Mitochondria are key for many cellular processes. One of the most important mechanisms is oxidative phosphorylation (OXPHOS) resulting in the production of cellular energy in the form of ATP. The OXPHOS system consists of five multiprotein complexes (I–V) and two mobile electron carriers (coenzyme q and cytochrome c) embedded in the lipid bilayer of the mitochondrial inner membrane.^1,2 The mitochondrial genome encodes 13 essential polypeptides of the OXPHOS system and the necessary RNA machinery (two ribosomal RNAs and 22 transfer RNAs (tRNA)). The remaining structural proteins and proteins involved in import, assembly and mitochondrial DNA (mtDNA) replication are encoded by the nucleus and specifically targeted to the mitochondria. OXPHOS disease is characterised by a wide variety of clinical symptoms, in which one or more organs can be involved, and by genetic and clinical heterogeneity.^2,3 With an estimated total number of about 1500 nuclear mitochondrial genes of which 600 have been identified so far,⁴ this complicates the process of identification of the underlying genetic defect. Although mutations in the mtDNA tRNA genes have been reported far more often than other mutations in mtDNA protein-coding genes,² this figure is highly biased by a preferential screening of these genes.

In this study, the complete mtDNA was screened for heteroplasmic mutations using denaturing high-performance liquid chromatography (DHPLC) analysis in a group of 116 unrelated patients suspected for OXPHOS disease but without the common mutations for mitochondrial encephalomyopathy, lactic acidosis and stroke-like syndrome (MELAS) m.3243A→G, myoclonic epilepsy, ragged red fibres (MERRF) m.8344A→G, Leigh/neuropathy, ataxia and retinitis pigmentosa m.8993T→G/C or large deletions. For this group of patients, we report that the ND5 gene is a commonly mutated gene.

PATIENTS AND METHODS

A group of 116 patients was selected by clinical features such as progressive encephalomyopathy, stroke-like episodes, neuropathy and laboratory (lactic acidosis, OXPHOS enzyme deficiencies) or morphological (light and electron microscope) investigations. The presence of the more common mutations MELAS (m.3243A→G), MERRF (m.8344A→G) and Leigh/neuropathy, ataxia and retinitis pigmentosa (m.8993T→G/C) was excluded for these patients in muscle DNA, using methods described previously.^5,6 In addition, the presence of deletions was excluded by long-range PCR (Expand Long PCR, Roche, Mannheim, Germany) in the muscle tissue.⁷

Clinical description of patients and families with ND5 mutations

In family 1, a girl (patient 1) presented with a Leigh syndrome that included a discrete pyramidal syndrome of the legs, a right-sided hemidystonia and a strabismus divergens at the end of her first year. She had increasingly severe headaches. At 5 years of age, she had an exercise intolerance that interfered with her normal daily activities. She had normal growth parameters. The MRI (T2 weighted) of the cerebrum showed increased signal intensities in the global pallidus and caudate nucleus in the cerebral peduncles and in the peri-aequaductal region. The EEG had a slow background pattern and diffuse intermittent slow activities compatible with diffuse functional defects of the cerebrum. Lactate was increased in serum (3.8 mmol/l, normal <2.5 mmol/l) and in cerebrospinal fluid (CSF; 6.2 mmol/l, normal <2.5 mmol/l). In muscle tissue the complex I and III activities were low, and in fibroblasts only complex I activity was decreased. Morphological studies showed an increased succinate dehydrogenase staining but no ragged red fibres.

In family 2, a male patient (patient 2) presented with ataxia, pronounced internuclear ophthalmoplegia and progressive walking problems starting at the age of 3 years. He had increased lactate levels in blood and CSF. The complex I and V activities in muscle was reduced (50%). The MRI of the brain (T2 weighted) showed bilateral, symmetrical hyperintensities in the midbrain and pontine region. Skeletal muscle histology was normal as was enzyme histochemistry apart from a slight increase in the succinate dehydrogenase staining of the subsarcolemmal region of the muscle fibres. On the basis of these results, he was diagnosed as having Leigh-like syndrome. There were no other affected relatives.

