Main

The pediatrician is quite familiar with Down Syndrome, a chromosomal disorder associated with three copies of chromosome 21 (Trisomy 21). Down Syndrome generally occurs in one of 660 live births and is recognized by clinicians because of a constellation of physical findings. But why are individuals with Down Syndrome affected with mental retardation, congenital heart disease and thyoid dysfunction among other disorders? The answer lies not with an abnormal gene or set of genes but with an abnormal dosage of a subset of the genes normally present on chromosome 21 in the critical Down Syndrome region (1).

The concept of gene dosage imbalance is not unique to Down Syndrome (2). However, until recently, it was not generally appreciated that increased dosage of a functionally normal gene could be critical to the manifestations of a Mendelian syndrome. Dosage abnormalities arise by DNA rearrangements, such as duplication or deletion of a DNA segment, rather than by point mutation of a gene. DNA rearrangements may be a major mutational mechanism responsible for some Mendelian traits. These important concepts of gene dosage and DNA rearrangements being causative of human genetic disease have been uncovered through studies of the common inherited peripheral neuropathy known as Charcot-Marie-Tooth disease or CMT.

CMT was described by Jean Martin Charcot, his student Pierre Marie, and independently by Howard Tooth in 1886 (3,4). It presents clinically with distal muscle atrophy and weakness with initial primary involvement of the peroneal nerve. A steppage gait results from doriflexor weakness with patients often complaining of tripping and falling secondary to catching the dropped foot in the swing-through phase of the gait. Interestingly for pediatricians, idiopathic toe walking of childhood is sometimes an historical precedent. Physical examination is remarkable for absent deep tendon reflexes, difficulty while toe walking and great difficulty with heel walking. Occasionally, hypertrophic peroneal and ulnar nerves may be palpated as they pass close to the skin surface and sometimes the greater auricular nerve may be visualized as it passes over the sternocleidomastoid muscle. Charcot, Marie and Tooth recognized the more frequent occurrence in siblings as well as previous and subsequent generations thereby noting the inherited nature of the condition years before Mendel's laws were rediscovered. We now know that the most frequently observed inheritance pattern is autosomal dominant, although X-linked, autosomal recessive and sporadic forms exist. CMT is one of the most common genetic diseases affecting approximately 1 of every 2,500 individuals.

The CMT polyneuropathy syndrome can be further differentiated into two major forms based on electrophysiologic studies. Type 1 CMT (CMT1), the demyelinating form, presents with reduced motor nerve conduction velocity (NCV); with values in the 20-30 m/sec range rather than the normal greater than 42 m/sec. Type 2 CMT (CMT2) primarily affects the axon (5) and usually has normal or slightly reduced motor NCV with decreased amplitudes.

THE CMT1A DUPLICATION

Genetic linkage studies indicated that the predominant form of type 1 CMT (CMT1A) maps to the proximal short arm of chromosome 17 in band 17p11.2 (6). Further mapping efforts revealed an unusual and unexpected set of data. Genetic markers tightly linked to the CMT1A locus appeared to reveal three alleles in affected individuals rather than the expected two (7). We hypothesized that this resulted from inheriting a duplication of the involved locus on one homologue of the chromosome 17 pair. This duplication hypothesis was confirmed by six independent molecular methodologies (7). Two of the methods, (i) the detection of a 500 kb patient-specific junction fragment by pulsed-field gel electrophoresis (PFGE) (8) and (ii) the identification of three copies of this locus by fluorescence in situ hybridization (FISH) analysis of interphase chromosomes (9), are being used extensively in the molecular diagnosis of this condition. The latter technique is the first example of the application of a FISH-based microscopic technique to the diagnosis of a Mendelian condition. The CMT1A duplication was specifically associated with reduced motor NCV (7,10,11) and thus acted as a biological marker for the disease. In 100 U.S. families (11) and in over 1000 European patients (12) studied, the duplication accounted for 70% of CMT1 patients; and it also accounted for 90% of sporadic cases in one study (13).

