Mitochondrial DNA transfer RNA gene sequence variations in patients with mitochondrial disorders

Transcription

1 Brain (2001), 124, Mitochondrial DNA transfer RNA gene sequence variations in patients with mitochondrial disorders Damien Sternberg, 1 Evi Chatzoglou, 2 Pascal Laforêt, 2 Guillemette Fayet, 2 Claude Jardel, 1 Patricia Blondy, 1 Michel Fardeau, 3 Serge Amselem, 4 Bruno Eymard 3 and Anne Lombès 2 1 Service de Biochimie B AP-HP, 2 INSERM U523, 3 Institut Correspondence to: A. Lombès, INSERM U523, Institut de de Myologie, Hôpital Pitié-Salpêtrière, Paris and 4 Service Myologie, Hôpital Pitié-Salpêtrière, Paris cedex 13, de Biochimie AP-HP and INSERM U468, Hôpital Henri France Mondor, Créteil, France a.lombes@myologie.chups.jussieu.fr Summary Many different pathogenic mutations in the mitochondrial (mt) transfer RNA (trna) genes have been reported for patients with mitochondrial encephalomyopathy. Although some of them are recurrent, most have only been described once and appear to be restricted to one patient or to one family. The incidence of mt trna gene alterations is not known, even though the frequency of some recurrent mutations has been analysed both in patients and in the general population. In this study, we describe the results of stepwise screening for sequence variations in the mt trna genes of 166 patients selected according to several criteria. Extensive sequence analysis of the trna genes was performed using denaturing gradient gel electrophoresis. A total of 31 patients (19%) were found to harbour significant levels of a pathogenic mutation, thus confirming the importance of mt trna mutations in human pathology. Forty-three different sequence variations were found, illustrating the great diversity of the mtdna sequence in humans. The functional assessment of all these sequence variations represented a difficult task; it was mostly based on indirect data, such as the phylogenetic conservation of modified nucleotides and the proportions of variant species in different tissues of the index case or in blood of maternal relatives. Direct demonstration of a correlation between the proportion of heteroplasmic sequence variations and the cytochrome c oxidase defect was performed at the single muscle-fibre level. Eleven heteroplasmic sequence variations were found, six of which are new mutations. One is a known Caucasian polymorphism but the other 10 are pathogenic. Two of them are the well-known pathogenic MELAS (mitochondrial myopathy, encephalopathy, lactic acidosis and stroke-like episodes) (A3243G) and MERRF (myoclonic epilepsy with ragged-red fibres) (A8344G) point mutations. They were found in 23 patients. The eight other mutations were restricted to one patient. The pathogenic nature of these mutations was demonstrated directly for five of them and hypothesized from indirect arguments for the other three. Thirty-two homoplasmic sequence variations were found. Twentynine were considered to be polymorphisms, even though 15 of these were found for the first time in our patients and two others had been reported previously as pathogenic. The pathogenic nature of three homoplasmic variants remains questionable. Keywords: mitochondrial DNA; mitochondrial encephalomyopathy; mitochondrial transfer RNA genes Abbreviations: COX cytochrome c oxidase; COX presence of a normal cytochrome c oxidase histochemical reaction; COX defective cytochrome c oxidase histochemical reaction; DGGE denaturing gradient gel electrophoresis; MELAS mitochondrial myopathy, encephalopathy, lactic acidosis and stroke-like episodes; MERRF myoclonic epilepsy with raggedred fibres; mtdna mitochondrial DNA; mt trna transfer RNA encoded by the mtdna; PCR polymerase chain reaction; RRF ragged-red fibres Introduction The diagnosis of human mitochondrial disorders relies upon a combination of different approaches, including clinical analysis, investigation of lactate metabolism in vivo, biochemical measurement of respiratory chain activities, morphological analyses and molecular genetic studies. Each Oxford University Press 2001 diagnostic element (lactate accumulation, morphological alterations of mitochondria, biochemical defects of the respiratory chain and identified genetic alterations) may be absent in some patients. Over the past 12 years, numerous genetic causes of human mitochondrial disorders have been

2 Mt trna genes in mitochondrial disorders 985 described (Zeviani et al., 1997; Simon and Johns, 1999; DiMauro and Andreu, 2000). Most of them have involved mitochondrial DNA (mtdna), affecting both transfer RNA (mt trna) and structural genes. Most of the pathogenic point mutations in the mtdna genes have been described only once and appear to be restricted to one patient or one family. The global incidence of these alterations is not known, even though the frequency of some recurrent mutations in the trna genes has been analysed both in patients and in the general population (Chinnery et al., 1997; Majamaa et al., 1998). The pathogenicity of an mtdna point mutation may be inferred from its absence in healthy subjects and by the phylogenetic conservation of the mutated nucleotide. The very high sequence diversity of human mtdna (Cann et al., 1987) complicates this evaluation. To overcome this problem, various methodological approaches have been discussed in the literature (Torroni and Wallace, 1994; Wallace, 1994), but a clear demonstration of the deleterious potential of new mtdna mutations has not always been provided. Most of the pathogenic mtdna mutations are heteroplasmic, i.e. mutated mtdna molecules coexist with a residual number of wild-type mtdna molecules. Assessment of heteroplasmy can be made by different techniques, which vary in sensitivity and the degree to which they are quantitative. When a heteroplasmic mutation is pathogenic, its proportional representation should correlate with the severity of the disease (Chinnery et al., 1998). Such a correlation can be analysed in family members as well as in the different tissues of the patients (if the mutation has a skewed tissue distribution). Some pathogenic mutations are homoplasmic, i.e. all of the mtdna molecules have the same mutation. This is most frequently the case in Leber s hereditary optic neuropathy (Wallace et al., 1988). Pathogenic homoplasmic mutations generally have incomplete penetrance, as shown by their presence in the healthy maternal relatives of the patients. Demonstration of the deleterious potential of such a mutation relies on the analysis of large groups of patients and of healthy controls. There should be a significant statistical relationship between the mutation and the disease, but also multiple independent occurrences of the mutation in different mtdna lineages. Cybrid technology has proved to be an essential tool for demonstrating the deleterious consequences of both heteroplasmic and homoplasmic mutations, as it allows the mtdna defects to be investigated in the absence of the influence of nuclear DNA-encoded factors (King and Attardi, 1989; Chomyn et al., 1991). However, it is highly demanding of both time and labour. Single muscle-fibre analysis represents an elegant and rapid alternative for the demonstration of the deleterious potential of original heteroplasmic point mutations when these mutations are associated with a cytochrome c oxidase (COX) defect in scattered muscle fibres (Moraes et al., 1993). We have looked for mt trna gene alterations in 166 patients who were referred to our laboratory for the diagnosis of a mitochondrial disorder. Extensive sequence analysis of Table 1 Classification of the patients screened for mt trna mutations according to the results of muscle investigations Muscle investigations Results of trna screening RRF / RC defect Screening 1 Screening 2 COX (n 166) (n 48) ND ND 23/0 0/0 44/19 12/2 0 or ND 59/6 31/2 0 19/1 2/0 0 0 or ND 21/1 3/0 The total numbers of patients screened are shown in the headings of the third and fourth columns. The number of patients with a pathogenic mutation follows the number of patients screened in each patient category. present; 0 absent; ND not determined; RC respiratory chain. the trna genes was obtained by the use of denaturing gradient gel electrophoresis (DGGE). The screening of patients was performed in two steps. We first analysed the trna Leu(UUR) and trna Lys genes in 166 patients. We then analysed the other 20 trna genes in 48 patients for whom the first step did not disclose the causative mutation. A total of 31 patients (19%) were found to harbour significant levels of a pathogenic mutation, thus showing the importance of mt trna mutations in human pathology. The complete set of mt trna genes was characterized in these 31 patients. Forty-three sequence variations from the human mtdna consensus sequence (Anderson et al., 1981) were found, thus confirming the high diversity of mtdna sequences in humans. We report the data that allowed the functional assessment of the mutations. They were either indirect (phylogenetic conservation of the mutated nucleotide, tissue distribution and familial transmission of the mutation) or direct (correlation between proportion of mutated species and biochemical defect in single muscle fibres). Some mutations illustrate pitfalls in interpretation. We discuss the frequency and localization of the pathogenic mt trna mutations and the relevance of criteria such as pedigree analysis and muscle morphology in making the decision to screen trna genes. Patients, material and methods Patients The mt trna gene sequence was determined in a total of 166 independent patients (32 children below the age of 15 years and 134 adults). Thirty-five of these patients have been described previously (Sternberg et al., 1998). Knowledge of muscle morphological characteristics was essential for our screening decisions. The distribution of the patients according to the presence of morphological and biochemical mitochondrial alterations is shown in Table 1. Screening for mutations in the trna Leu(UUR) or trna Lys gene was carried out on DNA from white blood cells or skin fibroblasts of 23 patients selected on the sole basis of their

3 986 D. Sternberg et al. clinical presentation. These patients had no muscle biopsy. Progressive encephalopathy was the predominant symptom in 19 of the patients. Progressive myoclonic epilepsy or stroke-like episodes, reminiscent of the myoclonic epilepsy with ragged-red fibres (MERRF) syndrome and the mitochondrial myopathy with encephalopathy, lactic acidosis and stroke-like episodes (MELAS) syndrome were observed in 11 cases. The disease affected multiple organs and/or was associated with elevated levels of lactate in the blood or CSF in 14 patients. A muscle biopsy was performed in 143 patients (Table 1). Abnormal accumulation of mitochondria [ragged-red fibres (RRF)] and/or scattered muscle fibres with a defective cytochrome c oxidase histochemical reaction (COX ) was found in 103 of these muscle biopsies (72%). A significant defect of one or several complexes of the muscle respiratory chain was found in 63 patients (44%), 19 of whom had no typical morphological alterations of muscle mitochondria. The search for an mt trna mutation was undertaken despite the absence of morphological or biochemical alterations of mitochondrial in 21 patients, being motivated either by the presence of a similar disease in maternal relatives (eight cases) or by the clinical profile of the patient (chronic ophthalmoplegia or progressive encephalopathy). For all 143 patients, genetic screening was carried out on total muscle DNA. Southern blot analysis was always performed first to exclude the existence of mtdna size rearrangement or depletion. Screening of the mt trna genes We used a two-step screening strategy. First, we scanned the trna Leu(UUR) and trna Lys