Mitochondrial calcium uniporter in Drosophila transfers calcium between the endoplasmic reticulum and mitochondria in oxidative stress-induced cell death

Mitochondrial calcium plays critical roles in diverse cellular processes ranging from energy metabolism to cell death. Previous studies have demonstrated that mitochondrial calcium uptake is mainly mediated by the mitochondrial calcium uniporter (MCU) complex. However, the roles of the MCU complex in calcium transport, signaling, and dysregulation by oxidative stress still remain unclear. Here, we confirmed that Drosophila MCU contains evolutionarily conserved structures and requires essential MCU regulator (EMRE) for its calcium channel activities. We generated Drosophila MCU loss-of-function mutants, which lacked mitochondrial calcium uptake in response to caffeine stimulation. Basal metabolic activities were not significantly affected in these MCU mutants, as observed in examinations of body weight, food intake, body sugar level, and starvation-induced autophagy. However, oxidative stress-induced increases in mitochondrial calcium, mitochondrial membrane potential depolarization, and cell death were prevented in these mutants. We also found that inositol 1,4,5-trisphosphate receptor genetically interacts with Drosophila MCU and effectively modulates mitochondrial calcium uptake upon oxidative stress. Taken together, these results support the idea that Drosophila MCU is responsible for endoplasmic reticulum-to-mitochondrial calcium transfer and for cell death due to mitochondrial dysfunction under oxidative stress.

