E. Nelis, P. de Jonghe, V. Timmerman
6.1.4 Genetics and pathomechanism
Molecular genetic studies have shown that autosomal dominant CMT1 is genetically heterogeneous and four loci have been mapped: CMT1A to chromosome 17p11.2 [164], CMT1B to 1q22-q23 [16], CMT1C to 16p13.1 [144] and CMT1D to 10q21.1-22.1 [166]. The CMT1 causing genes for all these loci have been identified over the last decade and are PMP22, MPZ, LITAF and EGR2, respectively. Furthermore, mutations in NEFL, originally associated with CMT2, are also found in autosomal dominant CMT1 pa-tients [61]. In the following paragraphs, we will discuss how positional cloning strategies and candidate gene approaches have contributed to the identification of the loci and genes for autosomal dominant CMT1.
z The CMT1A duplication and HNPP deletion
The CMT1A locus was assigned to chromosome 17p by Vance et al. [164]
in an American CMT family. At that time detailed genetic maps were not available and the genetic markers ± restriction fragment length poly-morphisms ± used in positional cloning strategies had a low informativity.
However, thanks to the availability of multigenerational pedigrees, others confirmed linkage to the CMT1A locus in several large families, and re-fined it to a large region (30cM) on chromosome 17p11.2-p12 [90, 92, 111, 120, 155]. In 1991, two research groups independently identified a tandem duplication of a chromosomal segment in 17p11.2-p12 in CMT1A patients [82, 121]. This duplication occurred de novo via an unequal crossing-over event and was subsequently transmitted to the next generations. The de novo appearance of the CMT1A duplication is a frequent finding in iso-lated CMT1 patients [55]. These de novo duplications are usually of pater-nal origin and arise from unequal crossing-over events during male sper-matogenesis [105]. In addition, a few duplications of maternal origin have been described [18, 85]. Large genetic epidemiological studies estimated that the frequency of the CMT1A duplication in dominant CMT1 cases is about 71% CMT1 [102, 171]. Physical mapping experiments were per-formed and the size of the CMT1A duplication region was estimated to be 1.5Mb [113, 122]. The unequal crossing-over event in CMT1A occurs
dur-ing meiosis via a chromosomal misalignment of large repeat elements, called the ªCMT1A-REPsº, flanking the CMT1A region (Fig. 6.1) [113].
With the exception of a few rare cases, the CMT1A duplication always has the same size, i.e., 1.5Mb, suggesting that the repeat sequences in this re-gion influence this constant DNA rearrangement. A patient, mosaic for the CMT1A duplication, was reported to have a reversion of the 1.5Mb CMT1A duplication in several somatic tissues [79].
Interestingly, Chance et al. [27] reported patients affected with HNPP to have a reciprocal deletion of the 1.5Mb CMT1A region on 17p11.2-p12.
Subsequent studies confirmed that the large majority of HNPP patients car-ried the 1.5Mb deletion [102]. At the molecular level this implies that three copies of the CMT1A-REP sequence are located on the CMT1A dupli-cation chromosomes while only one copy is present on the HNPP deletion chromosomes. It is unknown why HNPP seems to be much rarer than CMT1A. It has been hypothesized that a significant proportion of HNPPs are never diagnosed because the course is mild and patients may present only once in their life with a pressure induced palsy.
Further detailed analysis of the CMT1A duplication/HNPP deletion mu-tations demonstrated a ªhotspotº region of frequent unequal crossing over within the CMT1A-REP elements in a cohort of unrelated CMT1A and HNPP patients of North American [124], European [81, 156] and Japanese [174] origin. The sequence within the hotspot of recombination showed 98% identity between the proximal and distal CMT1A-REP elements [66, 123]. Sequence comparison of the low copy CMT1A-REPs revealed a mari-ner transposon-like element (MITE) near the hotspot of unequal recombi-nation. However, it is unlikely that the MITE codes for an actively tran-scribed transposase since the open reading frame contains several frame-6 Charcot-Marie-Tooth disease type 1 and hereditaryneuropathywith liabilityto pressure palsy z 97
Fig. 6.1. Model of unequal crossing over between misaligned CMT1A-REPs leading to the CMT1A duplication (1) and HNPP deletion (2). The blue boxes represent the proximal CMT1A-REP, the bright blue boxes the distal CMT1A-REP flanking the 1.4 Mb region, whereas the dark blue boxes represent the PMP22 gene
shift mutations [124]. Physical mapping strategies using large insert clone contigs (P1 and bacterial artificial chromosomes: PACs and BACs) were performed to delineate the complete nucleotide sequence of the CMT1A/
HNPP genomic segment to 1,421,129 bp of DNA. This latter study did not only allow the identification of several genes in the region, but also ex-plains an evolutionary mechanism relevant for the formation of the CMT1A-REP by DNA rearrangement during primate speciation [59].
