The G13513A Mutation in the ND5 Gene of Mitochondrial DNA as a Common Cause of MELAS or Leigh Syndrome: Evidence From 12 Cases

Background  The number of molecular causes of MELAS (a syndrome consisting of mitochondrial encephalomyopathy, lactic acidosis, and strokelike episodes) and Leigh syndrome (LS) has steadily increased. Among these, mutations in the ND5 gene (OMIM 516005) of mitochondrial DNA are important, and the A13513A change has emerged as a hotspot.

Objective  To describe the clinical features, muscle pathological and biochemical characteristics, and molecular study findings of 12 patients harboring the G13513A mutation in the ND5 gene of mitochondrial DNA compared with 14 previously described patients with the same mutation.

Design  Clinical examinations and morphological, biochemical, and molecular analyses.

Setting  Tertiary care university hospital and molecular diagnostic laboratory.

Patients  Three patients had the typical syndrome features of MELAS; the other 9 had typical clinical and radiological features of LS.

Results  Family history suggested maternal inheritance in a few cases; morphological studies of muscle samples rarely showed typical ragged-red fibers and more often exhibited strongly succinate dehydrogenase–reactive blood vessels. Biochemically, complex I deficiency was inconsistent and generally mild. The mutation load was relatively high in the muscle and blood specimens.

Conclusion  The G13513A mutation is a common cause of MELAS and LS, even in the absence of obvious maternal inheritance, pathological findings in muscle, or severe complex I deficiency.

The G-to-A mutation at nucleotide position 13513 in mitochondrial DNA (mtDNA) affects an evolutionarily conserved amino acid (D393N) in the reduced form of nicotinamide adenine dinucleotide (NADH) dehydrogenase 5 (ND5), one of the 7 subunits of complex I encoded by mtDNA. This mutation was originally described in an adult patient with MELAS (a syndrome consisting of mitochondrial encephalomyopathy, lactic acidosis, and strokelike episodes)¹ and subsequently in a patient with an overlap syndrome of MELAS and Leber hereditary optic neuropathy.² This same mutation was later identified in children with Leigh syndrome (LS), a neurodegenerative disorder usually starting before 1 year of age and leading to death within months or years.³ We now confirm in a series of 12 patients that MELAS and LS are characteristically associated with the G13513A mutation and compare clinical, histochemical, biochemical, and molecular findings in the 2 conditions.

Table 1 summarizes the biometric features, clinical phenotypes, morphological features of muscle specimens, respiratory chain defects, and mutation load in our cohort of 12 patients and in 14 additional patients with the same mutation who were previously described in the literature. Patient 3 in our series has been the subject of a case report from Massachusetts General Hospital, Boston.⁹ Clinically, the 3 adult patients (patients 1-3) had typical MELAS features, whereas the 9 infants and children had typical LS. They are not described in detail, but we discuss the unusual clinical features and compare them with similar data already reported in other subjects with the G13513A mutation.

Serial, 8-μm-thick cross-sections of frozen muscle specimens were stained with the modified Gomori trichrome and screened for succinate dehydrogenase and cytochrome-c oxidase activities, as described previously.¹⁰ Enzymes of the mitochondrial respiratory chain and citrate synthase were analyzed in muscle extracts as described previously.¹¹

Total DNA was extracted according to standard protocol using a commercially available DNA isolation kit (PUREGENE; Gentra System, Inc, Minneapolis, Minnesota) and following the manufacturer's instructions. Protocols for DNA extraction from blood, urine, hair roots, and cheek mucosa have been described previously.¹²

For restriction fragment length polymorphism analysis, a 119–base pair fragment was amplified by polymerase chain reaction (PCR) using a reverse primer at nucleotide (nt) 13610-13593 and a mismatched forward primer at nt 13491-13512 (with G instead of A at nt 13510). In the presence of the wild-type base at nt 13513, the PCR-amplified fragment contains a unique restriction site for the endonuclease BbsI (GAAGACN₂/N₆), which is absent in the mutant genomes. The digested product was electrophoresed in 12% nondenaturing acrylamide gel and analyzed in a phosphoimager (Molecular Analyst; BioRad, Hercules, California) using image-quantifying software (ImageQuant; Molecular Dynamics Corp, Sunnyvale, California) to assess the percentage of the mutation.

Direct sequencing was performed in a genetic analyzer (ABI Prism 310; Perkin-Elmer Applied Biosystems, Foster City, California) using a kit (BigDye Terminator Cycle Sequencing Reaction Kit; Perkin-Elmer Applied Biosystems).