genes in all patients. When this approach revealed a point mutation that might be pathogenic, we continued the search for sequence variations in the remaining 20 trna genes. In this way, we were able to evaluate the complete set of mt trna genes. When no significant mutation was found in the trna Leu(UUR) and trna Lys genes, we completed the scanning of the 22 mt trna genes in 48 of the patients (Table 1). Forty-three of these patients had typical morphological and/or biochemical alterations of their muscle mitochondria and five had no such alterations (Table 1). These five patients had predominant progressive encephalopathy without muscle symptoms. Mt trna genes were scanned for mutations and heteroplasmy by DGGE of fragments amplified by the polymerase chain reaction (PCR). This method has been described previously (Sternberg et al., 1998). The sequence of each trna gene was established either by direct sequencing or, for homoplasmic variants, by a second DGGE of the PCR fragments mixed with reference PCR fragments prior to electrophoresis (Sternberg et al., 1998). Evaluation of mt trna gene sequence variations Previous descriptions and evaluations of the sequence variations found in our patients were searched for in the MITOMAP database (Kogelnik et al., 1997) and in the journal Neuromuscular Disorders. The degree of phylogenetic conservation of substituted nucleotides or disrupted nucleotide pairs was evaluated by using the information on trna genes from numerous species (Sprinzl et al., 1998). High conservation was defined as the conservation of the nucleotide or of Watson Crick pairing in all mammalian species and in 80% of all the species analysed. Low conservation was defined as the presence of one variant in a mammalian species. The degree of conservation was considered medium when it did not fit the definition of low or high conservation. The haplotype of the mtdna with trna sequence variations was established using 15 polymorphic sites described as haplogroup markers for Caucasians (Torroni et al., 1996). Ten sites were screened by restriction enzyme digestion of amplified mtdna fragments (1715 DdeI, 4529 HaeII, 4577 NlaIII, 7025 AluI, 8249 HaeIII, 9052 HaeII, BamHI, BstNI, HinfI). Three sites located near or in mt trna genes were screened by restriction enzyme digestion or by DGGE (4332 AvaII, AluI, DdeI). Two sites located in mt trna genes were screened by DGGE only (12308 HinfI, MspI). Heteroplasmy was identified by the presence of multiple bands on the DGGE profile and the ratio between wild-type and mutant bands was used to evaluate the mutation load. A restriction-based method was used to evaluate precisely the level of heteroplasmy in the tissues from Patients 1 10, 12, 14 21, and 30 31, maternal relatives of patients 28 and 31 and single muscle fibres from muscle biopsies. The mutation-dependent restriction site was either naturally present in the mtdna or was created by means of mispairing PCR. Primers, amplification conditions, fragment lengths and restriction enzymes are listed in Table 2. Amplified fragments were radiolabelled during the last cycle of amplification in order to avoid radioactive heteroduplex molecules and were then digested. Restriction products were separated on polyacrylamide gels and counted with a β-imager. In one family (Patient 33 and her maternal relatives, T5814C variation), the restriction analysis was performed on nonradioactive fragments to distinguish between homoplasmy and heteroplasmy. DNA from single muscle fibres was obtained as described by Moraes and colleagues (Moraes et al., 1993). Schematically, muscle sections 16 µm thick were stained simultaneously for succinate dehydrogenase and COX activity. COX-negative (COX ) and COX-positive (COX ) fibre segments were dissected with a tungsten needle under an inverted microscope (Moslemi et al., 1998). Total DNA from the single-fibre segments was extracted by alkaline lysis in 200 mm KOH, 50 mm dithiothreitol at 60 C for 30 min followed by neutralization with 1 volume of 900 mm Tris HCl, ph 8.3, 200 mm HCl. The level of heteroplasmy was evaluated by means of the restriction enzyme-based method described above. The proportions of the mutant mtdna in COX fibres were compared with those in COX fibres with

4 Mt trna genes in mitochondrial disorders 987 Table 2 Technical conditions for the analysis of mutated mtdna species by restriction fragment polymorphism analysis Patients Mutation Primers used Fragment size Annealing Restriction for PCR (base pairs) temperature enzyme ( C) 1 21 A3243G Wt: ApaI (GC) M: T8355C (GC) Wt: DraI (mg8359t) M: T8362G (ma8358c) Wt: BstNI M: G4332A Wt: MaeI (GC) M: G4450A Wt: * AsnI (mta4455at) M: T5692C Wt: HpaI M: T5814C Wt: HphI (GC) M: A10006G Wt: ScaI M: (mc10009g) GC primers having a GC clamp at the 5 end; m modified nucleotide incorporated in primer; Wt wildtype species; M mutated species; *Five cycles at 53 C, five cycles at 54 C and 20 cycles at 55 C. the Mann Whitney non-parametric test with a two-tailed P value. Values of P below 0.05 were considered significant. Results Sequence variations found in trna genes Forty-three different point mutations were detected. Nine were found in several patients but 34 were restricted to one patient and 22 of these were found in our patients for the first time. Eleven appeared to be heteroplasmic and 32 homoplasmic according to the DGGE pattern (Fig. 1). The clinical symptoms, muscle morphology and molecular analyses of the patients with a pathogenic or questionable mutation are summarized in Table 3. Twenty-seven patients had a heteroplasmic point mutation in the trna Leu(UUR) or trna Lys gene (16% of 166 patients). Twenty-one had the MELAS mutation (A3243G, Patients 1 21), two had the MERRF mutation (A8344G, Patients 24 25). Four had original mutations (T3258C, A3280G, T8355C and T8362G in Patients 22, 23, 26 and 27, respectively). Patient 22 (mutation