z CMT1A and related disorders ± peripheral myelin protein 22 gene (PMP22) (OMIM 118220, 118300)
One year after the identification of the CMT1A duplication, PMP22 was as-signed to the CMT1A region [89, 112, 154, 162]. Importantly, the smaller duplications and deletions reported in the literature, still contained the PMP22 gene [105, 161], supporting a gene dosage effect as the disease mechanism [83]. Furthermore, over- and underexpression of PMP22 has been confirmed at the transcript and protein level in CMT1A and HNPP patient nerve biopsies, respectively [163]. The dosage sensitivity of PMP22 transcripts and protein was nicely illustrated by the genotype-phenotype correlations in man and rodents over- and underexpressing PMP22 (re-viewed in [15, 101]).
Beside the CMT1A duplication and HNPP deletion, point mutations in the PMP22 gene can result in the following distinct phenotypes: classical CMT1 (CMT1A), HNPP and the more severe DSS and CH (IPNMDB;
http://www.molgen.ua.ac.be/CMTMutations/) (Table 6.1). Mutations that truncate or severely alter the protein sequence and are rapidly degraded, mimicking underexpression resulting from the deletion, are predicted to lead to an HNPP phenotype. Indeed, most loss-of-function mutations re-sult in HNPP, while gain-of-function mutations lead to CMT1 or DSS, either by an increased dosage of a normal PMP22 protein or by a toxic ef-fect of the mutated PMP22 molecule. In vitro studies have demonstrated that missense mutations lead to impaired intracellular trafficking of PMP22 resulting in an accumulation of the mutant protein in the endoplasmic reticulum and Golgi apparatus. The mutant protein also traps normal PMP22 resulting in a decreased amount of PMP22 available for incorpora-tion in the myelin membrane [95].
The PMP22 gene was first cloned as the human homologue of the mouse growth arrest-specific 3 gene (Gas3) [87]. The gene is located on mouse chromosome 11, a syntenic region to human chromosome 17p11.2. It en-codes a membrane protein comprising 2±5% of total peripheral myelin protein content [107]. PMP22 expression in the peripheral nervous system (PNS) is most likely regulated by axonal contact [141]. The PMP22 gene has two tissue-specific promoters, one being nerve-specific [145]. The PMP22 protein has four transmembrane domains, two extracellular loops, and cytoplasmic amino and carboxy termini. After synthesis in the rough endoplasmic reticulum, the majority of PMP22 becomes rapidly degraded
and only a small fraction is processed in the Golgi apparatus and trans-ported to the cell membrane [106]. Although PMP22 has been known for almost two decades, its function is still unclear. Initial in vitro studies showed that PMP22 is a growth arrest and apoptosis-specific protein [42, 86]. Mouse pmp22 has been detected during development and in distinct adult neural and non-neural tissues. The zebrafish orthologue of PMP22 also shows expression in embryonic neural crest cells, suggesting a role in the early development of the PNS [172]. Moderate overexpression of PMP22 can induce susceptibility to apoptosis in some cell types. When this apoptotic response is counteracted, PMP22 can still modulate cell shaping and cell spreading. Therefore, PMP22 may have an important role in Schwann cell differentiation and myelination [25]. In the adult PNS, PMP22 most likely functions as an integral membrane protein since it is confined to the compact myelin of Schwann cells [69, 140]. Co-immunopre-cipitation and confocal microscopy experiments demonstrated that PMP22 and myelin protein zero (MPZ/P0), the major component of the peripheral myelin membrane, form complexes suggesting a complementary role of both proteins in cell adhesion of compact myelin [33]. Mutated or overex-pressed PMP22 causes a ªgain-of-functionº endoplasmatic reticulum reten-tion phenotype. Recently, it was demonstrated that PMP22 associates in a specific, transient, and oligosaccharide processing-dependent manner with the lectin chaperone calnexin (CNX) [39]. In the Tr-j mouse, a prolonged association of mutant PMP22 with CNX was found. Since CNX and PMP22 co-localize in large intracellular myelin-like figures, sequestration of CNX in intracellular myelin-like figures may be relevant for the CMT pathology.