Table 1 summarizes the basic biometric data of our 12 patients, their clinical phenotypes, the salient pathological and biochemical features of the muscle samples, and the G13513A mutation loads in muscle samples and other more easily accessible tissues. For ease of comparison, the same information is provided about the 14 previously described patients harboring the same mutation.

Of the 3 patients with the MELAS phenotype, the most unusual was patient 1 because of the early onset of symptoms (ptosis and ophthalmoparesis evident at 18 months of age) and the lack of strokelike episodes despite the magnetic resonance imaging evidence of a lesion in the left cerebral peduncle. The other 2 patients had typical strokelike episodes, although the age at onset in patient 3 was unusually late (61 years). Two other patients described in the literature, the original patient of Santorelli et al¹ and patient 2 of Hanna et al,⁴ had late onset of strokes (43 and 52 years of age, respectively), but both had had lifelong neurosensory hearing loss.

Seven of the 9 patients with LS had a fairly typical clinical course, characterized by normal development during the first months of life (≤ 7 months), followed by loss of acquired skills, hypotonia, failure to thrive, nystagmus, and global developmental delay. Patients 4 and 5 had a relatively late onset, at 2 years and at 15 months of age, respectively. Nonneurological clinical features included cardiomyopathy (patients 6, 7, 9, and 12), hydronephrosis and a dysplastic kidney (patient 7), and dysmorphic features (hypertelorism, small chin, and prominent posteriorly rotated ears in patient 7). Four of these 9 patients died at 3 years or younger, whereas the oldest survivor was 5 years old at last follow-up.

Family history was unavailable in 1 adopted individual (patient 1) but showed no clinical involvement of maternal relatives in 7 of 11 patients. In 4 instances, DNA from accessible tissues of the mother alone (2 families) or from other maternal relatives (3 families) was available. The mutation was undetectable in 3 cases with no family history but was present in very low levels in 2 cases, one of whom also had no family history (patient 11). The mother of patient 8 (in whom the mutation was undetectable in multiple tissues) had a subsequent pregnancy. Not too surprisingly, no mutation was detected in amniocytes and cord blood or in the blood of the newborn infant, who had met developmental milestones at 5 months of age (Figure 1). Patient 10 was the 10th child of an asymptomatic mother whose first 9 children showed no evidence of the mutation. It is therefore likely that this mutation tends to occur de novo or is often present at concentrations low enough to be compatible with a subclinical status. This finding is confirmed by data in the literature.^1-8 Of 14 patients described with the G13513A mutation, family history findings were negative in 8, positive in 4, and undescribed in 2.

Histochemical findings in 7 patients with LS in whom morphological analysis could be performed showed mitochondrial proliferation in the muscle fibers and intramuscular vessels (strongly succinate dehydrogenase–reactive vessels [SSVs]). Figure 2 illustrates typical findings. One of the patients with LS (patient 10) was unusual in that she had significant mitochondrial proliferation, with ragged-red fibers (RRFs). Of the 3 patients with the MELAS presentation, 2 had RRFs, and the other did not.

Respiratory chain enzyme activities in both sets of patients were quite variable (Table 1). Of the 3 patients with MELAS presentation, one had normal activities, one had isolated decrease of complex I levels, and the third had decreased activities for complexes I, III, and IV. In the 7 patients with LS in whom muscle biochemical analysis could be performed, 2 had normal activities, 3 had isolated complex I deficiency, and 2 had multiple deficiencies. Patient 10 (who had pre-RRFs [rare occurrence of typical RRFs but frequent presence of subsarcolemmal mitochondrial proliferation]) also had markedly increased citrate synthase levels such that complex I, III, and IV deficiencies were even more severe when normalized to this marker enzyme.

Again, these findings correspond to those of the 14 previously reported patients. Complex I activity was decreased in muscle from 9 of 10 patients so studied, but the decrease was either unspecified or modest, except for 3 patients, who had 25%, 30%, and 40% residual activities.^1,3,7 Complex I activity was also decreased in the liver of 1 patient with MELAS¹ and 1 with LS⁵ and in cultured skin fibroblasts from 4 patients with LS.^3,5,7 The variable and generally small decrease of complex I activity makes biochemical findings a poor indicator of the underlying molecular defect. We have observed that the sensitivity increases when complex I activity is measured in isolation (ie, by the NADH–coenzyme Q10 reductase assay) rather than in association with complex III (ie, by the NADH–cytochrome-c reductase assay).

