E2 ubiquitin conjugase Bendless is essential for PINK1-Park activity to regulate Mitofusin under mitochondrial stress

Cells under mitochondrial stress often co-opt mechanisms to maintain energy homeostasis, mitochondrial quality control and cell survival. A mechanistic understanding of such responses is crucial for further insight into mitochondrial biology and diseases. Through an unbiased genetic screen in Drosophila, we identify that mutations in lrpprc2, a homolog of the human LRPPRC gene that is linked to the French-Canadian Leigh syndrome, results in PINK1-Park activation. While the PINK1-Park pathway is well known to induce mitophagy, we show that in the case of lrpprc2 mutants, PINK1-Park regulates mitochondrial dynamics by inducing degradation of the mitochondrial fusion protein Mitofusin/Marf. We also discover that Bendless, a K63-linked E2 conjugase, is a regulator of Marf, as loss of bendless results in increased Marf levels. We show that Bendless is required for PINK1 stability, and subsequently for PINK1-Park mediated Marf degradation under physiological conditions, and in response to mitochondrial stress as seen in lrpprc2. Additionally, we show that loss of Bendless in lrpprc2 mutant eye results in photoreceptor degeneration, indicating a neuroprotective role for Bendless-PINK1-Park mediated Marf degradation. Based on our observations, we propose that certain forms of mitochondrial stress activate Bendless-PINK1-Park to limit mitochondrial fusion, which is a cell-protective response.


Introduction
Mitochondria are dynamic organelles and their size varies in response to various cellular cues such as developmental signaling (1), metabolic needs (2) or toxin induced stress (2). This change in mitochondrial size is crucial for cellular adaptation under different physiological conditions. For example, upon amino acid deprivation, mitochondria undergo fusion, which results in increased ATP production (3) and protects mitochondria from autophagy (4). Changes in mitochondrial size requires regulation of GTPases essential for mitochondrial dynamics. While the Dynamin 1-like (DNM1/Drp1) protein mediates fission, Mitofusins (Mfn1 and Mfn2 in mammals, Marf in Drosophila) and Optic Atrophy 1 (OPA1) mediate the fusion of mitochondrial outer and inner membranes respectively. Several post-translational modifications, such as phosphorylation, acetylation and ubiquitination are crucial for the activity of these proteins, and thereby play an important role in determining mitochondrial size (5,6). Misregulation of these proteins, and consequently, mitochondrial dynamics, is associated with metabolic and neurodegenerative diseases (7).
The E3 ubiquitin ligase Parkin (Park in Drosophila, PARK2 in humans) and the kinase PINK1, which are linked to autosomal recessive early-onset Parkinsonism, are known to regulate mitochondrial quality control (8). Studies in human cancer cell lines have shown that dissipation of the mitochondrial membrane potential (MMP) can stabilize PINK1 on the outer mitochondrial membrane (OMM) leading to Park recruitment, polyubiquitination of OMM proteins and 3 mitophagy (9)(10)(11)(12). Several in vivo studies have also shown a conserved role for PINK1-Parkin in mitophagy (13)(14)(15)(16)(17)(18)(19)(20). While PINK1-Park mediated mitophagy has been extensively studied in cells, how the PINK1-Park pathway is activated under physiological conditions in vivo remains elusive (21). Additionally, in vivo studies suggest a pro-fission role of PINK1-Park (22)(23)(24)(25)(26), perhaps through the regulation of mitofusin levels (27). As most of these studies utilize PINK1 and PARK2 mutants to study defects in mitochondrial dynamics, the mechanism by which they are regulated in vivo under various physiological conditions remains unresolved. Additionally, it is unclear as to how the PINK1-Park pathway may activate mitophagy, alter mitochondrial dynamics or selectively target certain OMM proteins in response to mitochondrial stress.
To study the regulation of Mitofusin/Marf in vivo, we undertook an unbiased genetic screen in Drosophila. From this genetic screen, we discovered that mutations in lrpprc2 (ppr), a homolog of human LRPPRC that is required for mitochondrial mRNA stability and translation (28,29), results in activation of the PINK1-Park pathway. This activation then leads to proteasome-mediated Marf degradation, but not mitophagy. We also discovered that mutations in bendless (ben), which encodes a K63-linked E2 ubiquitin conjugase, is essential for Marf degradation in ppr mutants. Further, we demonstrate an essential role for Ben in regulating the stability of PINK1, which in turn is required for maintaining steady state Marf levels in healthy cells. Finally, we show that in ppr mutants, Ben suppresses excessive mitochondrial fusion and prevents neuronal death under mitochondrial stress.

