Modelling of BCS1L-related human mitochondrial disease in Drosophila melanogaster

Mutations in BCS1L are the most frequent cause of human mitochondrial disease linked to complex III deficiency. Different forms of BCS1L-related diseases and more than 20 pathogenic alleles have been reported to date. Clinical symptoms are highly heterogenous, and multisystem involvement is often present, with liver and brain being the most frequently affected organs. BCS1L encodes a mitochondrial AAA + -family member with essential roles in the latest steps in the biogenesis of mitochondrial respiratory chain complex III. Since Bcs1 has been investigated mostly in yeast and mammals, its function in invertebrates remains largely unknown. Here, we describe the phenotypical, biochemical and metabolic consequences of Bcs1 genetic manipulation in Drosophila melanogaster. Our data demonstrate the fundamental role of Bcs1 in complex III biogenesis in invertebrates and provide novel, reliable models for BCS1L-related human mitochondrial diseases. These models recapitulate several features of the human disorders, collectively pointing to a crucial role of Bcs1 and, in turn, of complex III, in development, organismal fitness and physiology of several tissues. Supplementary Information The online version contains supplementary material available at 10.1007/s00109-021-02110-1.


Introduction
Ubiquinol-cytochrome c oxidoreductase (also known as cytochrome bc 1 or complex III) is the central electron transfer complex of the mitochondrial respiratory chain (MRC). It catalyses the oxidation of reduced coenzyme Q (CoQ) coupled with reduction of cytochrome c. The fully functional complex III is a dimer of the holoenzyme (complex III 2 ), composed of three catalytic core subunits. One of them is mtDNA-encoded (MT-CYB) whereas two are nuclear (n)DNA-encoded (CYC1 and UQCRFS1) [2]. Eight additional subunits (UQCRC1, UQCRC2, UQCRH, UQCRB, UQCRQ, Subunit 9, UQCR10 and UQCR11) are not directly involved in catalysis but are however needed for the correct functioning of the enzyme. MT-CYB contains the two CoQ binding sites, Q o and Q i , and two b-type heme groups with different redox potentials (low redox potential b L and high redox potential b H ). CYC1 harbours a c-type heme centre (c 1 ) and UQCRFS1 (also known as Rieske Fe-S protein) contains a 2Fe-2S cluster. Complex III couples its redox reactions to proton pumping through the Q cycle [3].
The biogenesis of complex III has been extensively studied in yeast, and the underlying molecular mechanisms are emerging from different studies (reviewed by [5]). Moreover, it is now well known that the whole process requires a set of proteins, including enzymes and chaperones, which are not part of the complex but play a role in its biogenesis and assembly [5]. Among the complex III-associated assembly factors, BCS1L is the most studied. BCS1L is a member of the AAA + family (ATPases associated with diverse cellular activities), as it displays a typical AAA + domain. During complex III biogenesis, BCS1L acts at the later stages, carrying out the incorporation of the catalytic UQCRFS1 and the smallest supernumerary subunit UQCR11 into the complex [6][7][8]. Experimental evidence indicates that heptameric structures of BCS1L act on UQCRFS1 via protein translocation, promoting its insertion into the inner mitochondrial membrane (IMM) after being imported into the mitochondrial matrix [9][10][11].
Mutations in BCS1L are the most frequent forms of complex III deficiency in humans (OMIM no. 124000), [12]. The first pathogenic variants of BCS1L were described in 2001 [13]. Over the following two decades, several other mutant BCS1L alleles linked to complex III deficiency have been reported and associated with a set of mitochondrial disorders of varying severity, ranging from early onset, lethal diseases, to mild conditions with chronic clinical courses. For example, BCS1L founder variants restricted to Scandinavian areas cause GRACILE syndrome (growth retardation, aminoaciduria, cholestasis, iron overload, lactic acidosis and early death) [14,15,47], whereas others are associated with metabolic acidosis and liver failure [13,[16][17][18][19][20], sensorineural Björnstad syndrome [21,22] or sensorineural hearing loss with developmental delay, hypotonia and encephalopathy [23].
To date, the only available animal model for a BCS1L-related disorder is a knock-in mouse carrying the GRACILE syndrome founder missense mutation c.232A > G (p.Ser78Gly). The homozygous mutant mouse displays reduced growth, lactic acidosis and liver impairment, progressing from steatosis to fibrosis, leading to death at approximately 6 weeks of age [24]. However, for yet unknown and controversial reasons, young animals have been reported to have normal Rieske Fe-S incorporation and complex III activity [24]. Interestingly, a recent report [25] has shown that the original GRACILE mouse resulted from the additional presence of a mtDNA mutation combined with the BCS1L c.232A > G (p.Ser78Gly) mutation, so that when the GRACILE and the mtDNA mutations are separated, the mouse phenotype is milder and lacks the features that are believed to cause infantile death in humans. Recently, the bcs1l ortholog in zebrafish has been implicated in warranting physiological mitochondrial respiration and contributing to normal morphogenesis [26]. Finally, a yeast strain carrying a single point mutation (F342C) in the Bcs1-specific domain has also been reported [27].
Here, we describe the generation and characterization of the second metazoan model of BCS1L-related mitochondrial disease. Genetic tools were utilised to explore the phenotypical, molecular and biochemical consequences of the loss of the BCS1L homolog in the invertebrate Drosophila melanogaster, defining the homeostatic role of the protein-most likely as a cytochrome bc 1 assembly factor-in development and physiology of several tissues.

