Human mitochondrial disease-like symptoms caused by a reduced tRNA aminoacylation activity in flies

The translation of genes encoded in the mitochondrial genome requires specific machinery that functions in the organelle. Among the many mutations linked to human disease that affect mitochondrial translation, several are localized to nuclear genes coding for mitochondrial aminoacyl-transfer RNA synthetases. The molecular significance of these mutations is poorly understood, but it is expected to be similar to that of the mutations affecting mitochondrial transfer RNAs. To better understand the molecular features of diseases caused by these mutations, and to improve their diagnosis and therapeutics, we have constructed a Drosophila melanogaster model disrupting the mitochondrial seryl-tRNA synthetase by RNA interference. At the molecular level, the knockdown generates a reduction in transfer RNA serylation, which correlates with the severity of the phenotype observed. The silencing compromises viability, longevity, motility and tissue development. At the cellular level, the knockdown alters mitochondrial morphology, biogenesis and function, and induces lactic acidosis and reactive oxygen species accumulation. We report that administration of antioxidant compounds has a palliative effect of some of these phenotypes. In conclusion, the fly model generated in this work reproduces typical characteristics of pathologies caused by mutations in the mitochondrial aminoacylation system, and can be useful to assess therapeutic approaches.


INTRODUCTION
Aminoacyl-tRNA synthetases (aaRSs) constitute an ancient family of enzymes that catalyze the attachment of amino acids onto their cognate transfer RNAs (tRNAs). The enzymes carry out a two-step reaction that first condenses the amino acid with ATP to form the aminoacyl adenylate and then transfer the aminoacyl moiety to the tRNA 3 0 end (1). The aminoacyl-tRNA is then delivered to the ribosome by elongation factors for the decoding of the messenger RNA (mRNA) according to genetic code rules. In animals, as in the vast majority of eukaryotes, protein synthesis occurs simultaneously in the cytoplasm and some organelles that possess their own genome. Human mitochondria have a circular doublestranded DNA genome (mtDNA) that codes for 13 polypeptides that are components of the respiratory chain and the oxidative phosphorylation (OXPHOS), responsible for supplying energy to the cell. Additionally, human mtDNA codes for two ribosomal RNAs and the 22 mitochondrial tRNAs (mt-tRNAs) required to decode all human mitochondrial mRNA codons. To aminoacylate these 22 tRNAs, a whole set of nuclear-encoded aaRS needs to be imported and function inside the organelle.
Defects in elements involved in mitochondrial protein synthesis are related to a heterogeneous number of mitochondrial diseases, which show diverse clinical symptoms including deafness, blindness, encephalopathy and myopathy. More than 50% of the known mtDNA mutations are concentrated in tRNA genes and associated to a wide variety of ailments (2). For example, mutations in mt-tRNA Leu (UAA) gene cause mitochondrial encephalomyopathy, lactic acidosis and stroke-like episodes (MELAS) (3,4), in mt-tRNA Lys produce myoclonic epilepsy with ragged red fibers (MERRF) (5) and in tRNA Ser typically lead to deafness (6). A poor understanding of the pathophysiology of mitochondrial translation diseases, the wide variety of symptoms they cause and the technical difficulty of working with mutant mitochondria complicate the research on these disorders. For that reason, the construction of model systems is necessary to facilitate the characterization, diagnosis and development of therapeutic approaches.
Some animal models for neurological conditions caused by mutations in cytoplasmic aaRS have been generated. For example, mice and Drosophila melanogaster have been used as models to study the CMT symptoms caused by substitutions in the glycyl-tRNA synthetase gene (GARS) (24)(25)(26) and tyrosyl-tRNA synthetase gene (YARS) (27). Similarly, the sequence identity between some human and Saccharomyces cerevisiae mt-tRNAs (28,29) has allowed the generation of yeast strains with mutations in mt-tRNA that mimic some human neuropathies such as MELAS (30). Trans-mitochondrial cybrid cell lines (31) have also been used to study the biochemical and cellular consequences of point mutations and deletions of mtDNA, including those affecting tRNA genes (32). A few animal models for mitochondrial illnesses caused by mutations in nuclear-encoded components of mitochondrial gene translation exist, such as the D. melanogaster technical knockout (tko) that carries a point mutation in the MRPS12 nuclear gene encoding mitochondrial ribosomal protein S12 (33). Finally, allotropic expression of functional tRNA derivatives in cybrid cells holding MERRF or MELAS mutations (34)(35)(36)(37), as well as overexpression of mitochondrial aaRS (38,39), have been tested as therapeutic approaches. Blastocyst injection of ES cell cybrids has allowed the creation of heteroplasmic trans-mitochondrial mice bearing mutant mtDNA from donor cells (40)(41)(42)(43).
The aim of the present work is the development of a D. melanogaster model to study the cellular and molecular effects of a deficient mt-tRNA aminoacylation activity. To achieve this goal, and avoid the drawbacks of manipulating mtDNA-encoded tRNAs, we have decided to limit the function of a nuclear-encoded mitochondrial aaRS, seryl-tRNA synthetase 2 (DmSRS2) (Figure 1), by means of an RNA interference (RNAi) approach. We have previously reported the identification of SRS2 in D. melanogaster, and its preliminary comparison with a paralogous protein present in insects (SLIMP) (44).
Here we describe the generation of a D. melanogaster model for human mitochondrial disease caused by mitochondrial aminoacylation restriction. The model has been first characterized at the molecular level, showing a decrease in DmSRS2 expression and function, with a reduction in mt-tRNA Ser aminoacylation. Secondly, the effect of DmSRS2 general or tissue-restricted depletion has been analyzed. This insult compromises viability, longevity, motility and tissue development. At the cellular level, SRS2 silencing strongly affects mitochondrial morphology, biogenesis and function, and induces lactic acidosis and reactive oxygen species (ROS) accumulation. The later observation prompted us to investigate the effect of administering antioxidant molecules to the affected animals. We report that such a treatment has a palliative effect and reduces the severity of some of the phenotypes caused by the silencing of the enzyme.