Results:
Loss of ppr results in reduced Marf levels 4 To identify novel regulators of mitochondrial dynamics, we screened a collection of Drosophila X-chromosome lethal mutations (30,31). This collection was generated to identify mutants with neurodegenerative phenotypes and has previously uncovered mutations in Marf (32) and several other genes required for mitochondrial function (29,33). We tested these mutants for misregulation of Marf protein using an HA-tagged Marf genomic construct (Marf::HA). We used the FLP-FRT mediated mitotic recombination strategy to create mutant clones (non-GFP cells) in a heterozygous background (GFP expressing cells) in the developing wing disc epithelium (34).
This allowed us to compare Marf levels in mutant and wild type cells within the same tissue ( Figure S1A-A').
From this screen, we found that mutant clones of two independent ppr alleles (ppr A and ppr E ) show reduced Marf:HA levels compared to the surrounding wild type cells ( Figure 1A-A'', 1E, Figure 1B-B'). To further confirm these results, we used an independent Marf genomic rescue line, Marf::mCherry and found reduced Marf::mCherry staining in ppr A mutant clones ( Figure   S1C-C'). To test for the possibility that the reduction in Marf::HA or Marf::mCherry is caused by reduced mitochondrial content, we checked the levels of an OMM protein Tom20 using an endogenous tagged line (Tom20::mCherry). We did not observe any change in Tom20::mCherry staining in ppr A mutant clones ( Figure 1B-B'', 1E). Taken together, these data suggest that downregulation of Marf in ppr mutants is not due to reduced mitochondrial content.
Additionally, we also looked at other proteins involved in mitochondrial dynamics -Opa1 and Drp1-using genomic tags. While we found the levels of Opa1::HA to be slightly increased in ppr A mutant clones ( Figure 1C-C", 1E), Drp1::HA levels remained unaltered ( Figure 1D-D", 1E). As mutations in ppr/LRPPRC result in mitochondrial defects due to reduced stability of 5 mtRNA (28,29,37), reduced Marf levels in ppr mutants appears to be an adaptation to segregate defective mitochondria by suppressing their fusion.
Since reduced Marf is expected to suppress mitochondrial fusion, we sought to test mitochondrial morphology in ppr A mutant clones. The cells in wing discs are very compact, and hence it is difficult to study mitochondrial morphology. Hence, we created mutant clones in the peripodial membrane, which is a squamous epithelium overlying wing discs. We used anti-Complex-V staining to mark mitochondria. Interestingly, we found that mitochondrial size is increased in these ppr A mutant clones (Figure S1E-E", 1F). Similar increase in mitochondrial size has been observed in LRPPRC knockdown in mouse liver (35) and in C.elegans (36). As many studies have shown that mitochondrial stress can result in increased mitochondrial size (3,4,36,38), we suspect a similar mechanism results in increased mitochondrial size in ppr mutant cells, while an independent mitochondrial quality control mechanism may suppress their fusion by inducing Marf turnover.

UPS dependent Marf degradation in ppr mutants
Reduced Marf levels in ppr mutant clones could be increased protein turnover via selective autophagy or ubiquitin-proteasomal system (UPS). We tested the possibility of autophagic degradation of Marf. We fed chloroquine, an inhibitor of autophagosome-lysosome fusion (39), to larvae and found that Marf::HA levels remain reduced in ppr A clones ( Figure 2B-B', 2E).
Moreover, we found that the levels of p62, a protein degraded primarily via autophagy, was not altered in ppr A clones (Figure S1D-D'). Thus we conclude that autophagy is neither enhanced nor likely the cause of Marf reduction in ppr mutant clones. 6 To investigate the role of UPS in Marf downregulation in ppr mutants, we fed larvae with the proteasomal inhibitor MG132 (40,41). While in DMSO-fed larvae, ppr A mutant clones had lower levels of Marf:HA as compared to the neighboring wild type cells, MG132-fed larvae show no change in Marf::HA levels ( Figure 2A-A', 2C-C', 2E). We further expressed a dominant negative form of Pros 6 to inhibit UPS activity (42) and tested its effect on Marf::HA levels in ppr A mutant clones. Similar to MG132 treatment, we found that Marf::HA levels were restored in ppr A mutant clones upon Pros 6 1 overexpression ( Figure 2D-D', 2E). These results suggest that UPS-mediated degradation of Marf results in Marf downregulation in ppr A mutant clones.