CG4908 is the D. melanogaster ortholog of human BCS1L
The integrative orthology prediction tool DIOPT [28] assigns the protein coding gene CG4908 (hereafter named Bcs1 and Bcs1 protein) as the high score putative Drosophila melanogaster ortholog of Homo sapiens BCS1L. Two transcripts of Bcs1, sharing the CDS, encode the same 431 amino-acid protein (Bcs1). Global alignment of Drosophila melanogaster Bcs1 and Homo sapiens BCS1L revealed a high degree of conservation of the protein sequence (Fig. 1A). Indeed, 60% of the protein sequence was found to be identical between human BCS1L and D. melanogaster Bcs1, and three domains of the protein known to be crucial for its localization and function-the N-terminal transmembrane domain (TMD), the internal mitochondrial targeting sequence (MTS) and the C-terminal AAA + ATPase domain-are highly conserved (Fig. 1A). Importantly, almost all the sites known to be mutated in patients with BCS1L-related mitochondrial disease [13][14][15][16][17][18][19][20][21][22][23]47] are conserved in both species, and most of them are located into the three main domains (Fig. 1A). Both yeast and mammalian Bcs1 proteins were shown to be mitochondrial proteins, specifically located into the IMM [29,30]. Since the subcellular localization of Drosophila Bcs1 has never been experimentally tested and validated, we cloned CG4908 cDNA into the pAc5-STABLE2-neo expression vector with a HA-tag at the C-terminus (Bcs1-HA). Confocal imaging of immunostained S2R + cells using an antibody against HA-tag plus an antibody against a mitochondrial complex V subunit (ATP5A) lets us experimentally validate that Drosophila melanogaster BCS1L-homolog is targeted to mitochondria (Fig. 1B).
Partial loss of Bcs1 in the whole organism has detrimental effects on flies' development due to severe complex III deficiency We exploited the binary system UAS/GAL4 [31] to drive the post-transcriptional silencing of Bcs1 in Drosophila. Firstly, the act5c-gal4 construct was used to drive ubiquitous Bcs1 KD. Crossing of heterozygous act5c-gal4/CyO.GFP with UAS-shBcs1 homozygous flies resulted in the loss of about 50% of the progeny between the larval and the pupal stage, and in altered mendelian ratios. All the individuals reaching the pupal and the adult stages were heterozygous UAS-shBcs1/CyO.GFP and therefore unsilenced ( Fig. 2A,  B). In fact, act5c-gal4 > UAS-shBcs1 larvae were found in the progeny and screened by qPCR, documenting a loss of about 80% of the transcript in KD larvae compared to control larvae (Fig. 2C). This resulted in complete egg-to-adult lethality, with KD larvae failing to thrive and dying at the 3rd larval stage, as documented by the presence of fully developed mouth hooks (Fig. 2D). Bcs1 KD larvae were also smaller than controls (Fig. 2E) and showed significant locomotor defects, including lower motility and dramatically decreased peristaltic movements compared to control larvae (Fig. 2F).
Biochemical analyses of MRC enzymatic activities revealed a severe reduction in CIII activity (47% of control activity) with a small but significant reduction in NADH/ ubiquinone oxidoreductase (CI) activity (79% of control activity) (Fig. 3A). We also observed an unexpected, small but significant increase in cytochrome c oxidase (CIV) activity in Bcs1 KD larvae (118% of control activity) (Fig. 3A). In addition, BNGE of DDM-solubilized mitochondrial complexes showed that the functional, dimeric form of CIII (CIII 2 ) was