Mitochondrial Ca 2ϩ is a key regulator for cellular metabolic functions by activating Krebs cycle dehydrogenases, metabolite shuttle systems, and ATP synthase. However, inappropriately high Ca 2ϩ levels in the mitochondrial matrix threaten cell survival by increasing reactive oxygen species (ROS) 6 production and triggering mitochondrial permeability transition (mPT). The main route for Ca 2ϩ uptake into mitochondria is through mitochondrial calcium uniporter (MCU), a Ca 2ϩ -selective ion channel located at the inner mitochondrial membrane (1), which was originally identified as CCDC109A (2,3). Recent studies have demonstrated that MCU has a homopentameric structure in which the second transmembrane domain forms a hydrophilic pore across the membrane (4) and the N terminus domain modulates MCU activity by protein-protein interaction and binding divalent cations (5,6). To counteract the Ca 2ϩ influx via MCU, mitochondrial sodium-calcium exchanger (NCLX) provides mitochondrial Ca 2ϩ efflux routes by exchanging Ca 2ϩ for Na ϩ ions (7). Leucine zipper EF-handcontaining transmembrane protein 1 (LETM1) may also be involved in mitochondrial Ca 2ϩ influx or efflux, which is still in a controversy (8 -11).
Previous studies have identified multiple regulators of MCU activity, including mitochondrial calcium uptake 1 and 2 (MICU1/2), mitochondrial calcium uniporter regulator 1 (MCUR1), and essential MCU regulator (EMRE). EMRE is a 10-kilodalton mitochondrial inner membrane protein with a single transmembrane domain, which is an essential component that bridges MICU1/2 with MCU (12). The transmembrane helix of EMRE interacts with MCU, and the C terminus of EMRE binds to MICU1 (13). In addition, silencing of EMRE completely abolishes the channel activity of MCU (14). MICU1/2 are Ca 2ϩ -binding EF-hand-containing proteins residing in the mitochondrial intermembrane space. They can modulate MCU activity, depending on cytoplasmic Ca 2ϩ con-centration ([Ca 2ϩ ] i ) (15)(16)(17). MCUR1 is an inner mitochondrial membrane integral protein binding to MCU and regulates MCU-dependent mitochondrial Ca 2ϩ uptake (18) and Ca 2ϩ threshold for mPT (19).
Over the last few years, studies on the MCU complex have proposed diverse roles of MCU at the cellular and organism levels, using its genetic ablation and/or pharmacologic inhibition models. Although a loss-of-function mutation of MCU in mice exhibited negligible influence on metabolism and cell death (20), further studies revealed that MCU plays an essential role in heart rate acceleration during fight-or-flight response (21), ROS-mediated wound repair (22), and skeletal muscle trophism (23). In pancreatic ␤-cells, MCU silencing decreases mitochondrial ATP synthesis and impairs the metabolism-secretion coupling (11,24). In neuronal cells, suppression of MCU relieves ischemia and reperfusion injuries as well as Ca 2ϩ excitotoxicity (25,26). In cardiac cells, however, a loss-of-function mutation of MCU did not protect from ischemic injury due to [Ca 2ϩ ] i overload despite preserved mitochondrial membrane potential (⌬⌿ m ) and reduced ROS formation (27).
The main source of mitochondrial Ca 2ϩ is the endoplasmic reticulum (ER), which has a higher Ca 2ϩ level (Ͼ400 M) required for protein folding and Ca 2ϩ signaling. Ca 2ϩ release from the ER is taken up by mitochondria not only to avoid [Ca 2ϩ ] i accumulation but also to stimulate mitochondrial energy metabolism (28). This ER-mitochondrial interaction is mediated by physical contacts between the two organelles via mitochondria-associated ER membrane (MAM) (29,30). MAM is composed of several proteins, including inositol 1,4,5-trisphosphate receptor (IP 3 R), glucose-regulated protein 75 (grp75), porin, and MCU (31). The MAM formation can be expanded in pathologic conditions, such as obesity and diabetes, in which increased ER-mitochondrial Ca 2ϩ connection exacerbates ER Ca 2ϩ depletion and mitochondrial Ca 2ϩ overload (32).
Oxidative stress disrupts cellular Ca 2ϩ homeostasis and consequently induces cytotoxicity. Oxidative stress activates IP 3 R, stimulates ER Ca 2ϩ release, intensifies ER stress, and leads to apoptosis (33,34). More importantly, oxidative stress plays significant roles in the pathogenesis of various chronic diseases, including neurodegeneration (35). However, the pathophysiological role of Ca 2ϩ dysregulation by oxidative stress still remains elusive.
In this study, we generated Drosophila MCU loss-of-function mutants for the first time and characterized their phenotypes on metabolism, Ca 2ϩ handling, and cell death. We demonstrated that attenuated Ca 2ϩ transport from the ER to mitochondria in MCU mutants prevents ROS-induced mitochondrial dysfunction and cell death. Because the Drosophila system provides powerful tools for genetic studies, our loss-of-function mutants and transgenic flies for MCU and other components of MCU complex would provide crucial information for understanding the regulatory mechanism of mitochondrial Ca 2ϩ homeostasis.

Drosophila MCU null mutant is viable
According to a recent report, CG18769 is homologous to mammalian MCU (36). Human and mouse MCU protein sequences are similar to the protein sequence encoded by CG18769 (supplemental Fig. S1A). CG18769-encoded protein localizes to mitochondria, and silencing CG18769 decreases mitochondrial Ca 2ϩ entry (36,37). To assess the in vivo role of MCU, Drosophila MCU loss-of-function mutants were generated by P-element imprecise excision in the 5Ј-untranslated region of CG18769 using the P(GSV6)GS11565 fly line (Fig.  1A). From the 200 excision alleles obtained, we found MCU 52 mutant, which lacked 1,476 bp (3R14578001-14580477) that encoded the transcription start site and the first exon of MCU (Fig. 1, A and B). In the mutant, MCU mRNA and protein were not detected by quantitative RT-PCR (Fig. 1C) and immunoblotting (Fig. 1D). Wild-type MCU protein was expressed weakly in embryo stage but strongly in larva, pupa, and adult stages (Fig. 1E). MCU 52 mutant flies were viable and showed a similar survival rate compared with wild-type ones (Fig. 1E).