Fly maintenance and strains
Flies were maintained in 60% relative humidity in 12 h light/dark cycles at 18, 25 or 29 C depending on the experiment. Flies were fed with standard fly food except when testing the effect of antioxidant molecules in the phenotype. In these cases, the micronutrient supplement K-PAX (K-PAX Inc.) was added to fly's food at a final concentration of 6.9 mg/ml. To determine the chromosomal location of the UAS-RNAi transgenes and for the proper manipulation of the transgenic flies, the balancer line w; If/CyO; Ly/TM3-Sb was used. UAS and GAL4 lines used in this study were as follows: w; UAS-dicer-2, yw; actin5C-GAL4/TM6B, w; patched-GAL4, w; repo-GAL4/TM6B, yw; Mef2-GAL4, and the yw; nubbin-GAL4; UAS-dicer-2/CyO-TM6B that ensures co-segregation of nubbin-GAL4 with UAS-dcr2.

Generation of transgenic UAS-RNAi strains
A 543 bp fragment from the DmSRS2 cDNA was subcloned into pWIZ vector in an inverted repeat manner (45). Transgenic fly lines were obtained by microinjection of the construction into w 1118 embryos using standard procedures (46). One homozygous strain was obtained carrying the UAS-RNAi DmSRS2 transgene in chromosome II, which on crossing gave rise to the w; RNAi DmSRS2 strain 1-dcr2 homozygous strain, that expressed both the RNAi DmSRS2 and dicer-2 protein. In addition, an independent strain carrying a different UAS-RNAi DmSRS2 transgene in chromosome II (RNAi DmSRS2 strain 23003) and a line used to silence the respiratory chain subunit ND75 from complex I (RNAi ND75 ), were purchased from the VDRC stock centre (ID 23003 and 100733, respectively) (47). Induction of RNAi transgene expression was based on the UAS-GAL4 system (48).

Viability and life span determinations
To measure adult viability, crosses with the heterozygous actin5C-GAL4 driver were maintained at 25 and 29 C and progeny was counted to n > 150. The dsRNA was expected to be expressed in 50% of the progeny, while the remaining 50% should not produce it and was used as internal negative control. Adults with active RNAi DmSRS2 were counted and represented relative to the maximum expected viability, set as 100%. For life span experiments, crosses with repo-GAL4 driver were kept at 29 C and with Mef2-GAL4 driver at 18 C until adulthood to allow viability. For each experiment, !100 adults were collected, transferred to fresh food vials every two days, maintained at 29 C and counted daily. Survival curves were constructed and compared using the Log-rank (Mantel-Cox) method.

