the frequency of mEJP events ( Figure 3C), possibly due to the elevation of intracellular calcium levels following loss of Hsc70-5. More frequent synaptic failure was observed upon stimulation at 10 Hz ( Figure 3F). Failed responses that are caused by the depletion of synaptic vesicles are preceded by a gradual decline in EJP amplitude [46]. However, in our case, the amplitude of the remaining EJPs was not affected. Thus, failures may instead be caused by impaired propagation of action potentials as described for Atpα mutant [32].
Cellular analysis of larval NMJ in Hsc70-5 knockdown animals did not reveal gross impairments in the morphology of distal compartments (Fig. S2) or inhomogeneity of presynaptic membrane or synaptic footprints ( Figure 3B) that are biomarkers for early and late stages of synapse disassembly, respectively [25,26]. Importantly, defects in the availability of synaptic components, such as Csp, VGlut, and brp (Figure 1), were observed at a stage in which no major neurodegeneration is noted [47]. Hence, the protective effects of blocking mitophagy described in this study were likely related to mitochondrial function rather than susceptibility to cell death. At this stage of disease progression, RNAi against Atg1 was sufficient to reverse defects in development, regulation of synaptic vesicles, synaptic terminal stability, and function ( Figures. 7 and Figure 8). Furthermore, Atg1-RNAi restored mitochondria abundance. In addition, it also reversed the only putatively degenerative change we found in Hsc70-5 knockdown larvae: MT cytoskeletal alterations ( Figure 7B).

Epistatic interaction of Hsc70-5 with the autophagic machinery.
Mitophagy, a type of macroautophagy that targets specifically mitochondria, is increased upon partial loss of Hsc70-5 function in Drosophila and HSPA9 function in human fibroblasts [14]. Increased autophagic flux can be likened to a double- edged sword that protects neurons from chronic oxidative stress but accelerate mitochondrial loss under conditions of premature mitophagy or impaired mitochondrial biogenesis [13,48]. The beneficial effects of Park overexpression in mouse embryonic fibroblasts treated with siRNA against Hspa9 are dependent on intact autophagic machinery, suggesting a potential therapeutic benefit of upregulating autophagy [17]. However, the interpretation of suppressed apoptosis in HEK293 cells and tumor-derived SH-SY5Y cells is complicated by the fact that (like most tumor-derived culture cells) these cells primarily generate ATP via aerobic glycolysis and are therefore not dependent on oxidative phosphorylation [49]. Thus, the balance between costs and benefits of effective mitochondria removal may differ in vivo.
We demonstrate that Hsc70-5 knockdown in Drosophila causes a severe loss of synaptic mitochondria and cellular ATP depletion [13]. The Drosophila NMJ has proved to be a particularly useful model as it is easily accessible and morphologically complex. These characteristics allow us to model a highly active synaptic population that is vulnerable to impairment in intracellular trafficking because it is located very distantly to the neuronal soma. The in vivo data obtained in this study did not provide evidence for the therapeutic benefit of promoting autophagic flux following impaired mitochondrial function. Atg1 overexpression, which is sufficient to induce autophagy in Drosophila [50] did not rescue impaired locomotion or mitochondrial mass following Hsc70-5 knockdown in the symptomatic larval model ( Figure 7A). Besides, Atg1 overexpression did not reverse alterations in synaptic development at the larval NMJ ( Figure 8). In symptomatic flies, Atg1 overexpression exacerbated climbing defects and shortened lifespan caused by loss of Hsc70-5. The inability of Atg1 to exacerbate Hsc70-5 knockdownassociated mitochondrial abundance or cellular defects at the larval stage might be explained by the fact that overexpression of Atg1 alone did not reduce mitochondrial abundance ( Figure 7A). It was also possible that the loss of mitochondria upon Hsc70-5 knockdown was so extreme that further Atg1 overexpression was unable to exacerbate the already severe defect (Figures. 7 and Figure 8).
The analysis of Atg1 overexpression in larval stages was complicated by the fact that Atg1 overexpression alone caused a sluggish righting phenotype ( Figure 6A). Pink1 overexpression alone also caused a sluggish righting phenotype in larvae and reduced life expectancy in adult flies ( Figure 6). Nevertheless, both Atg1 and Pink1 overexpression exacerbated loss of Hsc70-5-associated defects in locomotion and longevity in adult flies. Future studies need to address the interaction between Hsc70-5 and regulators of mitochondrial quality control such as Pink1 and park and proteins involved in mitochondrial dynamics in more detail.
Using a functional genetic screen, we identified that knockdown of autophagy-related genes in elav>Hsc70-5-KK100233 ,tub-GAL80 ts flies rescued climbing defects, abnormal wing posture, and ATP levels in 4-d-old flies ( Figure  5B). However, knockdown of autophagy was detrimental in late-symptomatic 10-d-old flies.
How can these data be reconciled with reports suggesting that increased autophagy may be of therapeutic potential in PD [51]? Mitochondrial quality control mechanisms are very divergent. In particular, the molecular mechanisms underlying mitochondrial stress responses in Drosophila and mammals differ. While neither prkn nor pink1 knockout mice display noticeable behavioral or morphological changes under baseline conditions, loss of park or Pink1 in Drosophila is detrimental and results in decreased lifespan and apoptotic flight-muscle degeneration [6,7,52,53]. The mammalian system may be equipped with more elaborate compensatory mechanisms and display more functional redundancy, whereas flies are particularly vulnerable to overactivation of mitophagy as evident by the extreme depletion of mitochondria from synapses ( Figure  1B). Similar pathological induction of autophagy has been reported for hypoxic-ischemic brain injury and two toxinbased PD models in which blockade of autophagy proved to be protective [54].