MCU 52 mutant does not show significant metabolic phenotypes
To find the exclusive role of MCU in Drosophila physiology, we investigated metabolic phenotypes of MCU 52 mutant. First, the body weight of MCU 52 mutants was not different from that of wild-type flies in both sexes (Fig. 1F). The amount of food intake of MCU 52 mutant was also similar to that of the wildtype fly (Fig. 1G). The concentration of circulating sugars in the hemolymph of MCU 52 mutants was not significantly different from that in wild-type flies (Fig. 1H). We also did not observe any detectable changes in starvation-induced autophagy (Fig.  1I). Collectively, these results showed that loss of MCU does not alter basal metabolism in Drosophila.

The Ca 2؉ channel activity of MCU is conserved in Drosophila
To assess mitochondrial Ca 2ϩ uptake in a physiological context, we measured mitochondrial matrix Ca 2ϩ concentration ([Ca 2ϩ ] mito ) in a larval muscle expressing mitochondria-targeted ratio-pericam (MTRP), a genetically encoded Ca 2ϩ indicator targeted to the mitochondrial matrix (38). MTRP was specifically expressed in muscle tissues by Mef-Gal4 and UAS-MTRP ( Fig. 2A). The localization of expressed MTRP to mitochondria was confirmed with two mitochondria markers, streptavidin ( Fig. 2B) and ATP5a (supplemental Fig. S2A).
To confirm that Drosophila MCU is a functional mitochondrial Ca 2ϩ uptake route, we compared [Ca 2ϩ ] mito increase in muscle tissues of control, MCU 52 mutant, and MCU 52 mutant expressing exogenous Drosophila wild-type MCU (Fig. 2C). Caffeine was used to stimulate Ca 2ϩ release from the ER, inducing elevation of cytosolic (supplemental Fig. S3A) and mitochondrial Ca 2ϩ levels (Fig. 2D). In control larvae, [Ca 2ϩ ] mito was increased sharply after caffeine stimulation and slowly returned to the basal level (Fig. 2D). However, although caffeine-induced [Ca 2ϩ ] i changes were not significantly different between control and MCU 52 mutant larvae (supplemental Fig.  S3A), MCU 52 mutant failed to elicit any [Ca 2ϩ ] mito change upon caffeine stimulation (Fig. 2, D and E). Furthermore, when exogenous Drosophila MCU was overexpressed in MCU 52 mutant, the absence of caffeine-induced [Ca 2ϩ ] mito response in the mutant was completely rescued (Fig. 2, D and E). These results consistently demonstrate the indispensable role of Drosophila MCU in mitochondrial Ca 2ϩ uptake.

Drosophila MCU in ROS-induced cell death
To test whether Drosophila MCU is functionally equivalent to the mammalian MCU, we expressed human MCU in the muscle of MCU 52 mutants (Fig. 2F). Similar to the results with Drosophila MCU, overexpression of human MCU in MCU 52 mutant larvae resulted in full recovery of [Ca 2ϩ ] mito response upon caffeine stimulation (Fig. 2, G and H). These results demonstrate that Drosophila MCU (encoded by CG18769) is a genuine orthologue of human MCU.
MCU has a mitochondrial targeting sequence at its N terminus, two coiled-coil domains, two transmembrane domains, and the DIME motif. Previous studies showed that substitution of two acidic amino acids within the DIME motif resulted in a dominant-negative effect on the uniporter activity of mamma-lian MCU (39,40). To test whether the DIME motif in Drosophila MCU is also critical for its Ca 2ϩ channel activity, we generated a mutant of MCU in the DIME motif (MCU NIMQ ) (Fig. 2I). Transgenic expression of wild-type MCU in the muscle of MCU 52 mutant resulted in full recovery of mitochondrial Ca 2ϩ uptake, as shown (Fig. 2, D and E). In contrast, expression of MCU NIMQ failed to rescue the defective [Ca 2ϩ ] mito response of MCU 52 mutant (Fig. 2, J and K). Therefore, the DIME motif is essential for its Ca 2ϩ transport activity in Drosophila MCU.