Quantitative real-time polymerase chain reaction
Total RNA was extracted from third instar larvae with TRIzol (Invitrogen), digested with DNase I and cleaned with the RNeasy MinElute Cleanup kit (Qiagen). One microgram of total RNA was retrotranscribed into cDNA using oligo(dT) primers to perform quantitative real-time polymerase chain reactions (PCRs) by means of Power SYBR Green and a StepOnePlus Real-time PCR System (Applied Biosystems). cDNA templates were amplified with a pair of primers designed with the Primer Express Õ software (Applied Biosystems) to detect the DmSRS2 cDNA (5 0 CCGTTCTGCGACCATTCAT3 0 and 5 0 CAGCTTCGTCTCCGGTATCC3 0 ) and another to detect the Rp49 cDNA, used as endogenous control (5 0 TG CCCACCGGATTCAAGA3 0 and 5 0 AAACGCGGTTCT GCATGAG3 0 ). Standard curves were calculated for both primer pairs to ensure a high efficiency level. Twenty microliters of reactions were prepared following the manufacturer's instructions, using ROX as reference dye and the following conditions: 50 C for 2 min; 95 C for 10 min; 40 cycles (95 C for 15 s; 60 C for 1 min). Fold expression changes were calculated using the 2 ÀÁÁCT method, where ÁÁC T is the RNAi ÁC T [C T average for DmSRS2 À C T average for the reference gene (Rp49)] À the w 1118 control ÁC T [C T average for DmSRS2 À C T average for the reference gene (Rp49)]. The value obtained for control larvae is represented as 1 and the other values are represented relative to it.

Analysis of in vivo mt-tRNA Ser aminoacylation
Total RNA was extracted with TRIzol (Invitrogen) from third instar larvae with inactive or active RNAi DmSRS2 and 30 mg of total RNA were electrophoresed on highresolution acid gels, as described in (50) and transferred to a Hybond XL (GE Healthcare) membrane by vacuum gel drying transfer (51). Aminoacylated mt-tRNAs were analyzed by northern blot using the following radiolabeled probes: 5 0 TGGTCATTAGAAGTAAGTGCTAATTTA C3 0 for mt-tRNA Lys (CUU), used as a control, 5 0 TGGA GAAATATAAATGGAATTTAACC3 0 for mt-tRNA Ser (GCU) and 5 0 TGGAAGTTAATAGAAAATTAAATTC TATCTTATG3 0 for mt-tRNA Ser (UGA). Signals were digitalized using a PhosphorImager TM from a gel exposed storage phosphor screen and were quantified using the ImageQuant TM TL software (GE Healthcare).

Wing preparation and microscopy image analyses
Twelve or more adults were kept at room temperature in 75% ethanol, 25% glycerol for >24 h and wings were excised in cold phosphate buffered saline (PBS) and mounted in Faure´'s medium. Images were taken in a Nikon E600 microscope with an Olympus DP72 camera, and L3-L4 areas were measured from males and females separately with the ImageJ software (52). Images from whole flies were taken at 30Â with a MZ 16F Leica stereomicroscope equipped with a DFC 300FX camera.

Electron microscopy
Fat bodies from 8 to 10 third instar larvae were dissected in Schneider's medium and fixed in 2% glutaraldehyde in 0.1 M cacodylate buffer pH 7.2. Postfixation was done with 2% OsO 4 and 1.6% K 3 Fe(CN) 6 in cacodylate buffer. Sections were contrasted with uranyl acetate and visualized in a Jeol JEM 1010 electron microscope with a high-resolution digital camera. For mitochondrial surface determination, >70 mitochondria were measured, and for mitochondrial density calculation, 35 images at 20 000Â were analyzed for each sample. The surface occupied by glycogen was calculated taking >15 images at 20 000Â for each sample, and it was represented as the area covered by glycogen in mm 2 in 100 mm 2 of total area (subtracting the mitochondrial and lipid droplet surface). Image analyses were performed with the ImageJ software (52). mtDNA copy number determination mtDNA was quantified as described (44).

Lactate, glycogen and pH determination
Concentration of lactate in larval tissues was measured as described (53) with a Cobas Mira Plus analyzer (Roche). Concentration of glycogen in larval tissues was measured adapting the protocol described (54) to a Safire 2 fluorometer (Tecan Group Ltd.). For both experiments, 20 larvae were used for each determination, and results were normalized by total protein, measured using BCA Protein Assay Kit (Pierce). pH of larval homogenates was measured at 4 C with a pH meter GLP21 (Crison).

Oxygen consumption measurements
Oxygen consumption was measured with an Oxygraph-2 k (Oroboros) as described (44). Complex I respiratory control ratios (RCR) were calculated using the oxygen consumption in the presence and absence of ADP (GM D and GM). The oxygen flux values in different respiratory states were normalized dividing them by the mitochondrial rate calculated from the relative mtDNA quantification (55) considering the rate for control larvae as 1 and representing the other values relative to it.

ROS assays
Wing imaginal discs were incubated with dihydroethidium (Invitrogen), rinsed twice in Schneider's medium, fixed with 4% P-paraformaldehyde and washed with PBS (56). Superoxide anion accumulation was visualized by confocal imaging.