PINK1 and Park dependent Marf regulation in ppr mutants
Several E3 ubiquitin ligases have been linked to Mitofusin degradation. For example, Mitofusin degradation by HUWE1 occurs under genotoxic stress or on altered fat metabolism (43,44) while Mitofusin degradation by Park occurs upon mitochondrial membrane depolarization (45,46). In Drosophila too, HUWE1, MUL1 and Park have been shown to affect Marf levels (27,43,47).  Figure 3B-B', 3D). Since park is genetically downstream to Pink1 (48,49), we tested whether PINK1 is also required for Marf degradation in ppr mutant clones. We generated ppr A Pink1 5 double mutant clones and found that these clones do not show a reduction in Marf::HA levels ( Figure 3C-C', 3D), 7 suggesting that mitochondrial impairment in ppr mutant cells causes PINK1-Park activation, and subsequently, Marf downregulation. Our observations match previous reports of down regulation of Mfn1 and Mfn2 upon CCCP treatment as a mechanism to suppress mitochondrial fusion prior to PINK1-Park mediated mitophagy (45,46).

UPR mt is not sufficient to induce Marf downregulation in ppr mutants
The role of the PINK1-Park pathway in mitochondrial quality control is well known. However, the exact mechanism of PINK1-Park activation in in vivo contexts remains unclear. In cancer cell lines, dissipation of MMP and increased oxidative stress have been shown to activate PINK1-Park on the OMM leading to mitophagy (10,50). Therefore, oxidative stress or reduced MMP may activate PINK1-Park and subsequent Marf degradation in ppr mutants. However, we have shown earlier that ppr mutants do not have increased oxidative stress as compared to controls (29). We checked MMP in ppr A mutant clones using TMRE, a dye that reversibly stains mitochondria in a membrane potential-dependent manner. We observed that TMRE intensity in ppr A mutant clones is similar to that of wild type cells ( Figure S3A-A'). These observations rule out the possibility that PINK1-Park is activated due to oxidative stress or altered MMP in ppr mutants.
Mitochondrial unfolded protein response (UPR mt ), which is a cellular response to altered mitochondrial proteostasis, has been shown to activate PINK1-Park leading to mitophagy (51). Therefore, we first tested for UPR mt activation in ppr mutants. We determined the levels of Hsp60A which is reported to be increased due to elevated UPRmt (52). We found increased Hsp60A levels in ppr mutants as compared to controls, suggesting elevated UPRmt in ppr mutants ( Figure S3B-B'). Activation of UPR mt upon the loss of LRPPRC has also been observed in C.elegans and mammalian cells, and hence, it appears to be an evolutionarily conserved phenomenon (53). Increased UPR mt may induce PINK1-Park activity, which in-turn could lead to Marf downregulation. Therefore, we genetically suppressed the UPR mt response pathways and checked its impact on Marf in ppr mutants. Transcription factors Crc (homolog of ATF4), Dve and Foxo are known to mediate UPR mt (54)(55)(56). We generated ppr A mutant clones with either crc, foxo or dve knocked down. None of these interventions affected Marf::HA downregulation in ppr A clones, suggesting that the activation of these UPR mt pathways may not be causing PINK1-Park activation ( Figure S3C-E'). However, these interventions would not change the altered mitochondrial proteostasis in ppr mutants, which can activate PINK1-Park. Since, to the best of our knowledge, there is no reported method to suppress mitochondrial proteostasis defects, we asked whether the induction of mitochondrial proteostasis defects