In family 3, a female patient (patient 3) showed a MELAS/chronic progressive external ophthalmoplegia overlap phenotype. At the age of 11 years, she presented with exercise intolerance and a mild developmental delay. Brain MRI showed a subinsular cerebral infarct of the left hemisphere resulting in hemi-parkinsonism with a diminished facial expression, monotone speech and dystonia. She also had a mild external ophthalmoplegia and strabism. Biochemical studies in isolated mitochondria revealed a defect in the oxidation of pyruvate and glutamate, suspicious of a pyruvate dehydrogenase complex deficiency. The OXPHOS complexes were normal in muscle but complex III activity was found to be reduced in fibroblasts. In a muscle biopsy specimen taken 18 years after the first biopsy, the activities of the OXPHOS complexes were still normal (85–120%), except for complex I (58%). The mother and grandmother of the patient were both healthy.

In family 4, a male patient (patient 4) with a MELAS/Leigh phenotype displayed since 5 months of age a failure to thrive, psychomotor retardation and increased lactate values in both CSF (5.2 mmol/l) and blood (2.8–3.7 mmol/l). Further examination revealed a retinitis pigmentosa and microcytic anaemia. Compatible with a Leigh phenotype, T2-weighed hyperintensities were observed in both the globus pallidus and the dorsal part of the brain stem, the medulla oblongata and the cervical myelum. Light microscopy of muscle tissue showed an increase of the endomysial and perimysial connective tissue, and aspecific myopathic changes. Complex I activity was strongly reduced in muscle (8%). In fibroblasts, the activities of complexes I and III were lowered. The patient’s condition began to deteriorate from 7 months onwards and he died at 19 months of age, after a viral infection. It was difficult to obtain further clinical details from other family members since they were not very cooperative, but there seemed to be no other cases that showed symptoms of mitochondrial disease, including the mother and maternal grandmother.

In family 5, a boy died 5 days after birth because of respiratory complications (patient 5). He developed convulsions, persistent lactic acidosis, hypotonia and heart failure as a result of a cardiomyopathy. Autopsy of the brain showed bilateral aberrant substantia nigra, nucleus olivarus and vacuolar degeneration of the dorsal funiculus as is more often seen in metabolic diseases. Light microscopy of the heart muscle showed a reduced number of enlarged mitochondria compatible with a general mitochondrial disease. His twin brother (patient 6) died 2 hours after birth because of severe lung hypoplasia. Muscle biopsy revealed cytochrome oxidase-negative fibres but no other histological abnormalities. Autopsy of the brain showed vacuolar degeneration of the dorsal funiculus of the spinal cord. Complex I, III and IV activities were reduced in muscle. The younger brother of the twin (patient 7) was a dysmature boy who presented with cardiac and respiratory insufficiency and persistent lactic acidosis. Also in this younger brother, complex I, III and IV activities were reduced in muscle. Ultrasound revealed a severe hypertrophic cardiomyopathy and left ventricular failure. He died 15 days post partum. The three boys were diagnosed with Leigh’s disease. Their mother is healthy and has two healthy children. One of them has a father different from the deceased boys. None of the unaffected children could be tested owing to ethical reasons. Further family history is unremarkable.

Mitochondrial DNA analysis

Total DNA was extracted from blood and muscle biopsy specimens using the Wizard Genomic DNA Purification Kit (Promega, Leiden, The Netherlands) according to the instructions of the manufacturer. For extraction of DNA from fibroblasts and hair, the Puregene Kit (Gentra, Minneapolis, Minnesota, USA) and the Qiamp DNA Mini Kit (Qiagen, Venlo, The Netherlands) were used, respectively. DHPLC on a WAVE System (Transgenomic, Elancourt, France) was used to screen the mitochondrial genome, except for the D-loop region, for mutations as published previously⁸ or according to the MitoScreen Assay Kit (Transgenomic) that was essentially based on that protocol.

