Altered 2-thiouridylation impairs mitochondrial translation in reversible infantile respiratory chain deficiency

Childhood-onset mitochondrial encephalomyopathies are severe, relentlessly progressive conditions. However, reversible infantile respiratory chain deficiency (RIRCD), due to a homoplasmic mt-tRNAGlu mutation, and reversible infantile hepatopathy, due to tRNA 5-methylaminomethyl-2-thiouridylate methyltransferase (TRMU) deficiency, stand out by showing spontaneous recovery, and provide the key to treatments of potential broader relevance. Modification of mt-tRNAGlu is a possible functional link between these two conditions, since TRMU is responsible for 2-thiouridylation of mt-tRNAGlu, mt-tRNALys and mt-tRNAGln. Here we show that down-regulation of TRMU in RIRCD impairs 2-thiouridylation and exacerbates the effect of the mt-tRNAGlu mutation by triggering a mitochondrial translation defect in vitro. Skeletal muscle of RIRCD patients in the symptomatic phase showed significantly reduced 2-thiouridylation. Supplementation with l-cysteine, which is required for optimal TRMU function, rescued respiratory chain enzyme activities in human cell lines of patients with RIRCD as well as deficient TRMU. Our results show that l-cysteine supplementation is a potential treatment for RIRCD and for TRMU deficiency, and is likely to have broader application for the growing group of intra-mitochondrial translation disorders.


INTRODUCTION
Mitochondrial diseases are a large and clinically heterogeneous group of disorders that result from deficiencies in cellular energy production and affect at least 1 in 5000 of the population. The underlying genetic defect in many patients remains unknown and there are no effective treatments (1,2). Most mitochondrial diseases are progressive conditions and lead to premature death. However, there is a unique condition, reversible infantile cytochrome c oxidase (COX) deficiency [or reversible infantile respiratory chain (RC) deficiency, RIRCD; OMIM# 500009], caused by the homoplasmic m.14674T.C/G mutation in the mt-tRNA Glu gene, showing spontaneous recovery during early childhood (3 -5). Affected children uniformly present with severe muscle weakness, often requiring assisted ventilation in the first days or weeks of life. If they survive the first months of life, they improve spontaneously, and recover fully by 2 or 3 years of age. The m.14674T.C/G mutation is thought to impair mitochondrial translation, as reflected by ragged red fibres/COX-negative fibres and multiple RC defects in skeletal muscle. The steady-state level of mt-tRNA Glu was low in early biopsies (16 -30%), but a slight increase occurred in the followup muscle biopsies, when the children were almost asymptomatic and remained low (30 -60%) in primary fibroblasts (3). The slight recovery of the steady-state level of mt-tRNA Glu in the face of dramatic clinical improvement indicates that, either this mild increase is sufficient to regain normal mitochondrial translation or other mechanisms downstream of mt-tRNA Glu are responsible for the clinical and biochemical recovery. Low levels of mt-tRNA Glu in muscle from clinically healthy mothers strongly suggest that the down-stream effects are able to ameliorate both the biochemical and clinical phenotype.
Although previous data provide strong evidence for a pathogenic role of m.14674T.C/G, they do not explain why all patients develop severe isolated myopathy in the neonatal period and, most importantly, what triggers the timed spontaneous recovery. Another unanswered question is why clinical symptoms manifest only in 30% of individuals carrying the homoplasmic m.14674T.C/G (3). However, no clear-cut nuclear modifiers of mtDNA disease have been identified to date (6).
RIRCD is not the only reversible mitochondrial disease. Autosomal-recessive mutations in a tRNA 5-methylaminomethyl-2-thiouridylate methyltransferase (TRMU, OMIM * 610230, also known as MTU1, MTO2), which is responsible for the 2thiouridylation of mt-tRNA Glu , mt-tRNA Gln and mt-tRNA Lys , but not of any other mt-tRNAs cause a severe but reversible infantile hepatopathy (7,8). Infants with reversible hepatopathy develop symptoms between 2 and 4 months of age, but if they survive this phase of liver failure, they recover and develop normally (8). The disease course and age of manifestation in TRMU deficiency shows remarkable similarities to RIRCD (5).
Recently, autosomal-recessive mutations were reported in infantile partially reversible hypertrophic cardiomyopathy in the gene MTO1 (OMIM * 614667) encoding the enzyme that catalyzes the 5-carboxymethylamino-methylation (mnm5s2U34) of the same nucleotide (U34) of the wobble position that is affected in TRMU deficiency for mt-tRNA Glu , mt-tRNA Gln and mt-tRNA Lys (9). Mutations in the glutamyl-tRNA synthetase (EARS2, OMIM * 612799) cause early onset severe neurological disease (leukoencephalopathy involving the thalamus and brainstem with high lactate, LTBL) and 8 out of 12 patients showed clinical improvement and stabilization after 1 year of age (10).
The age-dependent, partially reversible clinical presentation and the impairment of mt-tRNA Glu strongly suggest a possible pathophysiological link underpinning the spontaneous improvement in these mitochondrial conditions. We hypothesize that an impaired 2-thiouridylation in infants contributes to the clinical manifestation of RIRCD, therefore decided to study whether down-regulation of TRMU recapitulates the biochemical defect in RIRCD. Defining the common mechanism would not only suggest new avenues for treatment in these reversible disorders, but could also have more general relevance for the growing group of intra-mitochondrial translation defects.