is sufficient to induce Marf degradation. To induce mitochondrial proteostasis defects, we expressed a mutant form of ornithine transcarbamylase (ΔOTC) that accumulates in an unfolded state and is shown to trigger UPR mt in flies (56). We expressed either ΔOTC or wild type OTC in the posterior half of the wing disc using En-Gal4 (En>Gal4/+; UAS-ΔOTC/+ or En>Gal4/+; UAS-OTC/+) and tested Marf::HA levels. We found neither OTC nor ΔOTC expression changed Marf::HA levels in the posterior half (marked by RFP) as compared to the anterior half of the wing discs ( Figure   S3F-F', S3G-G'). Although these observations do not rule out a role for mitochondrial proteostasis in activating PINK1-Park in ppr mutants, our data suggest that UPR mt is not sufficient to cause Marf degradation.
Bendless, a K63-linked E2 ubiquitin conjugase, is a regulator of Marf 9 To gain further insight into PINK1-Park activation and Marf degradation, we screened for a gene whose loss may cause a subtle increase in Marf levels as observed in park Δ21 and Pink1 5 mutant clones ( Figure S4A-A', S4B-B'). In our genetic screen, we found two independent alleles of bendless (ben A and ben B ) showing a subtle but consistent increase in Marf::HA levels in mutant clones ( Figure 4A-A', 4E, S4C-C'). This was also confirmed by western blot using whole larval extracts ( Figure 4G-G"). Ben is a fly homologue of the K63-linked E2 ubiquitin conjugase UBE2N/UBC13 with a marked similarity from yeast to humans ( Figure S4G). We ruled out the possibility that the increase in Marf::HA levels upon the loss of Ben is due to increased mitochondrial content by determining Tom20 levels, as there was no difference in Tom20::mCherry levels between ben mutant clones and controls ( Figure 4B-B', 4E). We also did not find an increase in Marf mRNA levels in ben mutants suggesting that the increase in Marf protein levels is not a consequence of increased transcription ( Figure 4F). These data suggest that Ben regulates Marf levels post-transcriptionally.
Next, we asked whether Ben is sufficient to induce Marf degradation. To test this, we generated a C-terminal V5-tagged Ben (UAS-ben::V5) transgenic line for tissue specific expression of ben and confirmed that the fusion protein is biologically functional by complementing the lethality associated with the ben A mutant allele ( Figure S4F, S4H). We then expressed ben::V5 in the posterior half of the wing disc using the En-Gal4 driver and compared the fluorescence intensities of Marf::HA in the posterior and the anterior halves. We found that ben::V5 overexpression did not affect the levels of Marf::HA ( Figure 4C-C', 4E). Additionally, we overexpressed an N-terminal HA-tagged Ben (UAS-HA::ben) using En-Gal4 and found no change in Marf::mCherry levels ( Figure S4E-E'). These data suggest that Ben is necessary but not sufficient for Marf downregulation. Since loss of ben, Pink1 or park results in mild upregulation of Marf, we hypothesize that Ben acts in the PINK1-Park pathway to regulate the steady state levels of Marf.

Bendless is essential for Marf downregulation in ppr mutants
Given that Marf undergoes proteolytic degradation in ppr mutants, we wanted to look at if Ben is involved in Marf degradation not only basally but under mitochondrial stress as well. We thus created ppr and ben double mutant clones and found that ppr A ben A or ppr A ben B double mutant clones showed no reduction in Marf::HA levels, unlike ppr mutant clones ( Figure 4D-D', 4E and S4D-D'). This suggests that Ben is essential for Marf degradation in ppr mutant cells.

