The MFN2 gene is responsible for mitochondrial DNA instability and optic atrophy ‘plus’ phenotype

Abstract

MFN2 and OPA1 genes encode two dynamin-like GTPase proteins involved in the fusion of the mitochondrial membrane. They have been associated with Charcot–Marie–Tooth disease type 2A and autosomal dominant optic atrophy, respectively. We report a large family with optic atrophy beginning in early childhood, associated with axonal neuropathy and mitochondrial myopathy in adult life. The clinical presentation looks like the autosomal dominant optic atrophy ‘plus’ phenotype linked to OPA1 mutations but is associated with a novel MFN2 missense mutation (c.629A>T, p.D210V). Multiple mitochondrial DNA deletions were found in skeletal muscle and this observation makes MFN2 a novel gene associated with ‘mitochondrial DNA breakage’ syndrome. Contrary to previous studies in patients with Charcot–Marie–Tooth disease type 2A, fibroblasts carrying the MFN2 mutation present with a respiratory chain deficiency, a fragmentation of the mitochondrial network and a significant reduction of MFN2 protein expression. Furthermore, we show for the first time that impaired mitochondrial fusion is responsible for a deficiency to repair stress-induced mitochondrial DNA damage. It is likely that defect in mitochondrial DNA repair is due to variability in repair protein content across the mitochondrial population and is at least partially responsible for mitochondrial DNA instability.

MFN2, mitochondrial DNA instability, dominant optic atrophy, mitochondrial fusion, Charcot–Marie–Tooth 2A

Introduction

Hereditary optic neuropathies are a heterogeneous group of degenerative disorders of the optic nerve. Autosomal dominant optic atrophy is mainly related to mutations in the optic atrophy 1 (OPA1) gene, which encodes a dynamin-like GTPase involved in the fusion of the inner mitochondrial membrane (Delettre et al., 2000). Mitofusin 2 (Mfn2) is one of the two mitofusin proteins also required for mitochondrial fusion. Mfn1 and Mfn2 are conserved integral outer mitochondrial membrane proteins, each consisting of a large GTPase domain and two heptad repeats, or putative coil-coiled domains, all of which face the cytoplasm (Koshiba et al., 2004; Meeusen et al., 2004; Song et al., 2009). MFN2 mutations are a major cause of primary axonal Charcot–Marie–Tooth disease type 2A (Zuchner et al., 2004), an autosomal dominant neuropathy that impairs motor and sensory neurons with the longest axons resulting in earliest symptoms in distal extremities. An increasing number of articles report families with overlaps in clinical manifestations of autosomal dominant optic atrophy and patients with Charcot–Marie–Tooth disease type 2A. Patients with autosomal dominant optic atrophy may present peripheral neuropathies whereas some patients with Charcot–Marie–Tooth disease have optic atrophy (Zuchner et al., 2006; Yu-Wai-Man et al., 2010). Recently, it has been shown that OPA1 mutations can be responsible for syndromic forms of autosomal dominant optic atrophy associated with sensorineural deafness, ataxia, axonal sensory motor polyneuropathy, chronic progressive external ophthalmoplegia and mitochondrial myopathy with cytochrome c oxidase (COX)-negative and ragged red fibres, called dominant optic atrophy ‘plus’ phenotypes (Amati-Bonneau et al., 2008; Hudson et al., 2008). These phenotypes are related to mitochondrial DNA instability resulting in multiple mitochondrial DNA deletions. Charcot–Marie–Tooth disease type 2A with MFN2 mutations is also known to be associated with a variety of additional symptoms such as optic atrophy, sensorineural hearing loss, spastic paraparesis, tremor of fingers, pes cavus, scoliosis or knee joint contracture (Chung et al., 2006; Verhoven et al., 2006; Del Bo et al., 2008). But, contrary to OPA1, mitochondrial DNA instability has never been reported in patients harbouring MFN2 mutations.

In this study, we identified a novel missense mutation in MFN2 in a large three generational family with optic atrophy associated with axonal neuropathy and mitochondrial myopathy. The optic atrophy ‘plus’ phenotype that we describe, is related to mitochondrial DNA instability resulting in multiple mitochondrial DNA deletions in skeletal muscle. Furthermore, fibroblasts bearing the MFN2 mutation display a respiratory chain deficiency, a fragmentation of the mitochondrial network, a decrease in Mfn2 expression and a delay in mitochondrial DNA repair after oxidative stress.