predominantly and severely affected in Bcs1 KD larvae (Fig. 3B). These results were further corroborated by immunovisualization of the BNGE of DDM-solubilized mitochondria from Bcs1 KD and control larvae using an antibody against cytochrome bc 1 core subunit 2 (UQCR-C2). As shown in Fig. 3C, Western blot analysis revealed severe reduction of the signal corresponding to CIII 2 in Bcs1 KD mitochondria. Furthermore, the accumulation of low-molecular-weight, UQCR-C2-containing sub-assembly species was detected in Bcs1 KD individuals, as also previously observed in yeast and humans. In order to validate the specificity of the effects of Bcs1 silencing on developmental arrest and CIII stability, we also analysed a different Bcs1 UAS-RNAi line (UAS-Bcs1-IR) expressing a 344-bp inverted repeated sequence of CG4908. When the UAS-Bcs1-IR homozygous flies were crossed with heterozygous act5c-gal4 > CyO.GFP, we observed a loss of about 50% of the progeny between the pupal and adult stages (figure S1A). Accordingly, we also observed altered mendelian ratios in the adults, as all of the progeny carried the CyO balancer chromosome (figure S1B), as in the cross between the act5c-gal4 > CyO.GFP and the UAS-shBcs1 lines. The size of GFP negative (act5c-gal4 > UAS-Bcs1-IR) larvae from the progeny was comparable to that of both act5c-gal4 > + and UAS-Bcs1-IR > CyO.GFP. Notably, when KD efficiency in act5c-gal4 > UAS-Bcs1-IR larvae was assessed by qPCR, Bcs1 expression levels were reduced by 57% compared to control larvae (S1C). Importantly, prior to further characterizing this second Bcs1 RNAi line, we checked whether the off-target gene (named Hoe2) reported in the VDRC database entry for line UAS-Bcs1-IR was affected by the expression of the Bcs1-IR sequence. No alterations were observed in Hoe2 expression (figure S1D). Next, by analysing the MRC of Bcs1-IR KD larvae by BNGE and immunoblot of BNGE as for the act5c-gal4 > UAS-shBcs1 KD line, Bcs1-IR KD larvae were found to have CIII 2 deficiency (figure S1E), associated with accumulation of sub-assembly CIII 2 species (figure S1F), similar to shBcs1 KD larvae. The fact that two independent RNAi lines for Bcs1 shared phenotypical and biochemical features suggests that these are indeed specifically linked to defects in Bcs1 expression. Further, the two UAS-RNAi lines resulted in slightly different KD efficiencies (~ 80% by UAS-shBcs1 construct and ~ 60% by UAS-Bcs1-IR construct) which, in turn, correlated with phenotypical traits (larval vs pupal lethality, respectively) and with the severity of CIII deficiency.
Lastly, we observed that Bcs1 KD larvae have a strongly increased glycolytic metabolism in response to mitochondrial dysfunction linked to altered biogenesis and stability of CIII, as suggested by a twofold increase in Impl3 (ortholog of mammalian lactate dehydrogenase, Ldh) expression levels and a threefold increase in total body lactate content (Fig. 3D, E). Accordingly, the expression levels of other three key enzymes of glycolysis (i.e., phosphofructokinase-Pfk, hexokinase A-Hex-A and pyruvate kinase-PyK) were significantly increased in Bcs1 KD larvae (Fig. 3D). Taken together, these data demonstrate that physiological levels of Bcs1 expression are essential for the developmental stages of Drosophila, and their loss is linked to altered biogenesis of MRC CIII 2 , mitochondrial dysfunction and metabolic switch from oxidative to glycolytic ATP production, causing lactate accumulation.