Sequence analysis

All putative nucleotide changes as identified by DHPLC analysis were followed by sequence analysis of heteroduplex fragments. Sequencing of the ND5 gene was performed using the BigDye terminator cycle sequencing methodology on an ABI3100 Genetic Analyzer (Applied Biosystems, Foster City, California, USA). The nucleotide positions of the reported mtDNA mutations correspond to the National Center for Biotechnology Information reference nucleotide sequence J01415.1. The amino acid positions correspond to the National Center for Biotechnology Information reference ND5-protein sequence AAB58953.1.

Restriction fragment-length polymorphism analysis with a fluorescently labelled primer was used to determine the degree of heteroplasmy of the ND5 mutations. Mutation-specific restriction enzymes were selected for each mutation.

Hydrophobicity prediction software

For the prediction of possible changes in hydrophobicity and changes in configuration of transmembrane domains of the ND5 protein due to sequence variation, we used TopPred online prediction software (http://bioweb.pasteur.fr/seqanal/interfaces/toppred.html).

RESULTS

A total of 116 patients with OXPHOS disease were screened for mutations in the mtDNA by DHPLC analysis. Fourteen likely pathogenic mutations were detected, one of which was a mutation in the ND1 gene, one in the ATP 6 gene, four in the ND5 gene and eight in four different tRNA genes, four of which were in the tRNA-Leucine-1 gene. In addition, we report a sequence variant in the ND5 gene, for which clinical significance is currently unclear. All nucleotide changes detected in the patients for the ND5 gene are reported in this paper, except for a silent mutation, which was considered to be a polymorphism. The four pathogenic ND5 mutations represent 27% of the mtDNA mutations and are thereby a common genetic cause of OXPHOS disease in the tested group of patients.

In patient 1, a new ND5 mutation, m.13511A→T (p.K392M), was detected in blood (65%), fibroblasts (53–65%) and muscle (72%). The healthy mother did not have this mutation in muscle, blood and hair roots, suggestive for a de novo mutation. Using the complex I pathogenicity scoring system,⁹ 25 of a maximum of 40 points was obtained for this variant. This was based on strong conservation (7 points; fig 1), heteroplasmy (5 points), reduction of complex I in muscle and fibroblasts (10 points) and absence in several tissues of the unaffected mother of the index patient (3 points). No significant change in hydrophobicity and transmembrane regions was predicted with the TopPred software (results not shown).

Patient 2 carried the recently reported pathogenic ND5 mutation m.13042G→A (p.A236T)^10,11 in blood (77%), muscle (84%) and fibroblasts (86%). The apparently healthy mother and maternal grandmother carried the mutation too, albeit in a lower percentage (mother: hair 25%, blood 11%; grandmother: blood <2%, muscle 4–6%). The independent identification of this mutation in two laboratories, including ours, adds another 5 points and results in a total pathogenicity score of 30.

The pathogenic mutation m.13513G→A (p.D393N)¹² was identified in the unrelated patients 3 and 4. The pathogenicity score for this mutation is 39.⁹ In patient 3, the m.13513G→A mutation was heteroplasmic in blood (4–6%), fibroblasts (1–5%) and muscle (13–15%), but was absent in fibroblasts of her mother and maternal grandmother. In patient 4, the mutation was heteroplasmic (11–17%) in blood, hair and muscle tissue, but was undetectable in cultured fibroblasts and was absent in blood and hair roots of his mother and maternal grandmother.

In patient 5 of family 5, the sequence variant m.12622G→A (p.V96I) was identified in the ND5 gene. The mutation percentage was 25% in skeletal and heart muscle, and <5% in fibroblasts. In his twin brother (patient 6) the mutation was also present in muscle with a mutation load of 30%, and in his younger brother this was 0–10% (patient 7). The muscle biopsy specimen of the last patient had a high fat content, which could explain the low mutation percentage compared with his brothers. It was not possible to determine the mutation percentage in the brain, the affected organ, as no biopsy specimens were available for DNA testing. The mother of the three boys had a very low mutation load (<5%) in blood, muscle and fibroblasts. A relatively low pathogenicity score of 18 points was obtained for this sequence variant. The majority of the points was based on the presence of a biochemical defect in muscle (8 points), heteroplasmy (5 points) and the very low mutation load in the unaffected mother (3 points). Only 2 of 10 points were obtained for conservation, since the amino acid isoleucine was present instead of valine in several mammals (fig 1). In addition, the substitution p.V96I was encountered once in a database (http://www.giib.or.jp) in an unaffected Japanese centenarian. No significant change in hydrophobicity and transmembrane regions was predicted with the TopPred software (results not shown).