T3258C) was examined at age 32 years after an unexplained episode of severe hyperlactataemia associated with mild intolerance of exercise. Her mother had died in early adulthood from an unexplained cause. The patient had mild lipidosis and normal respiratory chain activities in muscle, and lipidosis with a combined respiratory chain defect in the liver. Patient 23 (mutation A3280G) complained of mild muscle weakness and severe intolerance of exercise since her mid-thirties. She had also had acute episodes of cardiac insufficiency that had regressed spontaneously. Her only daughter had similar muscle symptoms. The patient had, at age 47 years, numerous RRF Fig. 1 DGGE patterns of some trna Leu UUR sequence variations. Lane 1M shows a destabilizing sequence variation (G3277A) in a homoplasmic state; lanes 2F, 2L and 2M (Patient 22), 3B and 3M (Patient 23) and 4B and 4M (Patient 1) show heteroplasmic patterns obtained from different tissues (see key below); the three patients carried the T3258C, A3280G and A3243G mutations, respectively. These three mutations result in very similar migration of homoduplex and heteroduplex molecules. H homoduplex molecules; h heteroduplex molecules; C control (reference sequence); B, F, L, M PCR obtained from blood, skin fibroblasts, liver biopsy, muscle biopsy respectively. and COX fibres and a combined defect of the respiratory chain in muscle. Patient 26 (T8355C) suffered from acute respiratory failure at age 53 years. He had had progressive external ophthalmoplegia since early adulthood. None of his relatives were affected. Numerous RRF and COX fibres and a combined defect of the respiratory chain were observed in his muscle biopsy. Patient 27 (mutation T8362G) suffered

5 988 D. Sternberg et al. Table 3 Characteristics of 36 patients with pathogenic or dubious mutations in a mitochondrial trna gene Mutation* Patient Affected Clinical Lactate # Muscle RC Percentage Associated trna maternal symptoms histology** of mutated gene relatives DNA species variations A3243G (h) 1 (M, 42) M (PEO, SM), E(A), H, DM, D, R (B) RRF, COX I III 0.60 (M), 0.21 (B) 2 (M, 25) (o) E (S), DM, D (B,CSF) RRF, COX I III 0.64 (M), 0.47 (B) 3 (F, 48) (S) M (PEO, SM), DM, D (B) RRF, COX I III IV 0.53 (M), 0.24 (B) 4 (F, 40) M (SM), E (A), DM, R, L (B,CSF) RRF, COX I III 0.54 (M), 0.36 (B) 5 (M, 13) M (PEO, EI) 0 (B) Type I, RRF II III 0.59 (M), 0.20 (B) 6 (F, 27) (M, S) E (S), D (B) Type I, COX Normal 0.80 (M), 0.33 (B) 7 (F, 40) (M, C) E (S), H, N, DM, D, R RRF, COX IV 0.71 (M) 8 (M, 29) (S) E (S) (B,CSF) RRF, COX 0.44 (M), 0.20 (B) 9 (M, 72) (S, M, o) E, N, D, R RRF, COX III IV 0.52 (M) 10 (M, 62) E (A, S), DM, D (CSF) RRF, COX I III 0.47 (M), 0.26 (B) 11 (M, 23) (S) M (PEO, SM), H RRF, COX 12 (M, 9) (o) M (EI), E (S) (B,CSF) RRF, COX I III 0.83 (M), 0.72 (B) 13 (M,?) (S, M, o) H, DM, D, R RRF, COX I III IV 14 (M, 16) (S) M (EI), H, E(S), D (B) RRF, COX III 0.69 (M), 0.39 (B) T5774C 15 (F, 56) E (A), D, R RRF, COX I 0.57 (M) C15946T 16 (M, 60) M (PEO, SM), H, DM, D, R 0 (B) RRF, COX I IV 0.53 (M), 0.33 (B) 17 (M, 3) (S, M, o) E (A, ME), R (B,CSF) Lipidosis III IV 0.23 (M) 18 (M, 7) (S, M, o) E (S), H (B,CSF) RRF, COX 0.62 (F) 19 (F, 28) (S, M, C) E (S), M (PEO, EI) 0 (B) RRF, COX III IV 0.35 (M) 20 (M, 29) M (PEO, EI) RRF, COX 0.59 (M) 21 (F, 52) (M, o) M (EI), E(A), DM, D (B) RRF, COX I IV 0.52 (M), 0.2 (B) T3258C (h) 22 (M, 32) (M) Lactic acidosis, M (EI) (B) Lipidosis Normal see Fig. 1 A3280G (h) 23 (F, 47) (C) M (SM), H RRF, COX I see Fig. 1 T4386C A8344G (h) 24 (F, 38) (S, o) E (A, ME), N, D RRF, COX I III 25 (M, 18) (S, M, o) M (SM), E (A, ME), N, D (B) RRF, COX T8355C (h) 26 (M, 64) M (PEO, SM, RM) RRF, COX I III 0.62 (M), 0.10 (B) T8362G (h) 27 (F, 53) (S) M (SM) RRF, COX III 0.67 (M) A5600G G4332A (h) 28 (M, 52) E (S), D RRF, COX IV 0.81 (M), 0 (B), 0 (F) T4336C (h) 29 (F, 46) E RRF I IV G4450A (h) 30 (M, 60) Splenic lymphoma 0 (B) RRF, COX Normal 0.44 (M), 0.75 (B) T10410C, T10457C,C15913T T5692C (h) 31 (M, 56) M (PEO, SM, EI), E (A), H, N 0 (B) RRF, COX IV 0.46 (M), 0 (B), 0 (F) G5698A (h) 32 (F, 33) M (PEO) RRF, COX Normal T4418C, A14693G T5814C (H) 33 (F, 55) M (EI) RRF, COX Normal A10006G (H) 34 (M, 48) (S) E, M (PEO), N, D, G (B) Lipidosis, COX Normal C12246A G10014A (H) 35 (M, 9) M (EI) (B) RRF, COX G1664A A8302T (H) 36 (M, 1) (S) E, L (B) Lipidosis IV *The mutation results in a heteroplasmic (h) or homoplasmic (H) DGGE pattern. Gender and age at diagnosis are indicated in parentheses: M male; F female. No affected relatives or no information available; S affected siblings; M affected mother; C affected children; o other affected maternal relatives. D deafness; DM diabetes mellitus; E encephalopathy (A ataxia; ME myoclonic epilepsy; S stroke-like episodes); G gut involvement; H heart involvement (myocardiopathy or conduction defects); L liver involvement; M myopathy (EI exercise intolerance; PEO oculomotor myopathy; RM respiratory myopathy; SM skeletal myopathy); N neuropathy; R renal symptom. # High lactate level; 0 normal lactate level; not determined; B blood; CSF cerebrospinal fluid. **RRF a significant number of ragged-red fibres was found on muscle sections after staining with Gomori s trichrome; RRF a significant number of fibres containing mitochondrial aggregates were demonstrated by modified Gomori s trichrome and succinate dehydrogenase staining but no true ragged-red fibres were present; COX a significant number of COX-negative fibres were found after COX staining; type I significant increase in the proportion of type I muscle fibres. RC respiratory chain. Respiratory chain complexes that had a significantly decreased activity as assessed by spectrophotometric assays are indicated; not determined. (M) muscle; (B) blood, (F) fibroblasts. trna gene sequence variations other than the very common Caucasian polymorphisms T10034C, T10463C, A12308G, C15904T, A15924G and G15928A are shown.

