CPEO (chronic progressive external ophthalmoplegia) is a common mitochondrial disease phenotype in adults which is due to mtDNA (mitochondrial DNA) point mutations in a subset of patients. Attributing pathogenicity to novel tRNA mtDNA mutations still poses a challenge, particularly when several mtDNA sequence variants are present. In the present study we report a CPEO patient for whom sequencing of the mitochondrial genome revealed three novel tRNA mtDNA mutations: G5835A, del4315A, T1658C in tRNATyr, tRNAIle and tRNAVal genes. In skeletal muscle, the tRNAVal and tRNAIle mutations were homoplasmic, whereas the tRNATyr mutation was heteroplasmic. To address the pathogenic relevance, we performed two types of functional tests: (i) single skeletal muscle fibre analysis comparing G5835A mutation loads and biochemical phenotypes of corresponding fibres, and (ii) Northern-blot analyses of mitochondrial tRNATyr, tRNAIle and tRNAVal. We demonstrated that both the G5835A tRNATyr and del4315A tRNAIle mutation have serious functional consequences. Single-fibre analyses displayed a high threshold of the tRNATyr mutation load for biochemical phenotypic expression at the single-cell level, indicating a rather mild pathogenic effect. In contrast, skeletal muscle tissue showed a severe decrease in respiratory-chain activities, a reduced overall COX (cytochrome c oxidase) staining intensity and abundant COX-negative fibres. Northern-blot analyses showed a dramatic reduction of tRNATyr and tRNAIle levels in muscle, with impaired charging of tRNAIle, whereas tRNAVal levels were only slightly decreased, with amino-acylation unaffected. Our findings suggest that the heteroplasmic tRNATyr and homoplasmic tRNAIle mutation act together, resulting in a concerted effect on the biochemical and histological phenotype. Thus homoplasmic mutations may influence the functional consequences of pathogenic heteroplasmic mtDNA mutations.
CPEO (chronic progressive external ophthalmoplegia) is one of the most common mitochondrial disease phenotypes in adults. In approx. 50% of CPEO patients, mtDNA (mitochondrial DNA) analysis of skeletal muscle tissue reveals single large-scale mtDNA deletions which usually arise sporadically . However, CPEO may also be due to maternally inherited or sporadic mtDNA point mutations, or to mutations in nuclear genes resulting in autosomal disorders of intergenomic communication [2,3]. The identification of the specific pathogenic disease-defining mtDNA point mutation often remains a challenge, particularly when several mtDNA sequence variants are present. Owing to the high variability of the mitochondrial genome some earlier established criteria for pathogenicity of mtDNA mutations, for example, the absence in a control population, might fail in cases of previously undescribed private mtDNA mutations. Moreover, even heteroplasmy is no longer a clear criterion for pathogenicity, since many heteroplasmic polymorphisms have been reported , and various homoplasmic pathogenic tRNA mtDNA point mutations have been identified to date [5–9].
A 22-year-old male patient of Turkish ethnic origin (Patient A) presented with a medical history of slowly progressive upper eye lid ptosis, which was first observed at the age of 15 years. Later, the patient developed incomplete external ophthalmoplegia, as well as exercise intolerance. His previous past medical history was uneventful. All family members lived in a secluded village in the province of Sakarya (Adapazari), Eastern Turkey, and were reported to be free of neuromuscular symptoms. The patient underwent a neurological investigation, open skeletal muscle biopsy, including histochemical, biochemical and molecular genetic investigations, as well as genetic analyses of leucocyte and buccal mucosa DNA. Blood for genetic analysis was obtained from all accessible family members (mother, two sisters and one brother).
To compare functional features of mtDNA mutations in Patient A, we obtained corresponding histological and biochemical data from Patient B (female, aged 37 years at time of open skeletal muscle biopsy), presenting with MELAS (mitochondrial encephalomyopathy with lactic acidosis and stroke-like episodes) due to the tRNALeu(UUR) A3243G mtDNA point mutation. In Patient B, only skeletal muscle tissue was available for analysis. Skeletal muscle biopsies from 43 neurologically unaffected patients (25 male and 18 female, age range 3–77 years) with no histological evidence for a neuromuscular disorder were used as controls.
