Discussion We performed WES in 303 unrelated Japanese patients with CMT among a first case series and successfully identified the MME gene as a novel AR‐CMT disease‐causing gene using an overlap‐based strategy. After mutation screening analyses on MME in a second and third case series of unrelated patients, we finally identified mutations in the MME gene in 10 unrelated patients with CMT. WES is one of the most effective methods for identifying pathogenic mutations in Mendelian disorders.27, 28, 29 Abundant bioinformatic tools that efficiently prioritize pathogenic mutations or strategies that find disease‐causing genes have been developed. Three main analytical strategies—linkage‐based, de novo–based, or overlap‐based strategies—are widely used for gene research after WES.30, 31 The overlap‐based strategy helps to identify candidate genes by focusing on variants shared among multiple unrelated patients. The ESVD system we have developed based on this strategy is capable of simultaneously executing variant filtering and detecting the variants shared by disease samples and consequently contributed to the identification of the MME gene. This system could effectively identify novel causative genes in other Mendelian disorders. Our study revealed various recessive MME mutations: nonsense, missense, splice site, and deletion. The results of the RNA analysis obtained from patients with the splice site or nonsense mutations demonstrated aberrant splicing or lack of RT‐PCR fragments, respectively. Moreover, we confirmed the complete absence of detectable NEP by immunohistochemical and western blotting in the sural nerve of P1 with the c.654+1G>A mutation. These findings suggest that these mutations lead to a lack of NEP protein expression in the PNS attributed to a premature stop codon or nonsense‐mediated mRNA decay. Similarly, two nonsense mutations (p.Gln221X and p.Trp606X) may cause nonsense‐mediated mRNA decay. A missense mutation (p.Cys621Arg) in P4 was predicted to be damaging during in silico analysis, and cysteine 621 was highly conserved across species. Expression of NEP in the sural nerve was partially reduced. These findings suggest that this missense mutation may lead to abnormal axonal transport or fast degradation of the mutated protein. The 3‐bp in‐frame deletion (c.1231_1233delTGT) that eliminates cysteine at codon 411 was predicted to be deleterious using in silico analysis, and cysteine 411 was also highly conserved across species, suggesting that this deletion may affect the function of NEP. The c.439+2T>A and c.655−2A>G mutation causing in‐frame exon skipping may lead to the large structural alterations and affect the expression and function of NEP. Taken together, our results suggest that all the recessive MME mutations in these 10 unrelated patients with CMT could be loss‐of‐function mutations, although the effects of expression and function of NEP on some mutations (c.1231_1233delTGT, c.439+2T>A, and c.655−2A>G) need further analysis. The result of the haplotype analysis that P1 and P3 shared the same haplotype indicates a possible ancestral founder effect for the c.654+1G>A mutation. In contrast, P2 did not share this particular haplotype and an MAF of this variation in the HGVD database is 0.002 (genotype count: G/G = 299, G/A = 1, and A/A = 0), suggesting that the mutation results from independent mutational events. Clinically, all the patients with the MME mutation had a similar clinical and electrophysiological phenotype consistent with a late‐onset axonal‐type motor and sensory neuropathy. Although 1 patient (P5) had borderline motor nerve conduction velocity (NCV; 37.4 m/sec) of the median nerve between demyelinating and axonal form, we considered that his NCSs showed a primary axonal neuropathy with secondary demyelination because motor NCV of the ulnar nerve was 45.5 m/sec (>38 m/sec) and compound muscle action potential (CMAP) and sensory nerve action potential (SNAP) in the lower limb nerves were not evoked. In 2 patients, larger myelinated fibers were markedly decreased in peripheral nerves, and occasionally thin myelin sheaths without onion‐bulb formation were present, suggesting that the pathological process is primarily axonal degeneration. The human MME gene maps to chromosomal region 3q25.1 to q25.2, is composed of 24 exons, and is highly conserved among mammalian species.32 NEP is a type 2 MME consisting of 742 amino acids and has a molecular weight ranging from 85 to 110kDa depending on differences in its glycosylation. NEP is an ectoenzyme with the bulk of its structure, including the active site, facing the extracellular space,8 as shown Figure 4B. NEP is widely expressed in many tissues, including the nervous system,33, 34 and degrades a number of substrates such as enkephalins, substance P and atrial natriuretic peptide, and, most notably, the “Alzheimer peptide,” Aβ. Moreover, MME is also abundantly expressed in the kidney, and previous studies have suggested that MME might play a role in the normal physiological function of podocytes and was involved in various renal diseases.35, 36 However, renal failure and nephrotic syndrome were not observed in 8 patients evaluated, except for 2 who had mild proteinuria (Supplementary Table 3). These data indicated that loss‐of‐function mutations in the MME gene may be not sufficient to cause congenital renal disease. In the CNS, NEP is mainly located on neurons, particularly in the striatonigral pathway, hippocampus, and cortical regions,37, 38 and it is localized along axons and at presynaptic sites.39 These observations suggest that after synthesis in the soma, NEP is anterogradedly transported to axon terminals, where it may play an important role in neuronal function. In general, NEP knockout mice developed normally, but showed