MECHANISM FOR CMT1A DUPLICATION AND HNPP DELETION

The CMT1A duplication was independently identified by Prof. Dr. Christine van Broeckhoven and colleagues in Antwerpen, Belgium. They identified the first case of de novo duplication associated with sporadic CMT1 and showed by the segregation of marker genotypes that the de novo duplication occurred by unequal crossing over (14). This prompted us to search this region of the human genome for unusual features that might mediate the unequal crossing over. Physical analysis revealed that the duplicated segment was 1.5 megabases (Mb) or 1.5 million base pairs in length (15) and flanked by a 24 kb direct repeat we termed CMT1A-REP (15,16) (Fig. 1). The DNA sequences of the CMT1A-REP repeats are greater than 98% identical. Subsequent experiments showed that the duplication results from an homologous recombination event between a flanking proximal (or centromeric) CMT1A-REP misaligned with a distal (or telomeric) CMT1A-REP (15,17). This unequal crossing over event predicted a potential reciprocal recombination product resulting in deletion of the 1.5 Mb region. The deletion was subsequently shown to be associated with a distinct but related demyelinating peripheral neuropathy known as hereditary neuropathy with liability to pressure palsies or HNPP (18,19). Substantial experimental evidence supports the unequal crossing over model as outlined in Figure 2. The predicted structure of the CMT1A duplication has been recently directly visualized by FISH on stretched chromosome fibers (20). We identified a recombination hotspot associated with the unequal crossing over within the CMT1A-REP that is located near a mariner-like transposable element (17). Analysis of recombination products within the recombinant CMT1A-REP suggested double strand breaks initiating the recombination and indicated the strand exchange occurs in stretches of continuous sequence identify (21).

Figure 1
figure 1

Structure of CMT1A duplication and HNPP deletion. Chromosome 17 homologues are depicted with filled circle representing the centromere, filled rectangles corresponding to the flanking CMT1A-REP repeat, and the open oval depicting the PMP22 gene. The chromosome 17p short arm is not drawn to scale and the 1.5 Mb tandem duplication and 1.5 Mb deletion are not visible by conventional clinical cytogenetics techniques. Examples of the structure from a normal individual, as well as a CMT1A duplication patient and HNPP deletion patient, are given. Note a normal individual has 2 copies of PMP22 while a CMT1A duplication patient has three copies and an HNPP deletion patient has one PMP22 copy. Also note that the CMT1A duplication chromosome has three copies of the CMT1A-REP repeat while the HNPP deletion chromosome retains one copy (see Fig. 2 for mechanism).

Full size image
Figure 2
figure 2

Mechanism for formation of CMT1A duplication and HNPP deletion by reciprocal recombination. The proximal (centromeric) CMT1A-REP repeat is depicted as a filled rectangle while the distal (telomeric) copy is shown as an open rectangle. PMP22 is depicted as an open oval. The letters A, B, C and D refer to unique DNA sequences flanking the CMT1A-REP repeat while A′, B′, C′, and D′ refer to the same sequences on the homologous chromosome. The circled 1 and 2 refer to the recombination event and the products of reciprocal recombination. The recombinant CMT1A-REP is half-filled and half-open.

Full size image

PMP22 GENE DOSAGE

When we initially described the CMT1A duplication, four potential models for how a chromosomal duplication might affect a "CMT gene" and lead to disease were proposed. One of the models proposed a dosage sensitive gene within the genomic region that was duplicated. An initial hint that gene dosage was important came from the observation that a patient homozygous for the duplication had a more severe phenotype than her heterozygous sibling and parents (7). Support for a gene dosage model was obtained by investigating a patient with a cytogenetically visible chromosomal duplication of the proximal short arm of chromosome 17 (22). The hypothesis was if such patients with three copies of this genomic region were shown to have reduced NCV as part of their complex clinical phenotype, then a gene dosage effect would be responsible for the phenotype. If the NCV were normal in such patients then the gene dosage model would essentially be ruled out. Indeed, the dup(17)p12 patient had markedly reduced NCV in all nerves tested, therefore strongly supporting the gene dosage model (22). Subsequently, several patients were described with three copies of 17p12 secondary to visible interstitial tandem duplications (22,23), inverted duplication (24), or unbalanced translocations (25–27). Each of these chromosomal syndrome patients, as expected, had a complex clinical phenotype; however, in every case with trisomy for the 17p12 CMT1A region there was the reduced NCV associated with CMT1.