Statistical analyses
All the statistical analysis were performed using the software GraphPad Prism version 5.00, except the twoway analysis of variance (ANOVA) test that was performed with the SPSS 15.0 software.

DmSRS2 knockdown causes an aminoacylation deficiency
To partially reduce the levels of the D. melanogaster SRS2 in an inducible and regulated manner, we used RNAi expression under the control of the UAS-GAL4 system. We used two fly strains holding two RNAi transgenes designed to specifically target two different regions of the DmSRS2 mRNA (RNAi DmSRS2 strain 1-dcr2 and RNAi DmSRS2 strain 23003). When the RNAi DmSRS2 strains were crossed with the actin5C-GAL4 driver, the expression of the RNAi transgenes was constitutively and ubiquitously induced in the offspring. Because GAL4 activity is temperature dependent, crosses with actin5C-GAL4 were maintained at the highest temperature that allowed viability of the progeny, to ensure high efficiency of the RNAi silencing. Thus, the cross with RNAi DmSRS2 strain 1-dcr2 was maintained at 29 C, while the one with RNAi DmSRS2 strain 23003 was kept at 25 C. Both RNAi transgenes produced a reduction in DmSRS2 mRNA levels in larvae (Figure 2A), with different efficiencies depending on the strain used: RNAi DmSRS2 strain 1-dcr2 showed a mild effect with limited reduction to 0.796 ± 0.040, and RNAi DmSRS2 strain 23003 displayed a strong effect with a marked decrease to 0.163 ± 0.001, while the mRNA levels of the cytosolic DmSRS (DmSRS1) and the DmSRS2 paralogous SLIMP did not change significantly (Supplementary Figure S1). The availability of two strains with different efficiencies to silence DmSRS2 allowed us to investigate the range of phenotypes resulting from different degrees of silencing. This is reminiscent of the different severity levels and variety of symptoms that occur in most mitochondrial diseases.
To prove that the RNAi DmSRS2 was decreasing the function of the enzyme, the relative levels of aminoacylated and deacylated mitochondrial tRNA Ser (mt-tRNA Ser ) from larvae were analyzed by northern blot (Figure 2B), and the levels of in vivo aminoacylation under general RNAi induction were quantified and compared with RNAi inactive larvae ( Figure 2C).
In Figure 2B, intensity of the upper bands (corresponding to aminoacylated tRNAs) for the two mt-tRNA Ser isoacceptors (GCU and UGA) was reduced when RNAi was functional (+), while mt-tRNA Lys (CUU) was maintained completely aminoacylated, showing a single band (Supplementary Figure S2). RNAi DmSRS2 (+) larvae from strain 1-dcr2 showed a moderate decrease ( Figure 2C) in mt-tRNA Ser (GCU) aminoacylation level to 82.10 ± 0.01%, while larvae from strain 23003 showed a strong reduction to 50.44 ± 3.31%. Levels of mt-tRNA Ser (UGA) aminoacylation were similar in both strains when RNAi was functional: 37.88 ± 5.3% (strain 1-dcr2) and 38.06 ± 4.14% (strain 23003). Indeed, we observed a reduction of the mitochondrial encoded proteins ND1 (MT-ND1) and COX2 (MT-CO2) in knockdown larvae from RNAi DmSRS2 strain 1-dcr2 at 29 C (5.2% and 43.6%, respectively) and strain 23003 at 25 C (48.7 and 51.3%, respectively) compared with the control w 1118 (Supplementary Figure S3). These results confirmed that DmSRS2 was the D. melanogaster mitochondrial seryl-tRNA synthetase, as its silencing specifically diminished the amount of seryl-mt-tRNA Ser and mitochondrially encoded proteins. Therefore, the RNAi of DmSRS2 was confirmed to be a useful approach to limit the mitochondrial translation capacity in a controlled manner.