a greater sensitivity to endotoxin shock,40 elevated microvascular permeability,41 enhanced aggressive behavior, and altered locomotor activity.42 It was demonstrated that NEP knockout mice have increased levels of Aβ peptides in the brain, and administration of the NEP inhibitor, thiorphan, to rats led to increased Aβ levels.43, 44 Moreover, NEP knockout mice exhibited amyloid‐like deposits with signs of neurodegeneration in the hippocampus and behavioral deficits.13 These findings suggest that the decline of NEP is an important factor in the progression of Alzheimer's disease.45 In contrast, Thomas et al showed that the learning abilities were not reduced in older NEP knockout mice; rather, they were significantly improved, and Aβ deposits could not be detected by immunohistochemical methods, in spite of the elevated Aβ levels in the brains of these mice.46 In humans, no mutation in the MME gene linked to familial Alzheimer's disease has been reported, although some nucleotide repeat polymorphisms have been reported to be associated with susceptibility to sporadic Alzheimer's disease.47, 48 We hypothesized that human NEP deficiency would lead to cognitive impairment and increased Aβ plaque accumulation. Contrary to our expectation, neuropsychological testing revealed no obvious cognitive deficits in 9 patients evaluated by MMSE scores. Moreover, amyloid PET imaging, which is capable of visualizing amyloid accumulation even in mild cognitive impairment or preclinical stages of Alzheimer's disease,49 was conducted and P1 showed no obvious 11C‐PIB retention. These clinical data indicated that loss‐of‐function mutations in the human MME gene are not sufficient to cause early‐onset familial Alzheimer's disease, and NEP deficiency may be compensated for by other Aβ‐degrading enzymes, such as insulin‐degrading enzyme or endothelin‐converting enzyme, in vivo.11 In the PNS, NEP has been found in neonatal and early postnatal Schwann cells in the pig or rat sciatic nerve,6, 7 and NEP expression increased after axonal damage in adult rat Schwann cells in the sciatic nerve.50 Although there are no reports that MME mutations result in peripheral neuropathy, a previous study revealed that after chronic constriction injury of the sciatic nerve, MME knockout mice were more sensitive to heat and mechanical stimuli than were wild‐type mice, and developed edema and changes in limb temperature resembling human complex regional pain syndrome with increases in substance P and endothelin 1 in sciatic nerves.51 Interestingly, a recent study revealed the NEP is transported by antero‐ and retrograde axonal flow in rat sciatic nerves.52 In addition, our patients showed axonal neuropathy. These findings suggest that NEP could play a role in peripheral nerve development and axonal regeneration. However, as shown in Figure 6B, our immunohistochemical data revealed that NEP expression in the myelin sheath is considerably more than in the axon. For example, the mutations of peripheral myelin protein zero (MPZ) usually causes demyelinating neuropathy; however, sometimes axonal neuropathy phenotype is also shown, particularly in late‐onset form.53 In our study, given that all patients with MME mutations showed late‐onset CMT, we believe that similar degeneration process may be present in both conditions by impairing Schwann cell–axonal interactions. Further studies are needed to clarify the role of MME in the PNS, to identify the molecular pathomechanism underlying CMT, and to develop an effective treatment for the disease. NEP has received considerable attention as a therapeutic target for Alzheimer's disease. Fortunately, Iwata et al have successfully developed a new gene delivery system by an adeno‐associated virus that can achieve comprehensive gene expression of NEP in the brain of young NEP‐deficient mice, eventually decelerating Aβ accumulation and alleviating cognitive dysfunction.54 This gene therapy may be applied for the prevention and treatment of patients with not only Alzheimer's disease, but also CMT caused by MME mutations in the near future. Finally, we calculated the molecular diagnosis rate by only focusing on the AR mode of inheritance. Although only 17 of 77 (22%) patients received a molecular diagnosis by our microarray or NGS, the identification of the MME gene resulted in an improved molecular diagnosis rate of 35% (Fig 7). Interestingly, this result indicates that MME is the most frequent cause of AR‐CMT2 in Japan. Figure 7 Rate of molecular diagnosis categorized according to inheritance modes. The upper table indicates the number of patients with molecular diagnoses by modes of inheritance in 726 patients. The pie chart below shows the rate of molecular diagnosis focused on the AR mode of inheritance. Only 17 of 77 (22%) patients received a molecular diagnosis in known CMT or IPN genes. Ten patients (13%) received a new molecular diagnosis, MME gene mutations. Rate of molecular diagnosis was increased to 35%. AR = autosomal‐recessive; CMT = Charcot–Marie–Tooth; IPN = inherited peripheral neuropathy. In conclusion, for the first time, we identified MME as the most common causative gene for AR‐CMT using WES followed by the ESVD system with an overlap‐based strategy. Our genetic and immunohistochemical data strongly support the finding that the MME gene is a novel causative gene for AR‐CMT. All 10 patients with the MME mutation had a similar phenotype with a late‐onset axonal‐type motor and sensory neuropathy, but with no evidence of Alzheimer's disease, and we propose this new classification as AR‐CMT type 2T. Identification of the MME mutations responsible for AR‐CMT can improve the rate of molecular diagnosis, understanding of the molecular mechanisms of CMT, and establishment of effective therapeutic approaches.