6 Mt trna genes in mitochondrial disorders 989 from isolated limb weakness. Numerous RRF and COX fibres were observed in her muscle biopsy. Her sister had similar symptoms. Five patients had a heteroplasmic point mutation in a trna gene other than trna Leu(UUR) or trna Lys (10% of 48 patients). The mutations were located in trna Gln (G4332A, T4336C), trna Met (G4450A) and trna Asn (T5692C, G5698A). Two of these were new mutations (G4332A and G5698A). Patient 28 (mutation G4332A) has been described more extensively elsewhere (Bataillard et al., 2001). He had an acute stroke-like episode at age 47 years. He had been deaf since his twenties but was otherwise healthy. None of his relatives had similar symptoms. Patient 32 (mutation G5698A) had mild external ophthalmoplegia. None of her relatives had similar symptoms. Thirty-two homoplasmic sequence variations were found in 42 patients. The number of variants in each trna gene was variable: five in the trna Thr gene; four in each of trna Lys and trna Gly ; three in trna Arg ; two in each of trna Ala, trna Cys, trna Asp and trna Ser (AGY) ; and one in each of trna Val, trna Leu(UUR), trna Gln, trna Ser (UCN), trna His, trna Leu (CUN), trna Met and trna Glu. Five trna genes displayed no sequence variation (trna Phe, trna Trp, trna Ile, trna Tyr and trna Pro ). One (trna Asn ) showed two heteroplasmic mutations but no homoplasmic variation. The heteroplasmic and homoplasmic sequence variations were equally distributed in regions coding for the different domains of the trnas (aminoacyl arm, pseudocytosine arm, anticodon arm, dihydropyridine arm, supplementary arm and intermediate nucleotides) (Fig. 2). Evaluation of sequence variations in the mt trna genes Comparison with previous reports Two heteroplasmic mutations had been demonstrated previously to be pathogenic (the MELAS mutation in the trna Leu(UUR) gene and the MERRF mutation in the trna Lys gene). One heteroplasmic mutation (T4336C) had been reported to be a common Caucasian polymorphism (Leroy and Norby, 1994). Its heteroplasmy therefore appeared to be coincidental. We have published the G4450A mutation and demonstrated its pathogenic character (Lombès et al., 1998). The seven remaining heteroplasmic mutations had either not been described earlier or had been reported once (T5692C) but without direct assessment of their deleterious potential (Seibel, 1994). The situation was similar for the 32 homoplasmic-looking sequence variations. Fifteen had been reported as polymorphisms, either frequent (T10034C, T10463C, A12308G, C15904T, A15924G and G15928A) or relatively rare (T4386C, C5633T, C7476T, G7521A, A8308G, T10410C, A12171G, C12246A and C15946T). Fifteen sequence variations (A8302T, G1664A, G3277A, T4418C, A5600G, T5774C, T7581C, T8337C, A8348G, G10014A, A10042G, T10457C, A12234G, A14693G and C15913T) were found in our patients for the first time but we had already reported eight of them elsewhere (Sternberg et al., 1998). Two had been reported as pathogenic: A10006G in the homoplasmic state (Lauber et al., 1991) and T5814C in the heteroplasmic state (Manfredi et al., 1996; Santorelli et al., 1997). Arguments from the phylogenetic conservation of mutated nucleotides A mutation has a higher probability of being clinically relevant when conserved nucleotides are modified. The conservation of each nucleotide position is indicated in Fig. 2. Heteroplasmic variations modified 7, 3 and 1 nucleotides with high, medium and low conservation, respectively, versus 5, 7 and 20 for homoplasmic-looking sequence variations (Fig. 2). The distribution of mutated nucleotides in high-/ medium- or low-conservation classes differed significantly between heteroplasmic- and homoplasmic-looking sequence variations (two-sided P 0.01, Fisher s exact test). Haplotyping the mtdna We established the mtdna haplotype for most sequence variations (Fig. 2). The occurrence of a particular mutation on different mtdna haplotypes has been considered an indirect argument for its pathogenicity (Torroni and Wallace, 1994; Wallace, 1994). Even though this analysis is not immediately relevant for a mutation that is restricted to one patient, it will help in evaluating them in the future. Familial and tissue distribution patterns of the mutations Tissue distribution of the heteroplasmic mutations was analysed in Patients 22 (T3258C; Fig. 1), 23 (A3280G; Fig. 1), 26 (T8355C; Table 3), 28 (G4432A; Table 3), 30 (G4450A; Table 3) and 31 (T5692C; Table 3) and in 12 patients with the A3243G MELAS mutation (Table 3). All mutations had a skewed tissue distribution. It was extreme in the patients with the G4432A, T5692C and T8355C mutations, who had no trace of the mutation in tissues other than muscle. Blood samples were obtained from the mothers of Patients 28 and 31, who had a heteroplasmic mutation. In both cases, the mutation (G4332A and T5692C, respectively) was absent. Blood samples from three healthy maternal relatives of Patient 33, who had the homoplasmic T5814C mutation, had the mutation in a homoplasmic state, as confirmed by DGGE and RFLP (restriction fragment length polymorphism) analysis (Fig. 3).

7 990 D. Sternberg et al. Fig. 2 trna positions at which a nucleotide change occurs in one or several patients, shown on a canonical cloverleaf trna model. Circles indicate the positions of nucleotides involved in heteroplasmic point mutations (in italics in the accompanying table); squares indicate the positions of nucleotides involved in homoplasmic mutations. The number of occurrences was calculated from 75 patients with the exception of the numbers followed by an asterisk, which were calculated from 166 patients. l, m, h low, middle and high phylogenetic conservation, respectively, as defined in the text (Patients, material and methods); outgroup haplotype not classifiable in one of the Caucasian haplogroups as defined by Torroni and colleagues (Torroni et al., 1996); nd not determined. Conclusion from single muscle-fibre analyses We had appropriate samples to analyse the correlation between the proportion of the mutation (percentage of mutated mtdna species) and residual COX activity in single muscle fibres for six mutations (Table 4). This approach clearly demonstrated the deleterious effect of five heteroplasmic mutations: T8355C and T8362G in the trna Lys gene, G4332A in the trna Gln gene, G4450A in the trna Met gene and T5692C in the trna Asn gene. In a previous study in cybrids, mutation G4450A had been demonstrated to be pathogenic (Lombès et al., 1998) but the four remaining mutations had never been studied. The A10006G mutation looked homoplasmic after DGGE analysis. Analysis of its proportion in single muscle-fibre segments demonstrated its heteroplasmic state in some fibres. However, there was no significant relationship between the COX defect and a high percentage of mutated DNA species (Table 4). Discussion Mutations of the mt trna genes are recurrent causes of mitochondrial diseases Over the past 7 years, samples from 578 patients with a suspected respiratory chain disorder have been referred to our laboratory. Their clinical symptoms were highly heterogeneous, as were the investigations performed in the referring centres. We believe that this group of patients is representative of the different ways by which a suspicion of mitochondrial disease may be reached, because of the number of patients and of their diverse origins. A major deletion or significant depletion was shown by Southern blot analysis in 95 (16%) of these 578 patients. From the remaining 483 patients we selected 166 (34%) for the mt trna gene screening and identified a pathogenic mutation in 31 of them (19% of the selected patients, 5% of all patients referred). This gives a minimum estimate of the relative frequency of