EMRE is required for MCU activity in Drosophila
Eye-specific expression of MCU using Gmr-Gal4 led to glazed phenotypes in the eye with partial loss of pigmentation

Drosophila MCU in ROS-induced cell death
and irregular ommatidial array (Fig. 3A). The activity of MCU appeared critical for these effects because knockdown of MCU completely suppressed the eye phenotype (Fig. 3A). Consistently, a higher MCU expression with two copies of UAS-MCU transgene resulted in more severe phenotypes (Fig. 3A). Overexpression of EMRE, an essential component of the MCU complex, alone did not elicit defective phenotypes (Fig. 3A). Unexpectedly, co-expression of MCU and EMRE led to lethality. However, when those flies were reared at 23°C instead of 25°C, few managed to survive with defective eyes marked with black mass tissues, indicating that MCU and EMRE have a strong genetic interaction in vivo (Fig. 3A).
To confirm the role of EMRE in Drosophila MCU activity, we used the UAS-EMRE RNAi fly line to knock down EMRE expression. Silencing of EMRE led to impairment of mitochondrial Ca 2ϩ uptake (Fig. 3, B and C), closely resembling the result obtained from MCU 52 mutant (Fig. 2, D and E), implying that EMRE is required for mitochondrial Ca 2ϩ uptake. Furthermore, expression of MCU in EMRE knockdown fly failed to increase the caffeine-induced [Ca 2ϩ ] mito response (Fig. 3, B and Drosophila MCU in ROS-induced cell death C). These results suggest that EMRE interacts with MCU genetically and is an essential and irreplaceable component for Drosophila MCU activity.

Loss of MCU provides resistance to oxidative stress
Oxidative stress is the most common pathogenic mediator for various diseases and is also involved in aging processes. To address whether mitochondrial Ca 2ϩ uptake via MCU has any pathogenic role in inducing cell death under oxidative stress conditions, we first investigated the survival rates of wild-type fly and MCU 52 mutant fed on hydrogen peroxide (H 2 O 2 )-containing food. Dihydroethidium (DHE) staining was used to detect the level of ROS in the thorax of flies. Elevated level of ROS was detected in the flies fed on 1% H 2 O 2 -containing food for the past 72 h, indicating that feeding H 2 O 2 induced oxidative stress in the fly (supplemental Fig. S4A). Interestingly, MCU 52 mutant flies survived significantly longer than wildtype flies when fed on 1% H 2 O 2 -containing food, whereas normal food caused the mutants to survive slightly less than wildtype flies. These results suggest that MCU 52 mutant flies are more resistant to oxidative stress than wild-type flies (Fig. 4A).
To demonstrate whether ROS-induced apoptosis was attenuated by loss-of-function mutations of MCU, we performed a TUNEL assay under oxidative stress conditions. Wild-type flies showed strong TUNEL signals by 2% H 2 O 2 treatment for 3 h, which was markedly reduced in MCU 52 mutants (Fig. 4B). Additionally, we checked hid5ЈF-WT-GFP reporter expression and cleaved caspase-3 as a marker of early and late phases of apoptosis, respectively (41). Wild-type flies showed up-regulated GFP reporter expression from hid5ЈF-WT enhancer and increased cleaved caspase-3 staining, but MCU 52 mutants displayed markedly reduced signals for both apoptotic markers ( Fig. 4 (C and D) and supplemental Fig. S4 (B and C)). Consistently, we also confirmed the reduction of H 2 O 2 -induced apoptotic cell death by knockdown of MCU in Drosophila S2 cells (Fig. 4, E and F). Taken together, these results indicate that loss of MCU endows resistance to oxidative stress.