Stimulation of autophagy under conditions of oxidative stress
Oxidative and nitrosative stress is associated with various neurodegenerative diseases, including PD [55,56]. Pathogenic mechanisms include excitotoxicity, endoplasmic reticulum stress, protein aggregation, and damage to mitochondria, which are generators and targets of reactive oxygen species (ROS). Flies affected by Hsc70-5 loss displayed increased vulnerability to oxidative stress ( Figure 9A). These effects may be mediated through the HSPA9 interaction partner PARK7 [57]. PARK7 associated with autosomal-recessive early-onset familial PD is important for response against oxidative stress [58,59]. Loss of PARK7 renders cells vulnerable to oxidative stress [57,60,61]. Notably, the interaction between HSPA9 and PARK7 is vital for the control of oxidative stress in hematopoietic stem cells [57].
Cross-talk between autophagy and ROS/nitrosative stress in the context of cell signaling and pathological protein damage is considered a significant obstacle for the therapeutic modulation of autophagy [62,63]. Thus, it is noteworthy that suppression of autophagy was beneficial in the context of combined mitochondrial dysfunction and oxidative stress in the symptomatic adult model ( Figures. 7 and Figure 9) presented in this study. The in vivo evidence presented here suggested that reduced rates of autophagy might be protective for neurons compromised by pathologically increased levels of mitophagy. However, we observed that detrimental side effects exceeded protective benefits in the long-term (Figure 9).

Fly culture conditions
Flies were raised on standard cornmeal/agar medium. To circumvent pupal lethality caused by Hsc70-5 knockdown and analyze behavioral defects in adult flies, we utilized the Gal4/ Gal80 system (elav>Hsc70-5 KK100233 ,tub-GAL80 ts ). This allowed us to achieve late-onset conditional knockdown by raising larvae at 18°C before transferring them to 25°C at 5 d AEL. Flies were kept at 25°C during development for analysis of wing phenotype, climbing defects, longevity, ATP levels from heads, and temperature-induced paralysis at appropriate ages.