Bendless is required for PINK1 stability and activity
To understand the mechanism of Ben mediated Marf degradation, we first looked at subcellular localization of Ben using the Ben::V5 tagged line. We overexpressed Ben::V5 using Act-Gal4, and performed cell fractionation from whole larval extracts. We found that in addition to the cytoplasmic fraction (marked by the presence of Tubulin) Ben::V5 is also present in the mitochondria enriched fraction (marked by presence of ComplexV) ( Figure 4H). Mitochondria Ben might regulate the activity PINK1-Park pathway to degrade Marf. To study the role of Ben in the PINK1-Park pathway, we tested the functional interaction between ben and Pink1 with respect to Marf degradation. Since PINK1-Park activity is suppressed by PINK1 degradation (57), PINK1 overexpression may activate PINK1-Park mediated Marf downregulation in the wing disc. We overexpressed Pink1 in the posterior half of the discs (En-Gal4/UAS-Pink1, UAS-RFP or UAS-GFP) and checked the levels of Tom20::mCherry, Complex-V and Marf::HA.
We observed no change in Tom20::mCherry and Complex-V levels ( Figure 5A-A', 5D, S5A-A'), but a marked reduction in Marf::HA levels ( Figure 5B-B'), suggesting that PINK1 is both necessary and sufficient to downregulate Marf. Pink1 overexpression in the wing discs may not have a significant impact on mitophagy and thus the mitochondrial content (Tom20::mCherry and Complex-V) is not affected. To test the functional interaction between Ben and PINK1, we created ben A mutant clones in both wild type and in Pink1 overexpression backgrounds and found that Pink1 overexpression does not induce Marf::HA downregulation in ben A mutant clones ( Figure 5C-C', 5D) or in ben A mutant wing discs ( Figure S5B-B'). These data suggest that Ben is necessary for PINK1 activity to cause Marf downregulation.
To understand how Ben may regulate PINK1 activity, we checked the effect of loss of Ben on PINK1 levels. We performed western blots using whole larval extracts from control and ben A mutants containing genomic tagged PINK1::Myc. We found a significant downregulation of full length PINK1::Myc in ben A mutants, but an increase in low molecular weight PINK1::Myc bands, suggesting that Ben is required for stability of full length PINK1 ( Figure 5E-F'). The low molecular weight bands might be products of PINK1 degradation by mitochondrial proteases as described by Thomas et.al. (58). Taken together, our data suggests that Ben is required for the stability of PINK1 and mediates the homeostatic turnover of Marf .

Ben regulates mitochondrial dynamics under mitochondrial stress
An RNAi-based screen in larval fat bodies has shown ben RNAi leads to enlarged mitochondria, similar to when Pink1 or park are knocked down (59). This would be in accordance with our observation wherein loss of ben results in Marf upregulation. To better characterize mitochondrial morphology we looked at Complex-V antibody staining in mutant larval muscles.
In this tissue, loss of ben shows a similar filamentous network as seen in wildtype ( Figure 6A-B, S6A-B). However, loss of ben in ppr mutants exacerbates mitochondrial morphology defects seen in ppr. On loss of ppr alone larval muscles show distinctive large globular mitochondria along with filamentous and ring-shaped mitochondria ( Figure 6C, S6C). ppr ben double mutants rarely show filamentous mitochondria ( Figure 6D). Instead, we observed a significant increase in the size and frequency of large globular and ring-shaped mitochondria as compared to ppr ( Figure 6D). We also observed that in lesser frequency ppr A ben A double mutants mitochondria form clusters, especially around the nucleus which is not observed in either ppr A or ben A mutants ( Figure 6D, S6D). These results suggest that Ben is required to suppress the hyperfusion of defective mitochondria in ppr and thereby regulates mitochondrial quality control.

Loss of Bendless accelerates photoreceptor degeneration in ppr mutants
Mutations in human LRPPRC cause Leigh Syndrome, a neurometabolic disease (60). Earlier work has shown that mutations in ppr cause activity induced retinal degeneration (29). As mutations in ben exacerbates the mitochondrial morphology phenotypes in ppr mutants, loss of ben may accelerate the degenerative phenotype. To test this hypothesis, we made eye specific ppr, ben and ppr ben double mutant clones using the ey-FLP system (30). We found that ppr mutant and ben mutant eyes show normal morphology but ppr ben double mutant eyes show severe retinal degeneration in young flies suggesting that loss of ben can accelerate retinal degeneration ( Figure 6E-H). This result suggests that Marf regulation by Ben is a neuroprotective mechanism.