MCU-dependent Ca 2؉ uptake contributes to mitochondrial dysfunction by oxidative stress
To more clearly demonstrate whether oxidative stress can increase [Ca 2ϩ ] mito in Drosophila, we applied tert-butyl hydroperoxide (t-BuOOH) to larval muscle and applied H 2 O 2 to S2 cells. First, in control larval muscle in vivo, [Ca 2ϩ ] mito was increased after treatment of t-BuOOH from 1 to 100 mM in a dose-dependent manner, suggesting that oxidative stress can increase [Ca 2ϩ ] mito (supplemental Fig. S5A). Compared with in vitro experiments, a higher dose of t-BuOOH is required to induce rapid oxidative stress on inner muscle tissue under the cuticle. Treatment of t-BuOOH (35 mM) on control flies

Drosophila MCU in ROS-induced cell death
rise in S2 cells upon H 2 O 2 stimulation (Fig. 5C). Consistent with the in vivo muscle data, MCU dsRNA-treated S2 cells showed reduced H 2 O 2 -induced mitochondrial Ca 2ϩ uptake (Ϫ63%) in comparison with control (Luciferase dsRNA-treated) (Fig. 5, C and D). To estimate the functional deterioration of mitochondria as a result of [Ca 2ϩ ] mito overload, we monitored mitochondrial membrane potential (⌿ mito ) by using potentialsensitive JC-1 dye. Oxidative stress by H 2 O 2 treatment (3 mM) elicited depolarization of ⌿ mito in S2 cells (Fig. 5E). Intriguingly, knockdown of MCU strongly prevented H 2 O 2 -induced ⌬⌿ mito collapse in S2 cells (Fig. 5, E and F). Based on these results, we suggest that mitochondrial dysfunction and apoptotic cell death induced by oxidative stress are related to MCUmediated [Ca 2ϩ ] mito overload.

IP 3 R and MCU participate in oxidative stress-induced ERmitochondria Ca 2؉ transfer
Oxidative stress is reported to activate IP 3 R, a Ca 2ϩ channel in the ER, resulting in ER Ca 2ϩ release (33,34). Released Ca 2ϩ from the ER provides high Ca 2ϩ level in microdomains of ERmitochondrial junction called MAM. IP 3 R is a component of tethering structure between the ER and mitochondria (31). Therefore, we investigated the involvement of IP 3 R in ER-mitochondria Ca 2ϩ transfer under oxidative stress.
Muscle-specific expression of exogenous MCU using Mef-Gal4 was lethal in pupa stage of the transgenic fly (Fig. 6A). However, this lethality was blocked by knockdown of IP 3 R, suggesting that MCU and IP 3 R have a strong genetic interaction in vivo (Fig. 6A). To confirm the role of IP 3 R in mitochondrial Ca 2ϩ uptake, we compared [Ca 2ϩ ] mito increase induced by oxidative stress between control (MefϾ) and muscle-specific IP 3 R knockdown flies (MefϾIP 3 R RNAi). The IP 3 R knockdown flies showed significantly reduced mitochondrial Ca 2ϩ uptake (Ϫ41.2%) upon t-BuOOH stimulation in comparison with control (MefϾ) (Fig. 6, B and C). Additionally, when we silenced IP 3 R in MCU transgenic flies (MefϾMCU, IP 3 R RNAi), the [Ca 2ϩ ] mito induced by t-BuOOH was reduced by 34.7% when compared with that of MefϾMCU flies (Fig. 6, B and C). This was consistent with our previous results, where knockdown of IP 3 R blocked the lethality induced by MCU overexpression (Fig. 6A). In S2 cells, H 2 O 2 -induced [Ca 2ϩ ] mito increase was also strongly inhibited (Ϫ64.0%) by transfection of IP 3 R dsRNA (Fig. 6, D and E). These results strongly indicate that both IP 3 R and MCU are critical for the Ca 2ϩ transfer from the ER to mitochondria.
Finally, we examined the involvement of IP 3 R and MCU in oxidative stress-induced toxicity using transgenic flies that overexpress or silence superoxide dismutase 1 (Sod1). First, the degenerated eye phenotype in the flies expressing MCU using Gmr-Gal4 driver, as shown in Fig. 3A, was restored by Sod1 expression but was exacerbated by Sod1 knockdown (Fig. 6F). As stated above, the lethality of MCU overexpression using Mef-Gal4 was prevented by the knockdown of IP 3 R (Fig. 6A). However, interestingly, simultaneous knockdown of both Sod1 and IP 3 R failed to rescue the lethal phenotype of MCU overexpression using Mef-Gal4 (Fig. 6G). These results suggest that endogenous ROS plays an imperative role in the IP 3 R-and MCU-mediated mitochondrial Ca 2ϩ overload and toxicity.