All examinations were conducted according to the Declaration of Helsinki (2000) of the World Medical Association and the guidelines of the local ethics committee of the University of Bonn. Informed consent was obtained from all patients.
Skeletal muscle histology
Consecutive cryostate sections (6 μm) of open vastus lateralis muscle biopsies were processed for standard histological and enzyme/histochemical analysis, as described previously . Quantitative histochemical single-fibre analyses of COX (cytochrome c oxidase) and SDH (succinate dehydrogenase) activities were performed using fibre-specific grey-level determination of 12-bit video images for Patient A [11,12].
Respiratory-chain enzyme activities [rotenone-sensitive NADH: CoQ1 oxidoreductase (complex I), COX (complex IV)], and CS (citrate synthase) in skeletal muscle homogenate were measured spectrophotometrically, as described previously . Results were given in units/g of wet weight and were corrected for potential fibre-type variations or adaptive mitochondrial proliferation by normalization of data for the mitochondrial matrix marker enzyme CS.
Molecular genetic analyses
Genomic DNA was extracted from approx. 30 mg of skeletal muscle specimens and buccal mucosa with the QiaAmp DNA Mini Kit (Qiagen, Hilden, Germany). Genomic DNA was isolated from 10-ml aliquots of EDTA-anti-coagulated blood using a salting-out method . mtDNA was screened for large-scale rearrangements by Southern blotting of PvuII-cleaved total DNA from skeletal muscle using digoxigenin-labelled isolated human mtDNA as the full-length mtDNA probe (all restriction endonucleases were from New England Biolabs, Frankfurt, Germany). The presence of duplications was excluded by digestion with PmeI and BamHI. For direct sequence analysis, 28 overlapping PCR fragments were amplified from the whole mitochondrial genome of skeletal muscle DNA and sequenced using a commercial sequencing service (MWG Biotech, Ebersberg, Germany).
To quantify A3243G and G5835A mutation loads, we used different PCR-RFLP (restriction-fragment-length polymorphism) assays. For quantification of G5835A, we used a forward primer, 5′-GCACCCTAATCAACTGGCTTC-3′, and a reverse mismatch primer (mismatch underlined), 5′-CTAAAGACAGGGGTTAGGGCT-3′, to introduce a novel BanII restriction site in the wild-type allele. To introduce a novel BceAI restriction site in the mutant allele, we designed a forward primer, 5′-AGCTAAGCACCCTAATCAACTG-3′, and a mismatch reverse primer with a T7-extension site (italic), 5′-TAATACGACTCACTATAGGGAAATCTAAAGACAGGGGTTAGGCCG-3′. For quantification of A3243G, a PCR product was amplified using a forward primer, 5′-GAAAGGACAAGAGAAATAAGGCCT-3′, and a reverse primer, 5′-GGAATTGAACCTCTGACTGTAAAG-3′. PCR products were digested with HaeIII.
Amplification conditions were as follows: 95°C for 10 min; 33 cycles of 95°C for 15 s, 55°C for 30 s and 72°C for 40 s; and finally 72°C for 7 min. We separated restriction fragments on 10% polyacrylamide gels and the fragments were revealed with SYBR Green I staining (Sigma–Aldrich, Steinheim, Germany). The proportions of mutant and wild-type mtDNA were calculated from band intensities with local area background subtraction (Image J analysis software, http://rsb.info.nih.gov/ij).
