The Human Research Ethics Committee of the Children’s Hospital at Westmead approved this research.

A cohort of 12 patients was selected for YARS2 screening. Selection criteria included presence of anaemia, with or without lactic acidosis (10/12 patients had lactic acidosis), with or without skeletal myopathy (6/12 patients displayed myopathy/hypotonia), and a demonstrated respiratory chain enzyme deficiency. Nine of the 12 patients screened were of French origin, one was Italian and the other two were Australians of Lebanese origin, unrelated to the two original families for which a YARS2 mutation was previously described 16 . Clinical histories for patients for whom we identified pathogenic YARS2 mutations are given below (patients 4, 5 & 6) and previously reported cases are reviewed (patients 1, 2 & 3). Clinical features of the other patients are presented in Additional file 1.

Patient 1 and 2 are siblings of consanguineous parents and have been described previously 16 . Briefly, within the first 3 months of life P1 developed transfusion dependent sideroblastic anaemia and had persistent lactic acidaemia and hypertrophic cardiomyopathy. The cardiomyopathy spontaneously resolved. He had progressive exercise tolerance and at 17 years he developed dysphagia and acute respiratory difficulties and succumbed at 18 years of age. P2 also developed transfusion dependent sideroblastic anaemia in infancy and had persistent lactic acidaemia and progressive muscle weakness, similar to P1. At 15 years she developed dysphagia and had a gastrostomy inserted. Remarkably, at 17 years of age P2 showed improved muscle strength and stamina, and no longer required blood transfusions (previously being given every six weeks). She is currently 20 years old. Patient 3 16 , born to consanguineous parents, had delayed motor milestones and developed sideroblastic anaemia at 7 years, as well as lactic acidaemia. Her parents refused to allow her to have blood transfusions and she was treated with complementary therapies. At 24 years she had a mild skeletal myopathy, anaemia and mild peripheral muscle weakness.

Patient 4, the second child of non-consanguineous parents of Lebanese origin, is not known to be related to the other patients described here. There were no antenatal concerns. She was born at term by normal delivery with good Apgar scores. She had low birth weight of 2.7 kg but there were no immediate postnatal problems. She fed well by formula and was gaining weight at home.

There is a family history of a maternal aunt with sickle cell anaemia. The parents have a healthy 3-year old daughter, and there is a history of a prior fetal demise at 12 weeks gestation.

The proband presented at 8 weeks of age in acute hypotensive shock needing cardiopulmonary resuscitation following a few days of rhinorrhoea, loose stools and associated lethargy and feeding difficulties. A venous blood gas at the time of arrest showed pH 6.56, bicarbonate of 4.2 mmol/L, BE -28.8, anion gap of 30 and lactate of 27 mmol/L (normal range 0.7 – 2.0). A CT scan of the brain was normal. Urine metabolic screening showed increased lactate and ketones. Plasma amino acids, triglycerides, free fatty acids, uric acid, blood ammonium and serum transferrin isoforms were normal. There was a slightly low plasma total carnitine level noted.

She had associated multisystem problems. These included significant concentric left ventricular hypertrophy with altered echotexture, and abnormal diastolic function. She had an initial mild pericardial effusion which later required pericardial drainage. She also had hepatomegaly with raised transaminases and coagulopathy. She had possible seizures, managed with anticonvulsants. A subsequent MRI/MRS of the brain showed cerebral atrophy with normal spectroscopy and no lactate peak.

She had anaemia, with the lowest haemoglobin being 42 g/L, and required intermittent packed cell transfusions. Bone marrow aspirate showed reduced erythropoiesis with prominent red cell vacuolation in the red cell precursors and ringed sideroblasts. There was no evidence of a storage disorder on bone marrow biopsy.

She did not tolerate weaning from mechanical ventilation. Impaired cardiac output in conjunction with the above associated issues led to eventual decision for withdrawal of active intensive care. She succumbed at three months of age.

Subsequent to identification of the p.Phe52Leu YARS2 mutation, they have had two pregnancies which underwent prenatal testing, and were then terminated as both were found to be homozygous for the mutation.

Patient 5 is one of four siblings to non-consanguineous Lebanese parents (although they are from the same village in Northern Lebanon). Her sister was diagnosed with sideroblastic anaemia in Lebanon in her mid 20s and had been transfusion dependent for around 10 years. She ultimately died of cirrhosis induced by iron overload 10 years after commencing blood transfusions. This individual was reported to be an otherwise fairly healthy young woman without any other medical concerns.