Discussion
In this study, we established a Drosophila model system to understand functional roles of MCU in mitochondrial Ca 2ϩ homeostasis in vivo. By generating and characterizing a null mutant of MCU (MCU 52 ), we investigated the physiological roles of MCU in Drosophila. MCU 52 mutant did not show significant changes in body weight, metabolism, and autophagic flux compared with wild-type flies. However, caffeine-induced [Ca 2ϩ ] mito increase was abolished in MCU 52 mutant larval

Drosophila MCU in ROS-induced cell death
muscle, which was completely rescued by transgenic expression of either human or Drosophila MCU. In addition, the DIME amino acid motif, which forms the ion selectivity filter of the MCU channel, was indispensable for the activity of Drosophila MCU, suggesting that MCU is evolutionarily highly conserved. In MCU 52 mutant larval muscle and MCU-silenced S2 cells, exogenous ROS-induced [Ca 2ϩ ] mito increase, ⌬⌿ mito dissipation, and cell death were prevented. Suppression of IP 3 R, which is a Ca 2ϩ release channel in the ER, also protects from ROS-mediated mitochondrial Ca 2ϩ overload and cytotoxicity.
These results demonstrate the critical role of Drosophila MCU in Ca 2ϩ transfer from the ER to mitochondria, contributing to oxidative stress-induced mitochondrial dysfunction and apoptotic cell death.
Mitochondrial Ca 2ϩ is a crucial regulator in energy metabolism, Ca 2ϩ sequestration, and cell death. However, our Drosophila MCU loss-of-function mutant did not exhibit significant metabolic phenotypes. These unexpected results can be explained by undefined compensatory mechanisms for mitochondrial Ca 2ϩ uptake, such as MCU-independent slow Ca 2ϩ