Discussion:
To identify novel regulators of mitochondrial fusion in an in vivo system, we screened fly mutants for altered Marf levels. We found that mutations in ppr, which result in mitochondrial dysfunction, causes reduction in Marf levels ( Figure 1A). We found that in ppr mutants, Marf is degraded by the UPS in a PINK1-Park dependent mechanism ( Figure 2C-D, 3A-C). In the screen, we also identified mutations in ben causing subtle Marf upregulation ( Figure 4A, 4G).
We found that Ben is essential for PINK1 activity ( Figure 5C), regulates Marf levels ( Figure 4D) and mitochondrial morphology ( Figure 6D) in ppr mutants. We also found that a loss-of-function mutation of both ppr and ben in the eyes results in accelerated retinal degeneration ( Figure 6H) indicating that under mitochondrial stress Ben mediated regulation of mitochondrial dynamics is a protective mechanism ( Figure 7B).
Increased mitochondrial size has been observed in several mitochondrial diseases (61); however, it is not clear as to how these abnormal mitochondria contribute to disease progression.
Mitochondrial fusion occurs in a bid to increase oxidative phosphorylation under various cellular and mitochondrial stresses, a response termed as stress induced mitochondrial hyperfusion [SIMH (3)] (62,63). We propose that reduced ETC activity and mitochondrial stress in ppr (28,29,36), can induce SIMH ( Figure 6D, S1E). Mitochondrial fusion has been observed upon loss of ppr homologs in C.elegans, mouse and human cell lines (35,64) as well as in other mutants where the ETC is compromised (52,65,66). Since SIMH increases ATP synthesis and inhibits mitophagy (3,4,67,68), increased mitochondrial size appears to be a compensatory adaptation in ppr mutants in response to a bioenergetic deficit or mitochondrial stress.
Despite the increased mitochondrial size in ppr mutants ( Figure S1E), we observed Marf downregulation. We hypothesize that, while an adaptive mechanism may induce SIMH (cellular response), MQC may induce Marf degradation to suppress fusion of dysfunctional mitochondria (mitochondrial response) ( Figure 7B). As proposed earlier (62,63), this condition appears to be a conflict between the bioenergetic adaptations and the MQC mechanisms. Alternatively, increased mitochondrial size in ppr mutant cells may induce PINK1-Park to suppress further mitochondrial fusion by Marf degradation. A similar hypothesis was also proposed by Yamada et al. (69). We found that Marf degradation in ppr mutant clones in developing wing primordium is dependent on Park ( Figure 3B). We also observed a subtle increase in Marf levels in park and Pink1 mutants ( Figure S4A-B) (27). This suggests that the PINK1-Park plays a homeostatic role in Marf turnover in wild type tissue, while mitochondrial impairments -as in ppr mutants (29)may further amplify its activity to reduce Marf levels ( Figure 1A) possibly to segregate damaged mitochondria (9,45,46). The remarkable discovery by Narendra et al. that CCCP, which dissipates MMP, induces PINK1-Park-dependent mitophagy in cancer cells provided an unparalleled assay to investigate the mechanism further (10,70). However, we observe PINK1-Park activation in the absence of severe mitochondrial depolarisation ( Figure S3A) in ppr mutants results in Marf degradation and not mitophagy ( Figure 1A-B). We also find the PINK1 overexpression is sufficient to induce Marf degradation without triggering mitophagy ( Figure   5A-B, S5A). In vivo studies have shown PINK1-Park to function both in mitophagy (13)(14)(15)(16)(17)(18)(19)(20) and mitochondrial dynamics (22)(23)(24)(25)(26), but the physiological or cellular contexts that may determine various downstream activities of PINK1-Park are not known (21,71,72). Thus ppr mutants provide a novel and physiologically relevant in vivo system to study PINK1-Park mediated Marf regulation under mitochondrial stress.