2-Thiouridylation pattern in RIRCD patient cells
To investigate whether the homoplasmic m.14674T.C/G mt-tRNA Glu mutation impairs 2-thiouridylation of mt-tRNA Glu in fibroblasts and myoblasts of a patient with RIRCD myopathy, we performed high-resolution northern blots by incorporating N-acryloylamino phenyl mercuric chloride (APM) into the gels, which enabled us to separate thiolated and non-thiolated tRNA species (11). We used probes for the three mt-tRNAs (Glu, Lys, Gln) undergoing 2-thiouridylation by TRMU, and also probed for cytoplasmic tRNA Lys , and 5S rRNA as nonthiolated controls in RIRCD cells, TRMU-deficient patient cells and normal controls. We studied both steady-state levels and level of thiolation. The relative steady-state level of mt-tRNA Glu was reduced in both myoblasts ( Fig. 1A and B) and fibroblasts (Supplementary Material, Fig. S1A and B) from RIRCD patients, as shown previously (3). Steady-state levels of mt-tRNA Lys and mt-tRNA Gln in RIRCD fibroblasts were also slightly decreased, while this reduction was subtle in myoblasts. The TRMU patient's myoblasts showed an increase in the steady-state level of mt-tRNA Glu and mt-tRNA Lys (Fig. 1A and B), but there was no change of steady-state levels of the three thiolated mt-tRNAs in fibroblasts (Supplementary Material, Fig. S1A and B).
Down-regulation of TRMU (siRNA) decreased 2-thiouridylation and steady-state level of mt-tRNA Glu in RIRCD patient cells To investigate whether an additional impairment of 2thiouridylation compromises the mitochondrial translation defect in RIRCD, we down-regulated TRMU in fibroblasts and myoblasts of a patient. We used the siRNA, which showed the most prominent decrease of TRMU protein on immunoblotting (11). After siRNA-mediated down-regulation, both RIRCD and control fibroblasts (Supplementary Material, Fig. S2C) and myoblasts (Fig. 2C) showed low levels of thiolation of mt-tRNA Glu and mt-tRNA Lys , when compared with treatment with non-targeting siRNA (NT). Thiolation of mt-tRNA Gln was low before siRNA treatment and down-regulation of TRMU caused only a minor change in both patient and control (Fig. 2C). Cytoplasmic tRNAs were not fully thiolated; however, down-regulation of TRMU did not alter thiouridylation of cytoplasmic tRNAs (Fig. 2C). A part of mt-tRNA Glu , mt-tRNA Lys and mt-tRNA Gln always remained unthiolated. Down-regulation of TRMU further significantly compromised the steady-state level of tRNA Glu in RIRCD myoblasts compared with non-targeting siRNA-treated cells ( Fig. 2A and B).