Patients and methods

Patients

The pedigree of the Tunisian family is shown in Fig. 1. Affected and non-affected individuals included in this study were examined by a neurologist and an ophthalmologist. All clinical data are summarized in Table 1. Blood and tissue samples were obtained after patients had given informed consent. The index case was a 54-year-old female (Patient I-1), who presented with proximal and distal weakness of lower limbs associated with visual and hearing impairment. Symptoms rapidly worsened and at 60 years of age, she was wheelchair bound and exhibited a complex phenotype associating CNS and PNS disorders. She complained of cramps and pains in lower extremities. Clinical examination showed a sensory and cerebellar ataxia with pyramidal syndrome, generalized areflexia and generalized weakness in limb and axial muscles. Laboratory investigations revealed an increase in plasma lactate concentrations (3.6 mM/l, normal <2.1 mM/l) with a high lactate/pyruvate ratio (lactate/pyruvate ratio = 59, normal <21). Serum kinase was also increased (666 U/l, normal <140 U/l). Electrophysiological study revealed a sensory axonal motor neuropathy, predominantly sensory. Ophthalmological examination showed a moderate optic atrophy with a bilateral central scotoma. Visual acuity was 20/63 bilaterally. Audiogram revealed a bilateral sensorineural hearing loss. Brain MRI revealed multiple periventricular white matter lesions and severe diffuse cerebral atrophy (Fig. 2). She died at 69 years of age.

Figure 1

Pedigree of the family. Solid and hatched symbols represent clinically affected individuals. + = MFN2 wild-type allele; − = MFN2 mutated allele.

Figure 2

Brain MRI data from Patient I-1. Axial T₂-weighted brain MRI showing hyperintense lesions in periventricular and deep white matter with diffuse cerebral atrophy.

Table 1

Clinical data of affected members

Patient	Age at exam (years)	Optic atrophy onset	Axonal neuropathy			Onset of neurological symptoms	Additional symptoms			Brain MRI	Muscle biopsy
			Visual acuities (left eye, right eye)	Electrophysiology	Clinical symptoms		Deafness	Clinical myopathy	Others		OXPHOS deficiency	Histology	mtDNA multiple deletions	mtDNA depletion
I-1	60	?	20/63, 20/63	Sensory motor	+	54	+	Limb girdle weakness	Pyramidal and cerebellar syndrome Cataracts	Periventricular white matter lesions, diffuse cerebral atrophy	No	RRFs/COX	Yes	No
II-4	45	Early childhood	20/200, 20/200	ND	+	−	−	−	Cataracts	ND	ND	ND	ND	ND
II-5	44	Early childhood	20/63, 20/100	Sensory	+	20	−	Limb girdle weakness	−	Slight atrophy of occipital cortex	ND	ND	ND	ND
II-7	34	Early childhood	20/200, 20/200	Sensory motor	+	14	−	−	−	Unremarkable	ND	ND	ND	ND
II-8	40	Early childhood	20/63, 20/63	Sensory	+	30	+	Limb girdle weakness	−	Unremarkable	No	Lipid accumulation, RRFs/COX	Yes	No
II-9	38	−	−	Sensory motor	−	−	−	−	−	ND	ND	ND	ND	ND
II-11	34	Early childhood	20/32, 20/32	ND	+	20	−	−	−	ND	ND	ND	ND	ND
II-12	26	Early childhood	?	Sensory	+	10	−	−	−	Unremarkable	ND	ND	ND	ND
III-7	19	Early childhood	20/50, 20/50	ND	−	−	−	−	−	ND	ND	ND	ND	ND
III-14	16	Early childhood	20/200, 20/200	ND	−	−	−	−	Learning difficulties	Unremarkable	No	Non-specific	No	No
III-15	14	8	20/200, 20/200	ND	−	−	−	−	Learning difficulties	Unremarkable	ND	ND	ND	ND
III-16	6	−	−	ND	−	1,5	−	−	Psychomotor regression, spacticity	Periventricular and subcortical leucodystrophy	Complex II deficiency	Lipid accumulation, decrease of SDH activity	No	No

COX = cytochrome c oxidase; mtDNA = mitochondrial DNA; ND = not done; OXPHOS = oxidative phosphorylation; RRFs = ragged-red fibres; SDH = succinate dehydrogenase; ? = unknown.

Patient I-1 had 13 children among whom six had an early onset bilateral optic atrophy confirmed by an ophthalmological examination (Patients II-4, II-5, II-7, II-8, II-11 and II-12). Visual acuities were from 20/32 to 20/200 (Table 1). Patient II-5 had colour vision deficiency with tritanopia, and fundus examination showed a bilateral temporal optic atrophy. Visual field test revealed a bilateral central scotoma. All patients complained of neuromuscular symptoms, with an onset between 10 and 30 years of age, and had clinical signs of sensory ataxic neuropathy. In the following generation, three children (Patients III-7, III-14 and III-15) presented with an optic atrophy confirmed by fundus examination. Patient III-14 had a temporal atrophy. Neurological examination of Patients III-14 and III-15 was normal. Patient III-16 was born at term, eutrophic, after an uncomplicated pregnancy. Psychomotor development was initially normal but deteriorated at the age of 18 months with loss of walking and sitting posture. At 2 years of age, clinical examination showed lower limbs and axial hypertonia. Laboratory investigations revealed an increase in plasma lactate concentrations (3.4 mM/l, normal <2.1 mM/l) with a high lactate/pyruvate ratio (lactate/pyruvate ratio = 42, normal <21). After the initial regression, the child progressed slowly but developed a spastic paraparesis. At 6 years of age, he was able to say a few words and to walk with help.