The G13513A mutation was initially identified by PCR/restriction fragment length polymorphism analysis, but all cases were then confirmed by sequence analysis results. Table 1 shows the mutation loads in all available tissues. Somewhat surprisingly, the degree of heteroplasmy in skeletal muscle (mean of 10 patients, 74%; range, 41%-89%) was not much higher than in blood (mean of 6 patients, 65%; range, 32%-90%). The mutation load (95%) was also very high in the single sample of urinary sediment so studied. Thus, when this mutation is suspected, blood seems to be an adequate tissue to confirm the diagnosis. As previously discussed, the situation may be different for the asymptomatic maternal relatives. Although only 2 of the 4 mothers studied harbored the mutation, this was barely detectable in their blood, whereas the urinary sediment and, in 1 case, the cheek mucosa, had higher mutation loads. Although most cases are probably sporadic, maternal inheritance and the attendant risk of recurrence of the disease should be based on analysis of urinary epithelium cells if muscle is not available.

Data from the literature confirm the heavy (mean of 13 patients, 62%) but highly variable (range, 29%-90%) mutation loads in muscle, but showed generally lower loads (mean of 9 patients, 29%) in blood, ranging from undetectable⁴ to 73%.¹

Six other mutations in ND5 have been associated with disease (Table 2), including 3 (A13514G, G13063A, and A12770G) with typical MELAS,^8,13,15 1 (G13042A) with an overlap syndrome of MELAS and myoclonus epilepsy with RRF,¹⁴ and 2 (A13045G and A13084T) with a combination of MELAS and LS features.^15,16 Although it is hardly surprising that the clinical phenotypes would be similar to those observed in carriers of the more common G13513A mutation, the results of morphological analysis of muscle samples were normal in 5 of the 7 patients described (2 patients had the same A13514G mutation⁸). One patient had scattered RRF,¹³ and another had pre-RRF and SSVs.¹⁴ Respiratory chain activities in muscle were normal in 1 patient, whereas complex I activity was variably decreased (15%-88% of normal) in the others. Mutation loads in muscle (mean of 7 patients, 73%; range, 48%-90%) tended to be higher than in patients with the G13513A mutation. Family history was negative in 5 of the 7 cases, but the mutation was detected in the mother of one of these patients.

The ND5 gene of mtDNA is clearly a hot spot of pathogenic mutations, but the G13513A mutation is by far the most frequent, accounting for 26 of 33 patients (79%). It is therefore important to consider this mutation in patients with MELAS, LS, or overlapping features of the 2 syndromes. Clues that suggest this molecular diagnosis in patients with MELAS rather than the more common A3243G mutation in transfer RNA^Leu(UUR) include (1) a relatively late age at onset; (2) frequent lack of family history suggesting maternal inheritance; (3) pre-RRFs and SSVs; and (4) normal or only modestly decreased complex I activity. Conversely, finding pre-RRF or SSVs in muscle biopsy specimens from children with LS should alert to the possibility of ND5 (or ND3) mutations; the morphological features of muscle are usually normal or nonspecifically altered in LS owing to adenosine triphosphatase 6 mutations (maternally inherited LS) or to mutations in nuclear genes. In counseling families with LS and the G13513A mutation, it is important to examine multiple accessible tissues, especially urinary sediment, from the mother because, although the mutation often occurs de novo, it may occur in deceivingly low or undetectable levels in blood.⁶

Correspondence: Salvatore DiMauro, MD, Department of Neurology, Columbia University Medical Center, 3-313A Russ Berrie Medical Science Pavilion, 1150 St Nicholas Ave, New York, NY 10032 (sd12@columbia.edu).

Accepted for Publication: October 27, 2007.

Author Contributions:Study concept and design: Shanske and DiMauro. Acquisition of data: Coku, Lu, Ganesh, Krishna, Tanji, Bonilla, Naini, Hirano, and DiMauro. Analysis and interpretation of data: Shanske, Coku, Lu, Naini, Hirano, and DiMauro. Drafting of the manuscript: Shanske, Coku, Lu, Krishna, Tanji, Bonilla, Naini, and DiMauro. Critical revision of the manuscript for important intellectual content: Ganesh and Hirano. Obtained funding: DiMauro. Administrative, technical, and material support: Shanske, Coku, Lu, Krishna, Hirano, and DiMauro. Study supervision: Shanske, Tanji, Bonilla, Naini, and DiMauro.

Financial Disclosure: None reported.

Funding/Support: This work was supported by grant HD32062 from the National Institutes of Health and by the Marriott Mitochondrial Disorders Clinical Research Fund.