DmSRS2 silencing influences viability and tissue development
To study the phenotype caused by DmSRS2 depletion, we crossed the actin5C-GAL4 driver with the two transgenic RNAi DmSRS2 strains. At 25 C, 21.8% of strain 1-dcr2 pupae, and 1.4% of strain 23003 pupae, hatched successfully. At 29 C, temperature at which GAL4 activity is maximum (57), no adult animals could be observed to hatch from pupae ( Figure 3A). As expected for an aaRS, silencing of DmSRS2 impedes viability.
We used tissue-restricted RNAi induction to investigate effects caused by DmSRS2 silencing without compromising the viability of the organisms (Figure 3B and C). To express the RNAi in the wing imaginal disc region, which afterward develops into the adult wing blade and hinge, we crossed the RNAi DmSRS2 strains with nubbin-GAL4; UAS-dcr2 driver ( Figure 3B). As shown in the right panel, a 32.3% (25 C) and a 97.7% (29 C) of RNAi DmSRS2 strain 1-dcr2 flies showed tissue damage in wings, while all the RNAi DmSRS2 strain 23003-dcr2 flies presented wing defects both at 25 and 29 C. Although the wings conserved their general structure, the blade was unable to unfold and develop completely (left panel).
To analyze the cellular effects of DmSRS2 silencing in the wing, we crossed the RNAi transgenic lines with patched-GAL4 driver at 29 C, which allows for a better definition of the RNAi expressing region ( Figure 3C). The gene patched is expressed in the anteroposterior border of the wing disc, which gives rise to the wing area limited by longitudinal veins 3 (L3) and 4 (L4) (58). A partial or total loss of the anterior cross vein (acv) was observed in 4 and 15% of the wings from flies emerging from the crosses with strain 1-dcr2 and strain 23003, respectively. Moreover, in these animals, the L3-L4 wing area underwent a significant narrowing of 88.58 ± 2.13% (strain 1-dcr2) and 80.83 ± 2.54% (strain 23003), compared with the parental strain (patched-GAL4). The reduction in L3-L4 area was not due to a decrease in cell size, but in cell number, although there was no evidence of apoptosis when wing imaginal discs were subjected to Caspase-3 immunofluorescence (data not shown). Because the L2-L3 area, contiguous to L3-L4, did not show a significant decrease (data not shown), the effect observed in the L3-L4 region was considered cell autonomous.

Life span and motility are reduced by DmSRS2 knockdown
With the purpose of disrupting Drosophila mitochondrial translation in tissues typically affected in patients with mitochondrial pathologies, DmSRS2 silencing was restricted to neural and muscle cell types using repo-GAL4 and Mef2-GAL4 drivers, respectively (Figure 4). In Figure 4A, the survival curve for RNAi DmSRS2 strain 1-dcr2 flies with a repo-GAL4 driver at 29 C was significantly different to the parental line repo-GAL4, with a reduction in half-life from 32 to 21 days. Furthermore, repo-GAL4; RNAi DmSRS2 strain 23003 individuals were RNAi DmSRS2 larvae from strains 1-dcr2 and 23003 crossed with actin5C-GAL4 driver at 29 and 25 C, respectively. DmSRS2 mRNA levels were normalized using Rp49 mRNA as reference. Graph gives average with SEM from three independent experiments, and statistical significance is calculated by Student's t-test (*P < 0.05; ***P < 0.001). The mRNA level in control larvae is established as 1 and the other values are relative to this. (B) Aminoacylated and deacylated mitochondrial tRNAs (mt-tRNAs) were detected by northern blot from larvae with inactive (À) and active (+) RNAi for DmSRS2, coming from strain 1-dcr2 and 23003 crossed with actin5C-GAL4 driver at 29 and 25 C, respectively. Thirty micrograms of total RNA were loaded into high-resolution acid gels, and probes were designed to specifically detect the two mitochondrial tRNA Ser isoacceptors (GCU and UGA) and the mitochondrial tRNA Lys (CUU) as control. (C) The graph shows, for each lane from panel B, the relative abundance of aminoacylated  The left panel shows images of adults from the crosses between nubbin-GAL4; UAS-dcr2 driver and RNAi DmSRS2 strain 1 and strain 23003, which suffer severe wing damage, and control flies (w 1118 ). Scale bars correspond to 500 mm. Graph on the right represents the proportion of adults that exhibit wing defects when crosses are kept at 25 and 29 C. (C), Patched-GAL4 driver is crossed at 29 C with RNAi strains to restrict the DmSRS2 depletion in the region flanked by longitudinal veins L3 and L4. The images on the left show wings with a partial or total loss of the anterior cross vein (marked with an arrowhead) and a reduction in the L3-L4 area, compared with the parental line (patched-GAL4). Scale bars correspond to 500 mm. Graph on the right shows the averages with SEM of all the L3-L4 area measurements in percentage, compared with the parental line and statistics are performed by two-way ANOVA test (***P < 0.001). unable to hatch from pupal stage, at any tested temperature (18, 25 and 29 C).

