Florin Sasarman 1, Hana Antonicka 1 and Eric A. Shoubridge 1,2, INTRODUCTION

Transcription

1 The A3243G trna Leu(UUR) MELAS mutation causes amino acid misincorporation and a combined respiratory chain assembly defect partially suppressed by overexpression of EFTu and EFG2 Florin Sasarman 1, Hana Antonicka 1 and Eric A. Shoubridge 1,2, 1 Montreal Neurological Institute, 3801 University Street, Montreal, QC H3A 2B4, Canada and 2 Department of Human Genetics, McGill University, Montreal, QC H3A 2B4, Canada Received June 10, 2008; Revised and Accepted August 22, 2008 Human Molecular Genetics, 2008, Vol. 17, No doi: /hmg/ddn265 Advance Access published on August 27, 2008 The majority of patients with MELAS (mitochondrial encephalomyophathy, lactic acidosis, stroke-like episodes) carry a heteroplasmic A3243G mutation in the mitochondrial trna Leu(UUR). The mutation prevents modification of the wobble U base, impairing translation at UUA and UUG codons; however, whether this results in amino acid misincorporation in the mitochondrial translation products remains controversial. We tested this hypothesis in homoplasmic mutant myoblasts isolated from a MELAS patient and investigated whether overexpression of the mitochondrial translation elongation factors could suppress the translation defect. Blue-Native gel electrophoretic analysis demonstrated an almost complete lack of assembly of respiratory chain complexes I, IV and V in MELAS myoblasts. This phenotype could be partially suppressed by overexpression of EFTu or EFG2 but not EFTs or EFG1. Despite the severity of the assembly defect, overall mitochondrial protein synthesis was only moderately affected, but some anomalously migrating translation products were present. Pulse-chase labeling showed reduced stability of all mitochondrial translation products consistent with the assembly defect. Labeling patterns of the translation products were similar with [ 3 H]-leucine or [ 3 H]-phenylalanine, showing that loss of the wobble U modification did not permit decoding of UUY codons; however, endoproteinase fingerprint analysis showed clear evidence of amino acid misincorporation in three polypeptides: CO III, CO II and ATP6. Taken together, these data demonstrate that the A3243G mutation produces both loss- and gain-of-function phenotypes, explaining the apparent discrepancy between the severity of the translation and respiratory chain assembly defects, and suggest a function for EFG2 in quality control of translation elongation. INTRODUCTION MELAS (mitochondrial encephalomyophathy, lactic acidosis, stroke-like episodes) is one of the most common mitochondrial diseases. In over 80% of patients it is caused by a A3243G mutation in the mitochondrial trna responsible for decoding leucine at UUR codons (1), and another 10% have a T3271C mutation in the same gene (2). The carrier frequency of the A3243G mutation alone has been estimated at 4 18 per in northern European populations (3 5). Like many mutations causing mitochondrial respiratory chain dysfunction, the A3243G mutation can produce a range of clinical phenotypes in addition to MELAS, including CPEO (chronic progressive external ophthalmoplegia) (6,7) and MIDD (maternally inherited diabetes and deafness) (8 10). All A3243G MELAS patients described are heteroplasmic for the mutation, and the relative proportion of mtdnas carrying the mutation correlates with the age of onset and severity of disease (6,11). Early studies in cybrid cells showed that as little as 5% of wild-type mtdna was sufficient to completely rescue cellular oxygen consumption, demonstrating an extremely high threshold for expression of this mutation (12,13). The majority of MELAS patients have, however, well below 95% mutant mtdnas, suggesting that studies in cybrid cells To whom correspondence should be addressed. Tel: þ ; Fax: þ ; eric@ericpc.mni.mcgill.ca # The Author Published by Oxford University Press. All rights reserved. For Permissions, please journals.permissions@oxfordjournals.org

2 3698 Human Molecular Genetics, 2008, Vol. 17, No. 23 may not be readily extrapolated to the in vivo condition. In addition, evidence of respiratory chain dysfunction has been reported in the skeletal muscle (14) and brain (15) at very low levels of heteroplasmy (,10%), indicating that there may in fact be no threshold for expression of the mutation, at least in some tissues. Even in cybrid cells there is variability in the severity of the enzymatic deficiency. Using cytochrome c oxidase (COX) as a marker of the respiratory chain deficiency, residual enzyme activities between 2 and 67% of control have been observed at 90 95% mutant loads, depending on nuclear background (16,17). Further complicating the picture, recent studies of respiratory chain assembly in a patient with 90% of mutant mtdnas showed a severe combined enzyme deficiency in brain tissue, an isolated complex I defect in heart and skeletal muscle, and no apparent defect in liver (18). Although much progress has been made in characterizing the biochemical defect in tissues and cybrid cells, the mechanisms of pathogenesis remain controversial. Several studies have demonstrated a decrease in the steady-state levels of the amino-acylated mutant trna (19 22), a result attributed, at least in part, to the propensity of the mutant to dimerize (22). However, the extent of the translation defect is variable, ranging from severe in some nearly homozygous cybrids (12), to moderate in others (13,20), and in any case it does not correlate well with the proportion of UUR codons of the individual polypeptides (21). The latter observation was explained by the discovery that the mutation prevents the 5-taurinomethyl modification which occurs post-transcriptionally at the wobble U position of the trna Leu(UUR) anti-codon (19). This severely disrupts decoding of UUG codons by the mutant trna, but only partially affects decoding at UUA (23). While the above results suggest loss-of-function of the mutant trna as the pathogenic mechanism for MELAS, misincorporation of leucine at non-cognate codons, thus gain-of-function, has also been proposed as a potential pathogenic mechanism (19,24). Certain wobble base modifications restrict the pairing capacity of the first position of the anti-codon, such that pairing with only a subset