Down-regulation of TRMU impaired mitochondrial translation in RIRCD myoblasts
We reported previously that mitochondrial translation is normal in both fibroblasts and myoblasts of RIRCD patients studied by 35 S-methionine pulse labelling (3). However, down-regulation of TRMU by siRNA resulted in an impairment of mitochondrial translation in RIRCD myoblasts, while mitochondrial translation in controls was slightly increased, perhaps indicating a compensatory mechanism ( Fig. 3A and B).

Down-regulation of TRMU decreased mitochondrial protein levels in RIRCD myoblasts
Down-regulation of TRMU in RIRCD myoblasts resulted in a severe decrease of protein levels of the mitochondrial complex IV subunits COX I, COX II and also for NDUFB8, representing  to the gels to separate thiolated and unthiolated tRNA species was performed and probed for mt-tRNA Glu , mt-tRNA Lys , mt-trNA Gln , cytoplasmic tRNA Lys and 5S rRNA in immortalized human myoblasts of patients with RIRCD, TRMU deficiency and control cell lines. Results derive from two independent experiments, all representative blots were used for all tRNA probes following each other. (B) Quantification of the northern blots shows relative steady-state levels of the tRNAs and (C) the percentage of thiolated tRNA species compared with the whole amount of each tRNAs. For each sample the signal corresponding to the amount of tRNA was normalized to the signal corresponding to the amount of 5S RNA. The total levels of each of the four thio-modified tRNAs in the control cells were set arbitrarily to 100%. The values in the histogram are averages of two measurements, one corresponding to the signal from the gel without APM and the other to the total signal (thiolated plus unmodified) from the gel containing APM. The quantification of the modification is presented at the bottom panel and is expressed as a percentage of the thiolated signal from the thiolated + non-thiolated signals.  Blue native polyacrylamide gel electrophoresis (BN-PAGE) and in-gel activity of oxidative phosphorylation complexes BN-PAGE and 'in gel' activity measurement detected slightly reduced complex I and IV in untreated RIRCD myoblasts compared with controls, and this was the only 'cellular phenotype' of a defective mitochondrial translation (Fig. 3D). Downregulation of TRMU resulted in a further decrease of complex I and IV in RIRCD cells, but also led to a decrease in controls. There was an additional 70 kDa complex II intermediate noted in TRMU down-regulated cells, similarly to previously reported data in TRMU-deficient human primary fibroblasts (11). Complex III remained unchanged. In addition, non-specific complex V assembly intermediates were detected in TRMU down-regulated cells, which we consider to be a non-specific finding (Fig. 3D).
Thiolation of mt-tRNA Glu or the m.14674T>C mutation may affect EARS2 and MTO1 gene expression To explore potential compensatory mechanisms, we studied the effects of TRMU siRNA on the glutamyl-tRNA synthetase (EARS2), and MTO1, another enzyme affecting the 5-carboxymethylamino-methylation of the same nucleotide (U34) of the wobble position of mt-tRNA Glu , mt-tRNA Gln and mt-tRNA Lys . Gene expression levels of both EARS2 and MTO1 were reduced in RIRCD compared with controls, and further decrease was detected after TRMU depletion, suggesting that deficient 2-thiolation may alter these other important mt-tRNA Glu modifying factors (Fig. 3E). Depletion of TRMU resulted in increased expression of the gene-encoding cystathionase, the enzyme responsible for cysteine production. The decreased EARS2 and MTO1 gene expression did not result in significant protein reduction, possibly due to the short period of the siRNA experiment (Fig. 3F).
Importantly, the higher level of MTU1 gene expression in the RIRCD patient cells may be a compensatory change which further confirms the link between the two reversible mitochondrial conditions. Moreover, RT -PCR of skeletal muscle of a TRMU patient and early muscle biopsy of an RIRCD patient showed significantly higher MTU1 gene expression, which decreased in parallel with clinical recovery in a follow-up muscle.