Single-fibre PCR analysis
We used 10 μm sections of unstained frozen skeletal muscle tissue from Patient A located in-between the muscle sections stained for COX and SDH activity. Fibre types of single fibres were distinguished based on myosin ATPase staining at pH 4.6 and 9.4. For single-fibre PCR analysis, we randomly selected 20 type I fibres with comparable SDH activity quantified by fibre-specific grey-level determination to examine the dependency of the cellular biochemical function on the mutation load. Laser microdissection of muscle fibres was performed as described previously . To determine the G5835A mutation load in single fibres, we used a nested PCR technique. First, we amplified a PCR product spanning the mutation site using the forward primer, 5′-CCTTACCACGCTACTCCTACCTATCTCC-3′ (nucleotides 5462–5489), and the reverse primer 5′-CTCCAGCTCATGCGCCGAATAATAG-3′ (nucleotides 5985–5961). In the second round of nested PCR, we subjected PCR products of the first round to mismatch PCR-RFLP with digestion of the wild-type allele, as described above.
Total RNA was isolated from a small piece of skeletal muscle tissue from Patient A, 30–50 mg samples of muscle tissue from four healthy control subjects and cultured human osteosarcoma cells (143B cell line). To maintain charging of the tRNAs with amino acid and separate tRNA from amino-acylated tRNA, RNA was isolated and subjected to electrophoresis under acidic conditions on long gels, as described previously . Regions of mtDNA encompassing mitochondrial tRNA genes were amplified from DNA from the 143B cells by PCR and used as probes. For each tRNA, two different probes were used, either specifically designed for wild-type or mutant tRNA. Primer sequences used for the design of tRNALeu(UUR), tRNAVal, tRNATyr and tRNAIle probes can be obtained from the corresponding author on request. For the normalization of tRNAs to a non-mitochondrial small RNA species, blots were hybridized to a 5S rRNA probe (5′-GGGTGGTATGGCCGTAGAC-3′) which was end-labelled using [γ-32P]ATP (5000 Ci/mmol; Hartmann Analytic, Braunschweig, Germany) and T4-PNK (T4-polynucleotide kinase) . Purified PCR products (QIAquick PCR purification columns; Qiagen, Hilden, Germany) were radiolabelled with [α-32P]dCTP (5000 Ci/mmol; Hartmann Analytic) using a random-primed DNA labelling kit (Roche Applied Science, Mannheim, Germany), and unincorporated nucleotides were removed from the labelling reactions by gel filtration through ChromaSpin columns (Clontech, Heidelberg, Germany). Prehybridization, hybridization and analysis of the data by phospho-imaging was carried out as described previously . The two bands corresponding to amino-acylated and non-acylated tRNA were identified by comparison of native with alkali-treated de-acylated RNA samples from 143B cells and quantified individually in order to estimate the degree of loading with amino acid. Finally, the sum of both values was divided by the signal for 5S rRNA to obtain tissue levels of tRNAs.
Histochemical and biochemical evaluation of skeletal muscle showed an exceptionally severe mitochondrial pathology
As shown in Figure 1, back-to-back stained muscle biopsy sections from Patient A revealed a weaker overall COX staining, abundant SDH-positive COX-negative fibres (indicated by circles, ‘mosaic’ COX-staining pattern) and ragged-red fibres (asterisks) in comparison with Patient B with a similar mutation load of the heteroplasmic A3243G mtDNA point mutation (Figures 1b and 1e). The low overall COX staining of muscle fibres from Patient A is not a result of decreased mitochondrial content, as shown by the SDH staining intensities in Figures 1(c) and 1(f). Mitochondrial respiratory-chain enzyme activities in skeletal muscle homogenate of Patient A were compared with controls and Patient B (Table 1). In agreement with the histological stainings, the results showed a dramatic decline in absolute activities of complex I and complex IV in the biopsy of Patient A, but also when normalized for CS activity.