of bases is allowed at the wobble position (25,26). In the case of the trna Leu(UUR), the 5-taurinomethyl at the wobble position would dictate pairing with purines, and not with pyrimidines. It is conceivable that an unmodified wobble U in the mutant trna Leu(UUR) would permit decoding of the phenylalanine UUY codons, although in vitro studies on synthetic poly(uuc) mrnas revealed no decoding activity of the mutant trna (23). Misincorporation could also result from faulty aminoacylation of the mutant trna; however, examination of the charged trna in cybrid cells failed to find evidence for charging with an amino acid other than leucine (19). Further, a recent study in which two COX subunits, CO I and CO II, were immunocaptured from homoplasmic A3243G cybrid cells and analyzed by mass spectroscopy, identified only wild-type polypeptides (27), leading the authors to conclude that amino acid misincorporation does not contribute to pathogenicity. COX activity in these cells was, however, only 3% of control, and rapid degradation of mistranslated products could not be ruled out. We reasoned that part of the controversy over the pathogenic mechanism may have arisen because much of the data has been generated from cybrid cells, all of which are highly anaeuploid and genetically unstable. In the present study, we used diploid myoblasts, isolated from a muscle biopsy of a MELAS patient, to investigate the mechanism of pathogenesis. We show that when homoplasmic, the mutation produces a severe combined respiratory chain deficiency that results from both loss-of-function of trna Leu(UUR) and amino acid misincorporation, and that this phenotype can be partially suppressed by overexpression of two mitochondrial translation elongation factors. RESULTS The A3243G trna Leu(UUR) mutation causes a severe combined respiratory chain deficiency Myoblast clones that were either homoplasmic for the A3243G mutation (MELAS myoblasts) or homoplasmic wildtype were isolated from a muscle biopsy culture that was established from a MELAS patient. Cloned myoblasts, homoplasmic for the A8344G trna Lys mutation associated with MERRF (myoclonic epilepsy with ragged red fibers) (28), were used as a disease control (MERRF myoblasts). To investigate the specific defect in mitochondrial protein synthesis caused by the A3243G mutation, MELAS and MERRF myoblasts were pulse-labeled with a mixture of [ 35 S]-methionine and [ 35 S]-cysteine in the presence of emetine, an inhibitor of cytoplasmic translation. The overall rate of protein synthesis in mitochondria of MELAS myoblasts was 70% of control (range 60 80%) as measured by the total incorporation of the radioactive label (Fig. 1); however, the translation of some polypeptides such as ND5, and especially ND6, was disproportionately decreased. Additionally, we observed anomalous migration patterns for some mitochondrial translation products (identified with asterisks in Fig. 1, 5 8), suggestive of either amino acid misincorporation or premature translation termination. Translation of both ATP6 and ATP8 was increased relative to control. In contrast to the results obtained in MELAS myoblasts, translation of all the mitochondrial polypeptides was severely impaired in MERRF myoblasts, as has been reported previously (28) (Fig. 1), and the described premature translation termination product of the CO I subunit of COX was clearly evident (29). To investigate the effect of these translation defects on respiratory chain function, we first assayed COX/citrate synthase activity in whole cell extracts of MELAS and MERRF myoblasts as a marker of respiratory chain activity. The three mtdna-encoded COX subunits, CO I, II, III, are synthesized at rates of approximately 1, 0.6 and 0.9 in MELAS myoblasts relative to control (Table 1), whereas synthesis of the same subunits in MERRF myoblasts is,10% of control (CO I and III), and 0.3 of control for CO II. Despite the striking difference in the severity of the translation defects in the COX subunits between the two different trna mutations, the per cent residual activity of COX was indistinguishable in the MELAS ( ) and MERRF ( ) myoblasts (n 6). This discrepancy could be due to either decreased levels of fully assembled COX enzyme or to the presence of non-functional COX complexes in the MELAS myoblasts.

3 Human Molecular Genetics, 2008, Vol. 17, No Figure 1. Synthesis of mitochondrial polypeptides is only mildly affected in MELAS myoblasts. Homoplasmic wild-type (WT) and homoplasmic mutant (MELAS and MERRF) patient myoblasts were pulse-labeled in the presence of emetine and the mitochondrial translation products were run on a 15 20% polyacrylamide gradient gel, which was processed for fluorography as detailed in the Materials and Methods section. The seven complex I subunits (ND), one complex III subunit (cyt b), three complex IV subunits (CO) and two complex V subunits (ATP) are indicated on the left. Anomalously running bands in the mutant cells are indicated on the right: the single band below ND1 and the doublet above ND6 in the MELAS myoblasts are identified with asterisks, and the CO I premature termination product in the MERRF myoblasts is identified as pmerrf (29). To distinguish between these two possibilities, we investigated the assembly of all respiratory chain complexes in MELAS and MERRF myoblasts by Blue-Native PAGE electrophoresis. This analysis revealed an almost complete absence of fully assembled COX and complex I in MELAS myoblasts, only trace amounts of fully assembled complex V and slightly decreased levels of complex III (Fig. 2A and B). As expected, complex II was unaffected. In MERRF myoblasts, the four respiratory chain complexes containing mtdna-encoded subunits were undetectable (Fig. 2C). We conclude that the combined respiratory deficiency in MELAS is due to a failure to assemble the enzyme complexes. Partial suppression of the respiratory chain deficiency by overexpression of mitochondrial translation elongation factors EFTu and EFG2 Based on the observation that the translation elongation factor EFTu is a universal suppressor of many different trna mutations in yeast [(30) and references therein], we tested whether the severe phenotype of the MELAS myoblasts could be rescued by EFTu overexpression, or by overexpression of the other known