Oxidative phosphorylation spectrophotometric measurements

Enzymatic spectrophotometric measurements of the oxidative phosphorylation respiratory chain complexes and citrate synthase were performed at 37°C on muscle crude homogenates and fibroblasts according to standard procedures (Rustin et al., 1994). Proteins were measured according to Bradford microassay (Bradford, 1976) and results were expressed as nM/min/mg of proteins.

Muscle histopathology and ultrastructure

Muscle samples were frozen in cooled isopentane and stored in liquid nitrogen for histological and histo-enzymatic analysis including Gomori modified trichrome staining, COX activity, succinate dehydrogenase activity and double COX/succinate dehydrogenase staining according to standard protocols. A fragment of muscle was also fixed in 2% glutaraldehyde and processed for ultrastructural analysis by electron microscopy.

Mitochondrial DNA molecular analysis

Total DNA was extracted using standard phenol chloroform procedure. Long-range polymerase chain reaction (PCR) and Southern blot analysis were performed as previously described (Moraes et al., 1989; Paul et al., 1996). Mitochondrial DNA quantification in muscle was performed by real-time quantitative PCR as described by Rouzier et al. (2010). Primer sequences and PCR conditions are available on request. The relative amount of deleted mitochondrial DNA was determined using real-time fluorescence PCR (He et al., 2002). For DNA damage and repair assay, high molecular weight DNA was isolated from fibroblasts by using a QIAGEN Genomic-tip 20/G kit as described by the manufacturer.

Sequencing of nuclear genes

The coding regions of POLG1 (NM_002693.2), SLC25A4 (ANT1) (NM_001151.3) and PEO1 (Twinkle) (NM_021830.3) genes were sequenced as previously described (Naimi et al., 2006). Sequencing of the OPA1 gene (NM_130833.1) was performed as previously reported (Amati-Bonneau et al., 2008). For POLG2 (NM_007215.3), RRM2B (NM_015713.3) and MFN2 (NM_014874.3) analysis, all exons were amplified with intronic primers, which are available on request. PCR products were purified with ExoSAP-IT enzyme (USB), processed with an ABI PRISM® dRhodamine Terminator Cycle Sequencing Ready Reaction kit (Applied Biosystems) and analysed on an ABI 3130 automated sequencer (Applied Biosystems).

Cell culture

Skin punches were obtained from Patients II-5 and II-8 after informed consent. Primary fibroblast cultures were established using standard procedures in RPMI supplemented with 10% foetal bovine serum, 45 µg/ml uridine and 275 µg/ml sodium pyruvate. Cultures were incubated at 37°C with 5% CO₂.

Polarographic study

Polarographic studies of intact cell respiration and of mitochondrial substrate oxidation by digitonin (0.004%)-permeabilized cells (30–50 µg protein) were carried out in a 250 µl chamber equipped with a Clark electrode (Hansatech Instruments Ltd) containing 0.3 M mannitol, 10 mM KCl, 5 mM MgCl₂, 1 mg/ml bovine serum albumin and 10 mM KH₂PO₄ (pH 7.4) (Rustin et al., 1994).

Cell viability assay

Primary fibroblasts were grown in Dulbecco's modified eagle medium (DMEM) supplemented with 10% foetal bovine serum, 4.5 g/l glucose, 584 mg/l l-glutamine, 1% penicillin/streptomycin (DMEM-glucose) (Lonza) or in glucose-free DMEM supplemented with 10% foetal bovine serum, 5 mM galactose, 584 mg/l l-glutamine and 5 mM sodium pyruvate (DMEM-galactose) (Invitrogen).

Twelve-well culture plates were seeded with a constant number of cells. Twenty-four hours later (time zero), cells were detached by trypsin and counted using a haemocytometer. The value at time zero was considered the 100% value of the number of viable cells. Culture medium was replaced by DMEM-glucose or DMEM-galactose for different times before new cell counts.

Deconvolution microscopy and mitochondrial network analysis

Mitochondria were labelled using MitoTracker® Green 100 nM (Molecular Probes). Images were acquired with an inverted wide-field Leica (DMI6000B, Microsystems) equipped with a Roper CoolSnap HQ2 camera (Roper Scientific), a high-sensitivity CCD camera for quantitative fluorescence microscopy. Metamorph® 7.7 software (Molecular Devices) was used for image acquisition and Huygens software (Scientific Volume Imaging) for deconvolution. Imaris 7.1.1® software (Bitplane) was used for 3D processing and morphometric analysis.

Western blotting

Total protein extracts (5–20 µg) were separated on acrylamide–sodium dodecyl sulphate gels and transferred to polyvinylidene fluoride membranes (Millipore). Specific proteins were detected by using anti-Mfn1 (1:1000 dilution; Tebu-Bio, #H00055669-A01), anti-Mfn2 (1:4000 dilution; Abcam, #56889), anti-OPA1 (1:4000 dilution; BD Biosciences, #612606), anti-β-tubulin (1:10 000 dilution; Sigma-Aldrich, #T4026) and anti-porin antibodies (1:10 000 dilution; Mitosciences Euromedex, #MSA03). Anti-mouse secondary antibody (Dako, #P0447) was used at 1:10 000 and signals were detected using an Enhanced Chemiluminescence system (ECL Plus, Amersham).