Drosophila MCU in ROS-induced cell death
channels or exchangers. It is also conceivable that rapid changes in [Ca 2ϩ ] mito via MCU may not be required for maintaining basal metabolism and daily activities in Drosophila except for exogenous stress or emergency crisis. Mouse MCU knock-out models with outbred CD1 background also did not show any significant phenotypes except for reduced abilities to perform strenuous work (20). However, strangely, MCU deletion within a C57BL/6 background resulted in embryonic lethality (42). This discrepancy suggests that a compensatory mechanism that is absent in C57BL/6 background exist in CD1 mice and allows MCU-independent mitochondrial Ca 2ϩ uptake during animal development. By contrast, up-regulation of MCU augments mitochondrial Ca 2ϩ uptake, leading to aberrantly high [Ca 2ϩ ] mito level, and this stress accelerates further superoxide production through activation of the electron transport chain or other mechanisms. Together with increased [Ca 2ϩ ] mito and oxidative stress in mitochondrial matrix, this facilitates opening of mPT and cytochrome c release, leading to apoptotic cell death. In our study, ectopic overexpression of MCU in Drosophila muscle led to pupal lethality (Fig. 6A). Additionally, MCU overexpression in Drosophila eye using Gmr-Gal4 driver resulted in severely destroyed ommatidial array (Fig. 3A). These results imply that overexpression of Drosophila MCU accelerates mitochondrial Ca 2ϩ overload that is detrimental to various tissues and ultimately impairs a viability of organism.
EMRE is an essential auxiliary subunit containing a mitochondrial targeting sequence at its N terminus, a single transmembrane domain located at inner mitochondrial membrane, and an aspartate-rich C terminus (12). The MCU complex requires EMRE for its reconstitution in mammalian cells, but not in Dictyostelium discoideum (43). In mammalian cells, suppression of EMRE abrogates MCU-mediated Ca 2ϩ currents demonstrated by mitoplast patch clamp experiment (12,14). However, functional consequences of EMRE overexpression have not been studied yet (44). In our study, EMRE showed a strong genetic interaction with MCU in Drosophila. Co-expression of Drosophila EMRE and MCU in fly muscle led to lethality in larva stage. Furthermore, depletion of EMRE in larval muscle abolished caffeine-induced [Ca 2ϩ ] mito increase regardless of the expression level of MCU, indicating that EMRE is required for MCU activity from human to Drosophila.
Recent studies reported that MCU is involved in Ca 2ϩ excitotoxicity of cortical neurons (26) and oxidative stress-induced cell death of primary cerebellar granule neurons (45). Oxidative stress is proposed as the main causative factor for neurodegenerative, cardiovascular, and other mitochondria-related diseases (46). Previous studies reported that oxidative stress increases [Ca 2ϩ ] i (47)(48)(49), although there are controversies whether the source of Ca 2ϩ is from the extracellular environment or intracellular stores. We observed marked and sustained [Ca 2ϩ ] mito rises by exogenous oxidative stress inducers, such as H 2 O 2 and t-BuOOH in Drosophila, which have not been investigated previously. Oxidative stress produced by endoplasmic reticulum oxidase 1␣ (ERO1␣) can stimulate Ca 2ϩ release from the ER by activating IP 3 R (33). Loss of Ca 2ϩ store in the ER by ROS induces ER stress due to impaired Ca 2ϩsensitive chaperone activities and further ROS production by induction of C/EBP homologous protein (33). In addition, depletion of Ca 2ϩ in the ER stimulates Ca 2ϩ influx from outside of the cell via store-operated Ca 2ϩ entry (50). Both Ca 2ϩ release from the ER and influx from extracellular environment burden cells with the pathology of mitochondrial Ca 2ϩ overload.
Our study clearly showed that loss-of-function mutation of MCU abrogated [Ca 2ϩ ] mito increases triggered by t-BuOOH in fly muscle. Moreover, knockdown of IP 3 R also significantly attenuated oxidative stress-induced [Ca 2ϩ ] mito rises, which explains the protective roles of IP 3 R RNAi against lethality in MCU-overexpressed flies (Fig. 6A). Oxidative stress can increase ER Ca 2ϩ release by activating not only IP 3 R but also ryanodine receptor (RyR), both of which are main Ca 2ϩ release channels in the ER (33,51). Although it has been known that RyR plays more important roles than IP 3 R in Ca 2ϩ release from the sarcoplasmic reticulum in skeletal muscle, the muscle tissues from larva, pupa, and adult Drosophila express IP 3 R, which is critical for the muscle development (52). In this study, we could not investigate the interaction between RyR and MCU for oxidative stress-induced mitochondrial Ca 2ϩ overload and lethality because all of the flies with muscle-specific knockdown of RyR died before larval stage. However, we still consider that RyR is another strong candidate involved in mitochondrial Ca 2ϩ overload by oxidative stress in muscle.
In summary, we have established a Drosophila system to study the MCU complex in vivo, demonstrating that oxidative stress induces Ca 2ϩ release from the ER and mitochondrial Ca 2ϩ uptake via MCU, resulting in mitochondrial dysfunction and cell death in vivo. Intriguingly, a recent study showed that ROS modifies MCU directly by S-glutationylation and consequently influences the channel activity of MCU (53). These pieces of evidence indicate both direct and indirect regulation of the Ca 2ϩ channel activity of MCU by ROS. Further genetic studies would enable us to discover novel relationships connecting mitochondrial Ca 2ϩ homeostasis and other cellular activities. In addition, studying the in vivo role of MCU and its related genes in Drosophila will further extend our knowledge of their pathophysiological significance.