Specific silencing of Bcs1 in brain and dopaminergic neurons causes encephalopathic phenotypes and neuromotor dysfunction in flies
Given the strong phenotype observed in ubiquitous KD individuals, we sought to overcome developmental lethality through the tissue-specific silencing of Bcs1, by specifically knocking-down gene expression in fly homolog tissues that are mostly affected in patients, such as the brain. The CNSspecific knockdown mediated by crossing homozygous UAS-shBcs1 flies with homozygous elav-gal4 flies resulted in the reduction of 60% of Bcs1 expression in the head (Fig. 4A). Knockdown flies were affected by partial pupal lethality, with nearly 50% of the expected adults missing. In fact, numerous fully developed individuals failed to emerge from the puparium, and some of them got stuck while hatching, eventually dying (Fig. 4B).

Specific silencing of Bcs1 in muscle causes myopathic phenotypes flies
The striking phenotypes resulting from CNS-specific KD of Bcs1 prompted us to downregulate Bcs1 expression in fly mesodermal derivatives, using the how24b-gal4 driver. After crossing heterozygous how24b-gal4 > TM3 with homozygous UAS-shBcs1, we detected a 50% reduction by qPCR in Bcs1 expression in the larval body wall muscles (Fig. 6A). Although almost all KD larvae grew and reached the pupal stage, fully developed individuals (Fig. 6B) failed to hatch (Fig. 6C), even over several days. Indeed, all the adult individuals in the progeny were UAS-shBcs1 > TM3 (Fig. 6D). Remarkably, crossing of homozygous how24b-gal4 with homozygous UAS-shBcs1 flies confirmed the observed lethality, with the total absence of adults in the progeny (Fig. 6E). Thus, muscle-specific KD of Bcs1 in Drosophila leads to lethality at the latest pupal stages, likely due to inability of the flies to perform muscle contractions and coordinate movements needed to hatch.

Bcs1 silencing in the fat body has minor effects on organismal fitness in flies
Since the most severe cases of BCS1L-related mitochondrial diseases affect liver function, mostly leading to fatal metabolic syndromes, we knocked-down Bcs1 expression in the fat body, the fly functional analogue of liver and white adipose tissue, by crossing homozygous ppl-gal4 flies with homozygous UAS-shBcs1 flies. Although the expression of Bcs1 was drastically reduced (20% of the control) in the abdominal fat body of the progeny (Fig. 7A), the specific KD of the gene by ppl-gal4 did not lead to major effects during development, since similar to controls, a high percentage of laid eggs (85% vs 90% in control cross) reached the adult stage (Fig. 7B). However, we observed reduced lifespan in Bcs1 KD flies, with median survival rates between 44 and 67% of controls in male and female flies, respectively (Fig. 7C). We also detected impaired β-oxidation by quantifying the lipid content (triacylglycerides, TAGs) in the fat body of the adults (Fig. 7D). Indeed, in spite of having slightly but significantly lower TAG content in the fat body under normal feeding conditions compared to controls, Bcs1 KD adults retained higher content of TAG after 24 h of starvation, when mobilization of fatty acids from TAG is needed to fuel β-oxidation, preserving oxidative metabolism (Fig. 7D). Taken together, these results suggest that in the fat body of invertebrates, Bcs1 has a dispensable role during development but is essential to preserve mitochondrial β-oxidation and viability in the adult stage.