DISCUSSION

We detected 14 likely pathogenic mutations in a group of 116 patients, who were screened for mutations in the mtDNA using DHPLC analysis and for whom common mutations were excluded. The ND5 gene turned out to be a commonly mutated gene accounting for 27% of the mutations (n = 4), leaving out the sequence variant m.12622G→A for which the clinical significance is currently unclear (see below). The frequent mutation of the ND5 gene is in line with previous reports that mutations in the ND subunits of complex I play an important role in OXPHOS diseases.^13,14 For patients with paediatric OXPHOS disease, mtDNA mutations accounted for 20–25% of the cases, of which about 6% are in the ND1, ND3, ND5 or ND6 genes.¹⁵ Our patient population consisted of 56% children. For 14% of these patients an mtDNA mutation was detected, including all four ND5 mutations that are reported in this study (6%). Altogether, this indicates the importance of the ND5 gene in OXPHOS disease in childhood.

The ND5 protein is one of the 46 subunits that constitute the OXPHOS complex I. Seven subunits are encoded by mitochondrial DNA, the other subunits by nuclear genes. The ND5 protein is a hydrophobic polypeptide, which is located peripherally in the complex.¹⁶ Although the ND5 gene is the largest of the mitochondrially encoded complex I genes (1811 nt), this alone does not explain the increased number of mutations in this gene compared with other mitochondrial genes, since we found no mutation in the second and third largest mitochondrial genes COI (1541 nt) and ND4 (1377 nt). For comparison, the total size of the tRNA genes together is 1489 nt, and the number of mutations in this study is eight, which also indicates an increased sensitivity of the tRNA genes, especially tRNA-Leucine-1, for mutations compared with other genes. This is most likely explained by the fact that most of the nucleotides of tRNA genes are important to form higher-order structures and interactions in the tRNA molecules.

The most frequently reported mutation in the ND5 gene until now is the m.13513G→A mutation.^14,17 In this study, two families with this mutation were detected. In addition, the new mutation m.13511A→T, and the previously reported mutation m.13402G→A and the sequence variant m.12622G→A in ND5 are reported. The variant m.13511A→T has never been reported as a polymorphism according to the MITOMAP (http://www.mitomap.org/), mtDB (http://www.genpat.uu.se/mtDB/) and mtSAP (http://www.giib.or.jp) databases and was not present in the other 116 patients with mitochondrial disorders who were analysed by DHPLC analysis. The pathogenicity of the new variant was evaluated using the scoring system of Mitchell et al.⁹