In steady state conditions, PINK1 is imported into the mitochondria and cleaved by mitochondrial peptidases, it then retro translocates to the cytoplasm and is degraded by UPS to limit PINK1-Park activity (58,73,74). Loss of MMP, increased oxidative stress or increased UPR mt stabilizes full length PINK1, which then recruits Park leading to ubiquitination of OMM proteins and mitophagy (9,10,51,75,76). Given no change in MMP ( Figure S3A) and oxidative stress in ppr mutants (28,29,37), we suspected that mitochondrial proteostasis activates PINK1-Park to downregulate Marf. However, activation of UPR mt by ΔOTC expression did not result in Marf degradation suggesting that activation of UPR mt alone may not be sufficient to activate PINK1-Park mediated Marf degradation in vivo ( Figure S3G). Identification of additional factors leading to PINK1-Park activation for Marf degradation in vivo requires further investigation.
In most cells, PINK1 activity is maintained at low levels via its constant turnover by mitochondrial proteases and the UPS (57,77). For example, CHIP-mediated K48-ubiquitination promotes PINK1 turnover (78), while BAG2, a chaperon, prevents ubiquitination and promotes PINK1 stability (79,80). We found that Marf degradation in ppr mutants or by Pink1 overexpression is completely suppressed in the absence of the K63-linked E2 conjugase Ben ( Figure 4D, S4D and 5C). Previous studies have observed that the mammalian homolog of Ben, UBE2N, is dispensable for mitophagy but facilitates the clustering of mitochondria during CCCP induced mitophagy (81)(82)(83). We also found that the loss of ben does not alter developmental mitophagy during larval midgut remodeling ( Figure S5C), which has been shown to be dependent on PINK1-Park (13,14). 16 We hypothesize three possible mechanisms through which Ben regulates Marf degradation. One, given that Park activity is regulated by K63 ubiquitination (84), Ben may ubiquitinate Park. Two, Ubiquitin C-terminal hydrolase L1 (UCHL1), which suppresses Marf degradation (85), is K63 ubiquitinated leading to its autophagic degradation (86). Thus, Ben might mediate K63 ubiquitination and degradation of UCHL1. K63 ubiquitination of PINK1 by the Traf6-SARM1 complex is shown to stabilize PINK1 (87). Therefore, Ben may stabilize PINK1 by K63 ubiquitination. The fact that loss of ben results in reduced PINK1 levels ( Figure 5E-E'), suggests Ben is likely to increase the stability of PINK1 by K63 ubiquitination. Indeed human PINK1, in cell culture systems, is known to be ubiquitinated at K137 by both K48 and K63 linkages (88).
While K48 chains are linked with PINK1 degradation; the significance of K63 linkage is not obvious. K63 ubiquitination is suggested to protect proteins from proteasomal degradation (89).
In conclusion, Ben-PINK1-Park regulation of Marf appears to be a homeostatic function which is further activated in response to aberrant mitochondrial function. ppr ben double mutants show aberrant mitochondrial morphology in larval muscles and severe retinal degeneration as compared to ppr mutant eyes indicating a protective role for Ben ( Figure 6). Given that mutations in LRPPRC result in Leigh syndrome, it would be crucial to check the activation of Ben-PINK1-Park in Leigh syndrome and other mitochondrial diseases. Indeed, altered mitochondrial dynamics has been reported in many mitochondrial diseases (61,(90)(91)(92). It is possible an adaptive response in these diseases can modify mitochondrial dynamics in a Ben-PINK1-Park-dependent mechanism. Thus, further studies on the mechanisms of Ben-PINK1-Park activation will be crucial for understanding mitochondrial quality control in mitochondrial disease.        Upon mitochondrial stress (as we show in case of mutation in ppr) results in presence of impaired mitochondria, we predict that the increase in Ben-PINK1 mediated Marf degradation 23 keeps the damaged mitochondria isolated and thus can be repaired or sent for mitophagy. In the absence of Ben however Marf is not removed from impaired mitochondria and results in fusion of impaired mitochondria with the healthy pool.