Treatment of cells with hydrogen peroxide

Primary fibroblasts were plated in triplicate in 100-mm diameter culture sterile dishes at 60–70% confluence 20 h before treatment. H₂O₂ (30%, Sigma Aldrich) was diluted into phosphate buffered saline and the concentration was determined by absorbance at 260 nm as described (Shull et al., 1991). Monolayer cultures were exposed to 150 µM H₂O₂ for 30 min at 37°C in serum-free medium. Medium culture was replaced by complete medium culture and incubated for the indicated times. Control monolayers were mock-treated with corresponding serum-free medium alone (Yakes and Van Houten, 1997). Before genomic DNA extraction, medium culture was removed and plates were rapidly frozen in liquid nitrogen and stored at −80°C.

DNA damage and repair assays

Measurement of nuclear and mitochondrial DNA damage was performed using quantitative PCR that amplifies long DNA targets. The assay was performed essentially as described previously (Santos et al., 2006; Hunter et al., 2010). DNA lesions, including H₂O₂-induced damage, block the progression of the Taq DNA polymerase resulting in a decreased amplification of a target sequence (Yakes and Van Houten, 1997; Ballinger et al., 1999). For the PCR to be quantitative, it is necessary to amplify a given amount of template at a cycle number that is within exponential range of the PCR. Therefore, for each sample, three separate PCR reactions were performed with three different cycle numbers to determine quantitative conditions. PCR conditions are available on request and the following primers were used to amplify a 15.6-kb fragment of human mitochondrial DNA (sense primer 5′-CCC ACA GTT TAT GTA GCT TAC CTC CTC A-3′ and reverse primer 5′-TTG ATT GCT GTA CTT GCT TGT AAG CAT G-3′) and a 10 kb fragment of the human nuclear HPRT gene (sense primer 5′-GAA CGT CTT GCT CGA GAT GTG ATG AAG GAG-3′ and reverse primer 5′-TCT CCC ACC CAT ACT GGC AAA ACT TAA GCC-3′).

PCR quantification was normalized with a small target PCR not affected by H₂O₂ treatment. A human 172 bp mitochondrial DNA fragment was amplified using a sense primer (5′-GAA TTG TGT AGG CGA ATA GG-3′) and a reverse primer (5′-CTA CAC AAT CAA AGA CGC CC-3′). Primers used for the small nuclear HPRT amplicon (141 bp) were: 5′-TCA CAT TGT AGC CCT CTG TG-3′ (sense primer) and 5′-ACA CAA TAG CTC TTC AGT CTG-3′ (reverse primer). All PCR reactions were quantified using Quant-iT™ PicoGreen dsDNA reagent (Invitrogen), which fluoresces upon binding to DNA. Five microlitres of each PCR reaction were diluted in 1× Tris–EDTA buffer (45 µl) containing 1× Quant-iT™ PicoGreen dsDNA reagent. Standard curve was performed as described by the manufacturer. Fluorescence quantification was performed with a LightCycler® LC480 apparatus. Relative PCR products of mitochondrial DNA and nuclear DNA were normalized to mitochondrial DNA copy number and total amount of nuclear DNA, respectively. The relative PCR product, which represents the relative level of oxidative DNA damage, was calculated by dividing the raw fluorescence value of a H₂O₂-treated sample by that of the corresponding H₂O₂-untreated sample.

Results

Mitochondrial myopathy with multiple mitochondrial DNA deletions

Muscle biopsy was performed in four patients after informed consent (Patients I-1, II-8, III-14 and III-16). Muscle analysis of the index case (Patient I-1) showed typical features of mitochondrial myopathy including intracellular lipid accumulation with numerous ragged-red and COX-negative fibres (45%) (Fig. 3A and B). Electron microscopy showed altered morphology of mitochondria and cristae organization, and paracristalline inclusions (Fig. 3C). Muscle biopsy of Patient II-8 revealed similar findings with 20% of COX-deficient fibres. In the following generation, muscle analysis of Patient III-14, a 16-year-old, revealed non-specific histological changes such as a marked variability of fibre size. Muscle analysis of Patient III-16, a 4-year-old, showed lipid accumulation and increased mitochondrial contingent with a major decrease of succinate dehydrogenase activity (not shown). Biochemical investigation revealed a respiratory chain deficiency in Patient III-16 (Table 1). The two older subjects (Patients I-1 and II-8) carried multiple mitochondrial DNA deletions in muscle identified by both long range PCR and Southern blot analysis (Fig. 3D and E). The measurement of the relative amount of deleted mitochondrial DNA, using real-time fluorescence PCR, revealed 49% of deletions in Patient I-1 and 23% in Patient II-8 (He et al., 2002). Multiple mitochondrial DNA deletions were not found in the two younger subjects (Patients III-14 and III-16). The determination of relative mitochondrial DNA copy number was performed by real-time quantitative PCR in muscles of the four patients without finding any depletion (Table 1).