8 Mt trna genes in mitochondrial disorders 991 Table 4 Comparison of the percentages of mutated mtdna species found in COX-negative or in COX-positive muscle fibre segments in biopsies of patients and 33 Patient Mutation COX fibre segments* COX fibre segments* P 26 T8355C 82.9 [ ] (n 11) 44.9 [ ] (n 15) P T8362G 79.3 [ ] (n 12) 49.3 [ ] (n 13) P G4332A 99.3 [ ] (n 13) 70.5 [ ] (n 11) P G4450A 74.4 [ ] (n 11) 13.1 [ ] (n 11) P T5692C 88.2 [ ] (n 11) 28.3 [ ] (n 12) P A10006G 86.6 [ ] (n 9) 87.9 [ ] (n 9) P 0.40 n number of fibre segments analysed;*average proportion of mutated species [95% confidence interval]; P 0.05 (two-tailed nonparametric Mann Whitney test) was considered significant). Fig. 3 Familial transmission of the T5814C sequence variation in the trna Cys gene. Electrophoretic pattern of fragments generated by HphI restriction. Mutation T5814C abolishes the restriction site at nucleotide 5814, whereas the restriction site at nucleotide 5892 is maintained. Wild-type mtdna generates three fragments, of 73, 78 and 66 bp (including a 40 GC clamp), whereas mutated mtdna generates only two fragments, of 144 and 73 bp. The genealogical tree indicates the origin of the blood DNAs analysed in the lower part of the figure. The black circle indicates Patient 33. M.W. molecular weight; U.C. uncut. pathogenic mt trna point mutations as causal genetic defects in the population of patients probably presenting with a mitochondrial disease. The signs associated with a mutation in a trna gene are not specific There are no precise epidemiological data on mt trna gene point mutations and there is no general consensus for a diagnostic protocol. We used a two-step strategy for the genetic screening because it allowed us to analyse a greater number of patients and to avoid the application of very restrictive criteria for their inclusion. Clinical symptoms alone were clearly insufficient to orient the screening, as demonstrated by finding only one mutation in the 44 patients for whom investigations were decided despite the absence of significant mitochondrial alterations in muscle (Table 1). Interestingly, 11 of these patients had progressive encephalopathy resembling the MERRF or MELAS syndromes. Maternal transmission of the disease would obviously be a very important selection parameter. However, none of our patients with a pathogenic mutation had a clearly maternal pedigree, although 19 out of 31 of these patients had affected maternal relatives (Table 3). In contrast, the results of muscle biopsy proved to be a very useful selection parameter for the screening of mt trna mutations. A pathogenic mutation was found in 29 of the 103 patients (28%) who had significant mitochondrial morphological alterations (RRF and/or COX fibres) and in 20 of the 63 patients (32%) in whom a significant respiratory chain defect was observed (Table 1). Moreover, typical muscle morphological alterations or a defective respiratory chain were noted in the muscle fibres of 94% (30 out of 32) and 85% (23 out of 27) of the patients with a pathogenic mutation in an mt trna gene, respectively. The two patients with no typical morphological alterations did, however, have muscle lipidosis. The first (Patient 17) was 3 years old and had no muscle symptoms (Table 3). The second (Patient 22) had severe lactic acidosis and mild intolerance of exercise. Both young age and a mildness of muscle symptoms are therefore compatible with an mt trna mutation with isolated lipidosis in muscle. The presence of RRF and/or COX muscle fibres is, however, clearly non-specific as 39 patients in our study had typical morphological alterations in their muscle but no relevant mutation in their 22 mt trna genes (Table 1). Similar morphological alterations have been associated with sequence alterations in the mtdna structural COX genes (Comi et al., 1998; Hanna et al., 1998; DiMauro and Andreu, 2000). Mutations in structural genes other than COX genes were associated with RRF and hyperpositive COX muscle fibres (Andreu et al., 1999; DiMauro and Andreu, 2000; Musumeci et al., 2000). Disturbances in mtdna maintenance of various origins (Chariot and Gherardi, 1991; Hirano et al., 1994; Nishino et al., 1999) result in focal mtdna alterations that may not be revealed by Southern blot analysis of whole muscle DNA.

9 992 D. Sternberg et al. The tissue distribution of heteroplasmic mutations may be skewed and lead to falsenegative diagnosis The choice of the best tissue to be analysed is an important issue in screening for mtdna mutations, especially because the tissue distribution of heteroplasmic mutations was skewed in most patients. This was even more evident in patients without affected maternal relatives (Table 3). Analysis of mtdna in blood samples from patients is very convenient and has proved to be useful (Hammans et al., 1991) but would have resulted in false negative results in Patients 28 and 31. Such a misdiagnosis may have happened in the 23 patients in whom only blood DNA was investigated, but this is unlikely because these patients had a generalized disorder. The relative contributions of the different trna genes to mitochondrial pathology defines a strategy for screening The usual gap between the number of requests for genetic testing and the means to meet them makes selection criteria a common problem in diagnostic centres. The first step in trna gene screening revealed a pathogenic mutation in the trna Leu(UUR) and trna Lys genes in 27 (16%) of the 166 patients (Table 1), whereas the second step revealed a pathogenic mutation in only four (8%) of the 48 patients analysed. The proportion of abnormal trna genes among the total number of trna genes analysed was 8% in the first screening step and 0.4% in the second. These results confirm the previous observation that the trna Leu(UUR) and trna Lys genes are hotspots for mutation (Moraes et al., 1993; Schapira, 1997). The A3243G MELAS mutation is responsible for two-thirds of the