|.||Enzyme activity (unit/g of wet weight) .|
|Respiratory chain enzyme .||Controls (means±S.D.) .||Patient A .||Patient B .|
|NADH:CoQ1 reductase (complex I)||1.42±0.82||0.09||0.68|
|COX (complex IV)||9.41±2.87||1.34||4.31|
|.||Enzyme activity (unit/g of wet weight) .|
|Respiratory chain enzyme .||Controls (means±S.D.) .||Patient A .||Patient B .|
|NADH:CoQ1 reductase (complex I)||1.42±0.82||0.09||0.68|
|COX (complex IV)||9.41±2.87||1.34||4.31|
Histochemical stainings of serial sections of skeletal muscle biopsies of Patients A (a–c) and B (d–f) under identical conditions
mtDNA sequence analysis revealed three novel and potentially pathogenic tRNA mtDNA point mutations
MtDNA large-scale rearrangements were ruled out. Direct sequence analysis of the mitochondrial genome isolated from skeletal muscle tissue of Patient A revealed three novel and putative pathogenic tRNA mtDNA point mutations among a multitude of mutations when compared with the reference sequence . For details of the mutations, please see Figure 2.
Three novel tRNA mtDNA mutations residing in the tRNA
Tyr, tRNA Ile and tRNA Val genes of Patient A
We identified a novel G→A exchange at position 5835 of mtDNA which was found to be heteroplasmic in skeletal muscle. The G5835A mutation is located in the mitochondrial tRNATyr gene and causes a C→U transition at an evolutionary highly conserved position in the T-stem of the tRNATyr (Figure 2a). The mutation has not been described previously in human mtDNA databases and is not present in more than 2500 controls (compare with MITOMAP, http://www.mitomap.org/; Human Mitochondrial Genome Database, http://www.genpat.uu.se/mtDB/).
We also detected a previously undescribed homoplasmic mutation, del4315A, in the mitochondrial tRNAIle gene affecting the tRNA TψC loop (Figure 2b).
Finally, a not-yet-reported homoplasmic U→C transition was detected at position 1658 in the mitochondrial tRNAVal gene, affecting the tRNA T-stem at an evolutionary not-well-conserved position (Figure 2c). Further sequencing revealed two novel homoplasmic mutations in protein-coding genes: a C11972T transition in the ND4 gene and a A14539G transition in the ND6 gene (compare with MITOMAP, http://www.mitomap.org/; Human Mitochondrial Genome Database, http://www.genpat.uu.se/mtDB/), both genes encoding subunits of complex I. There is a high probability that C11972T and A14539G do not have functional consequences, since they do not result in amino acid changes. A known heteroplasmic D-loop polymorphism was detected at position 72, resulting in a T→C transition in skeletal muscle mtDNA. Sequencing revealed the following homoplasmic polymorphisms and were in agreement with the mitochondrial haplogroup ‘H’: 73A, 2706A and 7028C. More than 400 of the publicly available complete mitochondrial coding region sequences belong to haplogroup ‘H’. None of the five novel mutations mentioned above has been reported in these phylogenetically related individuals, or in more than 2100 other unrelated sequences. The further identified polymorphisms C64T, A263G, ins310C, A750G, A1438G, A4769G, A8860G, A15326G, A16269G and T16519C did not allow a more precise classification of the sample to the subgroups of the ‘H’ haplogroup described to date .
Mismatch PCR-RFLP analysis demonstrated, in agreement with the sequence analysis (Figure 3a), the presence of G5835A in 69±3% of skeletal muscle bulk mtDNA from Patient A (Figures 3b and 3c, lane 2). This mutation was restricted to muscle, and not present in buccal mucosa and blood (Figures 3b and 3c, lanes 3–4). Analysis of blood samples of the patient's two sisters, one brother and mother did not reveal detectable amounts of the heteroplasmic mutation (Figures 3b and 3c, lanes 5–8). These findings suggest a mutational event after germ-layer differentiation; however, a mutation arisen early in embryogenesis or in the maternal germ line cannot be excluded. All other mutations were found to be homoplasmic in the blood of the patient, one sister and the mother (other relatives were not analysed for these mutations).