Investigation of 2-thiouridylation in control and patient skeletal muscle
While steady-state levels of all mt-tRNAs, but not the cytoplasmic tRNA Lys , increased gradually by age in human skeletal muscle (Fig. 4A, B and D), the rate of thiolated/non-thiolated tRNA species for mt-tRNA Lys , mt-tRNA Glu and mt-tRNA Gln showed no change by age in normal skeletal muscle ( Fig. 4C and E). Skeletal muscle of a patient with TRMU deficiency showed impaired thiolation, but slightly increased mt-tRNA steady states, most likely reflecting compensation, suggesting that TRMU defect is not restricted to liver (Fig. 4E).
Follow-up skeletal muscle biopsies of two previously reported RIRCD patients (3) were studied and showed very low levels of thiolated mt-tRNA Glu , and also slightly lower levels of mt-tRNA Lys and mt-tRNA Gln in the symptomatic phase of the disease (1 months, 5 years 4 months of age) (Fig. 4F, I and H, K). The second patient had an unusually long symptomatic phase causing symptoms until at least 7 years of age. There was a 20% increase in the thiolated mt-tRNA Glu levels in both patients between early (1 month, 5 years 4 months), and follow-up biopsies after clinical recovery (8 years 9 months and 14 years of age) ( Fig. 4H and K). Both mt-tRNA Lys and mt-tRNA Gln showed an increase in thiolation status between the early and late biopsies, while thiolation of the cytoplasmic tRNAs did not change. In addition, mt-tRNA steady states were also lower in the symptomatic phase, further compromising mt-tRNA function ( Fig. 4G and J). Repeated analysis was not possible because of the small amount of available skeletal muscle.
In vitro L-cysteine supplementation resulted in improved mitochondrial respiratory function We investigated whether addition of L-cysteine, a substrate of TRMU, required as a source of sulphur for thio-modification, has an effect on TRMU function. In vitro supplementation of RIRCD cells with L-cysteine rescued slightly reduced complex I and IV on BN-PAGE ( Fig. 5A and B) and significantly increased 'in gel' enzyme activities, both in RIRCD and in control myoblasts ( Fig. 5C and D). Down-regulation of TRMU resulted in a further decrease of complex I and IV in RIRCD patient cells, and these changes were completely prevented by adding 5 mM L-cysteine to the culture medium (Fig. 5E). The positive effect of L-cysteine on mitochondrial translation was also confirmed by normal immunoblotting of mitochondrial proteins (COX I, COX II and NDUFB8) in RIRCD cells, if L-cysteine was added to the cell culture medium during TRMU down-regulation (Fig. 5F). Cysteine supplementation also improved the respiratory chain enzymes in MTO1-and TRMUdeficient fibroblasts. Our data indicated around a 20 and 30% improvement of complex I and complex IV, respectively, in both patient cell lines. This tendency was more noticeable in TRMUdeficient cells, where all complexes improved after the treatment ( Fig. 6A and C).
RIRCD cells and controls. Complex III was normal, but we also detected some additional bands by complex V antibody. Immunoblotting with complex II antibodies was used as a loading control.