mitochondrial translation elongation factors, viz. EFTs, EFG1 or EFG2. For this purpose, we transduced MELAS and wild-type myoblasts with retroviral vectors containing the cdnas for each elongation factor, confirming their overexpression by immunoblot analysis (except EFG2 for which antibodies were not available) (Fig. 3A). When compared with wild-type myoblasts overexpressing the same translation factor, COX activity increased to % (n ¼ 11) of control levels in MELAS myoblasts overexpressing EFTu and to % (n ¼ 7) in patient myoblasts overexpressing EFG2, which represents a fold increase from the residual COX activity in untransduced cells (Fig. 3B). The partial suppression of COX deficiency in MELAS myoblasts overexpressing EFTu or EFG2 was paralleled by an increase in the amount of assembled enzyme (Fig. 2A and B) and increased steady-state levels of both mitochondrial- and nuclear-encoded OXPHOS subunits, as illustrated by CO II and CO IV, two subunits of COX (Fig. 4). The steady-state level of EFTs was also increased in MELAS cells overexpressing EFTu (Fig. 3A). Overexpression of EFG2 produced a substantial increase in the levels of fully assembled complexes V and III, the latter to control levels, and a modest increase in assembled complex I. Overexpression of either EFTu or EFG2 restored the overall incorporation of radioactivity to control levels (123 and 100%, respectively), including a noticeable increase in the synthesis of ND6 in the case of EFG2; however, the anomalously migrating bands were still present (Fig. 5). These findings seem to be specific to the MELAS mutation, as no suppression of the defective phenotypes was observed in MERRF myoblasts (Fig. 2C, and data not shown). No suppression of the defects in respiratory chain assembly, COX activity or mitochondrial translation was seen by overexpression of EFTs or EFG1 in either MELAS or MERRF myoblasts. Decreased stability of the mitochondrial translation products in MELAS myoblasts Unassembled respiratory chain subunits are generally degraded rapidly by the AAA ATP-dependent protease, the quality-control machinery of the inner mitochondrial membrane (31). The severe respiratory chain assembly defect in MELAS myoblasts thus predicts an enhanced rate of degradation of the mitochondrial translation products. To test this hypothesis, wild-type and MELAS myoblasts were pulselabeled with a mixture of [ 35 S]-methionine and [ 35 S]-cysteine in the presence of anisomycin, a reversible inhibitor of cytoplasmic translation, and chased for various amounts of time (Fig. 6A). After 8.7 and 17.5 h, the radioactivity present in MELAS myoblasts was 40% of control, indicating a shorter half-life of the mitochondrial translation products in the MELAS myoblasts. Similar experiments in MERRF myoblasts (Fig. 6B) showed only trace amounts of all mitochondrial translation products after 17.5 h, in accord with the complete lack of assembled OXPHOS complexes in these cells. It is also apparent from the chase that the anomalously migrating bands (those identified with asterisks and, as seen in Figs 6 and 7, CO III) in the MELAS cells degrade very rapidly, within 2.2 h.

4 3700 Human Molecular Genetics, 2008, Vol. 17, No. 23 Table 1. Maximal theoretical fold change in labeling intensity upon misincorporation of leucine and phenylalanine in the mitochondrial translation products of MELAS myoblasts Subunits Number of codons MELAS/WT labeling intensity Experimental value (exp) a Theoretical values Loss-of-function Gain-of-function Leu Leu Phe MetþCys Phe at Leu Leu at Phe UUR CUN UUY 35 S 3 H-Leu b 3 H-Phe c 3 H-Leu d 3 H-Phe e ND No incorporation of label CO I ND cyt b ND ND CO III CO II ATP ND ND ND4L ATP a Ratio of measured incorporation of 35 [S]-(methionine and cysteine) between MELAS and wild-type myoblasts (reflects difference in labeling between mutant and wild-type cells when an amino acid presumably not involved in the misincorporation process is used). b Ratio of expected leucine incorporation between MELAS and wild-type myoblasts if all UUR codons are misread: [CUN/(UURþCUN)](exp). c Ratio of expected phenylalanine incorporation between MELAS and wild-type myoblasts if all UUR codons are misread by trna Phe : [(UURþUUY)/ UUY](exp). d Ratio of expected leucine incorporation between MELAS and wild-type myoblasts if all UUY codons are misread by trna Leu(UUR) : [(UURþCUNþUUY)/ (UURþCUN)](exp). e No incorporation of radioactive phenylalanine in MELAS myoblasts if all UUY codons are misread by trna Leu(UUR). Amino acid misincorporation in the mitochondrial translation products of the MELAS myoblasts The presence of several anomalously migrating mitochondrial translation products in MELAS myoblasts led us to suspect amino acid misincorporation as a mechanism that might reconcile the modest reduction in translational efficiency with the severe combined respiratory chain assembly defect. Misincorporation of amino acids during mitochondrial translation in MELAS myoblasts could result from faulty incorporation of other amino acids at leucine UUR codons, the inappropriate incorporation of leucine at non-cognate codons, or both. To test this hypothesis, we first compared the protease digest profiles [endopeptidase fingerprint analysis (32)] of the mitochondrial translation products in MELAS and wild-type myoblasts. Cells were pulse-labeled for an hour with a mixture of [ 35 S]-methionine and [ 35 S]-cysteine in the presence of emetine and PAGE was performed to separate the labeled mitochondrial translation products in the first dimension. Individual lanes were then excised from the gel and run in the second dimension, while being subjected to in-gel digestion with either endoproteinase Glu-C, which cleaves at the C-terminal end of glutamate residues, or with alkaline protease, which cleaves preferentially at the C-terminal end of large hydrophobic amino acids (Fig. 7). To ensure identical digest conditions, each second dimension gel included two slices of the first dimension gel run alongside: one containing labeled translation products from wild-type, the other from MELAS myoblasts. Overlapping parts of gel slices containing the mitochondrial translation products from a given cell line were run in different gels to confirm the reproducibility of the digest profiles. As seen from