Neuropathologic abnormalities can be sufficiently characteristic to suggest the genetic basis of some hereditary neuropathies such as those associated with mutations in MPZ, GJB1, GDAP1, MTMR2, SH3TC2, PRX, FGD4, and LMNA. We analyzed the morphologic features of 9 sural nerve biopsies from 6 patients with mutations of mitofusin 2. All patients presented in early childhood with axonal neuropathies designated as mild or severe motor and sensory neuropathy. In all cases, there was a marked decrease in density of myelinated fibers, mainly of large diameter fibers. These changes were more marked in the second biopsies of 3 patients that were performed from 7 to 19 years after the first biopsies. Neurophysiologic findings were most suggestive of axonal degeneration, but some onion bulbs were present in all cases. Axonal mitochondria were smaller than normal, were round, and were abnormally aggregated. These changes may result from abnormal mitochondrial fusion and fission. The results suggest that these clinical and pathological features may be sufficiently characteristic to suggest the diagnosis of mitofusin 2-related neuropathy.
A clinical syndrome of severe hereditary motor and sensory neuropathy of neuronal type with onset in early childhood (EOHMSN) was first reported by Ouvrier et al (1) and Gabreels-Festen et al (2). The usual clinical phenotype of this disorder is onset in the first 5 years of life with rapid progression such that most patients are almost completely paralyzed below the knees and elbows by their teens. All patients described had a severe axonal neuropathy by electrophysiological and/or neuropathologic criteria.
Six of 8 typical cases of EOHMSN from Sydney (including case 7 of Ouvrier et al ) have been found to carry mutations in the gene that encodes the mitochondrial fusion protein mitofusin 2, MFN2. Mutations in this gene have more commonly been associated with Charcot-Marie-Tooth (CMT) type 2A, which usually has an onset after 5 years of age (3-6). There have been relatively few previous reports of the pathological findings in nerves in MFN2-related neuropathies (4, 5). We present the histologic findings in sural nerve biopsy in 6 cases.
Materials and Methods
All subjects presented with an early-childhood neuropathy, which was designated mild (case 1) or severe (cases 2-6). All were sporadic cases. The screening of the MFN2 gene for mutations was performed as previously described (7). Two subjects carried compound heterozygous (recessive) mutations (cases 4 and 6; Table 1). Three had only 1 (dominant) mutation (cases 2, 3, and 5). In case 2, the clinical and neurophysiologic examinations of both parents and 2 siblings were normal. The child with a milder clinical phenotype (case 1) had a homozygous MFN2 mutation.
Pathological examination in all 6 cases included light and electron microscopic (EM) analysis of sural nerve, which was biopsied after informed consent of the parents. A second biopsy was performed on cases 1, 5, and 6. Standard techniques were used on paraffin sections. A portion of each biopsy was fixed in buffered glutaraldehyde, embedded in Epon, and prepared for light (semi-thin sections) and EM examinations. To study axonal mitochondria, several longitudinal sections of each nerve biopsy were systematically examined by EM and compared with longitudinal sections of 15 normal and abnormal sural nerves selected from among biopsies performed in our department during the past 30 years. From adults, there were 4 normal nerves, 3 LMNA-related axonal HMSN, 3 chronic inflammatory demyelinating polyneuropathies, 1 polyneuropathy associated with renal insufficiency, and 1 axonal polyneuropathy of unknown cause. From children, there were 3 normal nerves from patients aged 8 months, 5 years, and 9 years.
To determine myelinated fiber densities, photomicrographs of transverse sections of the nerves were taken by EM at a magnification of 2,000×. Negatives were directly scanned onto a computer and quantitatively analyzed using the Matrox Inspector software (Matrox, Montréal, Canada). A threshold was defined for each micrograph, leading to a binarized image. Myelinated fibers were automatically detected using the blob tool of the software, and the number of myelinated fibers per square millimeter was determined by the software. Unmyelinated fibers were identified by visual inspection, and their numbers per squared millimeter were determined by the Matrox Inspector software. Onion bulbs and regenerative clusters were identified by visual inspection of transversal sections of the biopsies. Onion bulbs were defined as one or more concentric layers of flattened Schwann cell processes. Regenerative clusters were defined as a group of 3 or more closely apposed small myelinated and unmyelinated axons.