Figure 3

Muscle analysis from Patient I-1. (A and B) Histopathology with Gomori modified trichrome (A) showing ragged-red fibres and COX/succinate dehydrogenase stain and (B) revealing COX-deficient fibres, which are recognized by the prevalent blue stain. (C) Ultrastructure of skeletal muscle showing accumulation of lipid droplets and abnormal enlarged mitochondria with paracristallin inclusions. Original magnification: ×12 000. (D and E) Molecular analysis. Long-range PCR (D) and Southern blot (E) analysis revealing multiple deletion bands in addition to wild-type fragments. C = control individual; M = lambda HindIII/EcoRI (left) and 1-kb DNA ladder (right); mtDNA = mitochondrial DNA; P = patient.

Identification of a novel mutation in the MFN2 gene

Analysis of genes involved in multiple mitochondrial DNA deletions (POLG1, POLG2, SLC25A4, PEO1, RRM2B and OPA1) revealed no mutation. The link between mitochondrial deficient fusion and mitochondrial DNA instability, and the recent identification of MFN2 mutations in patients with Charcot–Marie–Tooth disease type 2A and optic atrophy prompted us to analyse this gene. We identified a novel heterozygous missense mutation: c.629A>T (p.D210V) in the index case (Patient I-1). The identified mutation co-segregated with the disease and was found in all affected individuals (Fig. 1). The neurological examination of Patient II-9, who carries the mutation, was normal. Nevertheless, electrophysiological study revealed a sensory motor neuropathy (Table 1). The mutation changes a highly conserved Asp into a non-polar Val and was not present in 400 healthy control chromosomes (Fig. 4). The affected amino acid is located within the GTPase domain of Mfn2, which is essential for the function of mitofusins (Koshiba et al., 2004). Last, in silico study by Polyphen predicted this variant to be probably damaging (http://genetics.bwh.harvard.edu/pph/).

Figure 4

Identification of a novel MFN2 mutation. (A) Sequence analysis of MFN2 showing the c.629A>T (p.D210V) mutation (arrow). (B) Alignments of 10 vertebrate amino acid sequences of Mfn2 including Asp in position 210 (red rectangle).

Respiratory chain deficiency and fragmentation of the mitochondrial network in patient fibroblasts

We analysed fibroblasts of two affected individuals bearing the MFN2 c.629A>T mutation (Patients II-5 and II-8). Spectrophotometric analysis revealed a complex IV deficiency in both cases (Table 2). To analyse cell respiration, we measured basal rates of oxygen consumption and mitochondrial substrate oxidation. All factors tested were enhanced for both patients (Table 3). The increase of the oxygen consumption, which occurs under conditions in which mitochondrial mass is unaltered, has previously been reported in fibroblasts of patients with Charcot–Marie–Tooth disease type 2A bearing R364Q or A166T mutations (Loiseau et al., 2007). Multiple mitochondrial DNA deletions were not observed and the determination of relative mitochondrial DNA copy number was performed by real-time quantitative PCR without finding any depletion (not shown).

Table 2

Enzymatic analysis of cultured skin fibroblasts

OXPHOS activity
OXPHOS complexes	I	II	III	IV	V	CS
Control values (nM/min/mg of proteins)	9.0–27.1	18.5–35.0	57.4–176.2	109.9–252.7	22.0–46.2	74.7–161.1
Patient II-5	17.2	26.2	87.7	45.8	24.3	85.7
Patient II-8	15.9	22.9	196.9	57.9	27.2	118.1

Ratios	IV / I	IV / II	IV / III	IV / V	IV / CS
Control values	5.02–21.78	3.09–9.60	0.98–3.76	2.83–7.98	1.04–2.71
Patient II-5	2.66	1.74	0.52	1.88	0.53
Patient II-8	3.64	2.52	0.29	2.12	0.49

Spectrophotometric analysis of the respiratory chain enzyme activities. Results are expressed as extreme absolute values or absolute values for controls or patients, respectively.

Lowered values are shown in bold.

OXPHOS = oxidative phosphorylation.

Table 3

Polarographic analysis of the respiratory chain

	Respiration	Substrate oxidation
		Malate	Succinate	G3P
Control values (nM O2/min/mg of proteins)	5.90–13.80	7.90–16.60	7.50–15.80	4.90–13.50
Patient II-5	19.59	18.66	22.20	16.97
Patient II-8	17.62	15.52	19.39	15.86

Results are expressed as extreme absolute values or absolute values for controls or patients, respectively.