the fingerprints generated with both peptidases, the digest profiles of CO III, CO II and ATP6 are clearly different in MELAS, as compared with wild-type myoblasts. The CO III polypeptide contains seven glutamates, so a full digest with endopeptidase Glu-C would create eight products. Three of these do not contain methionine or cysteine (not detectable in our assay), three contain one methionine (most likely not detectable in our assay), while one peptide contains a methionine and a cysteine (molecular weight of 6.0 kda), and the final peptide contains seven methionines (molecular weight of 7.5 kda). Considering the sizes of the predicted peptides, the positions of the glutamates and the content of methionine/cysteine residues in the CO III polypeptide, it is clear that under the conditions used in our assay, endopeptidase Glu-C cleaves the nascent CO III polypeptide in wild-type cells only partially, creating two products of an apparent molecular weight of 18.9 and 7.2 kda (Fig. 7). The predicted prominent cleavage site is at position E64. Under the conditions of our assay (Fig. 7), we detected three additional peptides in the MELAS myoblasts of an apparent molecular weight of 14.7, 12.5 and 10.7 kda. The wild-type sequence of CO III contains three leucine residues, at positions 112, 137 and 168, all encoded by UUA codons. The misincorporation of glutamate at L112 in the CO III polypeptide would create a Glu-C partial digest of 13 and 16.9 kda; into L137 would create a digest of 14.3 and 15.6 kda, and into L168 would create a digest of 10.8 and 19.1 kda. It is not possible to conclude with certainty which, if any, of the three scenarios is more likely, as the digest is partial, and the apparent molecular weights of the mitochondrial translation products are often quite different from that predicted by the amino acid sequence. It is also possible that

5 Human Molecular Genetics, 2008, Vol. 17, No Figure 2. Partial suppression of the defect in assembly of OXPHOS complexes in MELAS myoblasts overexpressing EFTu or EFG2. Homoplasmic wild-type (WT), homplasmic MELAS (A and B) and homoplasmic MERRF (C) myoblasts were transduced with retroviral vectors carrying translation elongation factor EFTu (Tu), EFTs (Ts), EFG1 (G1) or EFG2 (G2) and analyzed by BN-PAGE. The blots were incubated with antibodies against specific subunits of each of the five OXPHOS complexes (Co I V). The blots in (B) represent longer exposures of the corresponding blots in (A), emphasizing the partial rescue of assembly by EFTu and by EFG2. Subcomplex indicates the presence of partially assembled complexes I or V in patient myoblasts. misincorporation of another amino acid at these or other positions facilitates digestion with Glu-C. Nonetheless, these results clearly demonstrate that amino acid misincorporation occurs in at least some of the mitochondrial translation products in MELAS myoblasts. To test whether inappropriate decoding of phenylalanine UUY codons by the mutant trna could also occur in MELAS myoblasts, we compared labeling patterns after pulse-labeling the mitochondrial translation products with [ 3 H]-leucine or [ 3 H]-phenylalanine (Fig. 8). The maximal theoretical change in labeling intensity was calculated for misincorporation of leucine and phenylalanine during mitochondrial protein synthesis in MELAS cells (Table 1). For both mutant and wild-type cells, the relative labeling of the individual polypeptides accurately reflects their leucine and phenylalanine content: ND4, for instance, which labeled several times more intensely with [ 3 H]-leucine than with [ 3 H]-phenylalanine contains 96 leucine and 20 phenylalanine residues (Fig. 8, Table 1). On the other hand, subunits containing a comparable number of leucine and phenylalanine residues, such is the case of CO I (62 and 41, respectively), labeled to a similar extent with both [ 3 H]-leucine and with [ 3 H]-phenylalanine. A comparison of the overall incorporation of radioactivity in MELAS versus wild-type myoblasts after Figure 3. Partial rescue of COX enzymatic activity in MELAS myoblasts overexpressing EFTu or EFG2. (A). Overexpression of the mitochondrial translation elongation factors in homoplasmic wild-type (WT) and homoplasmic MELAS myoblasts was confirmed through immunoblot analysis using polyclonal antibodies against each of the factors with the exception of EFG2. A monoclonal antibody against the 70 kda subunit of complex II was used to confirm equal loading. (B). Histogram showing the activity of COX normalized to citrate synthase in wild-type (WT) and homoplasmic MELAS myoblasts overexpressing the various translation factors. The COX activity of wild-type myoblasts before transduction was arbitrarily set at 100%. labeling of the mitochondrial translation products with [ 35 S]-methionine, [ 3 H]-leucine or [ 3 H]-phenylalanine showed no quantitative or qualitative differences in the resulting labeling patterns. These observations suggest that the mutant trna is unable to decode UUY codons and that phenylalanine is not the major amino acid involved in the misincorporation process in MELAS myoblasts. DISCUSSION The results of this study clearly demonstrate that the A3243G mutation produces a severe combined respiratory chain defect in myoblasts, as evidenced by an almost complete lack of assembly of complexes I, IV and V, and a slight decrease of assembled complex III. A similar pattern has been reported in the frontal cortex of a A3243G MELAS patient (18). This severe assembly defect occurs despite a modest reduction in the overall rate of mitochondrial protein synthesis. However, translation of some polypeptides is disproportionately decreased, and there is clear evidence of amino acid misincorporation in others. Thus, the mutation produces both loss- and gain-of-function biochemical phenotypes, which may explain why patients carrying a relatively low proportion of mutants