z CMT1B and related disorders ± myelin protein zero gene (MPZ) (OMIM 118200)
The myelin protein zero gene (MPZ) is located on chromosome 1q22-q23. It encodes a 219 amino acid, 28±30 kDa glycoprotein that accounts for more than 50% of total PNS myelin protein. The protein called P0 has one trans-membrane domain, an extracellular aminoterminus and an intracellular car-boxy-terminus. Crystallographic 3-D structural analysis of the extracellular domain shows similarity to an immunoglobulin variable domain [137]. Dur-ing Schwann cell development, P0 is simultaneously induced with genes en-coding other myelin proteins, such as PMP22, myelin basic protein (MBP) and myelin-associated glycoprotein (MAG) [78]. P0 is upregulated at the on-set of myelination. As a compact myelin protein, P0 most likely acts as a `dou-ble adhesive protein'. It holds myelin together at the intraperiod line through interactions of its extracellular, selfadhesive immunoglobulin domain [137]
and at the major dense line via interactions of its cytoplasmic domain [40]. Apart from its structural role in myelination, P0 plays a regulatory role as well. P0-overexpressing mice show failure in axon sorting and a myelina-tion arrest at early mesaxon formamyelina-tion. In early developing Schwann cells, high P0 overexpression inhibits polarization of Schwann cell membranes into 6 Charcot-Marie-Tooth disease type 1 and hereditaryneuropathywith liabilityto pressure palsy z 99
appropriate functional domains, dynamic axonal interaction and Schwann cell membrane expansion required for appropriate axonal sorting and myeli-nation [176]. Most MPZ mutations cause a classical CMT1 phenotype (CMT1B). However, some MPZ mutations lead to a more severe DSS or CH phenotype (reviewed in [99]). Specific MPZ mutations are also associated with CMT2, the axonal form of CMT (Table 6.1). The Thr124Met mutation is associated with a distinct CMT2 phenotype with pupillary abnormalities and deafness [28, 36, 93, 135]. Mutant P0 could affect myelin formation in three ways: (1) by not reaching the myelin membrane, (2) by reaching the myelin membrane but having lost its adhesive properties, or (3) by reaching the myelin membrane and having a dominant negative effect on the wildtype P0. In vitro studies have shown that mutated P0 can indeed reduce adhesion [41, 173]. The complex-formation between PMP22 and P0 might clarify the remarkable similarity between the CMT1A and CMT1B phenotypes [33]. Al-terations in either protein may interfere with the normal association of P0 and PMP22 into one functional complex. The disturbed interaction would subsequently result in demyelination as a common pathological pathway in CMT1A and CMT1B.
z CMT1C ± lipopolysaccharide-induced tumor necrosis factor gene (LITAF) (OMIM 601098)
Two autosomal dominant CMT1 families were reported to have a locus map-ping to 16p13.1-p12.3 (CMT1C) [144]. Only recently, three missense muta-tions in LITAF were identified in the two previously known CMT1C families and confirmed in a third family of different origin [143]. The LITAF protein (also known as SIMPLE; small integral membrane protein of lysosome/late endosome) is widely expressed and encodes a 161-amino acid protein that might be involved in protein degradation pathways. The missense mutations (G112S, T115N, W116G) cluster in a domain of the LITAF protein relevant for peripheral nerve function [12]. The function of LITAF is currently unknown, but the subcellular localization and putative domains point to a ubiquitin-mediated lysosomal degradation protein. LITAF is located in the lysosome/
late endosome [94], its murine orthologue interacts with NEDD4 an E3-ubi-quitin ligase [60], and it contains a conserved motif known to interact with another protein involved in sorting of ubiquitinated proteins to the multi-ve-sicular body [157].
Further mutation analysis of LITAF in CMT1 patients is needed to esti-mate the frequency of LITAF mutations in the total CMT1 population, and to delineate the associated phenotype.
z CMT1D ± early growth response element 2 gene (EGR2) (OMIM 607678) The EGR2 gene is located on chromosome 10q21.1-q22.1 [62] and encodes a 51 kDa protein of 475 amino acids. EGR2 is the human homologue of the mouse Krox20 gene [30], with an overall amino acid identity of 89%
(100% in the zinc finger domain) [167]. EGR2 is a member of the EGR family. The EGR proteins encode transcription factors containing Cys2His2 zinc finger domains, which bind a GC-rich consensus binding site [147].
Analysis of homozygous and heterozygous Krox20 knockout mice has shown that Krox20 is important in the development and segmentation of the hindbrain [131, 146]. Surviving homozygous Krox20 knockout mice have hypomyelination of the PNS with Schwann cells blocked at an early stage of differentiation, causing a trembling phenotype [159]. Krox20 ex-pression is activated before onset of myelination in the PNS and is essential for the final differentiation of myelinating Schwann cells [181]. These data suggest that Krox20 and its human homologue EGR2 are transcription fac-tors required for the transactivation of PNS myelination-specific genes. In-deed, using microarray expression profiling, 98 known genes were identi-fied that were induced by Egr2 in Schwann cells [96]. The putative Egr2 target genes included myelin proteins and enzymes required for synthesis of normal myelin lipids. RT-PCR to monitor Schwann cell gene expression confirmed that Egr2 is sufficient for induction of target genes including MPZ, PMP22, MBP, MAG, GJB1 and PRX. Furthermore, a dominant-nega-tive inhibition of wildtype Egr2-mediated induction of essential myelin genes using EGR2 DNA-binding domain mutants was demonstrated [96]. A recent study, however, provides evidence that myelin-related signaling in Schwann cells can be independent of Krox20 [109].