The growth of control and patient fibroblasts was not significantly different in glucose medium. In a glucose-free medium containing galactose, cells are forced to rely predominantly on oxidative phosphorylation for ATP production because the carbon source feeds the glycolytic pathway with a low efficiency. Cells with severe respiratory chain defects fail to survive under these conditions while fibroblasts with milder deficiency are able to grow (Robinson et al., 1992). The growth in galactose medium was significantly slower than in glucose medium for both control and patient fibroblasts. Again, in galactose medium no significant difference was observed between controls and patients (not shown). The same results have been reported with autosomal dominant optic atrophy fibroblasts bearing OPA1 mutations (Zanna et al., 2008).

Considering that Mfn2 function is important for mitochondrial network organization, we compared the mitochondrial morphology of fibroblasts carrying the MFN2 mutation with the one of fibroblasts obtained from normal individuals. After staining with MitoTracker® and examination by fluorescence microscopy, control fibroblasts displayed a typical filamentous interconnected network (Fig. 5A). The MFN2 mutant fibroblasts presented with a fragmentation of the mitochondrial network and less connected mitochondria (Fig. 5B and C).

Figure 5

MFN2 mutation affects the mitochondrial network in patient fibroblasts. Representative images are shown. A colour code highlights the connectivity of the mitochondrial network. (A) Control individual. (B) Patient II-5. (C) Patient II-8. (D) The diagram bar shows the distribution of mitochondrial length per cell in two independent experiments. Results are the mean of at least 40 images.

Decreased expression of Mfn2 protein in patient fibroblasts

We analysed the expression of Mfn2, Mfn1 and OPA1, the main proteins involved in mitochondrial fusion, in fibroblasts lysates by western blotting. We compared the expression of these proteins in patient fibroblasts with those observed in controls and in one patient with Charcot–Marie–Tooth disease type 2A bearing the A166T mutation (Loiseau et al., 2007). We used β-tubulin and the mitochondrial porin protein as controls for quantitation (Fig. 6A). Normalization relative to porin showed a significant reduction of Mfn2 expression in patient fibroblasts compared to controls. Steady-state levels of Mfn1 and OPA1 proteins were similar in total cell extracts from controls and from all patient fibroblasts (Fig. 6B).

Figure 6

Expression level of OPA1, Mfn2 and Mfn1 proteins in fibroblasts from controls and patients carrying the D210V (Patients II-5 and II-8) or the A166T alleles. (A) Representative western blot of OPA1, Mfn2, Mfn1, β-tubulin and porin performed with fibroblast lysates obtained from controls and from patients carrying the indicated MFN2 mutations. (B) OPA1, Mfn2 and Mfn1 bands normalized to porin band intensity. Results are means ± SEM obtained from two independent experiments and resolved in at least three blots (***P < 0.01).

Mitochondrial DNA repair deficiency in MFN2 mutant fibroblasts after oxidative stress

It has been shown that mitochondrial fusion is required for mitochondrial DNA stability (Chen and Chan, 2010). We asked whether Mfn2 deficiency could influence mitochondrial DNA repair after oxidative stress because one cause of mitochondrial DNA instability could be a failure to repair damaged DNA. Fibroblasts from the control individual and Patients II-5 and II-8 were treated with H₂O₂ to introduce oxidative DNA lesions, followed by incubation to allow DNA repair. Lesions in mitochondrial DNA or the nucleus were assessed by a gene-specific quantitative PCR-based assay in which base lesions, abasic sites or strand breaks interfere with the amplification of long DNA targets. This assay has proven particularly useful in examining mitochondrial DNA damage and repair kinetics after H₂O₂ treatment (Santos et al., 2006). Long PCR products (15.6 kb for mitochondrial DNA and 10 kb for the nuclear HPRT gene) were used and normalization was performed with shorter control amplicons (172 and 141 bp, respectively). Control fibroblasts showed a capacity to repair stress-induced DNA lesions, which leads to a 70% recovery 4 h after H₂O₂ treatment while fibroblasts of patients showed a severe mitochondrial DNA repair deficiency with <20% of recovery 4 h after treatment (Fig. 7A). As expected, we observed a low sensitivity to H₂O₂ damage of nuclear DNA compared to mitochondrial DNA (Fig. 7B). This result is consistent with the higher degree of protection and the more efficient repair options that exist in the nucleus (Santos et al., 2003). We did not observe any difference in the amount of oxidative nuclear damage between patients and control immediately after H₂O₂ treatment and at later steps. Taken together, these results suggest that MFN2, through mitochondrial fusion, contributes to the stability of the mitochondrial genome by supporting its repair after oxidative stress.