Quantitative assessment of mitochondrial length in myelinated and unmyelinated fibers was performed on longitudinal sections of the nerves. The length or largest diameter of each visible mitochondrion was recorded in several axons using the Matrox Inspector software. Quantitative data from patients and controls were compared using the Mann-Whitney U test. More than 10 micrographs from each sample were analyzed. The ages of normal controls used for quantitative studies of fiber density and mitochondrial length were 5 years (used for the first biopsy of cases 1, 5, and 6 and for cases 3 and 4), 9 years (used for the second biopsy of cases 1 and 5), and 22 years (used for the second biopsy of case 6).
In all cases, there was a marked decrease in the density of myelinated fibers, mainly those with large diameters. The myelinated fiber densities ranged from 369/mm2 to 3890/mm2 (Table 2; Figs. 1-3). Serial biopsies in cases 1, 5, and 6 revealed a progressive decrease in density of myelinated fibers, with only small myelinated fibers remaining in the second biopsy. Large numbers of the remaining myelin sheaths were thinner than normal (Fig. 3). Occasional regenerative clusters were present, but there were fewer of them in the second biopsy in cases with serial biopsies; quantitative analysis of clusters did not show significant differences between cases and controls. All cases had neurophysiologic findings of axonal degeneration, but some onion bulbs were seen in all biopsies. These were usually around thinly myelinated axons (Fig. 3). In cases 3 and 4, these Schwann cell proliferations were found in 1.3% (3/231) and 3.8% (16/412) of myelinated fibers, respectively (Fig. 4). Loss of unmyelinated fibers was evident with some empty Schwann cell stacks, but overall densities of unmyelinated fibers were large in all patients (Table 2).
In all cases, unequivocal lesions of mitochondria were apparent only on longitudinal sections. Many axonal mitochondria appeared smaller than normal and had round or spherical instead of tubular shapes. They were abnormally aggregated and tended to accumulate at the periphery of the axons; this peripheral distribution was most clearly evident in residual large myelinated fibers (Fig. 5). The inner and outer mitochondrial membranes were irregular, and the cristae were often disrupted. No paracrystalline inclusions were noted. These abnormalities were observed in both myelinated and unmyelinated fibers (Fig. 6) and did not appear to be more frequent at the level of nodes of Ranvier. In all cases, some axons had a subpopulation of normal elongated and tubular mitochondria. There were no significant anomalies of mitochondria in Schwann cell, perineurial cell, or endothelial cell cytoplasm; other axonal components appeared normal.
The lengths of all visible axonal mitochondria were recorded in several axons from EOHMSN patients and compared with those in normal controls of similar ages. Individual data are presented in Table 3. In all cases, mitochondrial lengths were significantly reduced compared with those in controls. In nerve biopsies from adult patients with other conditions (see Materials and Methods section), mitochondrial lengths were preserved and were not significantly different from those in normal controls of similar ages when compared using the Mann-Whitney U test (data not shown).
Large numbers of myelinated fibers could be examined in longitudinal sections in cases 1, 3, 4, and 5 (77-106 fibers/patient). The proportion of fibers with round aggregated mitochondria was 13% in cases 1 and 3, 22% in case 4, and 19% in case 5. By contrast, occasional round and small mitochondria were seen in control nerves, but mitochondrial aggregation was never observed.
Three patients of this series (cases 2, 3, and 5) harbored heterozygous mutations of the MFN2 gene. Heterozygous MFN2 mutations have been found in 19% to 33% of families with autosomal dominant CMT2 (3-5). In addition, de novo heterozygous MFN2 mutations were found in several patients with sporadic early-onset peripheral neuropathies (4-6). On the other hand, cases 1, 4, and 6 in this study were found to harbor mutations on both alleles of the MFN2 gene. In 3 patients with apparently sporadic early-onset peripheral neuropathies and mutations on both MFN2 alleles (including cases 4 and 6 in this series), we observed that each mutation was inherited from an asymptomatic parent with no neurophysiologic evidence of peripheral nerve dysfunction; this indicates a recessive mode of inheritance (7). Therefore, cases with the clinical phenotype of EOHMSN may represent a subset of CMT type 2A with either dominant or recessive inheritance.