6 3702 Human Molecular Genetics, 2008, Vol. 17, No. 23 Figure 4. Increased steady-state levels of mitochondrial- and nuclear-encoded OXPHOS subunits in MELAS myoblasts transduced with EFTu or EFG2. Western blot analysis of homoplasmic wild-type (WT) and mutant (MELAS) myoblasts, either untransduced, or expressing EFTu or EFG2 with antibodies against mitochondrial or nuclear OXPHOS subunits, as indicated on the left. Equal loading was confirmed with antibodies against the 70 kda subunit of complex II and against porin, a non-oxphos protein. Figure 6. Decreased stability of nascent mitochondrial translation products in MELAS myoblasts. Homoplasmic wild-type (WT) and MELAS (A) or MERRF (B) myoblasts were labeled with [ 35 S]-(methionine and cysteine) in the presence of anisomycin and subsequently chased for various intervals of time, as indicated at the top of each gel. Equal amounts of protein (50 mg) were run on 12 20% polyacrylamide gradient gels, which were analyzed through direct autoradiography. The seven complex I subunits (ND), one complex III subunit (cyt b), three complex IV subunits (CO) and two complex V subunits (ATP) are indicated on the left of each gel. Asterisks and pmerrf indicate anomalously running bands in the mutant cells. Figure 5. An increase in overall mitochondrial translation in MELAS myoblasts expressing EFTu or EFG2. Pulse-labeled mitochondrial translation products in wild-type (WT) and mutant (MELAS) myoblasts before and after transduction with retroviral vectors expressing each of the four mitochondrial translation elongation factors were separated on a 12 20% polyacrylamide gradient gel and analyzed through direct autoradiography. The seven complex I subunits (ND), one complex III subunit (cyt b), three complex IV subunits (CO) and two complex V subunits (ATP) are indicated on the left. Anomalously running bands in the mutant cells are indicated on the right. develop clinical disease (14), and why there is apparently no threshold for expression of the mutation in some tissues (15). The rate of synthesis of most individual polypeptides in MELAS myoblasts ranges from around 60% of control to wild-type levels, with the exception of ND6 which is,10% of control, and ATP6 and ATP8, which are synthesized at approximately double the rate of control cells. The severe decrease in the rate of synthesis of ND6 is a recognized consequence of the A3243G mutation, and it correlates with the unusually high content of UUG codons in this polypeptide (23). This subunit is essential in complex I assembly (33), and a mutation in ND6 has been described in a patient with MELAS (34). Among the other complex I subunits, the synthesis of ND5 is the next most affected, being reduced by about half. ND5 is also essential for complex I assembly (35) and ND5 mutations have been associated with MELAS (36,37). The decreased synthesis of ND5 or ND6 alone can thus account for the lack of complex I observed in the MELAS myoblasts. The increased synthesis of ATP6 and ATP8 has been reported previously in MELAS cybrids (20) and in fibroblasts from patients with mutations in the mitochondrial translation elongation factors EFTs and EFG1 (38 40). This increase has been proposed to reflect preferential translation of bicistronic messages in situations where mitochondrial translation becomes impaired (38), although this cannot be universally true, as it does not occur in MERRF myoblasts. Despite the increased synthesis of ATP6 and 8, the levels of fully assembled complex V are severely reduced in MELAS myoblasts (Fig. 2). Similarly, synthesis of the three mtdna-encoded COX subunits is % of control levels, but COX fails to assemble to a significant extent in MELAS myoblasts, and the residual enzyme activity is 6% of control levels. These observations, together with the appearance of several anomalous mitochondrial translation products, some of which had been reported in A3243G cybrids (20,41), strongly suggested that amino acid misincorporation occurs during mitochondrial protein synthesis in MELAS myoblasts.

7 Human Molecular Genetics, 2008, Vol. 17, No Figure 7. Amino acid misincorporation into newly synthesized mitochondrial products in MELAS myoblasts. Homoplasmic wild-type (WT) or mutant (MELAS) myoblasts were labeled with [ 35 S]-(methionine and cysteine) in the presence of emetine and 50 mg of protein of each sample were run in the first dimension on a 15 20% polyacrylamide gradient gel. Individual lanes were excised from the gel (top) and run in the second dimension on a 15% polyacrylamide gel containing either endopeptidase Glu-C (Glu-C) or alkaline protease (Alk. Pr.), as detailed in the Results and Materials and Methods sections. The position of each polypeptide in the first dimension gel is indicated at the top of the figure: ND: complex I subunits, cyt b: complex III subunit, CO: complex IV subunits, ATP: complex V subunits. Arrows point to digest products that are present either only in the wild-type, or only in the MELAS myoblasts. Asterisks indicate anomalous translation products synthesized in the mutant cells. Molecular weight markers are indicated on the left of the figure. This idea has been tested previously (20,24,27), but with contradictory results. In one study, the incorporation of [ 3 H]-leucine and the proteolytic fingerprint pattern of the mitochondrial translation products were investigated in EBVtransformed B-cells, carrying 70% A3243G mutant mtdna (24). The authors measured a decrease in the incorporation of leucine in most mitochondrial translation products, and a proteolytic fingerprint pattern distinct from control cells, concluding that amino acid misincorporation had occurred in several polypeptides. The measured decrease in leucine incorporation in ND5 was around 30%, while in COI, around 20%. However, as can be calculated from the values listed in