DmSRS2 depletion alters mitochondrial morphology and biogenesis
After the characterization of tissue and cellular defects caused by DmSRS2 silencing, we aimed to determine the effects at the subcellular level. Fat bodies from control larvae, or from the cross between actin5C-GAL4 and RNAi DmSRS2 strain 1-dcr2 , at 29 C, were visualized by transmission electron microscopy (TEM). SRS2 silencing notably affected mitochondrial ultrastructure ( Figure 5A) compared with control mitochondria. Mitochondria under DmSRS2 depletion were characterized by swollen matrices with low electron density, an evident reduction of cristae and a significant enlargement ( Figure 5B), with an average mitochondrial surface of 0.920 ± 0.08 mm 2 , which represents a 66% increase relative to wt mitochondrial area (0.555 ± 0.05 mm 2 ).
To evaluate the effect on mitochondrial biogenesis of RNAi DmSRS expression, density of mitochondria was estimated by the quantification of relative mtDNA copy numbers. Larvae under DmSRS2 depletion showed an increase in mtDNA copy number ( Figure 5C left panel). We observed a rise to 169.30 ± 8.61% when using RNAi DmSRS2 strain 1-dcr2 as parental line (at 29 C) and to 116.60 ± 4.39% when using strain 23003 (at 25 C). These results are in agreement with an increase in mitochondrial number observed in the electron micrographs, in which RNAi DmSRS2 affected adipocytes possessed 8.94 ± 0.73 mitochondria/100 mm 2 of cell surface, compared with 7.01 ± 0.79 mitochondria/100 mm 2 of cell surface in wild-type cells ( Figure 5C right panel).

Mitochondrial translation deficiency leads to lactic acidosis and reduction of glycogen
One of the first signs of mitochondrial pathology in patients is the appearance of lactic acidosis, a physiological state caused by mitochondrial metabolism deficiency. Low levels of ATP synthesis result in the accumulation of cytosolic pyruvate, which is converted to lactate to satisfy energy demands (59). Simultaneously, inefficient mitochondrial respiration leads to the acidification of blood and tissues. Lactate concentration in samples from control flies had lactate values of 1.41 ± 0.11 nmol/mg protein, and 1.44 ± 0.11 nmol/mg protein, at 25 and 29 C, respectively ( Figure 6A). In larval tissues under RNAi DmSRS2 activation, this concentration rose to 2.44 ± 0.27 nmol/mg protein and 2.7 ± 0.34 nmol/mg protein at 25 and 29 C, for strain 1-dcr2. Similarly, strain 23003 had lactate values of 2.53 ± 0.3 nmol/mg protein at 25 C. In agreement with Values from at least four assays for each genotype were averaged, represented with SEM and compared by Student's t-test (**P < 0.01, ***P < 0.001). these findings, the pH of DmSRS2-silenced larvae decreased from 7.19 ± 0.04 in w 1118 , to 7.05 ± 0.07 in RNAi DmSRS2 strain 1-dcr2, and to 7.07 ± 0.02 in strain 23003, at 25 C. Moreover, larvae maintained at 29 C showed a decrease in pH from 7.12 ± 0.04 to 6.70 ± 0.12 in RNAi DmSRS2 strain 1-dcr2 at 29 C ( Figure 6B).
Larval glycogen levels for control flies were also measured, obtaining a concentration of 0.82 ± 0.07 mg/ mg protein at 25 C. The larval glycogen concentration at 25 C from strain 1-dcr2 and strain 23003 were 0.42 ± 0.06 mg/mg protein and 0.61 ± 0.04 mg/mg protein, respectively ( Figure 6C, left panel). These results are in agreement with a reduction in the area occupied by glycogen observed in the electron micrographs ( Figure 6C, right panel). Thus, the silencing of DmSRS2 leads to a statistically significant reduction in cellular glycogen.
Taken together, our results may indicate that an increase in lactate production and a reduction in stored glycogen are linked consequences of a decreased capacity for mitochondrial protein synthesis. and a loss of mitochondrial cristae (marked with arrowheads). Scale bars correspond to 1 mm, 500 nm and 200 nm, from left to right. (B) TEM images were used to measure mitochondrial surface in w 1118 and RNAi DmSRS2 strain 1-dcr2 fat bodies. Columns represent the mitochondrial surface mean with SEM, and Student t-test is performed to determine statistical significance (***P < 0.001). (C) The left graph shows the relative mtDNA copy number determination in control larvae (w 1118 ) and larvae from the cross between strains RNAi DmSRS2 1-dcr2 or 23003 and actin5C-GAL4 at