Histologic features of the biopsies were similar in all 6 cases of EOHMSN. In all biopsies, there was a marked loss of large myelinated fibers compared with age-matched controls, and all had occasional clusters of myelinated fibers. In 2 cases, there were prominent Schwann cell proliferations with onion bulbs. Such lesions have been previously described in MFN2-related neuropathies (5). It must be stressed, however, that most of these Schwann cell proliferations could only be detected by EM examination and were not easily seen on semi-thin sections. Recently, Lacour and colleagues (8) reported a case of MFN2-related neuropathy that, by electrophysiological and neuropathologic criteria, was found to be a demyelinating neuropathy. With respect to the distinction between axonal degeneration and demyelination, it should be noted that progressive axonal loss is a consistent feature in the course of any chronic progressive neuropathy. In severe forms of CMT with early onset, such as the EOHMSN syndrome, demyelinating lesions may not be recognized but may still contribute to the marked axonal loss. Because standard nerve conduction studies analyze large myelinated fibers, the marked loss of large myelinated fibers we observed could explain the electrophysiological findings suggestive of axonopathy. Electron microscopy is very useful for the detection of demyelinating lesions preferentially affecting small fibers. The unexpected observation of an increased number of unmyelinated fibers in all patients, particularly when a second biopsy had been performed, may indicate unsuccessful attempts to regenerate nerve fibers.
Previous studies have not emphasized mitochondrial abnormalities in MFN2-related CMT2A. Chung et al (5) reported findings in 7 nerve biopsies from patients with MFN2-related CMT2A but did not describe any ultrastructural abnormalities of mitochondria. Kijima et al (9) studied 2 nerve biopsies from a series of 7 cases but did not indicate whether mitochondria were analyzed by EM. Only Verhoeven et al (4) reported mitochondrial abnormalities in transverse EM sections of sural nerve biopsies from 2 children with MFN2-related CMT2A. They observed degenerative changes of axonal mitochondria, which appeared small and electron-dense in one patient and swollen in the other. Verhoeven et al also observed various abnormalities of mitochondrial cristae and, in 1 patient, a focal accumulation of degenerative mitochondria in axons. In their latter case, longitudinal sections were reported not to show any other abnormalities (in particular, the authors did not report any reduction of mitochondrial length). In our cases, most of the characteristic lesions observed were best detected on longitudinal sections and were not well seen on transverse sections. In longitudinal sections, the abnormal aggregation of these round mitochondria was very striking; in controls, the mitochondria appeared as randomly dispersed inside the axons. We also observed irregularities of the inner and outer mitochondrial membranes and frequent disruptions of mitochondrial cristae. Mitochondria are, however, known to be very fragile structures and very sensitive to fixation, and disruptions of cristae must therefore be interpreted with caution.
Mitofusin 2 is a large mitochondrial transmembrane GTPase with 2 coiled coil domains and 2 transmembrane spans. It is targeted to the outer mitochondrial membrane where it interacts with mitofusin 1 to regulate the mitochondrial network architecture by stimulating the fusion of mitochondria. Homo-oligomeric and hetero-oligomeric complexes formed by mitofusins 1 and 2 are essential for mitochondrial fusion; their terminal coil domain may function as a tether between adjacent mitochondria during the fusion process (10-12). A dynamic balance of fusion and fission events determines normal mitochondrial morphology and leads to the characteristic tubular appearance of mitochondria on longitudinal section of axons. In the peripheral axons of patients with MFN2 mutations, mitochondria were small and aggregated and spherical, rather than tubular in shape, abnormalities that suggest abnormal fusion secondary to the dysfunction of mitofusin 2. Some previous observations in cellular models of CMT2A have previously suggested such a dysfunction. For example, Baloh et al (13) expressed CMT2A-associated forms of mitofusin 2 in cultured dorsal root ganglion neurons and observed abnormal clustering of small fragmented mitochondria in both neuronal cell bodies and proximal axons. In embryonic fibroblasts from mitofusin 2-deficient mice, Chen et al (10) observed by EM that mitochondria appeared mostly as spheres and ovals of widely varying sizes. By contrast, no morphologic abnormalities of the mitochondrial network were observed by confocal microscopy in cultured skin fibroblasts from patients with MFN2-related CMT2A (14). One of the authors (J.-M.V.) studied these cultures by EM and was unable to find any mitochondrial anomalies (unpublished observations).