For the new m.13511A→T mutation, a pathogenicity score of 25 was found, which is in the range of probably pathogenic mutations.⁹ According to the TopPred prediction, the affected amino acid at position 392 is located in an external loop just after a transmembrane region. The change from lysine to methionine does not have a great impact on the local hydrophobicity and the external loop is still predicted (results not shown), but the amino acids at positions 392 and 393 are part of a putative quinone-reactive site of the enzyme,¹⁸ and the amino acid at position 393 is changed by the pathogenic m.13513G→A mutation. By a change of the amino acid at either position 392 or position 393, the putative quinone-reactive site could be lost and subsequently have a negative effect on the activity of the OXPHOS system. Although functional studies to support the pathogenicity of the new sequence variant have to be performed, we consider the mutation m.13511A→T in the ND5 gene to be the causative mutation in the OXPHOS disease in family 1. For comparison, the pathogenicity score for the m.12622G→A variant was 18, which is in the range of the possibly pathogenic group of sequence variants.⁹ The lower score is due to the fact that the conservation of the changed amino acid is not complete in mammals and the variant was reported once as a homoplasmic variant in an asymptomatic Japanese individual. It remains, however, difficult to judge the significance of information of non-curated databases. We therefore conclude that a causal relationship between the presence of the m.12622G→A sequence variant and the onset of mitochondrial disease cannot be entirely excluded at this moment. This is based on the fact that we detected the sequence variant in muscle biopsy specimens of children with a Leigh syndrome who died shortly after birth, whereas their healthy mother had only a low mutation percentage in muscle (<5%). In addition, although the mutation load in the patients was relatively low (25–30%), it is important to note that low mutation loads in ND5 can be associated with severe clinical phenotypes. Studies have demonstrated that ND5 synthesis is probably the rate-limiting step for the activity of complex I and consequently of respiration.^16,19,20 This may explain why mutations in this gene can cause a complex I defect even when present at a low mutant load.²¹ Clearly, functional studies or the detection of this mutation in other, independent, individuals with OXPHOS disease is needed to provide a conclusive answer on the pathogenicity of this sequence variant.

We looked for common clinical features among patients with an ND5 mutation. In previous reports, mutations in the ND5 gene have been associated with MELAS^12,22 or with various MELAS-overlap syndromes like MELAS–Leigh syndrome,¹⁴ MELAS–Leber hereditary optic neuropathy (LHON),²³ MELAS–MERRF¹⁰ and MELAS–Leigh–LHON.²⁴ More specifically, the m.13513G→A mutation has been recognised as a frequent cause of MELAS syndrome^23,25 and Leigh syndrome.^14,17,21 This mutation was detected in two families in this study. We compared the clinical aspects of the m.13513G→A mutation reported in literature with our two patients (table 1) and found that these patients differ from previous reports in the relatively low percentage of heteroplasmic mutations in muscle (13–16%) and other tissues (4–17%). Again, low mutation loads in the ND5 protein mutations can cause a complex I defect even when present at a low mutant load for reasons mentioned above. Still, no clear correlation exists between the degree of heteroplasmy in muscle for the m.13513G→A mutation, and the type and degree of complex deficiency or age at onset and severity of disease.

Clinical heterogeneity is also apparent for other ND5 mutations. Patient 5, with the m.13042G→A mutation, presented with a Leigh overlap syndrome and reduced complex I and V activities at the age of 4 years, whereas the first reported patient with this mutation was a boy who presented with a MELAS/MERRF overlap syndrome and severe complex I deficiency from the age of 17 years.¹⁰ Moreover, in the recent second report of the m.13042G→A mutation, a family was described in which people with the mutation presented with a prevalent ocular phenotype, including LHON-like optic neuropathy, retinopathy and cataract, but characterised also by strokes and early deaths.¹¹

For mitochondrial diseases it is not unusual that the same mutation in the same gene results in different phenotypes, possibly caused by additional nuclear-modifying genes or epigenetic factors.²⁶ Mutations in the ND5 gene are no exception to this rule. Despite this clinical heterogeneity, the common denominator in the clinical phenotype seems to be the MELAS or Leigh type of disease.

In conclusion, mutations in the ND5 gene are a common cause of OXPHOS disease. Although the ND5 gene is the largest mitochondrial gene, there is an increased sensitivity of this gene for mutations. Moreover, even low mutation levels in the ND5 gene seem to be able to compromise complex I function. In this study, complete screening of the mitochondrial genome using DHPLC analysis of 116 patients resulted in the detection of four families with an ND5 mutation, accounting for 27% of the total number of mtDNA gene mutations. The patients with ND5 mutations showed a broad spectrum of clinical phenotypes although MELAS and Leigh-like phenotypes seem to be the most common. We therefore conclude that for patients suspected of a mitochondrial disease, in particular patients with a Leigh-like or MELAS phenotype and complex I deficiency, initial screening for mutations in the ND5 gene should be performed for effective detection of a causative mutation.