8 3704 Human Molecular Genetics, 2008, Vol. 17, No. 23 Figure 8. Pulse-labeling of mitochondrial translation products with [ 3 H]-leucine or [ 3 H]-phenylalanine in wild-type and mutant patient myoblasts. Homoplasmic wild-type (WT) and MELAS or MERRF myoblasts were labeled with either [ 3 H]-leucine or [ 3 H]-phenylalanine, as indicated at the bottom of the gel. Equal amounts of protein (50 mg) were run on a 15 20% polyacrylamide gradient gel, which was processed for fluorography, as detailed in the Materials and Methods section. The 13 mitochondrially synthesized polypeptides are indicated at the left of the gel: ND: complex I subunits, cyt b: complex III subunit, CO: complex IV subunits, ATP: complex V subunits. On the right of the gel, anomalously running bands described in Figure 1 are indicated. Table 1, the maximum decrease in leucine incorporation that can take place due to loss-of-function of the trna Leu(UUR) in homoplasmic mutant cells is 9 and 11% for ND5 and CO I, respectively. The cells also showed a normal pattern of respiratory chain assembly, inconsistent with the reported translation defects. Amino acid misincorporation was also suggested by a report showing a 15% overall decrease in [ 3 H]-leucine incorporation into mitochondrial protein in cybrid cells carrying the A3243G mutation (20). In contrast, a recent study concluded that amino acid misincorporation did not occur in the mitochondrial translation products of cybrids homozygous for the A3243G mutation (27). Immunocapture of COX and mass spectroscopy analysis identified exclusively wild-type CO I and CO II in these cells (CO III was not detected) (27). As in the myoblasts studied here, the steady-state levels of CO I and CO II were,5% of control, implying that upon synthesis, these COX subunits are not incorporated into a functional complex and are rapidly degraded. This is most likely due to the fact that several of the polypeptides synthesized in the mitochondria of MELAS myoblasts contain an incorrect sequence, as revealed by their proteolytic fingerprint patterns (see Fig. 7). The rapid decay of these mutant proteins would then prevent their accumulation and detection through immunoprecipitation. Our results reveal amino acid misincorporation in CO II, CO III and ATP6, in the MELAS myoblasts. It is worth mentioning that, just as with ND5 and ND6, mutations in both CO II (42) and CO III (43) have been associated with MELAS. Amino acid misincorporation could occur in additional polypeptides, but this might not be detectable with the endopeptidases used in this study. In addition, the most anomalously running bands could be the result of premature termination of translation, but the endopeptidase digestion pattern of these products does not permit assignment to any particular translation product. The fact that different proteolytic fingerprint patterns are generated with the endoproteinase Glu-C in MELAS and wild-type myoblasts suggests that glutamate might be involved in misincorporation at leucine codons; however, it is also possible that misincorporation at other sites makes the glutamate residues more accessible to digestion with the endopeptidase. In any case, the difference in digestion patterns must reflect some qualitative difference in the nature of the polypeptides. An intriguing question about misincorporation is whether it happens in a conservative manner, that is, for a given codon which is mistranslated, is the same amino acid used at all times, in all cell lines harboring the A3243G mutation? In MELAS myoblasts, we always observed the anomalously running bands present at the same positions, with the same appearance (doublet versus single band), regardless of the radioactive amino acid used for labeling (see Figs 1, 5, 6, 8). Their protease digest profiles were also reproducible with both endopepetidases used in this study. Furthermore, other groups have reported abnormal translation products with similar migration patterns in cybrids carrying the A3243G mutation (20,41). Taken together, these observations suggest that conservative misincorporation is likely to occur as a consequence of the A3243G mutation in the trna Leu(UUR). The other common MELAS mutation, T3271C in trna Leu(UUR), also causes a moderate mitochondrial translation defect with anomalously running bands, decreased ND6 synthesis and complex I deficiency (44) suggesting that amino acid misincorporation might be a common theme to mutations in mitochondrial trnas that are associated with MELAS. In yeast, mitochondrial defects due to the yeast counterpart of the A3243G human mutation are rescued by overexpression of the translation elongation factor EFTu (30), as well as by the overexpression of the mitochondrial leucyl-trna synthetase (45). While EFTu appears to be a general suppressor of trna mutations in yeast (30,46,47), its overexpression had no effect on the mutant phenotype caused by a mutation in the mitochondrial trna Ser(UCN) in human cells (48). Nonetheless, the results obtained in yeast prompted us to investigate whether overexpression of any of the mitochondrial translation elongation factors could suppress the A3243G phenotype in myoblasts. We show that EFTu as well as the translocation factor EFG2, but not EFG1 or EFTs, can partially suppress this phenotype. The rescue of mitochondrial trna mutant phenotypes by the overexpression of EFTu can be attributed to its function in the translation process, namely providing charged trnas to the ribosomal A-site during elongation of peptide chains, and hence, promoting protein synthesis. Because of the direct interaction with the charged trnas, overexpression of EFTu might additionally contribute to the stabilization of the mutant trna. We previously showed that, in a patient with a mutation in the translocation factor EFG1, the levels of EFTu increase as part of an adaptive response in patient heart, the tissue least affected by the

9 Human Molecular Genetics, 2008, Vol. 17, No presence of the mutant protein (40). However, overexpression of EFTu in control cells has a dominant negative effect (38,40) on mitochondrial translation (Fig. 5), steady-state levels of COX subunits (Fig. 4), COX activity (Fig. 3B) and assembly of OXPHOS complexes (Fig. 2), suggesting that the relative ratios of the mitochondrial translation factors are important for efficient protein synthesis, and this may explain, in part, why suppression of the mutant phenotype is relatively modest. At present, it is unclear why mitochondria possess two translocation factors, EFG1 and EFG2, and the partial suppression of the MELAS biochemical defect by overexpression of EFG2 raises questions about the function of this factor in vivo. Overexpression of EFG2 does not rescue the translation defect in fibroblasts from patients with EFG1 mutations, indicating that the two factors have independent and nonoverlapping functions (39), consistent with the results we present here. Yeast deletion mutants of MEF2, the EFG2 homolog, do not have an obvious respiratory phenotype (49). Based on this study, one possibility would be that EFG2 is involved in quality control of mammalian mitochondrial