Background
Charcot-Marie-Tooth (CMT) disease is a heterogeneous group of inherited peripheral neuropathies with an estimated prevalence of 1 in 2,500[
1]. At present 40 different genes has been identified to cause inherited peripheral neuropathies[
2]. CMT is subdivided into type 1 and 2, depending on whether the motor conduction velocity (MCV) is less or above 38 m/s[
3,
4]. A third form has intermediate MCV[
5]. Intermediate CMT is defined as MCVs between 25 and 45 m/s. Distal Hereditary Motor Neuropathy (dHMN) is the spinal form of CMT[
6].
Twelve different genes have been described to cause CMT2. The majority of the mutations are reported in the
mitofusin 2 (
MFN2) gene[
2]. This gene is responsible for 18-23% of those with CMT2, and is probably the most common cause of CMT2 [
7‐
9]. A
MFN2 mutation has also been described in a single family with intermediate CMT[
9].
Mitofusins are evolutionary conserved GTPases of the mitochondrial outer membrane, and has an essential role for the controlled fusion of the mitochondria membrane[
10]. The MFN2 protein spans the mitochondrial outer membrane with a large N-terminal and a relatively short C-terminal exposed to the cytosol[
11]. The
MFN2 mutation so far has almost exclusively been missense mutations that affect the N- or the C-terminal of the MFN2 protein[
2,
9].
We analyzed 232 Norwegian CMT families for mutations in the MFN2 gene.
Discussion
We screened 232 unselected and unrelated Norwegian CMT families with available DNA. A total of 7
MFN2 missense mutations were identified in 8 families (Table
1). The literature describes three nonsense mutations and one deletion, but missense mutations in far the most common types of mutations in the
MFN2 gene[
2,
9,
19].
We identified 4
de novo mutations in the
MFN2 gene, paternity could not be confirmed in two families (family 3 and 4) due to lack of paternal DNA. Our 25-50% (2-4/8) frequency of
de novo mutations in an population sample corresponds well with the 34% and possibly 55% frequency found in a major collaboration on
MFN2 mutations[
2]. Two of our
de novo missense mutation occurred in codon 94. The literature describes 5 other families with
de novo mutations in codon 94 with confirmed paternity[
9,
20]. These mutations have also been described in families with different ethnicity[
2]. This strongly suggest that codon 94 is a hot spot for mutations. The
de novo c.1403G>A mutation has previously been described in three CMT families, but it was also found in a healthy control[
2,
21,
22]. The c.2113G>A was found in two unrelated Norwegian families, and has also been described by others[
2,
21]. The c.1709A>G in exon 15 was a novel missense mutations and a
de novo mutation found in the proband. Mutations in exon 15 have not previously been described. The other novel missense mutations were found in probands from CMT families with several affected, but we only had DNA from a single first cousin whom also carried the mutation. We excluded duplication of PMP22 and point mutation in
PMP22,
Cx32,
MPZ,
SIMPLE and
EGR2 genes in the affected with
MFN2 missense mutations.
Previous mutations in the
MFN2 gene have all been assigned to a CMT2 phenotype, except from a single family that was assigned intermediate CMT[
2,
7‐
9,
19‐
21]. Additional symptoms such as ataxia, optic neuritis, optic atrophy, pyramidal signs, scoliosis and tremor have also been described[
2,
23‐
25].
Missense mutations in codon 94, i.e. amino acid change Arg94Trp and Arg94Gln cause intermediate CMT, CMT2 or CMT2 with additional features[
9]. The proband in the family with intermediate CMT had motor CV <38 m/s in all tested motor nerves including median nerve and should therefore be classified CMT1. However, his affected father had motor CV >38 m/s, and the family was classified intermediate CMT. The missense mutation Arg94Trp occurred de novo in our family 1. Since we had no neurophysiological data on the probands affected daughter, we classified the family CMT1 rather than the more infrequent intermediate CMT. The missense mutation in codon 468, i.e amino acid change Arg468His was found in a proband with CMT2 and her father with Parkinson disease and distal neuropathy indicating co-segregation as well as in 1 of 260 control chromosomes[
21]. A patient with a severe CMT carried two mutations, the Arg468His in
MFN2 and Gln95stop in
ganglioside-induced differentiation-associated protein 1 (
GDAP1)[
22]. Two family members carried exclusively the missense mutation in the
MFN2 that caused the amino acid change Arg468His. One had a mild axonal neuropathy at age 56 years, while the other was unaffected at age 18 years[
22]. Three other family members with the Gln95stop in
GDAP1 were unaffected. The proband in our family 3 with the Arg468His amino acid change had reduced motor CV, while 1 of 250 control chromosomes had the missense mutation carried by an unaffected woman age 25 years.