DISCUSSION
The synthesis of the 13 mitochondrial-encoded proteins is a complex pathway, which requires 150 different proteins (ribosomal proteins, ribosomal assembly proteins, aminoacyl-tRNA synthetases, tRNA-modifying enzymes, tRNA methylating enzymes and several initiation, elongation and termination factors) involved in mitochondrial translation (12)(13)(14). Most of these gene defects result in histological (COX-deficient or ragged red fibres) and biochemical abnormalities (multiple respiratory chain defects) in affected organs. The clinical phenotypes are usually early-onset, severe and often fatal, implying the importance of mitochondrial translation from birth (15). Some of these conditions affect multiple tissues; however, tissue-specific manifestations have been reported for several mt-tRNA aminoacyl synthetases or mt-tRNA-modifying genes (13,14).
Based on the striking similarities between two clearly reversible mitochondrial conditions, RIRCD due to a homoplasmic mt-tRNA Glu mutation and reversible infantile liver failure due to TRMU deficiency, we hypothesized that the reversibility may be due to basic mechanisms involving mt-tRNA Glu . The importance of mt-tRNA Glu in reversible disease is also supported by the partial recovery of patients with mutations in other two recently identified mt-tRNA Glu -modifying genes (EARS2 and MTO1) (9,10). Reversibility (or even improvement) is an extremely rare event in severe childhood mitochondrial disorders and 2-thiouridylation may offer a common pathway; therefore, we studied whether modifying thiolation of the U34 position on mt-tRNA Glu , which is affected in reversible infantile liver failure, could contribute to the muscle-specific RIRCD, caused by the homoplasmic m.14674T.C/G mutation in mt-tRNA Glu .
The uridine at the first anticodon position (U34, wobble position) is present only in the anticodon of 3 mt-tRNAs (Glu, Lys and Gln). It is modified at carbons 2 and 5, and while carbon 2 is modified exclusively through thiolation (s 2 ), various methyl derivates can be found at carbon 5 (methylaminomethyl mmm 5 , carboxymethylaminiomethyl cmmm 5 , etc.) (16). The 2-thio group is required for the efficient codon recognition, and in the case of mt-tRNA Glu , it is necessary for the recognition by the glutaminyl-tRNA synthetase (17). The 2-thio group confers conformational rigidity, ensuring stable and accurate codon-anticodon pairing, and causes a steric repulsion with its 2 ′ OH group at the 3 ′ end of the tRNA, therefore an interaction between these two positions may be possible (17). Vice versa, an altered 2-thiouridylation may further impair the mutant but still functioning mt-tRNA Glu , possibly similar to its effect in the case of m.1555A.G (18).
We detected minor changes of 2-thiouridylation of mt-tRNA Glu in RIRCD fibroblasts and no change in myoblasts, suggesting that the homoplasmic m.14674T.C/G mutation per se does not affect thiolation of mt-tRNA Glu . To further explore the effect of an impaired 2-thiouridylation on function of the mutant mt-tRNA Glu , we depleted TRMU in primary patient cells in vitro. Down-regulation of TRMU resulted in defective 2-thiouridylation of all 2-thiolated mt-tRNAs (mt-tRNA Lys , mt-tRNA Glu , mt-tRNA Gln ) in both fibroblasts and myoblasts of a patient with RIRCD, as well as in controls. Importantly, the impairment of 2-thiouridylation of mt-tRNA Glu was most severe in RIRCD myoblasts, implicating that mutant m.14674T.C myoblasts are more sensitive for the 2thiouridylation defect of mt-tRNA Glu , triggered by the depletion of TRMU, than RIRCD fibroblasts or control myoblasts.
It was suggested previously that TRMU is not required for mitochondrial translation if steady-state levels of mt-tRNAs are normal (11). Furthermore, down-regulation of TRMU did not result in a further impairment of mitochondrial translation in fibroblasts or myoblasts carrying the m.3243A.G (MELAS) and m.8344A.G (MERRF) mutations (11). Our   Data were normalized to the complex II and are presented as the mean + SD (n ¼ 3). The control value obtained for the control untreated fibroblasts was represented as 100% and the value from the L-cysteine treated cells was expressed as a percentage of the control value. The asterisk denotes that the level of complex IV was significantly lower in the MTO1 and TRMU patient cells compared to control (P ≤ 0.004, ANOVA). The triangle indicates the significance after L-cysteine supplementation (P ≤ 0.006, ANOVA). results are supporting these previous studies that depletion of TRMU did not significantly alter mitochondrial translation on pulse labelling in controls, however, unlike in MELAS and MERRF myoblasts, down-regulation of TRMU resulted in an impaired mitochondrial protein synthesis in RIRCD myoblasts. This was further confirmed by a severe decrease of mitochondrial subunits (immunoblotting) and assembled complexes (BN-PAGE). The reasons behind these differences can be mutation specific, or other functions of TRMU may be involved (11).