Patient 5 was noted to be anaemic at the age of 23. An initial bone marrow performed in Lebanon reportedly showed a form of refractory anaemia for which she had four erythropoietin injections. Subsequently, she returned to Australia and a second bone marrow aspirate four years later demonstrated ringed sideroblasts. At that time, her lactate was elevated at 3.24 mmol/L with a pyruvate of 0.13 mmol/L. On review a year later, her lactate was 5.10 mmol/L and haemoglobin was low at 104 g/L. Creatine kinase was 30 U/L, at the lower end of the normal range.

Patient 5 was found to have scoliosis at the age of 11 years, and she went on to have titanium rods inserted which were subsequently removed secondary to infection. She has subsequently had a spinal fusion. She also has endometriosis.

Her 6-minute walk time was reduced at 353 m. Her pulmonary function tests demonstrated a mild restrictive defect with the FVC of 70% predicted and FEV1 of 70% predicted; the Diffusing lung capacity for carbon monoxide (DLCO) was normal. Her ECG and echocardiography were also normal.

On examination, there was no organomegaly, and her cardiovascular and respiratory examination was normal. She had a long, well-healed scar over her spine, with a slight residual right thoracic scoliosis. She had subtle bilateral hip flexor weakness, but her power otherwise was intact. Deep tendon reflexes and sensation were normal. There were no problems with gait or balance. Intelligence was normal.

Patient 6, the child of non-consanguineous parents of French origin required blood transfusion for sideroblastic anaemia in the first year of life, which resolved spontaneously. A bone marrow examination during the first year of life showed the presence of megakaryocytes, signs of erythroblastic dysplasia, and an excess of sideroblasts. A partial deficiency in complex IV was detected in lymphocytes (muscle biopsies were not performed), and abnormal assembly of RC complexes I, III, IV and V were observed on Blue Native-PAGE of fibroblasts. At 3 years of age there was a discrete thickening of the myocardium, which normalised by age 5 years. No skeletal myopathy was present, but she had nystagmus and strabismus. Liver function tests were normal. Mild but permanent lactic acidaemia (3 mmol/L) was found at 5 years. Lactic acidaemia was also present at 6 years and she reported fatigue on prolonged walking or climbing stairs, together with leg pain and muscle weakness. There was no evidence of anaemia. Intelligence was normal.

All YARS2 and PUS1 coding exons were PCR amplified from gDNA and Sanger DNA sequencing performed by Macrogen (Korea). mtDNA from p.Phe52Leu YARS2 patients was sequenced at the Lotterywest State Biomedical Facility Genomics, Royal Perth Hospital. mtDNA haplogroups were assigned using Haplofind 19 .

Recombinant YARS2 mutant proteins were generated to assess their tyrosylation activity. Recombinant wild-type and YARS2 variants were cloned, expressed and purified from E. coli as previously described 20 . In vitro tRNA^Tyr aminoacylation assays were performed using previously described methods 20 . Apparent kinetic parameters were determined from Lineweaver-Burk plots in the presence of 5 to 27 nM YARS2 (depending on the variant) and native E. coli tRNA^Tyr (Sigma), with concentrations ranging from 0.28 to 2.1 μM. Experimental errors for k_cat and K _m varied at most by 20%. Numerical values are averages of at least two independent experiments.

Primary patient and control fibroblast cell lines were routinely cultured in DMEM: F12 (Life Sciences) containing 20% Amniomax (Life Sciences) and 10% heat inactivated fetal bovine serum (FBS; In Vitro Technologies), penicillin (50 μg/ml) and streptomycin (50 μg/ml; Life Sciences) and primary myoblast cell lines were cultured in DMEM: F12, 20% Amniomax, 20% FBS, in a humidified incubator containing 5% CO₂ and maintained at 37°C. Fibroblasts were transduced and differentiated using a lentiviral MyoD vector as previously described 21 , except that cells were seeded onto plates coated with 0.15 mg/ml rat tail collagen (BD Biosciences). Myoblasts were differentiated in the same media as described for fibroblast transdifferentiation. Cells were harvested at day 6 of differentiation for enzyme assays, immunoblotting or DNA extraction.

Respiratory chain enzyme activities were determined as previously described 22 .