RNAi DmSRS2 compromises mitochondrial respiration and triggers ROS build-up
Because mitochondrial morphology, biogenesis and metabolism were altered in DmSRS2-silenced larvae, we decided to test if they were accompanied by a decrease in mitochondrial respiratory capacity. We monitored the mitochondrial oxygen consumption of larval tissues on addition of substrates and inhibitors of the respiratory chain and OXPHOS complexes ( Figure 7A). Complex I RCR [oxygen consumption under ADP addition (GM D ; state 3) divided by oxygen consumption limited by ADP (GM; state 2)] in larvae subjected to DmSRS2 general knockdown were lower than in control animals, indicating an uncoupling between the respiratory chain and OXPHOS (60). The RCR for RNAi DmSRS2 strain 1-dcr2 mitochondria showed a decrease to 2.45 ± 0.20, compared with control (3.10 ± 0.17) at 29 C and for RNAi DmSRS2 strain 23003 a reduction to 2.89 ± 0.07, in comparison with 3.95 ± 0.34 in w 1118 at 25 C ( Figure 7A, upper graphs). In larvae obtained from crosses between actin5C-GAL4 line and RNAi DmSRS2 strain 1-dcr2 (at 29 C) or strain 23003 (at 25 C), oxygen consumption values (normalized by mitochondrial density) showed a notable decrease, and thus the respiratory capacity per mitochondria is lower in RNAi DmSRS2 affected tissues ( Figure 7A, lower graphs). The increment in mitochondrial surface and number might be a compensatory response to the limited mitochondrial function caused by the DmSRS2 depletion.
To synthesize ATP by OXPHOS, the mitochondrial respiratory chain complexes transport electrons that are finally transferred to the molecular oxygen. When the respiratory chain and OXPHOS function incorrectly, ROS accumulate inside mitochondria and can induce oxidative stress. Taking into account that DmSRS2 is crucial for mitochondrial activity, we checked if the restricted expression of the RNAi DmSRS2 in the anteroposterior border of the wing imaginal disc (using patched-GAL4 driver) led to an increase in ROS. As shown in Figure 7B, wing imaginal discs exhibited a marked increase in superoxide anion restricted to cells under DmSRS2 interference at 29 C, comparable with that of cells affected by an RNAi against the ND75 complex I subunit of the respiratory chain, used as positive control.

Antioxidant treatment palliates defects caused by RNAi DmSRS2
Considering that reduction in mitochondrial tRNA Ser serylation disturbs organelle function and increases ROS production, we investigated whether supplementing flies' diet with an antioxidant cocktail (K-PAX; K-PAX Inc.) may have a palliative effect on affected flies. Indeed, adult flies with muscle-specific RNAi DmSRS2 silencing at 29 C underwent a significant improvement in longevity (a half-life increase from 6 to 12 days) when treated with the antioxidant mix ( Figure 8A). Similarly, supplementation with antioxidant molecules improved locomotion ability in both the RNAi DmSRS2 1-dcr2 and the RNAi DmSRS2 23003 strains ( Figure 8B). In the case of RNAi DmSRS2   Values from seven independent determinations were averaged, represented in the graph with SEM and evaluated by Student's t-test (*P < 0.05, **P < 0.01, ***P < 0.001). (B) Tissular pH from control and DmSRS2 ubiquitously and constitutively depleted larvae at different temperatures. Columns represent the mean of more than three replicates with SEM, where significance is determined by Student's t-test (*P < 0.05). (C) On the left graph, glycogen concentration in larval tissue subjected to RNA DmSRS2 general expression, compared with control (w 1118 ) larval tissue, at 25 C. Columns show the average with SEM from at least four measurements and analyzed by Student's t-test (*P < 0.1, **P < 0.01). On the right graph, electron micrographs from w 1118 and RNAi DmSRS2 strain 1-dcr2 larval fat bodies were used to determine the area occupied by glycogen.