Discussion
Among human mitochondrial diseases, complex III deficiency is one of the rarest. Over the past 20 years, the BCS1L gene has emerged as the main cause of complex III deficiency, leading to conditions such as GRACILE syndrome, Björnstad syndrome and severe metabolic syndromes [12]. Yeast has provided important information on the role of Bcs1 in the incorporation of the catalytic Rieske Fe-S subunit into complex III [6,8] and work on human disease in vitro models has confirmed this function [7]. Other studies have elucidated the molecular mechanism of UQCRFS1 loading into complex III by BCS1L, proposing and refining the model by which a homo-heptamer of BCS1L mediates the process by ATP-driven pushing of UQCRFS1 across the membrane into the pre-complex III [9][10][11]. While providing useful information on the molecular mechanisms underlying genetic defects in eukaryotes, yeast has obvious limitations as a model of human clinical conditions. In 2010, the first mouse model harbouring the GRACILE syndrome founder mutation in BCS1L was described [24]. As already mentioned, these animals recapitulate some of the clinical features of patients, but they are also very different from patients, especially for their baffling biochemical and molecular features, which appear normal at young age [24,32]. Young mice fail to display overt defects in complex III activity, and steady-state levels of UQCRFS1 are comparable to those of controls. By contrast, older, symptomatic animals start displaying the typical features of complex III deficiency that characterize the human condition [24].
Since Drosophila melanogaster harbours an uncharacterized homolog of BCS1L in its genome (Bcs1), we decided to expand the set of BCS1L-related mitochondrial disease models by generating models of Bcs1 deficiency in flies. Firstly, we were able to show that the protein is highly conserved, especially in domains that are known to be essential for its function. This analysis also highlighted that most of the sites affected by missense mutations are conserved between humans and flies. Further, we demonstrated that Bcs1 protein is a bona fide mitochondrial protein in Drosophila cells. Regarding the effects of dysfunctional Bcs1 in flies, we exploited the UAS/GAL4 binary system in order to drive the silencing of the gene by expressing a specific shRNA in different tissues and systems. We showed that the loss of most but not all of the transcript in the whole show mean and 95% CI, n = 60 animals. Statistical analysis was performed with one-way ANOVA with Dunn's multiple comparisons test (*p ≤ 0.05, **p ≤ 0.01; ****p ≤ 0.0001) organism causes developmental arrest between the third larval and the pupal stages and failure to reach the adult stage. This correlates with higher expression of key enzymes of the glycolytic pathway (Ldh, Pfk, Hex-A, PyK), together with a dramatic increase in total body lactate in larvae lacking physiological levels of Bcs1.These are typical markers of BCS1L-linked CIII deficiency and, more generally, of mitochondrial disease. We showed that this very severe clinical phenotype is linked to a profound defect in complex III amount/activity and accumulation of intermediate CIII assemblies, closely resembling what was previously observed in patients. Interestingly, the parallel reduction in CI activity confirms observations from patients suffering from complex III deficiency. In fact, the instability of CI is a Fig. 6 Muscle-specific KD of Bcs1 in D. melanogaster. A Bcs1 expression levels normalized by Rp49 reference expression levels and measured by qPCR in larval body wall muscles, expressed as relative quantity of template in the sample (RQ) in muscle-specific KD larvae (how24b-gal4 > UAS-shBcs1) compared to control larvae (how24b-gal4 > +). Data are plotted as mean ± S.D. (n = 3, Student's t test **p ≤ 0.01). B Morphological evaluation of muscle-specific Bcs1 KD pupae. p = pupa, h = head, th = thorax, ab = abdomen, ey = eyes, l = legs, w = wings Scale bar: 1 mm. C Relative percentage of egg to adult viability of muscle-specific Bcs1 KD cross (how24b-gal4/TM3 x UAS-shBcs1) and control cross (how24b-gal4/TM3 x w 1118 ), calculated at three developmental stages (eggs, pupae and adults) (n > 250, Chi-square test 14.66 df(2), ***p < 0.001). D Mendelian frequencies of adults obtained by crossing homozygous how24b-gal4 flies with homozygous UAS-shBcs1 flies (n = 156, Chi-square test 111.9, df(1), **** p ≤ 0.0001).E Representative image of the progeny obtained by crossing homozygous how24b-gal4 flies with homozygous UAS-shBcs1 flies. Control flies (left vial) hatch from the pupa and reach the adult stage whereas muscle-specific Bcs1 KD flies (right vial) fail to hatch and die at the latest pupal stage well-known phenomenon resulting from faulty CIII biogenesis in mammals [1,[33][34][35][36][37][38], as the assembly of the respiratory chain complexes appears to be strictly interconnected [39]. The severe effects described above are triggered by knocking-down Bcs1 expression by 80%, without completely abolishing its function. This observation further supports the crucial role of Bcs1 in metazoans and possibly explains why no loss of function alleles of BCS1L have ever been described in homozygous state. We also established a fundamental role of Bcs1 in the brain and muscle (mesodermal derivatives), since the specific silencing of the gene in such tissues was sufficient to compromise their function, phenocopying features observed in patients, including encephalopathy and muscle hypotonia. B=Notably flies partially missing Bcs1 in CNS and mesodermal derivatives also died early. In addition, even a reduction in Bcs1 expression Fig. 7 Fat body-specific KD of Bcs1 in D. melanogaster. A Bcs1 expression levels normalized by Rp49 reference expression levels and measured by qPCR in adult abdominal fat body, expressed as relative quantity of template in the sample (RQ) in fat body-specific KD adults (ppl-gal4 > UAS-shBcs1) compared to control adults (ppl-gal4 > +). Data are plotted as mean ± S.D. (n = 3, Student's t test ***p ≤ 0.001). B Relative percentage of egg to adult viability of fat body-specific Bcs1 KD cross (ppl-gal4 x UAS-shBcs1) and control cross (ppl-gal4 x w 1118 ), calculated at three developmental stages (eggs, pupae and adults), (n > 250, chi-square test 0.1039 df(2), ns = not significant. C Representative survival curves (Kaplan-Meier) of fat body control (ppl-gal4 > + , solid lines) and KD (ppl-gal4 > UAS-shBcs1, dotted lines) flies under standard culture conditions. Females and males are represented by red and blue lines, respectively. Statistical significance was tested by the log-rank (Mantel-Cox) test (****p ≤ 0.0001). D Quantification of TAG in control (ppl-gal4 > +) and Bcs1 KD abdominal fat body (ppl-gal4 > UAS-shBcs1), normalized to the protein content and expressed as relative quantity in the sample compared to control. Data are plotted as mean ± S.D. (n = 3, two-way ANOVA with Tukey's multiple comparisons ns = not significant, *p ≤ 0.05, ****p ≤ 0.0001) restricted to a specific population of neurons (dopaminergic) caused age-related neuromotor dysfunction. Lastly, we knocked-down the expression of Bcs1 in the fat body, i.e. the insects' functional homolog of liver and adipose tissue of mammals. Any early-onset deleterious effect of Bcs1 silencing was excluded, especially during development. Nevertheless, we observed signs of mitochondrial dysfunction in the tissue (impaired β-oxidation), leading to slow progression of phenotype over time, eventually resulting in early death. Thus, in contrast to humans and similar to mice, flies are not significantly affected by Bcs1 dysfunction in the fat body, suggesting that physiological differences between species might explain why patients suffering from GRACILE syndrome exhibit severe liver disease. Indeed, in the GRACILE mouse liver, disease does not play a prominent role in the development of the clinical phenotype. This is in line with our observations, as dysfunction of Drosophila Bcs1 in the fat body appears to have much less severe effects than in other tissues such as brain and muscle. Taken together, these observations point out to the possibility that liver-disease might be a specific effect of BCS1L dysfunction only in human GRACILE syndrome.
In conclusion, we have provided an extensive genetic and functional characterization of Bcs1 in Drosophila melanogaster. In addition, we have validated and characterized in depth for the first time a non-mammalian animal model of BCS1L-related mitochondriopathy, phenocopying and mimicking several patients' features. Drosophila is a promising tool to study the pathophysiology of BCS1L-related mitochondrial complex III deficiency, and to test novel treatment strategies, although the puzzling results of the GRACILE mutation in young mice plus the consistent absence of homozygous loss of function mutations of BCS1 in humans and the severity of partial loss of Bcs1 in flies should prompt to hypothesize an additional, yet unknown role of BCS1 that can account for its dispensability in CIII biogenesis in young mice, in contrast with its indispensability in humans and flies.

