A four-generation family from Yancheng (China) with autosomal dominant CMT was recruited in April 2007 at the Department of Paediatric Orthopaedics, The Children's Hospital of Soochow University (Suzhou, China) (Fig. 1). Blood samples were obtained from 7 affected individuals (3 females and 4 males) with CMT and 6 unaffected family members (II1, II4, II5, II8, II9 and III1) after obtaining written consent. The clinical features of the proband (III2) were analyzed retrospectively. A detailed physical examination was performed on all 7 patients. An electrophysiological examination was available for 3 patients (III3, III4 and IV1). Written informed consent was obtained from all participants and approval was obtained from the Ethics Committee of the Chinese Academy of Medical Sciences (Beijing, China).

Genomic DNA was extracted from peripheral blood of an affected family member (III2) using the QIAamp DNA Blood Mini kit (Qiagen GmbH, Hilden, Germany). The DNA was broken into fragments ranging from 180–280 bp using an ultrasonoscope (S2; Covaris, Inc., Woburn, MA, USA) (8). Adapter oligonucleotides from Illumina (single reads; Illumina Inc., San Diego, CA, USA) were ligated to the ends and fragments were amplified with the Paired-End Sequencing Library Prep kit (Agilent Technologies, Inc., Santa Clara, CA, USA) according to the manufacturer's protocol. Enrichment of coding exons was performed with SureSelectXT Human All Exon V4 (Agilent Technologies, Inc.). After hybridization of the sequencing primer, the bases were incorporated using the Illumina HiSeq2000 platform (Illumina, Inc.) for 90 cycles of sequencing per read to generate paired-end reads including 90 bp at each end and 8 bp of the index tag (9). Image analysis and base calling were performed using Illumina Pipeline (version 1.3.4; Illumina, Inc.).

Error assessment and base calling were also performed using Illumina Pipeline (version 1.3.4) to generate the primary data. Clean reads of 90 bp length were mapped to the GRCh37 reference human genome from the National Center for Biotechnology Information database (NCBI; http://www.ncbi.nlm.nih.gov/) using the Burrows Wheeler Aligner Multi-Vision software package (version, 0.6.2; Wellcome Sanger Institute, Cambridge, UK) (10). Single nucleotide variations (SNVs) and indels were identified using SOAPsnp software (version, 2.04; Beijing Genomics Institute, Shenzhen, China) and GATK Indel Genotyper (version, 3.4–46; http://www.broadinstitute.org/gsa/wiki/index.php/), respectively. Previously identified SNVs were filtered through the NCBI dbVar (https://www.ncbi.nlm.nih.gov/dbvar/browse/) and ExAC browser (http://exac.broadinstitute.org/). Candidate disease-associated genes were obtained using Phenolyzer software (http://phenolyzer.wglab.org/). The OMIM database (https://omim.org) was searched using ‘hereditary motor and sensory neuropathy’ as key words and the phenotypic series ‘Charcot-Marie-Tooth disease-PS118220’ (https://omim.org/phenotypicSeries/PS118220). Known disease-causing mutations were recorded in the Human Gene Mutation Database (HGMD) at the Institute of Medical Genetics in Cardiff, UK (http://www.ghmd.cf.ac.uk/).

The genomic DNA of the family members was isolated from peripheral blood using a standard SDS-proteinase K-phenol/chloroform method (11). The reference sequence of the candidate gene MFN2 was obtained from the UCSC Genome Browser (http://genome.ucsc.edu), and the candidate variant was confirmed by PCR-Sanger DNA sequencing. The PCR primers (forward, 5′-TGCTCCTCTGCTTAGTCA-3′; reverse, 5′-GAAACTGGCTGATCAAACGC-3′) were designed with Primer 3.0 software (http://primer3.ut.ee/). The PCR reaction contained 20–100 ng genomic DNA, 0.5 µl each primer (10 µM), 4 µl dNTP (2.5 mM), 12.5 µl 2X GC buffer I, 2.5 U LA Taq DNA polymerase (Takara Biotechnology Co., Ltd., Dalian, China) and deionized water to 25 µl. The PCR program was performed using the following conditions: 95°C for 3 min; followed by 38 cycles at 94°C for 30 sec, 58–60°C for 30 sec; 72°C for 50 sec; and a final extension at 72°C for 8 min. The amplicons were run on an ABI 3730 Sequence Detection System (Applied Biosystems; Thermo Fisher Scientific, Inc., Waltham, MA, USA). The sequencing data were analyzed using CodonCode Aligner (version 6.0.2.6; CodonCode Corporation, Centerville, MA, USA).

PCR-restriction fragment length polymorphism (PCR-RFLP) analysis was used to verify the candidate mutation in all individuals of the pedigree and 120 healthy Chinese controls (60 males and 60 females; age, 20–45 years) recruited between April 2016 and January 2018. Controls were from the Chinese Han population with no family history of muscle diseases. The first PCR was amplified longer fragments around the mutated site using the procedure stated above. Nested-PCR was performed using a forward (5′-TGCTCCTCTGCTTAGTCA-3′) and a mismatched reverse primer 5′-AATTTCAGTCGGTCTTCC-3′) using the thermocycling conditions stated above. Amplicons from the nested-PCR were digested at 37°C for 15 min by MspI (Takara Biotechnology Co., Ltd.). An 8% neutral polyacrylamide gel was used to distinguish alleles of the wild type and mutant type by loss or gain of the restriction site. Silver-staining (0.1% AgNO₃) was used at room temperature for 10 min prior to the final step of the chromogenic reaction.