Experimental procedures
Fly strains MCU 52 mutant and a revertant were generated using P-element excision of the GS11565 line obtained from the Kyoto Stock Center (Kyoto Institute of Technology, Kyoto, Japan). The revertant with a precise P-element excision was used as a wild-type control. The P-element excisions in MCU 52

Food intake assay
To measure feeding activity for 24 h, the food intake assay was conducted as described previously (55).

Measurement of trehalose and glucose
Trehalose and glucose in fly body fluid were measured as described previously (56). For each genotype, hemolymph from 5-7 wandering larvae was extracted by tearing up the cuticles. 1 l of hemolymph was diluted with 99 l of trehalase buffer (5 mM Tris, pH 6.6, 137 mM NaCl, and 2.7 mM KCl) and incubated at 70°C for 5 min. Then 40 l of diluted hemolymph was mixed with either 40 l of trehalase buffer or 40 l of trehalase solution. Trehalase solution was prepared by diluting 3 l of porcine trehalase (Sigma) in 1 ml of trehalase buffer. Samples were incubated at 37°C overnight, and glucose levels were measured using a glucose assay kit (Sigma-Aldrich).

Starvation-induced autophagy assay
Third instar larvae before wandering stage were rinsed with PBS and either starved in a double-distilled water-containing Petri dish or fed in a food-containing vial for 4 h. Then larvae were dissected and fixed in 4% paraformaldehyde. mCherry-ATG8a signals in larval fat body were observed under a confocal microscope.

TUNEL assay
To detect H 2 O 2 -induced cell death, wandering larvae were dissected and incubated for 3 h in Schneider's medium containing 2% H 2 O 2 . Dissected larvae were fixed in 4% paraformaldehyde (PFA) and washed with PBS. Samples were incubated in 0.1 M sodium citrate at 65°C, and cell death was detected using an in situ cell death detection kit (Roche Applied Science). After TUNEL reaction, the samples were stained by phalloidin and Hoechst to detect filamentous actin and nucleus, respectively. To detect in S2 cells, S2 cells were incubated for 6 h in Schneider's medium-containing 100 mM H 2 O 2 . After fixation with 4% PFA for 15 min and washing with PBS, cells were incubated in 0.1 M sodium citrate at room temperature for 10 min, and apoptosis was detected by using an in situ cell death detection kit (Roche Applied Science).

DHE staining
For ROS detection, adult fly thoraces were dissected in Schneider's medium, incubated for 5 min with 30 M DHE in the same medium, washed twice, fixed slightly with 4% PFA for 8 min, and rinsed with PBS.

Drosophila S2 cell culture and dsRNA bathing
Drosophila S2-DRSC cells were cultured and dsRNA bathing was conducted as described previously (57). For silencing mRNA expression of MCU and IP 3 R, dsRNA was synthesized and bathed. The following primers were used for dsRNA synthesis: MCU dsRNA, 5Ј-TAATACGACTCACTATAGGGTG-GAGGATGTGAAGAATCGC-3Ј and 5Ј-TAATACGACT-

Calculation of [Ca 2؉ ] mito increases
In every scattered plot, the y axis represents area under the curve of the [Ca2 ϩ ] mito imaging result under caffeine, t-BuOOH, or H 2 O 2 treatment.

Statistics
Values are presented as mean Ϯ S.D., and n is the number of independent experiments. p values were obtained by Student's t test or one-way analysis of variance, and Ͻ0.05 was considered to be significant.
Author contributions-S. C., K.-S. P., and J. C. conceived and designed the experiments; S. C., X. Q., S. B., H. Y., and J. K. performed the experiments; K.-S. P. and J. C. analyzed the data; and S. C., J. K., S. B., J. P., K.-S. P., and J. C. wrote the paper.