Fly stocks and maintenance
Flies were raised on standard cornmeal medium and maintained at 23 °C, 70% relative humidity on a 12-h-light and 12-h-dark cycle. The UAS fly strain (y [1]

RNA isolation, reverse transcription and qRT-PCR
Total RNA was extracted from 10 adults (whole body), 30 heads, 15 larval body wall preparations or 15 adult abdominal fat body preparations for each genotype (1:1 malesfemales) using TRIzol reagent (Thermo Fisher Scientific) according to the manufacturer's instructions. First-strand cDNA synthesis was performed with the GoScript reverse transcriptase kit (Promega). qRT-PCRs were performed in triplicate using a Bio-Rad CFX 96 Touch System (Bio-Rad) using GoTaq qPCR SYBR Green chemistry (Promega). The 2 −∆∆ Ct (RQ, relative quantification) method was used to calculate the relative expression ratio. Rp49 was used as endogenous control, and the oligonucleotides employed

Analysis of larval locomotor activity
Locomotor activity was assessed using 60-mm Petri dishes with graph paper (0.25 cm 2 grids) attached to the bottom. Larvae were put in the dish and their activity was recorded with a digital camera for 1 min. The number of lines crossed and peristaltic movements per minute were measured for each larva. A total number of 20 individuals per genotype were analysed.

Climbing test
The climbing test was performed using a modified version of the countercurrent apparatus originally designed by Seymour Benzer [40]. Twenty flies were placed into the first tube, tapped to the bottom and allowed to climb a 10-cm distance for 10 s. The flies that reached the 10 cm distance were shifted to a second tube, tapped again to the bottom and allowed to climb for further 10 s. The procedure was repeated five times. Then, the number of flies in each tube was counted. Climbing indexes were calculated as the weighted average of flies in the different tubes, divided by five times the number of flies in the test. A minimum number of 60 individuals per sex and genotype were analysed.

Lifespan assay
Flies were reared at standard low density, collected after hatching and divided into males and females over a 24-h window. Adults of the same sex were kept at a density of 10 per vial (for a total number of 50 individuals) at 23 °C. Flies were counted and transferred to fresh medium three times per week, with no anaesthesia.

Isolation of mitochondria
Mitochondria were prepared by differential centrifugation as described previously [41] with minor modifications. Larvae were homogenized with a Dounce glass potter and a loose-fitting glass pestle in 10 mL of isotonic isolation buffer (225 mM mannitol, 75 mM sucrose, 5 mM HEPES, 1 mM EGTA, pH 7.4) with 1% BSA. Samples were centrifuged at 1000 × g (Eppendorf 5810R) at 4 °C for 10 min. The supernatant was filtered through a 100-μm strainer (Corning) and centrifuged at 6,000 × g at 4 °C for 10 min. The mitochondrial pellet was washed in 10 mL of isolation buffer and centrifuged at 6000 × g for 10 min. The wash was repeated using 10 mL of isolation buffer without BSA and centrifuged at 7000 × g for 10 min. The mitochondrial pellet was re-suspended in minimal volume of isolation buffer without BSA. Protein concentration was measured by the Bradford assay (Bio-Rad protein assay).