Recent data suggest that the fusion-fission process is not only important for the maintenance of normal mitochondrial morphology but also required to keep a functional mitochondrial population in the cell. In skin fibroblasts from patients with MFN2-related CMT2A, Loiseau et al (14) observed a significant coupling defect (resulting in a reduced efficacy of oxidative phosphorylation) and a reduction of the mitochondrial membrane potential. Embryonic fibroblasts deficient for both mitofusins 1 and 2 have decreased cellular respiration and widespread heterogeneity of mitochondrial membrane potential (15). In cultured dorsal root ganglion neurons expressing CMT2A-associated forms of mitofusin 2, Baloh et al (13) observed a markedly reduced axonal mitochondrial transport that was not attributable to diminished ATP levels in the neurons. Taken together, these experimental data suggest that the presence of an effective mitochondrial network may be required to maintain functional peripheral nerve axons.
From a genetic point of view, the observation of similar mitochondrial abnormalities in patients with either dominant or recessive MFN2 mutations suggests that most MFN2 missense mutations are responsible for a loss of function resulting in defective mitochondrial fusion. In subjects with dominant heterozygous MFN2 mutations, the presence of a “wild-type” mitofusin 2 is probably not able to compensate the defect due to the EOHMSN-associated mutation. A recent study showing that in cells deficient for either mitofusin 1 or 2, wild-type mitofusin 2 cannot complement mutant mitofusin 2, whereas wild-type mitofusin 1 can. In the former case, mitochondria remain fragmented, whereas in the latter case, there is a restoration of mitochondrial tubulation (16). Complementation by wild-type mitofusin 1 occurs through the formation of hetero-oligomeric complexes between mutant mitofusin 2 and wild-type mitofusin 1, including complexes that form in trans between mitochondria. By contrast, homo-oligomeric complexes formed by mutated and wild-type mitofusin 2 are nonfunctional for mitochondrial fusion and fail to restore mitochondrial tubulation. Detmer and Chan (12) speculated that peripheral nerves may contain little or no mitofusin 1 expression to compensate for mutant mitofusin 2. The resulting defects in mitochondrial dynamics coupled with the extreme length of these axons may lead to neuronal dysfunction and axon degeneration.
Characteristic ultrastructural lesions of nerve biopsies in some patients can be very helpful for suggesting particular mutations in several genes. For example, i) numerous large onion bulbs corresponding to the abnormal proliferation of Schwann cells are associated with PMP22 mutations; ii) numerous clusters constituted by small myelinated fibers with scattered onion bulbs are associated with GJB1 mutations; iii) many outfoldings of myelin lamellae are associated with MTMR2, MTMR13, and FGD4 mutations; iv) SH3TC2 mutations are associated with characteristic anomalies of the cytoplasm of the unmyelinated Schwann cells; v) PRX mutations are associated with subtle anomalies of the nodes of Ranvier, and vi) LMNA mutations are associated with severe rarefaction of myelinated fibers without any sign of regeneration (17, 18). We conclude from the present findings that in EOHMSN, ultrastructural anomalies of axonal mitochondria may be predictive of the presence of MFN2 mutations and, therefore, that MFN2 should be added to this list.
The authors thank Phillipe Sindou, Laurence Richard, and Martine Piaser for excellent technical assistance.
- motor and sensory neuropathies, hereditary
- myelin sheath
- retrograde degeneration
- sural nerve
- neurophysiology-biologic function
- neuropathy, sensory
- neuropathies, inherited
- axonal neuropathy
- sh3tc2 gene
- mfn2 gene
- lmna gene