protein synthesis at the translocation step, possibly just before peptide bond formation. Another possibility is that it interacts with components of the translation machinery that have yet to be identified, such as, for instance, mammalian translational activator proteins. MATERIALS AND METHODS Cell culture The MELAS myoblasts used in this study were cultured from a muscle biopsy specimen taken from a 6 years old female patient who presented with stroke-like episodes and was diagnosed with MELAS. Myoblasts were purified by fluorescence-activated cell sorting (FACS) (50), cloned (28), and individual clones were tested for the A3243G mutation by RFLP analysis with HaeIII after last-cycle-hot PCR (51). Homoplasmic mutant and homoplasmic wild-type clones were then immortalized with a retroviral vector expressing the E6 and E7 genes of type 16 human papilloma virus and with a retroviral vector expressing htert, the protein component of human telomerase (52). Retroviral vectors and enzyme measurements The construction of the retroviral vectors containing the cdna sequence of the four mitochondrial translation elongation factors (EFTu, EFTs, EFG1 and EFG2), their transfection into the packaging cell line and the subsequent infection of patient cells have been previously described (40). COX and citrate synthase activities were measured in myoblast cell extracts according to methods described earlier (53). Blue-Native PAGE and immunoblotting BN-PAGE (54) was carried out as described previously. Briefly, mitoplasts, prepared from myoblasts by treatment with 1.2 mg digitonin per mg of protein (55), were solubilized with 1% lauryl maltoside, and samples containing 20 mg of protein were separated on 6 15% polyacrylamide gradient gels. Assembly of complexes II V was assessed by immunoblotting with monoclonal, commercially available antibodies (Molecular Probes), and assembly of complex I with a polyclonal, anti-nd1 antibody (a kind gift from A. Lombes, Paris). For immunoblotting, whole cells were solubilized in 1.5% lauryl maltoside, and 20 mg of protein were run on a 12% Tris Glycine SDS gel, and subsequently transferred to a nitrocellulose membrane. For immunodetection, the same antibodies described previously (38) were used, along with monoclonal antibodies against the CO II subunit of COX (a kind gift from N. Kennaway, Portland, OR, USA). Pulse-labeling of mitochondrial translation products In vitro pulse-labeling of mitochondrial translation with [ 35 S]-(methionine and cysteine) (PRO-MIX from GE Healthcare) for 1 h was performed as described previously (28). In experiments where the label was chased, cells were incubated for h in 40 mg/ml chloramphenicol prior to labeling, and emetine was replaced with anisomycin (at the same final concentration of 100 mg/ml). [ 35 S]-labeled products were detected either through direct autoradiography or fluorography. Pulse-labeling of mitochondrial translation products with [ 3 H]-leucine (GE Healthcare, 160 Ci/mmol, 5 mci/ml stock, 450 m Ci/ml final) or with [ 3 H]-phenylalanine (GE Healthcare, 123 Ci/mmol, 5 mci/ml, 450 m Ci/ml final) was performed in high-glucose DMEM lacking either leucine or phenylalanine (Wisent/Multicell). [ 3 H]-labeled products were detected by fluorography. Gels were fixed in an isopropanol-acetic acid solution, treated with Amplify (GE Healthcare), dried and exposed to preflashed Amersham Hyperfilm TM MP film (GE Healthcare) at 2808C, for 2 weeks. Endopeptidase fingerprinting The proteolytic digestion of in vitro synthesized mitochondrial polypeptides was performed as described in (32) with slight modifications. Mitochondrial translation products labeled with [ 35 S]-(methionine and cysteine) for 1 h were run on a 0.75 mm-thick, 15 20% polyacrylamide gradient gel. Gel slices were cut and soaked for 30 min in buffer A (125 mm Tris HCl (ph 6.8), 0.1% SDS, 1 mm EDTA, 10% glycerol). 1 mm-thick 15% SDS polyacrylamide gels were poured using preparative combs and 0.05% SDS in the stacking gel. Endoproteinases Glu-C (50 m g/ml) and alkaline protease (500 ng/ml) were prepared in buffer A, and 100 ml of either enzyme solution was run into the gel. Presoaked gel slices were positioned into the wells, overlayed with 150 ml of the appropriate endoproteinase and the electrophoresis was run in a classical manner, except that the run was paused for 30 min when the samples reached the bottom of the stacking gel. After the electrophoresis was finished, the gels were equilibrated in a solution of isopropanol:water:acetic acid (25:65:10), dried and exposed for direct autoradiography. FUNDING This research was supported by a grant from the CIHR to E.A.S. E.A.S. is an International Scholar of the HHMI.

10 3706 Human Molecular Genetics, 2008, Vol. 17, No. 23 ACKNOWLEDGEMENTS We thank Timothy Johns for help with the cell culture and Guy-Hellen Guercin for cloning the mitochondrial translation factors. Conflict of Interest statement. None declared. REFERENCES 1. Goto, Y., Nonaka, I. and Horai, S. (1990) A mutation in the trna(leu)(uur) gene associated with the MELAS subgroup of mitochondrial encephalomyopathies. Nature, 348, Goto, Y., Nonaka, I. and Horai, S. (1991) A new mtdna mutation associated with mitochondrial myopathy, encephalopathy, lactic acidosis and stroke-like episodes (MELAS). Biochim. Biophys. Acta, 1097, Chinnery, P.F., Johnson, M.A., Wardell, T.M., Singh-Kler, R., Hayes, C., Brown, D.T., Taylor, R.W., Bindoff, L.A. and Turnbull, D.M. (2000) The epidemiology of pathogenic mitochondrial DNA mutations. Ann. Neurol., 48, Schaefer, A.M., McFarland, R., Blakely, E.L., He, L., Whittaker, R.G., Taylor, R.W., Chinnery, P.F. and Turnbull, D.M. (2008) Prevalence of mitochondrial DNA disease in adults. Ann. Neurol., 63, Uusimaa, J., Moilanen, J.S., Vainionpaa, L., Tapanainen, P., Lindholm, P., Nuutinen, M., Lopponen, T., Maki-Torkko, E., Rantala, H. and Majamaa, K. (2007) Prevalence, segregation, and phenotype of the mitochondrial DNA 3243A.G mutation in children. Ann. Neurol., 62, Mariotti, C., Savarese, N., Suomalainen, A., Rimoldi, M., Comi, G., Prelle, A., Antozzi, C., Servidei, S., Jarre, L., DiDonato, S. et al. (1995) Genotype to phenotype correlations in mitochondrial encephalomyopathies associated with the A3243G mutation of mitochondrial DNA. J. Neurol., 