We analyzed all types of CMT families and found MFN2 missense mutation in CMT1, CMT2, intermediate CMT and dHMN patients.
The CV in the median motor nerve is used as reference for classifying CMT1, CMT2 and intermediate CMT [
3‐
5]. dHMN is defined with normal motor and sensory CV and chronic denervation on EMG[
6]. We do not have CV of the median motor nerve in the probands with CMT1. However, they all had reduced CV in motor nerves compatible with CMT1. The CV is equal in the ulnar and the median motor nerve. Thus, the proband with CV <38 m/s in the ulnar motor nerve was classified CMT1 (Family 1). We interpreted the marked reduction in motor CV of the tibial nerve as demyelination (Family 3). The slight reduction in motor CV of the peroneal nerve in Family 2 was interpreted as CMT2, as the CV cut off for the normal value is less in the peroneal than the median nerve, and an extrapolation with this proportional factor would increase the CV from 35.2 to 42.1 m/s, while a similar extrapolation in family 3 would increase the CV from 15.7 to 18.8 m/s (Additional file
2, table S3). Our neurophysiological data is far from optimal. The 232 consecutive unselected and unrelated CMT families with available DNA were from all regions in Norway, and were included into the study irrespective of the perfection of neurophysiological parameters, if the clinical or other parameters suggested it was CMT. An International study of mutations in the
MFN2, included 323 unrelated probands of whom 44 unrelated probands had an unknown neurophysiology[
9]. The lack of complete neurophysiological data makes the neurophysiological classification a challenge, along with the fact that CMAP were reduced in several recordings. Severely reduced CMAP and low CV in lower limbs suggest CMT2. However, the neurophysiological classification on CMT is based on CV alone not including CMAP or other neurophysiological features. For that reason we focused on CV in our interpretation. This provides a challenge in family 2, 3, 4 and 5 since neurophysiological data is only available from the lower limbs. Family 4 is diagnosed as dHMN based on a motor and a sensory CVs were normal, but the amplitudes of the sensory potentials were not measured which makes it difficult to judge whether the sensory nerves were involved or not.
The International study on MFN2 mentioned above included 323 unrelated probands, of whom 249 (77%) probands were diagnosed as CMT2, 20 (6%) had intermediate CMT, six (2%) had CMT1, three (1%) had dHMN and one (<1%) had hereditary sensory and autonomic neuropathy (HSAN, while 44 (14%) had an unknown neurophysiological phenotype. MFN2 mutations were found in 28 probands with CMT2 and in one family with intermediate CMT. We found the MFN2 mutation in 2 of 86 (2.3%) unrelated CMT1 families and in 1 of 15 (6.7%) unrelated dHMN families. These low frequencies of MFN2 mutations in CMT1 and dHMN made it unlikely for the International study to obtain MFN2 mutations in these CMT subtypes. Unfortunately, we do not have CV from the median nerve in the two families classified as CMT1. Thus, further studies are needed to confirm that MFN2 mutations can cause CMT1.
Clinically the age at onset was early in the CMT1 families (Family 1 and 3), while it was late in CMT2, intermediate and dHMN (Family 4-8), with one exception (Family 2). Additional symptoms such as ataxia, kyphoscoliosis and tremor were found three families (Table
2). The phenotype genotype correlation can be questioned in the proband with dHMN, since the classification was based on single motor and sensory nerve and he had multiple other disorders that might have caused his motor and sensory symptoms (for details see result section). However, since the motor and sensory CVs were normal, we characterized the proband as dHMN. The identified missense mutation c.1709A>G in exon 15 might be a polymorphism, although we did not find the mutation among 200 control chromosomes.
Point mutations in the MFN2 gene is likely to be the fourth most common cause to CMT after the duplication of PMP22 gene, and point mutations in the Cx32 and MPZ genes. We found that 2.3% of CMT1, 5.5% of CMT2, 12.5% of intermediate CMT and 6.7% of dHMN families have a point mutation in the MFN2 gene. We analyzed for mutations in five other common genes that may cause CMT1, CMT2, intermediate CMT or dHMN. However the exclusion analyses were not exhaustive since mutations in at least 40 genes can cause CMT. We suggest that genetic analyses of MFN2 should not be restricted exclusively to those with a CMT2 phenotype.
Competing interests
The authors declare that they have no competing interests.
Authors' contributions
GJB acquired the material, conceived the study, participated in the design of the study and drafted the manuscript. JCS, AL and HH carried out the molecular genetic studies and the sequence alignment. MBR conceived the study, participated in the design of the study and drafted the manuscript. All authors read and approved the final manuscript.