Immunoblotting was as previously described 16 , with the following modifications. Membranes were probed with 1:500 anti-OXPHOS (Abcam) for 2 h at room temperature or with 1:500 anti-YARS2 (N-terminal, Abgent), 1:10000 anti-porin (Abcam), 1:1000 anti-mitofilin (Abcam), 1:500000 anti-α-actinin-2 (a generous gift to the INMR from Prof Alan Beggs, Harvard Medical School, Boston Children’s Hospital, Division of Genetics), 1:500 anti-PGC1-α (Sigma) or 1:500 anti-TFAM (Abcam) overnight at 4°C. Densitometry was performed as previously described 16 and protein levels were normalised to porin protein levels.

DNA was isolated from patient fibroblast cell pellets and skeletal muscle tissues and sex- and age-matched controls using a QIAamp DNA mini kit (QIAGEN). Tissues were homogenized using a TissueRuptor (QIAGEN). Samples were treated with RNase A (QIAGEN).

A mitochondrial encoded gene (ND1) was quantified relative to a single copy number nuclear encoded gene (β2M) 23 to determine the mtDNA content in patient and control myoblasts, myotubes and muscle. These genes were amplified in separate reactions containing 40 ng total DNA, 1 Unit of Immolase™ DNA Polymerase (Bioline), and a final concentration of 1.5 mM MgCl₂, 500 pM of each primer, 2% DMSO and 1 M betaine. Standard curves were generated by amplifying known amounts of a pTOPO2.1/B2M or pTOPO2.1/ND1 plasmid. The cycle conditions were: 95°C for 12 minutes, (95°C for 15 s, 60°C for 15 s, 72°C for 20 s) × 35 cycles. Primer sequences were (5' to 3'): ND1_F – CATAAAACTCTTCACCAAAGAGCC; ND1_R – GGGGTTCATAGTAGAAGAGCGA; β2M_F – TGCTGTCTCCATGTTTGATGTATCT; β2M_R – TCTCTGCTCCCCACCTCTAAGT.

Generalised linear models for univariate analysis (SPSS v.19, IBM) were used to assess the effect of cell type (fibroblast, forced myotube), sample (patient P2, patient P6, control) and their interaction with normalised Complex IV activity and normalised Complex IV/CS activity. Generalised linear models for univariate analysis were also used to assess the effect of cell type (myoblast, myotube), sample (patient P2, control) and their interaction with relative mtDNA/nDNA. Where there was a significant interaction between cell type and sample, effects were analysed independently. No adjustments were made for multiple statistical analyses. An independent sample t-test (SPSS v.19, IBM) was used to test for differences in relative mtDNA/nDNA between patient and control muscle.

We have previously identified a p.Phe52Leu YARS2 mutation in three MLASA patients (P1, P2 and P3; Table 1) who demonstrated a muscle-specific RC deficiency affecting complexes I, III and IV 16 . Screening of two additional unrelated patients with similar clinical features identified Patients P4 and P5 as also being homozygous for the missense mutation c.156C > G (p.Phe52Leu). The non-consanguineous parents of P4 showed both were heterozygous for the c.156C > G substitution. Parental DNA was not available for P5.

^a Sibling of P1.

^b Values expressed as % residual activity.

^c Citrate synthase activity was expressed relative to protein.

CS = citrate synthase.

ND = not determined.

Despite similar RC enzyme results in muscle, clinical severity varied greatly among patients (Table 1). Patient P4 showed a severe phenotype and died at 3 months due to hypertrophic cardiomyopathy. P1 also had cardiomyopathy in infancy, however this resolved without specific therapy. P5 did not develop symptoms until adulthood, characterised by mild myopathy and transfusion-independent anaemia. Patient P2 presented at birth with a moderate myopathy and transfusion-dependent anaemia. Remarkably, at 17 years of age her clinical condition spontaneously improved, while her sibling, P1, had progressive muscle weakness, becoming wheelchair bound by 14 years and died of respiratory failure at 18 years of age. The four older YARS2 p.Phe52Leu patients all had normal cognitive function. All five of the homozygous p.Phe52Leu YARS2 patients are Australians of Lebanese origin consistent with our proposal that p.Phe52Leu represents a rare founder mutation 16 .

mtDNA sequencing revealed that although the patients were of the same ethnicity, they did not all have the same mtDNA haplogroup (Table 1; see Additional file 2 for a complete list of mtDNA variations). The 3 previously reported patients (P1,P2 & P3) all belonged to haplogroup K. P4, the most severely affected patient, belonged to haplogroup T and P5, the least severely affected patient belonged to haplogroup H.