Several mutations in EGR2 have been described in patients with different phenotypes, i.e., classical CMT1, DSS and CH [153, 166]. Some of these mu-tations were present in the homozygous and others in the heterozygous state (Table 6.1). One mutation causes a CMT1 phenotype with cranial nerve def-icits [108]. Clinical involvement of cranial nerves is unusual for CMT1 and may demonstrate a similar role for Krox20 and EGR2 in brainstem and cra-nial nerve development [108]. The effect of some mutations on the DNA binding capacities of EGR2 has been studied in order to correlate the residual DNA binding capacities to the clinical severity. The results confirm that the severity of CMT1D correlates with the residual amount of DNA binding of the mutated EGR2 [167]. The dominant nature of these mutations seems to be in contrast with the Krox20 heterozygous knockout mouse, which shows no phenotypical abnormalities [131]. This suggests that these mutations do not cause a loss of function, but rather have a dominant negative or a gain-of-function effect. Another mutation was shown to interfere with the binding of NGFI-A binding proteins, possible co-repressors of EGR2, prob-ably leading to increased transcription of EGR2 [167].
z CMT1F ± neurofilament light polypeptide gene (NEFL) (OMIM 607734) Mutations in NEFL were recently reported as a cause for autosomal domi-nant axonal CMT linked to chromosome 8p21 (CMT2E) [91]. Further anal-ysis of NEFL in CMT patients showed that patients with NEFL mutations have an early disease onset and are usually severely affected, with moder-6 Charcot-Marie-Tooth disease type 1 and hereditaryneuropathywith liabilityto pressure palsy z 101
ately to severely slowed NCVs. Based on the clinical characteristics, some of them were diagnosed with DSS [35, 61].
NEFL encodes a 62 kDa structural protein, which is one of the most abundant cytoskeletal components of neuronal cells [43]. NEFL assembles with neurofilaments of higher molecular mass, medium (NEFM) and heavy (NEFH) chain polypeptides, into intermediate filaments and forms an ex-tensive fibrous network in the cytoplasm of the neuron. Neurofilament ac-cumulation tightly correlates with radial growth of axons during myelina-tion. Neurofilaments determine the axonal diameter and, hence, the con-duction velocity of peripheral nerves [4, 75]. Nefl knockout mice do not have a CMT-like phenotype [180], while transgenic mice have a severe pe-ripheral neuropathy with massive motor neuron death [76].
z Roussy-Levy syndrome ± myelin protein zero gene (MPZ), peripheral myelin protein 22 gene (PMP22) (OMIM 180800)
Plant-Bordeneuve et al. [115] identified a point mutation in MPZ in the original family studied by Roussy and Levy. This finding shows that the family belongs to the CMT1B subtype. Other RLS patients, however, showed the CMT1A duplication [5, 152] (Table 6.1). These findings provide evidence against the RLS as a distinct genetic entity, but suggest a close re-Table 6.1. Genes associated with CMT1 and related peripheral neuropathies
Gene Locus Inheritance Phenotype
zPMP22 (duplication) CMT1A AD CMT1, RLS
zPMP22 (duplication) CMT1A AR DSS
zPMP22 (deletion) HNPP AD HNPP
zPMP22 CMT1A, HNPP AD CMT1, HNPP, DSS, CH
zPMP22 CMT1A AR DSS
zMPZ CMT1B AD CMT1, CMT2, DSS, CH, RLS
zMPZ CMT1B AR DSS
zLITAF CMT1C AD CMT1
zEGR2 CMT1D AD CMT1, DSS
zEGR2 CMT4E AR CH
zNEFL CMT2E AD CMT1, CMT2, DSS
zGDAP1 CMT4A AR CMT1, CMT2, DSS
zMTMR2 CMT4B1 AR CMT1, CH
zSBF2/MTMR13 CMT4B2 AR CMT1
zKIAA1985 CMT4C AR CMT1
zNDRG1 CMT4D AR HMSN-L
zPRX CMT4F AR CMT1, DSS
zCTDP1 CCFDN AR CCFDN
lation with CMT1. What causes the additional features of gait ataxia and essential tremor needs further clarification. Roussy-Levy syndrome (RLS) comprises clinical features of CMT1 combined with static tremor of the upper limbs and gait ataxia [127].