Figure 7

MFN2 mutation affects mitochondrial DNA repair following DNA damage induced by oxidative stress. (A) Mitochondrial DNA (mtDNA) repair activity after H₂O₂-induced DNA damage in control and patient fibroblasts. Long-range PCR was used to evaluate the oxidative damage, induced by H₂O₂ treatment, in mitochondrial DNA. The relative PCR amplification of a 15.6-kb mitochondrial DNA fragment was normalized to mitochondrial DNA copy number that was evaluated by PCR amplification of a 172-bp mitochondrial DNA fragment. (B) Nuclear DNA repair activity after H₂O₂-induced DNA damage in control and patient fibroblasts. Long-range PCR was used to evaluate the oxidative damage, induced by H₂O₂ treatment, in nuclear DNA. The relative PCR amplification of a 10-kb fragment of the HPRT gene was normalized to a 141-bp HPRT fragment. (A and B) Cells were exposed to 150 µM H₂O₂ for 30 min and either harvested immediately or allowed to recover in conditioned medium for the indicated times. Results represent the mean of relative PCR amplification ± SD of two independent experiments in which three PCRs per point were performed. Values were normalized to untreated cells and differences were analysed by Student's t-test (*P < 0.05). NT = non-H₂O₂-treated cells.

Discussion

In this study, we identified MFN2 as a new gene involved in mitochondrial DNA instability diseases. We describe a large family in which affected individuals carry a novel MFN2 mutation responsible for multiple mitochondrial DNA deletions in skeletal muscle. The clinical phenotype associated with this MFN2 mutation is unusual because patients present with optic atrophy in early childhood, associated with axonal neuropathy and mitochondrial myopathy in adult life. MFN2 mutations cause Charcot–Marie–Tooth disease type 2A, which is designated as hereditary motor and sensory neuropathy type VI (HMSN VI) when it is associated with visual impairment due to optic atrophy (Zuchner et al., 2006). Nevertheless, optic atrophy has never been reported as the main clinical feature associated with a MFN2 mutation despite the large inter- and intrafamilial clinical variability that characterizes Charcot–Marie–Tooth disease type 2A phenotypes (Chung et al., 2006; Verhoven et al., 2006; Del Bo et al., 2008). In the present family, the clinical presentation is close to autosomal dominant optic atrophy ‘plus’ phenotypes associated with OPA1 mutations, which is associated with, at the least, optic atrophy and muscular involvement ranging from non-specific myopathy to mitochondrial myopathy with COX-negative and ragged red fibres (Amati-Bonneau et al., 2008; Hudson et al., 2008). CNS and PNS can also be involved. Contrary to some of the patients with autosomal dominant optic atrophy, it is interesting to note that, in our family, none of the affected individuals presented with an external ophthalmoplegia. So far, OPA1 was the only gene involved in autosomal dominant optic atrophy ‘plus’ phenotypes with multiple mitochondrial DNA deletions. However, no mutation has been found in OPA1 and in other genes involved in mitochondrial DNA instability. By sequencing the MFN2 gene, we identified a novel heterozygous missense mutation, c.629A>T (p.D210V) located within the GTPase domain of Mfn2. Mutations in the GTPase domain of Fzo, the yeast orthologue of mitofusin 1 and Mfn2, inhibit mitochondrial fusion (Griffin and Chan, 2006) and the majority of missense mutations found in MFN2 in patients with Charcot–Marie–Tooth disease type 2A reside in the highly conserved GTPase domains (http://www.molgen.ua.ac.be/CMTMutations/Mutations/Default.cfm) (Cartoni and Martinou, 2009). The localization of p.D210V, the amino acid conservation, in silico predictions and the absence of this variant in 200 control individuals are strong arguments for its pathogenicity. Familial segregation is also in favour of the deleterious nature of p.D210V. A single individual bearing the MFN2 mutation (Patient II-9) had a normal clinical examination. However, this female, who has an affected child, presented with an axonal neuropathy at electrophysiology. This observation is not surprising since in some families, up to 25% of individuals harbouring MFN2 mutations may present only subclinical symptoms (Lawson et al., 2005). Another subject (Patient III-16), who carries the MFN2 mutation, presented with an unusually severe phenotype. Although we excluded mutations in nuclear genes involved in mitochondrial DNA multiple deletion syndromes, it is possible that a variant in an unknown modifier gene could explain the severity of the phenotype presented by this child.

Fusion controls mitochondrion morphology and fusion deficiency leads to severe defects in cell respiration and neurodegenerative diseases (Knott et al., 2008; Liesa et al., 2009). The association between inactivation of mitofusins and muscular disorders has been recently demonstrated by Chen et al. (2010) in a mouse model. Conditional inactivation of Mfn1 and Mfn2 in muscle causes myopathy with severe mitochondrial dysfunction, abnormal mitochondria and compensatory mitochondrial proliferation. Mutant mice have both quantitative and qualitative abnormalities of mitochondrial DNA (depletion and accumulation of deletions and point mutations). Our observation confirms that MFN2 mutations can be responsible for mitochondrial myopathy with mitochondrial DNA instability in humans. The accumulation of mitochondrial DNA deletions in skeletal muscle can be secondary to an increase in mitochondrial DNA damage, a defect in mitochondrial DNA repair and/or a failure to clear mitochondria with damaged DNA (Chen et al., 2010). By analysing the mitochondrial proteome of mutant mice, Chen et al. (2010) found that a lack of Mfn proteins causes protein heterogeneity from one mitochondrion to another and this might be the origin of mitochondrial DNA instability (Chen and Chan, 2010; Chen et al., 2010). If protein stoichiometries are improperly balanced, protein complexes involved in mitochondrial DNA maintenance, replication, repair or clearance may operate in an ineffective way. Results obtained with fibroblasts bearing the MFN2 p.D210V mutation are in favour of this hypothesis. We show for the first time that fibroblasts presenting with a reduced mitochondrial fusion also have a lower capacity to repair stress-induced mitochondrial DNA lesions compared to control cells. Further analyses will be necessary to determine whether these fibroblasts also show an increase in mitochondrial DNA damage and a clearance deficiency. But it is likely that the defect in mitochondrial DNA repair that we observed is due to variability in repair protein content across the mitochondrial population, and contributes to mitochondrial DNA instability.