heteroplasmic point mutations in trna genes, which confirms its particular status among trna gene mutations (Hammans et al., 1995; Majamaa et al., 1998). It was found in nine (35%) of the 26 patients with progressive encephalopathy and stroke-likes episode(s) resembling the MELAS syndrome. The MERRF mutation is less frequent but is still recurrent. It was found in two (40%) of the five patients who had progressive encephalopathy with myoclonic epilepsy that resembled the MERRF syndrome. In addition to these two recurrent mutations, four original pathogenic mutations were identified in the trna Leu(UUR) and the trna Lys genes, emphasizing the importance of scanning these two genes before all other trna genes. The decrease in screening efficiency between the first and second steps may have been due entirely to the hotspot character of these two trna genes, but it may also indicate that the criteria that have been used to select patients for the second step are not totally efficient (Hammans et al., 1991; Hirano et al., 1994; Comi et al., 1998; Nishino et al., 1999). Functional assessment of trna sequence variations may be difficult Most of the trna mutations (34 out of 43) were restricted to one patient and 21 were entirely new. This situation emphasizes not only the high sequence variability that is found in human mtdna but also the need for the correct evaluation of these sequence variations. None of the indirect evidence for the clinical relevance of a given mutation is absolute. The heteroplasmic state of a mutation is often considered to be good evidence. However, mutation T4336C, which is a common Caucasian polymorphism, was found to be heteroplasmic in one muscle sample (Leroy and Norby, 1994). Conversely, some pathogenic mutations may be nearly homoplasmic in affected tissues and result in a homoplasmic pattern in DGGE. The high degree of phylogenetic conservation of the mutated nucleotide is a relative criterion, as the A8344G (MERRF) mutation modifies a mildly conserved nucleotide while several well-known polymorphisms modify highly conserved nucleotides (Kogelnik et al., 1997). However, there is no example of a demonstrated pathogenic mutation involving a nucleotide with a low degree of phylogenetic conservation. Mutation A10006G has been suggested, but not demonstrated, to be pathogenic (Lauber et al., 1991). It involves a nucleotide with a low degree of conservation and we have demonstrated here its polymorphic nature by means of single muscle-fibre analysis. The clinical relevance of a mutation has also been assessed on the basis of its occurrence in diverse mtdna haplotypes (Brown et al., 1995). Because most mutations are restricted to one patient, the mtdna haplotype is useless in the first analysis even though it will help in further assessment in the case of recurrent mutations. Despite these difficulties, we were able to assess the clinical relevance of 40 of the 43 mutations found in our series of patients. Among the 11 heteroplasmic mutations, seven have been directly demonstrated to be pathogenic, either in previous reports [mutations A3243G (MELAS), A8344G (MERRF) and G4450A)] or in the present work (mutations G4332A, G4450A, T5692C, T8355C and T8362G). One heteroplasmic mutation proved to be a wellknown polymorphism (T4336C). The three remaining heteroplasmic mutations (T3258C, A3280G and G5698A) could not be analysed fully because of lack of appropriate samples. They were considered to be pathogenic from indirect criteria, such as their heteroplasmic state, their absence in already published series of controls and the high degree of conservation of the mutated nucleotide. Twenty-nine of the 32 homoplasmic mutations were considered to be polymorphisms. Six of them had been characterized previously as frequent polymorphisms (T10034C, T10463C, A12308G, C15904T, A15924G and G15928A). Ten others were found in association with a heteroplasmic mutation (T4386C, T4418C, A5600G, T5774C, A8308G, T10410C, T10457C, A14693G, C15913T and C15946T). Thirteen others involved a nucleotide with a low degree of phylogenetic conservation (Fig. 2), among which mutation A10006G was demonstrated to be nonpathogenic by single muscle-fibre analysis. Three sequence variations remain of dubious significance. The mutation T5814C exemplifies the difficulty of the

10 Mt trna genes in mitochondrial disorders 993 functional assessment of sequence variations. It has been reported in the heteroplasmic state and analysis of the mutant load in single muscle fibres has demonstrated the pathogenicity of the mutation (Manfredi et al., 1996; Santorelli et al., 1997). However, Patient 33 had the mutation in a homoplasmic state in muscle and her only symptom was isolated intolerance to exercise at age 55 years. It was not possible for us to analyse the mutation load in single muscle fibres from her, but we obtained blood samples from her healthy maternal relatives (brother, son and daughter) and found the mutation in the homoplasmic state. This contradiction between the data reported in the literature and ours suggests that some mt trna sequence variations have different severities in different individuals. This possibility is raised to explain the wide range of clinical symptoms associated, for example, with the MELAS A3243G mutation (Chinnery et al., 1997). It underlies our effort to characterize the mtdna haplotype and the 22 trna gene sequences fully in all the patients with a heteroplasmic mutation. Two other mutations (A8302T and G10014A) could not be characterized further. They were restricted to a particular patient and involved a nucleotide with a medium degree of conservation in one patient and one with a high degree of conservation in the other. These mutations appeared homoplasmic in total muscle DNA but may be heteroplasmic at the level of single muscle fibres and could cause the mosaic pattern of the COX defect. Samples from other tissues, from maternal relatives or repeated muscle biopsies are needed to complete the functional evaluation of these mutations. In conclusion, mt trna gene alterations are an important cause of mitochondrial disorders. The trna Leu(UUR) and trna Lys genes deserve priority in trna gene screening as their scanning may disclose frequent mutations as well as new ones. The screening of the 20 other trna genes, although less efficient, regularly reveals new mutations. However, the amount of work needed for the extensive screening of the 22 mt trna genes has precluded epidemiological studies to date. Technical progress will certainly reduce the cost of genetic screening and the time required for it. It will allow the identification of numerous original sequence variations, the functional assessment of which will represent an important challenge. Acknowledgements We would like to thank the physicians who supplied patient samples, especially Dr Marc Bataillard, Dr Jean-Marie Cuisset, Pr Philippe Gajdos, Dr Thierry Maisonobe, Dr Hélène Ogier, Dr Michel Parent and Dr François Ziegler, and Dr Gillian Butler-Brown for careful reading of the manuscript and useful comments. This work was supported by grants from FRM (Fondation pour la Recherche Médicale) and from AFM (Association Française contre les Myopathies). E.C. was the recipient of a postdoctoral fellowship from AFM. References Anderson S, Bankier AT, Barrell BG, de Bruijn MH, Coulson AR, Drouin J, et al. Sequence and organization of the human mitochondrial genome. Nature 1981; 290: Andreu AL, Hanna MG, Reichmann H, Bruno C, Penn AS, Tanji K, et al. Exercise intolerance due to mutations in the cytochrome b gene of mitochondrial DNA. N Engl J Med 1999; 341: Bataillard M, Chatzoglou E, Rumbach L, Sternberg D, Tournade A, Laforêt P, et al. Atypical MELAS syndrome associated with a new mitochondrial trna glutamine point mutation. Neurology. 