Quantification of the heteroplasmic G5835A tRNA
Tyr mtDNA mutation
Skeletal muscle single-fibre PCR analysis for Patient A showed that all SDH-positive COX-negative fibres, as indicated by comparison of COX- and SDH-activity (Figures 1b and 1c), had at least 98% of the G5835A mutation, which demonstrated a high threshold for phenotypical expression at the single-cell level (Figure 4).
Correlation of COX activities with G5835A tRNA
Tyr mtDNA mutation loads in single skeletal muscle fibres of Patient A
Levels of the mitochondrial tRNAs for tyrosine and isoleucine were severely reduced
In comparison with the nuclear-encoded 5S rRNA, which served as a loading control, we observed a 95% reduction in the tRNATyr amount in total skeletal muscle RNA of Patient A (Figures 5a and 5b). tRNAIle levels were found to be decreased by approx. 85%. In contrast, the tRNAVal amount was reduced by only approx. 60%. The levels of tRNALeu(UUR) were even slightly increased. The same results were independently obtained for each tRNA with two different probes designed for wild-type (Figure 5a) or mutant tRNA (results not shown). This approach rules out the possibility that lower tRNA signals in the patient's muscle might be caused by poorer hybridization of wild-type probes to mutant tRNAs. Evaluation of the ratio of amino-acylated compared with non-acylated tRNA showed that tRNATyr and tRNAVal were predominantly loaded with amino acid in control muscles, whereas tRNAIle and tRNALeu(UUR) were equally present in acylated and de-acylated forms. As shown in Figure 5(a), charging of tRNAIle with amino acids was clearly impaired in the patient's muscle, with less than 15% compared with approx. 40% in controls.
Steady-state tRNA levels in the skeletal muscle of Patient A
The diagnosis of mitochondrial disorders often requires a complex approach, including clinical evaluation, skeletal muscle histology, biochemical and, finally, molecular genetic analyses . However, assigning pathogenicity to mtDNA point mutations is often a challenge [24,25], especially when novel mutations are present in individuals with an ethnic background which has not been extensively studied with respect to haplogroup-defining mutations. Moreover, since various homoplasmic tRNA mtDNA point mutations have been shown to be pathogenic [5–9] and many apparently neutral polymorphisms are heteroplasmic , the heteroplasmic state of a mutation is no longer a major criterion for pathogenicity. Equally problematic is the criterion of degree of conservation among species, since, for example, the A3243G tRNALeu(UUR) mtDNA mutation with proven pathogenicity in humans is the normal genotype in the domestic dog, Canis familiaris.
This diagnostic challenge is amply illustrated in the genetic analysis of Patient A with CPEO in the present study in which we detected five novel mtDNA mutations. Two mutations were located in protein-coding genes and were silent mutations, i.e. did not result in an amino acid change. However, all further three novel tRNA mtDNA point mutations are potentially pathogenic. To address the pathogenic relevance of each of these mutations, we performed (i) single skeletal muscle fibre analysis and (ii) Northern-blot analyses as functional tests.