YARS2 and PUS1 mutation screening was performed on a cohort of ten French or Italian patients with demonstrated RC enzyme deficiency who also presented with anaemia and at least one of the other MLASA features, ie. lactic acidosis or myopathy. One patient, P6 (Table 1) had two novel YARS2 mutations, a heterozygous missense mutation c.572G > A (p.Gly191Asp) and a heterozygous nonsense mutation c.1078C > T (p.Arg360X). Patient P6 YARS2 variations were not present in dbSNP (NCBI), 1000 genomes (browser.1000genomes.org), or the ESP database (evs.gs.washington.edu). Screening of parental DNA showed that each parent carried one or the other of the mutations.

Patient P6 ([p.Gly191Asp];[p.Arg360X]) has mild lactic acidosis and myopathy, and only required blood transfusions during the first year of life (Table 1). Impaired complex IV enzymology was demonstrated via RC testing of lymphocytes and defective assembly of complexes I, III, IV & V on Blue Native-PAGE of fibroblast samples (data not shown).

YARS2 variants were identified in two other patients (P7 & P8) however they are unlikely to be pathogenic. These are described in Additional file 1. No PUS1 mutations were identified in this cohort of patients.

To determine the functional impact of the p.Gly191Asp YARS2 mutation on tyrosylation capacity, in vitro assays were performed using E. coli tRNA^Tyr as substrate. Gly191 lies within the catalytic domain of YARS2 24 , and the protein prediction program SIFT (http://sift.jcvi.org/) predicts the Gly191Asp variation to be damaging.

The aminoacylation assay measured incorporation of [¹⁴C]-tyrosine on native E. coli tRNA^Tyr by recombinant YARS2. The p.Gly191Asp mutation did not significantly affect the K _m of the reaction (Table 2), indicating that YARS2 interaction with tRNA^Tyr was apparently unaffected. However the catalytic rate (k_cat) was perturbed, resulting in reduced tyrosylation activity. The p.Gly191Asp variant resulted in a 38-fold decrease in catalytic efficiency (k_cat / K _m ) as compared to the wild-type enzyme, a greater reduction than the previously reported p.Phe52Leu mutation (9-fold decrease) 16 . This result was consistent with the SIFT prediction of the severity of the p.Gly191Asp variation on protein function.

^a Loss of efficiency is calculated relative to the wild-type YARS2.

The expression level of YARS2 and the RC complexes was investigated by immunoblotting of patient P6 fibroblast and myotubes (Figure 1a). Due to the muscle-specific nature of the YARS2 RC deficiency and the unavailability of P6 muscle samples, myotubes were generated by transdifferentiation of fibroblasts using a lentiviral vector expressing MyoD 21 , a master regulatory myogenic transcription factor. This model was validated by demonstrating a similar induction of mitochondrial RC expression levels via native and Myo-D forced myogenesis for patient P2, for whom a myoblast cell line was available (Additional file 3). Protein levels of YARS2 were not significantly altered in either patient P6 fibroblasts or MyoD-forced myotubes, relative to controls, suggesting that reduced YARS2 protein levels are not the cause of the muscle-specific phenotype. This was also the case for p.Phe52Leu YARS2 (P2), which is shown for comparison (Figure 1a). There was no evidence of a truncated YARS2 protein that might be expected to arise from the p.Arg360X mutation in P6. Sequencing of P6 cDNA showed that the c.1078C > T (p.Arg360X) mutant transcript was unstable and qPCR showed that total YARS2 mRNA levels were ~50% of controls (Additional file 4). Porin (VDAC), an outer mitochondrial membrane protein, and α-actinin-2, a muscle specific protein, were used as loading controls, and to demonstrate the efficacy of the transdifferentiation of fibroblasts into myotubes (Figure 1a). Immunoblotting of patient fibroblast samples showed RC complex I & IV levels were 50-60% of control levels (Figure 1a). The observed complex I & IV deficiencies were more severe in MyoD-forced myotubes (P2:25-50% of control levels; P6:5-10% of control levels) compared to fibroblasts, consistent with a muscle-specific phenotype.