At this point, a specific dosage sensitive gene within the CMT1A duplication region had not been identified. The duplicated region contained an estimated 30-50 genes and determining the dosage sensitive one(s) seemed a daunting task. These investigations were aided by studies of the mouse models for human inherited neuropathies. The Trembler (Tr) and TremberJ (TrJ) have distinct mutations in the same gene on mouse chromosome 11 and hence have allelic mutations. Furthermore, mouse chromosome 11 is syntenic to (i.e. has the same or similar genes as) human chromosome 17. These mice manifest reduced motor NCV and neuropathological features similar to CMT1, as well as the more severe, earlier onset peripheral neuropathy, known as Dejerine-Sottas syndrome (DSS). Drs. Ueli Suter, Eric Shooter, and colleagues were investigating genes involved in the response to peripheral nerve crush injury. Their strategy was to isolate genes differentially expressed in a library made from a crushed nerve and to compare these genes to those obtained from a library made from the contralateral noncrushed nerve. They identified a rat gene whose expression appeared to be shut off after crush and increased during remyelination (28). The mouse orthologue was identified and this gene was found to encode a peripheral myelin specific protein of approximately 22 kD named Pmp22. The Pmp22 gene, which mapped to mouse chromosome 11, was mutated in the Tr and TrJ mice (29,30). The mouse Pmp22 gene was used to clone the human PMP22 gene which was then mapped within the CMT1A duplication (31). Comparison of the amino acid sequences of human, rat and mouse PMP22 proteins, predicted from conceptual translation of the cDNA clones, revealed a high degree of evolutionary conservation: 87% identity between human and rat and 86% identity between human and mouse (31). A consensus sequence for N-linked glycosylation is conserved in all three species (31).

At this point an apparent conundrum existed wherein the mouse models resulted from Pmp22 point mutations while the majority of human CMT1A patients had a 1.5 Mb duplication including PMP22. We then screened for PMP22 point mutations in the CMT1 patients who did not have the CMT1A duplication. We identified a de novo PMP22 point mutation (S79C) associated with sporadic CMT1 that was then transmitted in a dominant manner (32) and have subsequently shown that heterozygous PMP22 point mutations could also be associated with a DSS phenotype (33). Subsequently, several other human neuropathic PMP22 point mutations have been described including mutations identical to the mouse models Tr (34) and TrJ (35) (Fig. 3). Compared with the duplication, PMP22 point mutations usually have more severe clinical consequences. As anticipated, a frameshift leading to a nonsense mutation and presumably causing a null allele (i.e. absence of a protein product), resulted in an HNPP phenotype. This type of mutation has consequences similar to the HNPP deletion in that both result in haploinsufficiency for PMP22 (36). Of note, no PMP22 coding region mutations were observed in seven families with a CMT1A duplication or in two CMT1A patients with a de novo duplication (37). Thus, the lack of PMP22 point mutations in the inherited duplication cases and especially in the de novo duplication patients further supported the assumption that DNA duplication, leading to increased dosage of PMP22, is the major mechanism for CMT1A.

Figure 3
figure 3

PMP22 point mutations and associated neuropathy. The hypothetical structure of the PMP22 protein is depicted with four transmembrane domains and individual circles representing the primary amino acid sequence. The numbers refer to the specific amino acid positions in which mutations have been shown to give an inherited neuropathy phenotype. Below are listed the specific amino acid mutations and associated phenotype. CMT1 (yellow), DSS (pink), HNPP (blue), and the mouse mutants Tr, TrJ (green). Note 11/23 PMP22 alterations represent de novo mutations associated with sporadic disease.

Full size image

Several animal models, including two mice models (38–40) and a rat model (41), have been genetically engineered to overexpress wild type PMP22 and these animal models faithfully recapitulate the electrophysiologic and neuropathologic phenotype of CMT1A. Similarly, a Pmp22 knockout mouse (42,43) and an antisense mouse model (44) display the HNPP phenotype. Furthermore, quantitative measurements of both PMP22 mRNA and protein in sural nerve biopsies from CMT1A duplication and HNPP deletion patients reveals the expected increased and decreased levels, respectively (45–48). Thus, substantial evidence supports that PMP22 is a dosage sensitive gene and the contention that CMT1A results from trisomic PMP22 while HNPP results from monosomic PMP22.

GENETIC HETEROGENEITY

Although the majority of CMT1 patients have the CMT1A duplication as the proximate molecular cause for their disease, it is now clear that genetic alterations of at least ten loci can be associated with CMT (49). Four genes with disease causing mutations have been identified (Table 1). Besides the PMP22 gene these include MPZ (50), encoding the major myelin protein P0, GJB1 or Cx32 (51) encoding the connexin-32 gap junction protein, and EGR2 encoding a transcription factor. Different MPZ mutations have been associated with either CMT1B, DSS or recently, congenital hypomyelinating neuropathy (CHN) (52). Cx32 mutations have been associated with X-linked CMT. EGR2, the human homologue of the mouse Krox-20 gene, is involved in late myelin gene expression, and different EGR2 mutations have been associated with either a CMT1, CHN or DSS phenotype (53,54). Thus, human inherited demyelinating peripheral neuropathies, myelinopathies, represent a spectrum of disorders resulting from detects in myelin structure, maintenance and/or formation. Identification of each of the genes with a disease causing mutation yields a critical reagent with which to study the biology of the normal human peripheral nerve.