242, Koga, Y., Akita, Y., Takane, N., Sato, Y. and Kato, H. (2000) Heterogeneous presentation in A3243G mutation in the mitochondrial trna(leu(uur)) gene. Arch. Dis. Child., 82, Kadowaki, T., Kadowaki, H., Mori, Y., Tobe, K., Sakuta, R., Suzuki, Y., Tanabe, Y., Sakura, H., Awata, T., Goto, Y. et al. (1994) A subtype of diabetes mellitus associated with a mutation of mitochondrial DNA. N. Engl. J. Med., 330, Reardon, W., Ross, R.J., Sweeney, M.G., Luxon, L.M., Pembrey, M.E., Harding, A.E. and Trembath, R.C. (1992) Diabetes mellitus associated with a pathogenic point mutation in mitochondrial DNA. Lancet, 340, van den Ouweland, J.M., Lemkes, H.H., Ruitenbeek, W., Sandkuijl, L.A., de Vijlder, M.F., Struyvenberg, P.A., van de Kamp, J.J. and Maassen, J.A. (1992) Mutation in mitochondrial trna(leu)(uur) gene in a large pedigree with maternally transmitted type II diabetes mellitus and deafness. Nat. Genet., 1, Chinnery, P.F., Howell, N., Lightowlers, R.N. and Turnbull, D.M. (1997) Molecular pathology of MELAS and MERRF. The relationship between mutation load and clinical phenotypes. Brain, 120 (Pt 10), Chomyn, A., Martinuzzi, A., Yoneda, M., Daga, A., Hurko, O., Johns, D., Lai, S.T., Nonaka, I., Angelini, C. and Attardi, G. (1992) MELAS mutation in mtdna binding site for transcription termination factor causes defects in protein synthesis and in respiration but no change in levels of upstream and downstream mature transcripts. Proc. Natl. Acad. Sci. USA, 89, King, M.P., Koga, Y., Davidson, M. and Schon, E.A. (1992) Defects in mitochondrial protein synthesis and respiratory chain activity segregate with the trna(leu(uur)) mutation associated with mitochondrial myopathy, encephalopathy, lactic acidosis, and strokelike episodes. Mol. Cell. Biol., 12, Chinnery, P.F., Taylor, D.J., Brown, D.T., Manners, D., Styles, P. and Lodi, R. (2000) Very low levels of the mtdna A3243G mutation associated with mitochondrial dysfunction in vivo. Ann. Neurol., 47, Dubeau, F., De Stefano, N., Zifkin, B.G., Arnold, D.L. and Shoubridge, E.A. (2000) Oxidative phosphorylation defect in the brains of carriers of the trnaleu(uur) A3243G mutation in a MELAS pedigree. Ann. Neurol., 47, Dunbar, D.R., Moonie, P.A., Jacobs, H.T. and Holt, I.J. (1995) Different cellular backgrounds confer a marked advantage to either mutant or wild-type mitochondrial genomes. Proc. Natl. Acad. Sci. USA, 92, van den Ouweland, J.M., Maechler, P., Wollheim, C.B., Attardi, G. and Maassen, J.A. (1999) Functional and morphological abnormalities of mitochondria harbouring the trna(leu)(uur) mutation in mitochondrial DNA derived from patients with maternally inherited diabetes and deafness (MIDD) and progressive kidney disease. Diabetologia, 42, Fornuskova, D., Brantova, O., Tesarova, M., Stiburek, L., Honzik, T., Wenchich, L., Tietzeova, E., Hansikova, H. and Zeman, J. (2008) The impact of mitochondrial trna mutations on the amount of ATP synthase differs in the brain compared to other tissues. Biochim. Biophys. Acta, 1782, Yasukawa, T., Suzuki, T., Ueda, T., Ohta, S. and Watanabe, K. (2000) Modification defect at anticodon wobble nucleotide of mitochondrial trnas(leu)(uur) with pathogenic mutations of mitochondrial myopathy, encephalopathy, lactic acidosis, and stroke-like episodes. J. Biol. Chem., 275, Janssen, G.M., Maassen, J.A. and van Den Ouweland, J.M. (1999) The diabetes-associated 3243 mutation in the mitochondrial trna(leu(uur)) gene causes severe mitochondrial dysfunction without a strong decrease in protein synthesis rate. J. Biol. Chem., 274, Chomyn, A., Enriquez, J.A., Micol, V., Fernandez-Silva, P. and Attardi, G. (2000) The mitochondrial myopathy, encephalopathy, lactic acidosis, and stroke-like episode syndrome-associated human mitochondrial trnaleu(uur) mutation causes aminoacylation deficiency and concomitant reduced association of mrna with ribosomes. J. Biol. Chem., 275, Wittenhagen, L.M. and Kelley, S.O. (2002) Dimerization of a pathogenic human mitochondrial trna. Nat. Struct. Biol., 9, Kirino, Y., Yasukawa, T., Ohta, S., Akira, S., Ishihara, K., Watanabe, K. and Suzuki, T. (2004) Codon-specific translational defect caused by a wobble modification deficiency in mutant trna from a human mitochondrial disease. Proc. Natl. Acad. Sci. USA, 101, Flierl, A., Reichmann, H. and Seibel, P. (1997) Pathophysiology of the MELAS 3243 transition mutation. J. Biol. Chem., 272, Yarian, C., Townsend, H., Czestkowski, W., Sochacka, E., Malkiewicz, A.J., Guenther, R., Miskiewicz, A. and Agris, P.F. (2002) Accurate translation of the genetic code depends on trna modified nucleosides. J. Biol. Chem., 277, Suzuki, T., Wada, T., Saigo, K. and Watanabe, K. (2002) Taurine as a constituent of mitochondrial trnas: new insights into the functions of taurine and human mitochondrial diseases. EMBO J., 21, Janssen, G.M., Hensbergen, P.J., van Bussel, F.J., Balog, C.I., Maassen, J.A., Deelder, A.M. and Raap, A.K. (2007) The A3243G trnaleu(uur) mutation induces mitochondrial dysfunction and variable disease expression without dominant negative acting translational defects in complex IV subunits at UUR codons. Hum. Mol. Genet., 16, Boulet, L., Karpati, G. and Shoubridge, E.A. (1992) Distribution and threshold expression of the trna(lys) mutation in skeletal muscle of patients with myoclonic epilepsy and ragged-red fibers (MERRF). Am. J. Hum. Genet., 51, Enriquez, J.A., Chomyn, A. and Attardi, G. (1995) MtDNA mutation in MERRF syndrome causes defective aminoacylation of trna(lys) and premature translation termination. Nat. Genet., 10, Feuermann, M., Francisci, S., Rinaldi, T., De Luca, C., Rohou, H., Frontali, L. and Bolotin-Fukuhara, M. (2003) The yeast counterparts of human MELAS mutations cause mitochondrial dysfunction that can be rescued by overexpression of the mitochondrial translation factor EF-Tu. EMBO Rep., 4, Koppen, M. and Langer, T. (2007) Protein degradation within mitochondria: versatile activities of AAA proteases and other peptidases. Crit. Rev. Biochem. Mol. Biol., 42, Cleveland, D.W., Fischer, S.G., Kirschner, M.W. and Laemmli, U.K. (1977) Peptide mapping by limited proteolysis in sodium dodecyl sulfate and analysis by gel electrophoresis. J. Biol. Chem., 252, Bai, Y. and Attardi, G. (1998) The mtdna-encoded ND6 subunit of mitochondrial NADH dehydrogenase is essential for the assembly of the