The protein sequences of human mitofusin 2 (NP_001121132.1) encoded by MFN2 and of the other homologous proteins from nine different animals were obtained from the NCBI Protein database (https://www.ncbi.nlm.nih.gov/protein/) in FASTA format. Multiple sequence alignment and conservative analyses were performed using MEGA software (version 7; Institute for Genomics and Evolutionary Medicine, Temple University, Philadelphia, PA, USA). The pathogenicity of several missense variants at the same location including that found in the participant family were respectively predicted using online tools PolyPhen-2 (http://genetics.bwh.harvard.edu/pph2/), Scale-Invariant Feature Transform (SIFT; http://sift.jcvi.org/), MutationTaster (http://www.mutationtaster.org/) and M-CAP (http://bejerano.stanford.edu/mcap/). The three-dimensional structures of normal and mutant mitofusin 2 were generated by homology modeling using SWISS-MODEL (http://swissmodel.expasy.org/) (12). The structure of mitofusin 2 was examined with the crystal structure of the truncated mitofusin 1 structure (pdb 5GOM). The interactions between the mutant amino acid (R397P) and the neighboring residues were exhibited and simulated by PyMOL (Schrödinger, LLC, New York, NY, USA; http://www.pymol.org/). Homology modeling based on the truncated mitofusin-1 (pdb 5GOM, unit 1A) solved with 2.8 Å resolution.

All 7 affected patients expressed manifestations of bilateral finger contractures and a gait abnormality because of pes cavus, lower limb weakness, foot drop and deformities. The proband (III2) initially presented with symptoms at age 4 years (Fig. 2). Patient (II3) had similar pes cavus, foot muscle atrophy, strephenopodia, and presented with symptoms at age 7 years. Patient (II6) had more critical symptoms, such as lower limb atrophy and limited walking ability after age 30 years with upper limb weakness. The present findings indicated that the clinical manifestations and developmental progression of all affected family members were similar, but the age of initial onset was slightly different, with men aged 3–4 and women 7–8 years. In addition, the male patients exhibited more severe symptoms than the female patients, and they lost their ability to walk independently after age 30. The electrophysiological examinations were performed on 3 patients (III3, III4, and IV1). The results (recorded in the case history without the electromyogram) exhibited milder NCV changes. Based on the results of the physical and electrophysiological examinations, the disease of this family was diagnosed as CMT2.

The raw data were filtered into clean data and >96.75% of reads had 99.9% accuracy to identify variations among the patients. Following mapping to GRCh37, 99.95% of the yielded reads were correctly matched, and average sequencing depth reached at least 126-fold. In total, >23,250 exonic and 84,189 intronic SNVs were identified by SAMtools, and highly frequent single nucleotide polymorphism (SNPs) were excluded through international criteria. A similar process was used to screen the 22,013 indels. The candidate variations were produced through three processes: i) Removing highly frequent SNPs, ii) reserving the variations from exons and 10 bp around splicing sites, and iii) eliminating synonymous mutations; eventually 2,223 variations remained. The inheritance model was considered for further advanced analysis to determine the potential disease-causing mutation. In this patient (III2), potential SNVs and indels were separately identified in 541 and 178 candidate genes according to the obvious inheritance pattern of autosomal dominance in this four-generation pedigree. Following several filtering processes, the novel mutation was ultimately identified as c.1190G>C in exon 10 of MFN2. This mutation led to substitution of arginine for proline at amino acid residue 397 (p.R397P).

The candidate mutation in the MFN2 gene was verified by Sanger sequencing. As a result, the variant from Sanger sequencing was completely consistent with that from the WES analysis (Fig. 3A). PCR-RFLP indicated that the wild-type allele failed to be digested by MSPI and the mutant allele was successfully separated into 19 and 86 bp fragments. All affected individuals (II3, II6, III2, III3, III4, III5, and IV1) carrying the heterozygous mutation (c.1190G>C) presented the genotype of three 105, 86 and 19 bp fragments; however, all normal individuals (II1, II4, II5, II8, II9, and III1) and normal controls presented one fragment of 105 bp (Fig. 3B). The close co-segregation of the phenotype and genotype suggested that the R397P variant may be the CMT causative mutation in the present family.

A sequence conservation analysis of mitofusin 2 indicated that the amino acid located in position 397 was highly conserved among 10 species (Fig. 3C). Substitution of the wild-type amino acid (arginine) with the mutant amino acid (proline) may change its biological function.

The c.1190G>C; p.(R397P) variation in MFN2 was not present in the HGMD database, indicating that this variation is a rare variation in MFN2. Furthermore, the SIFT PROVEAN, polyphen-2, Mutation Taster, and M-CAP programs were respectively used to predict the pathogenicity of this variation (Table I). The results from the bioinformatics analysis suggested that c.1190G>C in MFN2 described a disease-causing mutation.

According to the SWISS-MODEL prediction, R397, located in a critical position of the coiled-coil domain, contributed to terminate the helix bundle (Fig. 4A). However, the change in R397P led to extension of the helix bundle. R397 interacted with the R400, L401 and D67 residues through H-bonds, whereas the more hydrophobic residue proline resulted in the reformation of H-bonds with glutamine at position 387 (Fig. 4B).