Table 1 Myelin genes and inherited peripheral neuropathy clinical phenotypes
Full size table

CLINICAL IMPLICATIONS

The molecular findings from studies of the inherited peripheral neuropathies have resulted in the development of diagnostic tools that enable a precise diagnosis and improve counseling for recurrence and prognosis. DNA diagnostic testing for the CMT1A duplication, HNPP deletion and Cx32 mutations are commercially available. Molecular diagnostic testing should be considered in CMT1 and the related peripheral neuropathies HNPP, DSS and CHN (55). These disorders can present with extremely variable clinical phenotypes. The phenotype can vary even between identical twins with the CMT1A duplication (56). The detection of the CMT1A duplication, HNPP deletion, or Cx32 mutation in a sample of peripheral blood, amniocytes or chorionic villi, establishes the exact molecular form of the disease in a given family. This facilitates inclusion or exclusion of the diagnosis in other family members who are at risk, enables prenatal diagnosis, and often provides prognostic information (57).

The diagnostic of CMT1 and HNPP should be considered even in the absence of a family history, given the high mutation (duplication/deletion) rate in sporadic cases. Molecular diagnostic testing also should be considered in neuropathies that are generally assumed to be acquired. HNPP is usually characterized by episodes of numbers and by palsies following minor compression or trauma to the peripheral nerve, though severely affected patients have features overlapping with CMT1. Carpal tunnel syndrome (58) and other entrapment neuropathies also are frequent manifestations of HNPP. Because electrophysiologic studies frequently suggest a conduction block, HNPP can sometimes be difficult to differentiate from acquired neuropathies such as chronic immune demyelinating polyneuropathy (CIDP) (5) and multifocal neuropathy (59). Therefore, the HNPP deletion should be excluded in all patients with a recurrent demyelinating mononeuropathy or polyneuropathy of unclear etiology (55).

CONCLUSIONS

Remarkable progress has recently been made in elucidating the molecular genetic basis of inherited peripheral neuropathies (49,60–62). These studies have uncovered a novel mechanism of inherited disease: DNA duplication. Findings at the CMT1A locus have crystallized the concept of "gene dosage" or gene copy number effects. Moreover, the observations of CMT1 patients with cytologically visible duplications of 17p12 have bridged a gap in our understanding of the contribution of genes to chromosomal syndromes versus Mendelian disorders (61).

The identification of region specific repeats such as CMT1A-REP, that can lead to DNA rearrangements and human disease traits, has helped delineate the concept of genomic disorders (63). Genomic disorders are caused by an alteration of the genome due to a genomic architecture predisposing to gain or loss of dosage sensitive genes (63,64). The CMT1A-REP repeat appears to have been duplicated during primate evolution because humans and chimpanzee have two copies, whereas gorillas and other lower primates have only one copy (16,65). The evolution of the mammalian genome therefore might result in structural features that leave particular regions of the human genome susceptible to DNA rearrangements responsible of genomic disorders (63).

The identification of four different genes, PMP22, MPZ, Cx32 and EGR2, involved in myelinopathies has provided a molecular understanding for the basis of genetic heterogeneity i.e., mutations at different loci causing the same phenotype. Similarly, the identification of multiple mutations in each of these genes has given us insights into the variability of expression of mutations i.e., different mutations involving the same gene can cause distinct clinical phenotypes (Table 1). The identification of these genes has provided important tools to study the normal biology of the peripheral nerve. Moreover, the availability of objective molecular diagnostics has enabled a better understanding of the clinical overlap among inherited neuropathies, clearly documented the role of new mutations in sporadic disease, and given a better appreciation of the contribution of genetics to diseases assumed to be acquired.

Future studies will likely identify other genes important to peripheral nerve function that may be responsible for some rare forms of CMT or related neuropathies. Determination of the complete genomic sequence of the CMT1A region will delineate all the genes within the 1.5 Mb CMT1A duplication. Further investigations may determine whether any of these genes play a role in modifying the neuropathy phenotype. Finally, it remains to be determined how many other Mendelian disorders result from duplication and the effects of increased gene dosage.