Drosophila culture
Flies were cultured on standard media containing sucrose, malt, yeast and corn flour at room temperature. Crosses were maintained at 25℃. Crosses involving RNAi were maintained at 28℃. Drosophila larvae expressing UAS-Pros 6 1 were maintained at 25 o C till 3 rd instar stage, and were then transferred to 28 o C for 24 hours before dissection, to avoid cell death observed on prolonged inhibition of proteasomal activity. To activate the FLP-FRT system, heat shock was given during first instar larval stages at 37 o C for 1hr. Genotypes used are as listed in Table 1. For drug treatments 3 rd instar larvae were transferred to food containing 3mM chloroquine, 100µM MG132, or DMSO (vehicle control) for 24 hours prior to dissection. For western blot and qPCR, 3 rd instar larvae were used. We observed that development of ppr A mutant larvae is substantially delayed. Therefore, we used size matched 3 rd instar ppr A mutant larvae that are obtained after 14-15 days post hatching.

Generation of transgenic flies
ben sequence was amplified from genomic DNA. These PCR amplified ben ORF sequences were then inserted into a pUAST vector containing attB sites, flanking the insert using Opa1::3FLAG-2HA genomic construct was generated using the P(acman) system (93). Briefly, the 3FLAG-2HA tag was amplified from C-terminal tag fusion vector pL452-C-3FLAG-2HA and inserted at the C terminal of Opa1 through recombineering in the P(acman) clone CH322-27B08, which was subsequently injected into y 1 w 1118 ; PBac{y+-attP-3B}VK00033 flies.
Samples were mounted in Vectashield (VectorLabs -H100) and imaged under 40X or 63X oil immersion Leica Stellaris 5 or Olympus FV3000 confocal microscopes. Images were processed using Fiji. All antibody dilutions and the blocking solution were made in 1X PBST; details of antibodies and their dilutions used are listed in Table 2.

Eye phenotype imaging
Mutant eyes were created by crossing heterozygous mutant flies with w cl(1) FRT19A /Dp(1;Y); ey-FLP flies. The eye images were then acquired on a Leica M205FA Stereo Zoom microscope. 25 Thirty 3 rd instar larvae were collected and washed with chilled mitochondrial isolation buffer (210mM mannitol, 70mM sucrose, 1mM EGTA, 5mM HEPES, and 0.5% BSA). The larvae were then transferred into a 1.5ml centrifuge tube containing 250µl of chilled mitochondrial isolation buffer. Micro-pestles were used to homogenize the samples, keeping the sample on ice. 300 µl of mitochondrial isolation buffer was added to the lysate. Lysate was centrifuged at 200G for 5 mins at 4℃ to remove large debris. The supernatant was then centrifuged at 1500G for 5mins at 4℃ and the pellet was discarded. The supernatant was centrifuged at 8000G for 15mins at 4℃.

Mitochondrial fractionation
The supernatant containing the cytoplasmic fraction was processed for western blot by adding Laemmli buffer (0.004% bromophenol blue, 20% glycerol, 4% SDS and 0.125M Tris-HCl pH 6.8) having 5% beta-mercaptoethanol and heated at 98°C for 5mins. The pellet was resuspended in 550µl fresh mitochondrial isolation buffer and again centrifuged at 8000G for 15 mins at 4℃.
The pellet containing the mitochondrial fraction was resuspended in 100µl of Laemmli buffer having 5% beta-mercaptoethanol and heated at 98℃ for 5mins.

Western blot
3 rd instar larvae were crushed in RIPA lysis buffer [50mM Tris,150mM NaCl, 0.2% Triton X 100 and 1X protease and phosphatase inhibitor cocktail (Thermo Fisher -A32965,A32957 respectively)], followed by centrifugation at 16,000g for 10 mins at 4℃. Clear fat free supernatant was used for total protein estimation. Lysate was mixed with equal volume of 1X  (Fig. 5F). confocal microscope at 63X oil objective.

Mitochondrial morphology analysis
Wing discs immunostained for Complex-V were imaged using Leica Stellaris 5 confocal microscope at 63X oil objective. The mitochondria were segmented on Fiji using the Trainable Weka segmentation plugin (94). The segmented images were then used to find out mitochondrial area using Particle Analyze Tool on Fiji.
Blind test: For qualitative assessment of mitochondrial morphology in larval muscle, we renamed a set of images containing mitochondria from larval muscles with random numbers. The images from different genotypes (control, ben A , ppr A , and ppr A ben A ) were pooled and were assessed for the presence of different mitochondrial morphologies, including presence or absence of mitochondria network, large globular mitochondria, ring shaped mitochondria and mitochondrial aggregates. Multiple images were used for the assessment, 40 images from 11 larvae for control, 24 images from 7 larvae for ben A , 27 images from 7 larvae for ppr A , and 32 images from 9 larvae for ppr A ben A .

Statistics analysis
At least three independent experiments were used for all quantifications, the n values for each experiment is indicated in their respective figure legends. Two-tailed paired t-test was used to analyze data obtained from clonal analysis, One sample t-test was used to analyze the data in Supplementary Fig. 5. Two-tailed unpaired t-test was used to analyze all other data sets.
Significance of the data was represented as * for p<0.05, ** for p<0.01, and *** for p<0.0001.
Details of the test used and the significance is mentioned in respective figure legends.