An as yet uncharacterized function of TRMU in sulphurtrafficking was suggested previously (11). However, a defect in iron -sulphur (Fe -S) biosynthesis would not affect COX, which does not contain an Fe -S centre, therefore cannot explain the full biochemical phenotype caused by downregulation of TRMU.
It was suggested before that the 2-thiouridylation of mt-tRNA Glu affects not only the accuracy and efficiency of translation, but also important for the recognition of the tRNA by the mitochondrial glutamyl-tRNA synthetase (EARS2). A disturbance of this interaction, possibly altered by the m.14674T.C/ G mutation, would further contribute to the defect in mitochondrial translation in RIRCD (19). In support of this hypothesis, in RIRCD cells gene expression of EARS2 was lower than in controls and down-regulation of TRMU resulted in a further decrease of EARS2 gene expression. TRMU down-regulation also led to decreased EARS2 expression in controls, suggesting that thiolation may affect other mt-tRNA Glu modifications.
A synergistic effect of the yeast proteins involved in 2thiouridylation (MTU1) and methylaminomethylation (MTO1) of the U34 wobble nucleotide was suggested previously (19,20). We studied whether down-regulation of the 2-thiouridylation alters MTO1 in our cellular model and similar to EARS2, MTO1 expression was decreased in RIRCD myoblasts. TRMU downregulation resulted in a further decrease in MTO1 gene expression in RIRCD and also in control myoblasts, suggesting a link between the two modification steps of U34.
To explain the age-dependent, tissue-specific infantile presentation of reversible mitochondrial disease, we studied physiological or developmentally regulated changes in 2thiouridylation in skeletal muscle biopsies of patients of different age. Our results suggest that the level of thiolated and nonthiolated tRNA species in normal human skeletal muscle does not change by age; however, steady-state levels of mt-tRNAs increase during the first years of life. Most importantly, skeletal muscle of two RIRCD patients in the symptomatic phase showed clearly decreased thiolation and mt-tRNA steady-state levels which improved in parallel with the clinical recovery, providing experimental evidence for a role of thiolation in the reversibility.
The most exciting result of our study was the effect of in vitro L-cysteine supplementation. BN-PAGE showed minor abnormalities in RIRCD myoblasts, similar to a previous study (4), and a defect of complexes I and IV was more pronounced on the 'in gel' activity assay. Adding L-cysteine to the culture medium fully reversed this deficiency. Furthermore, L-cysteine prevented the decrease in respiratory complexes in TRMU down-regulated RIRCD cells and controls, supporting that low cysteine concentrations may play a role in triggering a reversible mitochondrial translation defect in vitro, and this can be rescued by L-cysteine supplementation. L-Cysteine supplementation led to an improvement in most respiratory chain complex activities in TRMU-and MTO1-deficient cells, indicating that the positive effect is not specific to the thio-modification.
Recent publications suggested a possible beneficial effect of supplementation with N-acetylcysteine, a precursor of sulphidebuffering glutathione in mice and patients with a rare mitochondrial condition, ethylmalonic encephalopathy due to mutations in the ETHE1 gene encoding a mitochondrial sulphur dioxygenase (21). A double-blind cross-over study on patients with mitochondrial myopathies showed that 30-day supplementation with a whey-based cysteine donor resulted in significantly reduced oxidative stress (22), and a recent paper reported lower levels of reduced cysteine and thiols in plasma of children with mitochondrial diseases, suggesting that relative thiol deficiency could be an important factor in paediatric mitochondrial conditions (23).
How does cysteine play a role in reversible mitochondrial disease in infants? TRMU protein requires sulphur for its activity supplied by the cysteine desulfurase enzyme. Since the availability of cysteine in the neonatal period is limited by the low activity of the cystathionase enzyme, dietary cysteine intake may be very important at this age. It was hypothesized that between 1 and 4 months of age inter-current illnesses, combined with reduced dietary cysteine intake, may compromise TRMU activity, resulting in decreased 2-thiolation (7). Decreased cysteine levels could reflect differences in nutrition, or could be due to other environmental, genetic or epigenetic factors (Fig. 7). Our data suggest that L-cysteine supplementation may potentially reverse the age-dependent clinical manifestation of RIRCD and TRMU deficiency. Further investigation of infantile cysteine levels may help to unveil these mechanisms which can have important implications in reversible mitochondrial disease, but also in other mitochondrial conditions.