Demonstrating a correlation between biochemical phenotype and mutation loads of heteroplasmic mtDNA point mutations in individual cells has been long used as a strong evidence for pathogenicity . In Patient A, we found that those muscle fibres that displayed a failure of oxidative phosphorylation (as indicated by negative COX staining in SDH-positive fibres) harboured very high levels of the heteroplasmic G5835A tRNATyr mutation. Thus the heteroplasmic tRNATyr mutation is proven to be functionally relevant and plays a primary role in creating the mosaic pattern in our histochemical COX muscle stainings. The 98% mutation load threshold for cellular biochemical impairment is, however, higher than that described for the A3243G tRNALeu(UUR) mutation  and the G12276A tRNALeu(CUN) mtDNA mutation [12,27], which suggests a mild functional effect of the mutation. This is in contrast with the severe biochemical alterations present in skeletal muscle homogenate of Patient A (Table 1) and with the histological findings of reduced overall COX staining intensity, a mosaic pattern of COX activity and abundant ragged-red fibres (Figure 1). The biochemical phenotype and general decrease in COX activity (including fibres with low G5835A mutation loads), apart from the mosaic pattern, cannot be explained by the presence of the heteroplasmic G5835A tRNATyr mtDNA mutation alone. Functional analyses of both further tRNA mtDNA mutations demonstrated that the tRNAIle mutation plays an addition role for pathogenicity. Since this mutation is homoplasmic, it affects all cells, which readily explains the overall decrease of the COX staining in the skeletal muscle biopsy. In this respect, Northern-blot analyses revealed that apart from tRNATyr, tRNAVal and tRNAIle also showed decreased steady-state levels, leaving a remaining 40% and 15% of the respective tRNAs as compared with controls. The latter value is similar to residual tRNA amounts reported in previous studies [6,7] for homoplasmic pathogenic tRNA mtDNA mutations (9–15% and 5–10% respectively). Impaired amino-acylation of tRNAIle was also observed, further stressing the pathogenic relevance. However, the presence of the homoplasmic tRNAIle mutation alone appears not to cause serious clinical symptoms. The maternal relatives of Patient A carried all homoplasmic mutations present in the patient, but no family member showed apparent signs of a neurological disorder. We suggest that the G5835A tRNATyr and del4315A tRNAIle mutation act in concert and lead to more severe problems for translation of mitochondrially encoded proteins. Since the tyrosine and isoleucine codons are present at high frequencies in most protein-coding mitochondrial genes, low levels of these tRNAs, in combination with a defect in charging of tRNAIle for amino acids, readily explain the severe biochemical and histological phenotype. In contrast, the 60% reduction of tRNAVal is probably not sufficient to contribute to an additional impairment of translation. This interpretation is in agreement with the conservation of sites among species: the tRNATyr mutation is located at a strictly conserved site and the tRNAIle mutation is located at a site which is conserved among primates, whereas the tRNAVal mutation occurred at a non-conserved site. Not surprising, Northern-blot analysis confirmed the considerable pathogenic role of the heteroplasmic G5835A tRNATyr mutation, since tRNATyr steady-state levels were found to be decreased by 95%. This is higher than the skeletal muscle bulk mtDNA mutation load which might indicate an effect of mutant tRNAs on the stability of wild-type tRNAs. Interaction between tRNA molecules can be a possible mechanism for this effect and has been described previously for dimerized A3243G tRNALeu(UUR) .
In conclusion, we demonstrated that two of five not-previously-described mtDNA mutations, the G5835A tRNATyr and del4315A tRNAIle mutation, are functionally relevant. For both mutations, this was proven by Northern-blot analyses, and for the heteroplasmic mutation with single-fibre analysis as well. Single-fibre results suggested a rather mild effect of the G5835A tRNATyr mutation, which was in contrast with severe histological and biochemical skeletal muscle findings. Concerted pathogenic effects of both mutations on the skeletal muscle phenotype can readily explain these findings. Our present study underlines the importance of assessment of both (i) the complete sequence information and (ii) the biochemical effects of each candidate pathogenic mutation in this process. The interaction of tRNA mtDNA point mutations shown in the present study provides a basis for understanding how homoplasmic mutations, specific for a family or even an entire haplogroup, may influence the functional effects of pathogenic heteroplasmic mtDNA point mutations.
This study was supported by the Deutsche Forschungsgemeinschaft (KU-911/15-1, SCHR-562/4-3) and the Center for Molecular Medicine Cologne (CMMC, C9). The technical assistance of Maria Bust, Ulrike Strube and Karin Kappes-Horn is gratefully acknowledged. We thank the patient and family members in Sakarya (Adapazari), Eastern Turkey, for their co-operation.
The nucleotide sequence data reported will appear in the DDBJ, EMBL, GenBank® and GSDB Nucleotide Sequence Databases under the accession number DQ473537.