Morphological analysis
Analyses of mitochondria and NMJs were performed as previously described [14]. The area fraction occupied by mitochondria was used as the index of mitochondrial mass. To quantify futsch loops, the number of loops located to the two most distal boutons at each terminal of the NMJ was scored. The total number of loops in these regions was normalized to the number of terminals.

Fat body assay for investigating autophagy
Assay to validate functionality of UAS-Atg1 overexpression or RNAi constructs against autophagy-related proteins were performed using a mosaic genetic analysis in larval fat body cells essentially as previously described [34,35]. Starvation of 2 nd instar stage larvae was performed for 5-6 h in fresh empty vials on a filter paper soaked with H 2 O. For imaging, 2 nd instar larvae that were well fed or starved for 4-5 h were cut open and turned inside out like a "sock" before removal of fat bodies. Fat body tissues were stained with DAPI for 2 min, briefly washed in PBS, and mounted on a glass slide using Vectashield and imaged immediately using a Leica SP5 II confocal imaging system (Leica Camera AG, Wetzlar, Germany) using a 40× Plan-Apochromat 1.4 N.A. oil objective. The entire procedure from fat body isolation to imaging was completed within 30 min.

ATP measurements
ATP levels in head homogenates were measured and normalized using a luciferase-based bioluminescence assay as described earlier [13]. Five heads of female flies were homogenized in 6 M guanidine-HCl (Sigma-Aldrich, SRE0066) and frozen in liquid nitrogen. Next, samples were boiled for 3 min, cleared by centrifugation at 14,000 g for 5 min, and diluted to measure protein concentration (1:10 diluted samples, Bradford Assay Kit; Sigma-Aldrich, B6916) and ATP level (1:2000 diluted samples, ATP Determination Kit Sensitive Assay; Biaffin GmbH & Co KG, LBR-P010). ATP levels were normalized to the protein concentration.

Locomotion analysis
Climbing assays were, unless otherwise noted, conducted as previously described [14]. Climbing male flies was monitored by analyzing their performance to climb 6 cm (challenging assay, used in Figures. 4E and Figure 5D) or 3 cm (less challenging assay used in Figure 9C for 10-d-old flies) within 14 s. A successful attempt was scored as 1, and failure to reach the top as 0. Each fly was assessed three times to calculate the average climbing score. At least 40 flies per genotype were analyzed. Larval locomotion was investigated by examining larvae crawling speed and righting assay. Larvae locomotion speed was quantified as previously described [25].

Longevity assay
Drosophila larvae were transferred from 18°C to 25°C to boost UAS-transgene expression 5 d AEL. Adult male flies were maintained at 25°C in groups of 20 or fewer. 100 flies were assayed per genotype. No anesthesia was used in survival experiments.

Rapamycin treatment
Rapamycin (LC Laboratories, 53,123-88-9) was dissolved in ethanol and added to standard fly food (Meidi LLC, V100) at appropriate concentrations (0.1 and 200 μM). For control food (0 μM), ethanol alone was added. 5 d AEL, larvae were transferred from 18°C to 25°C to induce transgene expression on 0 μM and 0.1 μM rapamycin food. Newly hatched male flies were continually maintained in 0 μM and 200 μM rapamycin food at 25°C for longevity assay.

Statistical analysis
Statistical tests were performed using PAST.exe software (http:// folk.uio.no/ohammer/past/index.html) unless otherwise noted. The standard error of mean (SEM) and standard deviation (SD) are shown as a box and black line. Data were first tested for normality by using the Shapiro-Wilk test (a = 0.05). Normally distributed data were analyzed either by Student's t-test (two groups) and Bonferroni comparison to adjust the P value or by a one-way analysis of variance followed by a Tukey-Kramer posttest for comparing multiple groups. Nonnormal distributed data were analyzed by either a Mann-Whitney test (two groups) or Bonferroni comparison to adjust P value or a Kruskal-Wallis H-test followed by a Dunn's test for comparisons between multiple groups. Statistical significance was defined as P < 0.05.