Cell culture and siRNA transfection
Fibroblast and myoblast cell cultures of two RIRCD patients, a TRMU-deficient and a MTO1-deficient cell line as well as controls (Supplementary Material) were obtained from the Biobank of the Medical Research Council, Centre for Neuromuscular Diseases, Newcastle, and were immortalized as described previously (24). Informed consent was obtained from all subjects. Fibroblasts were grown in high glucose Dulbeccos modified Eagle's medium (Sigma, Poole, UK) supplemented with 10% foetal bovine serum. Muscle cells were grown in skeletal muscle growth medium (PromoCell, Heidelberg, Germany), supplemented with 4 mM L-glutamine and 10% foetal bovine serum and cultured as recommended by the supplier. Stealth RNAi duplexes (TRMU HSS124809 or HSS124809 siRNA) were transiently transfected at a final concentration of 12 nm using Lipofectamine RNAiMAX (invitrogen), according to the manufacturer's specifications. Transfections were repeated on Day 3, cells were either harvested or again transfected on Day 6, with cells being harvested on Day 9.

Supplementation with L-cysteine
Myoblasts were grown in skeletal muscle growth medium (described above, 0.2 mM L-cysteine), supplemented with 5 mM L-cysteine (Sigma). Cells were left to grow for 5 -9 days. The medium was changed every 72 h.

Immunoblotting
For immunoblotting, protein extracts were prepared as described previously (15). Aliquots of total protein (5-20 mg) were loaded on 14% sodium dodecyl sulphate -polyacrylamide gels (SDS -PAGE), transferred to polyvinylidene fluoride membranes and subsequently used for detection of TRMU, with a polyclonal, affinity purified antibody (from Prof. E. Shoubridge) at a dilution of 1:1000. The blots were also probed with monoclonal antibodies recognizing mitochondrial COX I (Molecular Probes), COX II (Mitosciences) or NDUFB8 (Mitosciences), EARS2 (Abgent), MTO1 (Proteintech Group, Inc.) and b-actin (Sigma) according to the recommendations of the suppliers.

APM-northern blotting analysis
Isolation of RNA from both cells and tissues was carried out using Trizolw (Invitrogen) following the manufacturers recommendations. We performed northern blotting on APM containing gels, essentially using the method described previously (11). This is the standard method to separate thiolated and nonthiolated tRNA species (16). Following transfer to GeneScreen Plus membrane (Perkin Elmer), the presence of tRNA species was detected using 32 P-labelled PCR products as described previously (25). The probes for human mt-tRNA Glu , mt-tRNA Lys , mt-tRNA Gln , the cytoplasmic tRNA Lys and 5S RNA were generated using primers listed in the Supplementary material. Quantification of the radioactive signal was performed with imageJ software.

Pulse-labelling of mitochondrial translation products
In vivo 35 S-metabolic labelling studies were performed as described previously (11,26) with the following modifications. Cells, cultured to 60-70% confluency in T25 mm flasks, washed with phosphate-buffered saline (PBS; Sigma) and washed by incubating twice for 10 min at 378C/5% CO 2 in methionine/cysteine-free DMEM (Sigma, Poole, UK), with the media replaced between each incubation. Cells were then incubated for 15 min at 378C/5% CO 2 in methionine/cysteine-free DMEM supplemented with 5% (v/v) dialyzed FBS, 0.1 mg/ml emetine dihydrochloride (Sigma). Following addition of 200 mCi/ml 35 S-methionine/cysteine ( 35 S EasyTag EXPRESS; Perkin Elmer), cells were incubated for 15 min at 378 C/5% CO 2 , then washed twice with ice-cold DMEM supplemented with 7.5 mg/ml methionine. Cell pellets were prepared after washing once with ice-cold PBS. Radio-labelled proteins were then analyzed using SDS-PAGE as described previously (3).

RT -PCR
RNA was isolated from myoblasts after non-targeted and siRNA transfection (Arcturus PicoPure RNA isolation kit; Applied Biosciences). cDNA was prepared using 0.5 mg RNA and RT -PCR was performed with SYBR Green detection. Data were normalized to b-actin and evaluated by DDCt and standard curve analysis. Melting curves from PCR products showed a single peak and product sizes were confirmed with gel electrophoresis. Primer sequences used in RT -PCR reactions for human TRMU, EARS2, MTO1, cystathionase (CTH), and b-actin (ACTB) are listed in the Supplementary Material. . Schematic representation of cysteine sources for functional TRMU enzyme. The cystathionase enzyme or also called cystathionine gamma-lyase plays an essential role in cysteine production. However, in the early months of life the activity of this enzyme is low. Metallothionein which represents another cysteine source, although presents at high levels at birth, dramatically decreases in the neonatal period. Therefore, the production of this amino acid is limited. Dietary cysteine intake might play a crucial role for the normal TRMU enzyme activity within the first few months of life when combined with underlying genetic diseases.