Analysis of MRC activity
Prior to enzymatic MRC complex activity assays, isolated mitochondria were subjected to 3 freeze-thaw cycles in 10 mM ice-cold Tris hypotonic buffer (pH 7.6) performing three freeze-thaw cycles with liquid nitrogen to disrupt the mitochondrial membranes. The activities of mitochondrial respiratory chain complexes and citrate synthase (CS) were measured by spectrophotometry, as described previously [42], following modified versions of the procedures for the reduction of decylubiquinone and cytochrome c [43].

BNGE
Isolated mitochondria were resuspended in 1.5 M aminocaproic acid, 50 mM Bis-Tris/HCl pH 7.0. The samples were solubilized with 4 mg of n-dodecyl β-d-maltoside (Sigma) per mg of protein. After 5 min of incubation on ice, samples were centrifuged at 18,000 × g at 4 °C for 10 min. The supernatant was collected and resuspended with Sample Buffer (750 mM aminocaproic acid, 50 mM Bis-Tris/HCl pH 7.0, 0.5 mM EDTA and 5% Serva Blue G). Native samples were separated using NativePAGE 3-12% Bis-Tris gels (Thermo Fisher Scientific) according to the manufacturer's protocol.

Western blot and immunodetection
BN-PAGE gels were transferred to PVDF membranes in Dunn bicarbonate buffer (10 mM NaHCO 3 , 3 mM Na 2 CO 3 ) applying a constant voltage of 25 V at 4 °C for 1 h using a XCell II™ Blot Module (Thermo Fisher Scientific). For the immunodetection of specific protein targets, blotted PVDF membranes were blocked in 5% skimmed milk in PBS-T (0.1% Tween-20) at room temperature for 1 h and then incubated overnight at 4 °C with primary antibodies diluted in 3% BSA in PBS-T. After incubating the primary antibody, the PVDF membranes were washed three times with PBS-T for 10 min and incubated with the secondary HRP-conjugated antibody for 1 h at room temperature. Chemiluminescent signals were recorded using an Alliance Mini HD9 (UVITEC). The antibodies used were mouse monoclonal anti-ATP5A (Abcam, Ab14748) and rabbit polyclonal against D. melanogaster UQCR-C2 (custom-made and kindly provided by Dr. Edward Owusu-Ansah, Columbia University, NY). Densitometric analysis was performed using the Fiji software version 2.0.0.

Lactate quantification
Lactate content was measured using the Lactate-Glo Assay system (Promega). Briefly, 10 3rd instar larvae were homogenized in 150 μL of mammalian cell lysis solution on ice and centrifuged at 14,000 × g at 4 °C for 5 min to remove insoluble material. The supernatant was diluted 1:10 in HBSS (Thermo Fischer Scientific) and quantified according to the manufacturer's protocol. Lactate content was normalized by protein content assessed by Bradford method.

TAG quantification
For TAG quantification, freshly dissected 14 abdomens of 3-day-old adults (1:1 males-females) were analysed by coupled colorimetric assay (CCA) as previously described [45]. TAG content was normalized by protein content, as assessed by the Bradford method.

Morphological analysis of larval structures
Morphological analysis of larval mouth hooks was performed on whole-mount preparations, as described by Van Der Meer [46]. Samples were analysed with a Leica DMR microscope using 40 × DIC optics, and images were taken by a Nikon Eclipse 80i using a 40 × differential interference contrast (DIC) objective.

Statistical analysis
Statistical analysis was performed with GraphPad Prism Software, version 8.2.1. Statistical tests and significance are described in the figure captions.