Prior to our study, mitochondrial function had been examined in fibroblasts carrying MFN2 mutations, including point mutations in the GTPase domain (Loiseau et al., 2007; Amiott et al., 2008). These studies reported normal respiratory complex activity, mitochondrial DNA content and mitochondrial morphology in patient fibroblasts. This observation suggested that these cells express sufficient levels of Mfn1 to allow the formation of fusion-competent complexes. Indeed, it has been shown by using Mfn1^−/− and Mfn2^−/− cells that many Mfn2^{Charcot–Marie–Tooth disease type 2A} alleles promote mitochondrial fusion in Mfn2^−/− cells but are non-functional in Mfn1^−/− cells (Detmer and Chan, 2007). These results suggest that in most tissues of patients with Charcot–Marie–Tooth disease type 2A, mitochondrial fusion is supported by Mfn1/Mfn1 and Mfn1/Mfn2^{Charcot–Marie–Tooth disease type 2A} hetero-oligomeric complexes. However, in tissues with low Mfn1 expression, mitochondrial fusion would be essentially abolished. This hypothesis might explain the neuronal specificity of the disease. In this report, the situation is different because fibroblast analysis revealed a mitochondrial fusion deficiency with fragmented mitochondria leading to respiratory chain and mitochondrial DNA repair defects. This difference is difficult to explain at the moment but in fibroblasts of both patients we observed a lower level of Mfn2 expression compared to controls. In contrast, steady-state levels of Mfn2 protein seem to be similar in control cells and in Mfn2^{Charcot–Marie–Tooth disease type 2A} fibroblasts carrying classic mutations such as the mutation A166T and others (Amiott et al., 2008). The reduction of Mfn2 expression found in p.D210V fibroblasts could be involved in the observed phenotype. However, although most of the MFN2 mutations are missense, several mutations predicted to generate a truncated protein responsible for haploinsufficiency have been described and are associated with a ‘classical’ Charcot–Marie–Tooth disease type 2A phenotype (Liesa et al., 2009). Our results suggest that the observed effects are mutation-dependent and it would be interesting to test the role of the p.D210V mutant in a Mfn1/Mfn2 null cell system (Detmer and Chan, 2007).

As regards the consequences on the respiratory chain, the three primary Leber's hereditary optic neuropathy mutations involve complex I subunits and result in a complex I defect (Chevrollier et al., 2008; Yu-Wai-Man et al., 2011). Recently, it has been shown that complex I also seems to play a major role in the respiratory chain impairment found in fibroblasts from patients bearing different OPA1 mutations (Zanna et al., 2008). This finding establishes a supplementary link between these mitochondrial diseases characterized by a selective degeneration of retinal ganglion cells. Nevertheless, impairment in mitochondrial ATP synthesis driven by complex I substrates observed in OPA1 deficient fibroblasts is difficult to explain. A complex IV deficiency with a normal complex I activity has also been described in fibroblasts of patients carrying OPA1 or OPA3 mutations (Chevrollier et al., 2008). These results are similar to those observed with the MFN2 mutation segregating in our family. It is likely that oxidative phosphorylation complex activities vary according to patients with autosomal dominant optic atrophy and could be also modulated by modifier genes. Further work will be necessary to understand the bioenergetic consequences of both MFN2 and OPA1 mutations.

In conclusion, this observation makes MFN2 the sixth gene associated with ‘mitochondrial DNA breakage’ syndromes. The similarity of symptoms caused by MFN2 and OPA1 mutations supported the idea that these proteins are functionally similar and this is confirmed by our observation, which indicates that mutations in both genes can lead to mitochondrial DNA instability. In clinical practice, MFN2 has to be tested in autosomal dominant optic atrophy phenotypes and in patients carrying multiple mitochondrial DNA deletions.

Funding

This work was made possible by grants to V.P.-F. from the Association Française contre les Myopathies (AFM).

Acknowledgements

We thank Alexia Figueroa for technical help, Drs Robert Verdet, Guy Balmeyer and Prof. Christian Hamel for ophthalmological expertise.

Abbreviation

Abbreviation
• COX

  cytochrome c oxidase

• Mfn

  mitofusin

• OPA

  optic atrophy

• PCR

  polymerase chain reaction