We measured complex IV activity in myotubes derived from MyoD-forced myogenesis of fibroblasts from YARS2 patients P2 (p.Phe52Leu) and P6 ([p.Gly191Asp];[p.Arg360X) to confirm that the reduced RC complex levels (Figure 1a) correlated with a reduction in RC enzyme activity. Transdifferentiation of fibroblasts to myotubes resulted in a significant increase in complex IV activity relative to total protein in normal controls (P = 0.002; Figure 1b), consistent with the increased relative levels of mitochondrial RC subunits detected by Western blot (Figure 1a and Additional file 3). In contrast, for both the patient P2 & P6 cell lines there was a decrease in complex IV activity relative to total protein (P < 0.001; Figure 1b) and relative to citrate synthase (P ≤ 0.005; Figure 1c). Both patient P2 fibroblasts and patient P6 fibroblasts had significantly lower complex IV/CS activity compared to control fibroblasts (P = 0.02 and P < 0.001 respectively; Figure 1c). However, MyoD-forced myogenesis revealed a deficit in Complex IV/CS activity that was < 50% of control levels and could be considered diagnostic. These results were consistent with the more severely reduced levels of Complex IV observed in patient myotubes compared to fibroblasts by Western blot (Figure 1a).

The muscle-specific phenotype of YARS2 patients may be related to the increased requirements for RC complexes in muscle 25 26 . Thus, we compared the relative levels of mitochondrial RC expression levels in 10 μg of total protein from fibroblasts, MyoD-forced myotubes, and skeletal muscle (Figure 2). In control samples, there was a ~2-fold increase in the relative levels of all RC complexes with transdifferentiation from fibroblasts to myotubes, and a further ~2-fold incremental increase of RC expression levels between myotubes and skeletal myofibres (Figure 2). However, cells from Patient 2 failed to induce mitochondrial RC complex I and IV expression with myogenesis, either via forced myogenesis of fibroblasts (Figure 2) or indeed native myogenesis of myoblasts (Additional file 3). Skeletal muscle samples from Patient 2 demonstrated a severe deficiency of complex I and IV, with compensatory upregulation of complex II. Thus, our results suggest that patients with YARS2 mutations can generate sufficient RC complex I, and IV in fibroblasts, but not in muscle where there is increased demand.

There were higher levels of mitofilin and porin (VDAC) in control myotubes and muscle compared to fibroblasts, indicative of greater numbers (or size) of mitochondria with myogenesis (Figure 2); consistent with increased RC complex levels. Skeletal muscle from patient P2 showed higher levels of mitofilin and porin compared to control muscle, suggesting mitochondrial proliferation was upregulated in response to the RC deficiency. There was no evidence of a compensatory upregulation of YARS2 in patient muscle.

We further investigated the role of mitochondrial proliferation in YARS2 pathology. PGC1-α is a transcription co-activator which promotes mitochondrial biogenesis 27 and TFAM controls mtDNA copy number and transcription 28 . Elevated levels of PGC1-α and TFAM were seen in patient P2 muscle compared to controls (Figure 3a), consistent with induction of mitochondrial proliferation pathways. Interestingly, relative levels of mtDNA/nDNA were not significantly higher in patient P2 muscle compared to control muscle (P = 0.09; Figure 3c), despite our biochemical evidence showing higher relative levels of mitochondrial structural proteins and induction of PGC1-α and TFAM pathways.

We next assessed the relative levels of PGC1-α and TFAM with myogenesis in patient and control myoblasts. PGC1-α levels were similar in patient and control myoblasts and myotubes, and showed ~2-fold upregulation with myogenesis (Figure 3b). Levels of TFAM also increased incrementally with myogenesis, but showed 3-fold higher expression levels in patient myoblasts and myotubes, relative to controls (Figure 3b). Analysis of the relative levels of mtDNA/nDNA during myogenesis showed there was strong evidence of an interaction between cell type and sample (P < 0.001) so cell types were compared within samples. Myogenesis of control myoblasts induced a ~2-fold increase in the relative levels of mtDNA/nDNA (P < 0.001). However, myogenesis-induced upregulation of mtDNA proliferation was not observed for P2 (P = 0.22; Figure 3d). The elevated level of TFAM observed in patient muscle tissue and cells, does not appear to result in elevated mtDNA levels.