2001; 56: Brown MD, Torroni A, Reckord CL, Wallace DC. Phylogenetic analysis of Leber s hereditary optic neuropathy mitochondrial DNAs indicates multiple independent occurrences of the common mutations. Hum Mutat 1995; 6: Cann RL, Stoneking M, Wilson AC. Mitochondrial DNA and human evolution. Nature 1987; 325: Chariot P, Gherardi R. Partial cytochrome c oxidase deficiency and cytoplasmic bodies in patients with zidovudine myopathy. Neuromuscul Disord 1991; 1: Chinnery PF, Howell N, Lightowlers RN, Turnbull DM. Molecular pathology of MELAS and MERRF. The relationship between mutation load and clinical phenotypes. Brain 1997; 120: Chinnery PF, Howell N, Lightowlers RN, Turnbull DM. MELAS and MERRF. The relationship between maternal mutation load and the frequency of clinically affected offspring. Brain 1998; 121: Chomyn A, Meola G, Bresolin N, Lai ST, Scarlato G, Attardi G. In vitro genetic transfer of protein synthesis and respiration defects to mitochondrial DNA-less cells with myopathy-patient mitochondria. Mol Cell Biol 1991; 11: Comi GP, Bordoni A, Salani S, Franceschina L, Sciacco M, Prelle A, et al. Cytochrome c oxidase subunit I microdeletion in a patient with motor neuron disease. Ann Neurol 1998; 43: DiMauro S, Andreu AL. Mutations in mtdna: are we scraping the bottom of the barrel? [Review]. Brain Pathol 2000; 10: Hammans SR, Sweeney MG, Brockington M, Morgan-Hughes JA, Harding AE. Mitochondrial encephalopathies: molecular genetic diagnosis from blood samples. Lancet 1991; 337: Hammans SR, Sweeney MG, Hanna MG, Brockington M, Morgan- Hughes JA, Harding AE. The mitochondrial DNA transfer RNALeu(UUR) A G(3243) mutation. A clinical and genetic study. Brain 1995; 118: Hanna MG, Nelson IP, Rahman S, Lane RJ, Land J, Heales S, et al. Cytochrome c oxidase deficiency associated with the first stopcodon point mutation in human mtdna. Am J Hum Genet 1998; 63: Hirano M, Silvestri G, Blake DM, Lombès A, Minetti C, Bonilla E, et al. Mitochondrial neurogastrointestinal encephalomyopathy (MNGIE): clinical, biochemical, and genetic features of an autosomal recessive mitochondrial disorder. Neurology 1994; 44:

11 994 D. Sternberg et al. King MP, Attardi G. Human cells lacking mtdna: repopulation with exogenous mitochondria by complementation. Science 1989; 246: Kogelnik AM, Lott MT, Brown MD, Navathe SB, Wallace DC. MITOMAP: an update on the status of the human mitochondrial genome database. Nucleic Acids Res 1997; 25: Lauber J, Marsac C, Kadenbach B, Seibel P. Mutations in mitochondrial trna genes: a frequent cause of neuromuscular diseases. Nucleic Acids Res 1991; 19: Leroy D, Norby S. A new human mtdna polymorphism: trna(gln)/4336 (T C). Clin Genet 1994; 45: Lombès A, Bories D, Girodon E, Frachon P, Ngo MM, Breton- Gorius J, et al. The first pathogenic mitochondrial methionine trna point mutation is discovered in splenic lymphoma. Hum Mutat 1998; Suppl 1: S Majamaa K, Moilanen JS, Uimonen S, Remes AM, Salmela PI, Kärppä M, et al. Epidemiology of A3243G, the mutation for mitochondrial encephalomyopathy, lactic acidosis, and strokelike episodes: prevalence of the mutation in an adult population. Am J Hum Genet 1998; 63: Manfredi G, Schon EA, Bonilla E, Moraes CT, Shanske S, DiMauro S. Identification of a mutation in the mitochondrial trna cys gene associated with mitochondrial encephalopathy. Hum Mutat 1996; 7: Moraes CT, Ciacci F, Bonilla E, Jansen C, Hirano M, Rao N, et al. Two novel pathogenic mitochondrial DNA mutations affecting organelle number and protein synthesis. J Clin Invest 1993; 92: Moslemi AR, Tulinius M, Holme E, Oldfors A. Threshold expression of the trna(lys) A8344G mutation in single muscle fibres. Neuromuscul Disord 1998; 8: Musumeci O, Andreu AL, Shanske S, Bresolin N, Comi GP, Rothstein R, et al. Intragenic inversion of mtdna: a new type of pathogenic mutation in a patient with mitochondrial myopathy. Am J Hum Genet 2000; 66: Nishino I, Spinazzola A, Hirano M. Thymidine phosphorylase gene mutations in MNGIE, a human mitochondrial disorder. Science 1999; 283: Santorelli FM, Siciliano G, Casali C, Basirico MG, Carrozzo R, Calvosa F, et al. Mitochondrial trna cis gene mutation (A5814G): a second family with mitochondrial encephalopathy. Neuromuscul Disord 1997; 7: Schapira AH. Mitochondrial disorders. [Review]. Curr Opin Neurol 1997; 10: Seibel P, Lauber J, Klopstock T, Marsac C, Kadenbach B, Reichmann H. Chronic progressive external ophthalmoplegia is associated with a novel mutation in the mitochondrial trna(asn) gene. Biochem Biophys Res Commun 1994; 204: Simon DK, Johns DR. Mitochondrial disorders: clinical and genetic features. [Review]. Annu Rev Med 1999; 50: Sprinzl M, Horn C, Brown M, Ioudovitch A, Steinberg S. Compilation of trna sequences and sequences of trna genes. Nucleic Acids Res 1998; 26: Sternberg D, Danan C, Lombès A, Laforêt P, Girodon E, Goossens M, et al. Exhaustive scanning approach to screen all the mitochondrial trna genes for mutations and its application to the investigation of 35 independent patients with mitochondrial disorders. Hum Mol Genet 1998; 7: Torroni A, Wallace DC. Mitochondrial DNA variation in human populations and implications for detection of mitochondrial DNA mutations of pathological significance. [Review]. J Bioenerg Biomembr 1994; 26: Torroni A, Huoponen K, Francalacci P, Petrozzi M, Morelli L, Scozzari R, et al. Classification of European mtdnas from an analysis of three European populations. Genetics 1996; 144: Wallace DC. Mitochondrial DNA sequence variation in human evolution and disease. [Review]. Proc Natl Acad Sci USA 1994; 91: Wallace DC, Singh G, Lott MT, Hodge JA, Schurr TG, Lezza AM, et al. Mitochondrial DNA mutation associated with Leber s hereditary optic neuropathy. Science 1988; 242: Zeviani M, Fernandez-Silva P, Tiranti V. Disorders of mitochondria and related metabolism. Curr Opin Neurol 1997; 10: Received June 29, Revised November 29, Accepted January 5, 2001