DISCUSSION
As predicted, silencing of SRS2 expression by means of RNAi in D. melanogaster led to a decrease in the aminoacylation levels of the two mitochondrial tRNA Ser . This fact confirmed the DmSRS2 canonical function as mitochondrial seryl-tRNA synthetase in vivo [previously proposed by bioinformatic approaches in (44)], and allowed us to generate an animal model for human mitochondrial disease caused by aminoacylation deficiencies.
The stringency of the system was modulated using two different RNAi transgenic strains, crossing them with driver lines with various promoters and maintaining the animals at different temperatures. In that way, RNAi DmSRS2 strain 23003 resulted more efficient in targeting DmSRS2 mRNA and reducing serine-mt-tRNA Ser levels compared with RNAi DmSRS2 strain 1-dcr2 ( Figure 2). Accordingly, the constitutive or tissue-restricted silencing using strain 23003 produced strongest consequences in adult viability, longevity and tissue development (Figures 3 and 4).
This approach allowed us to induce DmSRS2 silencing, and produce flies with different levels of protein depletion leading to a range of phenotype severity, which permitted a better reproduction of the variability of symptoms displayed by patients with mitochondrial disorders. This is particularly true of those symptoms caused by mutations in mt-tRNAs that cause MELAS or MERRF.
These pathologies are characterized by lactate acidosis and mitochondrial respiratory chain dysfunction (16)(17)(18)20,22,23), traits also present in our Drosophila model ( Figures 6A, B and 7A). The drop in the complex I RCR, caused by an increase in the oxygen consumption in state 2, indicated an uncoupling between the respiratory chain and the OXPHOS, possibly due to an anomalous permeabilization of the mitochondrial inner membrane to protons, which could positively feedback the respiratory chain independently of the F 1 -F 0 -ATPase activity (60). Moreover, oxygen consumption values in the different respiratory states suffered a clear reduction when they were normalized taking into account the mitochondrial content (55). These results suggest that affected tissues in our model suffer from a defective mitochondrial respiratory capacity that is masked by an increase in mitochondrial biogenesis, a compensatory response also observed in patients with MELAS/MERRF symptoms (61)(62)(63). Intracellular ROS have been proposed as modulators of mitochondrial proliferation (64). Similarly, DmSRS2 deficiency might induce an increase in mitochondrial biogenesis stimulated by the accumulation in ROS ( Figure 7C).
The mitochondria of MELAS and MERRF patients present varying degrees of heteroplasmy, that is, the coexistence of different proportions of wild-type and mutant mtDNA populations in different tissues. Pathological symptoms appear when the mutant mtDNA copy number exceeds the level that guarantees the correct functioning of a tissue (threshold effect), leading to a complex variety of clinical manifestations with different levels of severity. This situation is again mimicked by our model, whose mitochondria suffer from a partial ablation of DmSRS2 activity. The affected organelles appear enlarged, with a decrease in the surface occupied by cristae, and electron pale matrices ( Figure 5). Apart from morphological abnormalities in mitochondria, the RNAi DmSRS2 prompted an increase in mitochondrial density, observed in the micrographs and confirmed by relative mtDNA quantification.
Taking into account the DmSRS2 silencing generated oxidative stress, and with the aim to asses a therapeutic approach to palliate the defects in longevity and locomotion ability caused by the muscular deficit of DmSRS2, antioxidants were added to flies' diet and we observed an amelioration in both phenotypes (Figure 8). The commercial antioxidant mix used (K-PAX; K-PAX Inc.) is a complex combination of compounds, and thus it is difficult to discuss the beneficial effect observed in this work. However, the significant palliative effect observed warrants a deeper analysis on the potential therapeutic effect of ROS inhibitors in diseases caused by mitochondrial malfunction.
In support of this conclusion it should be noted that, during the preparation of this manuscript, a fly model for ataxia with leukoencephalopathy caused by rearrangements on methionyl-tRNA synthetase 2 gene was published (21). The phenotype observed in our model closely coincide with the model from Bayat et al., with the same kind of mitochondrial morphology and respiratory defects, as well as a build-up in mitochondrial biogenesis and ROS accumulation, and a reduction in cell proliferation independent of apoptosis and cell growth events.
Our animal model is able to reproduce many traits that characterize mitochondrial disorders caused by mutations in the mitochondrial serylation apparatus. As example, an insertion in the mt-tRNA Ser that cause sensorineural hearing loss, results in a reduction in serylation efficiency, a moderate mitochondrial dysfunction, morphological alterations and lactate elevation (65,66). Mutations in mt-tRNA Ser related to Multisystem Disease with Cataracts (67) and deafness, retinal degeneration, myopathy and epilepsy (68) cause defects in mitochondrial function, abnormal mitochondrial morphology and proliferation, and those involved in MELAS/MERRF result in a group of features, such as pleomorphic mitochondria, increment in lactate, decrease in respiratory chain activity and increase in mitochondrial density (69), that coincide with the phenotypes observed in our model. On the other hand, similar symptoms have also been observed in patients with HUPRA syndrome (hyperuricemia, pulmonary hypertension, renal failure in infancy and alkalosis), which is caused by a mutation in the SRS2 gene (18).
The increasing number of pathogenic mutations found in mitochondrial aaRSs (15)(16)(17)(18)(19)(20)(21)(22)(23) justifies the generation of animal models for the study of these diseases and for the development of therapeutic strategies, among which treatment with antioxidant molecules should be considered.

SUPPLEMENTARY DATA
Supplementary Data are available at NAR Online: Supplementary Figures 1-3 and Supplementary Methods.