Charcot-Marie-Tooth Disease Type 1A -- Association with a Spontaneous Point Mutation in the PMP22 Gene

Evaluation of Patients

More than 100 unrelated patients with CMT underwent clinical and electrophysiologic evaluation. Studies of motor-nerve conduction velocity were performed with an electromyograph (model TD20, TECA, White Plains, N.Y.). Uniformly decreased motor-nerve conduction velocity was the electrophysiologic criterion for the diagnosis of CMT type 1. Conduction velocities were tested in the median and ulnar nerves, and in the peroneal nerve in some patients. The reductions in motor-nerve conduction velocities in patients with CMT type 1 were nearly identical in proximal and distal nerve segments and in bilateral nerves tested (Garcia CA, et al.: unpublished data).

Molecular analyses with several independent methods⁹ identified 32 unrelated patients with CMT type 1 but without the type 1A duplication (Wise CA, et al.: unpublished data) who were further analyzed for a mutation in the coding region of the PMP22 gene. A PMP22 point mutation was identified in one patient (Subject 7), who presented at 10 years of age with clinical symptoms of atrophy and weakness of the leg muscles, absent deep-tendon reflexes, and pes cavus. Motor-nerve conduction velocities were severely reduced: 9.7 and 9.8 m per second in the median and ulnar nerves, respectively. An analysis of family members revealed three with CMT type 1. Subject 6 presented with weakness of the intrinsic hand muscles in addition to the clinical findings present in his brother, Subject 7, and had motor-nerve conduction velocities of 9.0 and 12.0 m per second in the median and ulnar nerves, respectively. Subject 4 had clinical symptoms similar to those of his son, Subject 7, and nerve conduction velocities of 17.3 and 25.5 m per second in the median and ulnar nerves. The nerve conduction velocities in these patients with the PMP22 point mutation were close to or within the range observed in patients with the CMT type 1A duplication (10.4 to 42.0 m per second), but appear to lie at the lower end of the distribution²⁸. Other family members had no clinical symptoms of CMT type 1, and those evaluated electrophysiologically had normal nerve conduction velocities (>50 m per second) in the median and ulnar nerves.

Molecular Analysis

Figure 1. Figure 1. PCR Amplification and Heteroduplex Analysis of the PMP22 Coding Region.

In Panel A, four exons spanning the PMP22 coding region were individually amplified with genomic DNA and the indicated primer sets (1, 2, 3, and 4). The asterisk denotes the exon in which a point mutation was identified in Subject 7. F denotes forward, and R reverse. In Panel B, heteroduplex analysis of DNA from members of a family with CMT type 1 identified a mutation in PMP22 coding exon 3. Specific PCR products generated from each family member and from an unrelated control subject were combined, denatured, and annealed. When a mutation is present, the annealing of a mutant strand with the complementary wild-type strand results in heteroduplex DNA, which contains a local base mismatch at the site of mutation. The more slowly migrating heteroduplex band can be detected on an acrylamide gel that enhances its separation from the homoduplex band. Lanes 1 through 7 show the designated exon 3 duplex PCR products for the individual family members, as identified by the accompanying pedigree. A heteroduplex band (arrowhead) was detected in Subjects 4, 6, and 7, but not in the unaffected family members (Subjects 1, 2, 3, and 5). Lane 8 contains the negative control sample showing only the homoduplex band.

The coding region of the PMP22 gene was amplified by the polymerase chain reaction (PCR) with the use of 300 ng of genomic DNA and 1.0 micromole primers in a volume of 50 microliters. Figure 1A shows the sequences of primer sets 1, 2, 3, and 4 that were used to amplify individual coding exons of PMP22. PCR conditions for primer sets 1, 3, and 4 were as follows: initial denaturation at 94 °C for 2.5 minutes, followed by 35 cycles consisting of denaturation at 94 °C for 1 minute, annealing at 60 °C for 1 minute, and extension at 72 °C for 1 minute, and a final cycle of extension at 72 °C for 7 minutes. An annealing temperature of 55 °C was used for primer set 2.

For the analysis of heteroduplexes, 5 microliters each of PCR products derived from the individual family members and a nonpatient control was mixed, denatured at 95 °C for five minutes, and cooled to 35 °C. A negative control (10 microliters of control PCR product) was always included²⁹. Loading dye was added (2 microliters), and the samples were loaded on a 1.0-mm-thick gel measuring 21 by 40 cm (Bio-Rad, Richmond, Calif.) with mutation detection enhancement acrylamide (A.T. Biochem, Malvern, Pa.). The bands were visualized with ethidium bromide according to the manufacturer's instructions.

Templates for the direct determination of the nucleotide sequence of biotinylated PCR products were generated by two rounds of PCR amplification. The first round was performed as described for each exon, and a second round (15 cycles) used 0.2 microliter of the first-round PCR product and 0.1 micromole primers, with one primer biotinylated at the 5' end. The templates were isolated on streptavidin beads by combining 20 microliters of PCR product with 33.4 microliters of Dynabeads (Dynal, Oslo, Norway) according to the manufacturer's instructions. The beads were resuspended in 20 microliters of TE buffer (10 mM TRIS-HCl, pH 7.5, 1 mM EDTA), and 7 microliters was used in dideoxy sequencing reactions with the Sequenase 2.0 kit (U.S. Biochemical, Cleveland) with use of³⁵S-labeled deoxyadenosine triphosphate.

Paternity Testing

Paternity testing was performed by Southern blot analysis of DNA from Subjects 1, 2, 3, and 4 with use of single-locus probes D2S44, D4S139, D14S13, and D16S85 that provided a combined average power of exclusion of 99.93 percent.

To identify patients with CMT type 1 who did not have the type 1A duplication, a cohort of unrelated patients with CMT type 1 was screened for the presence of the 1.5-Mb duplication in 17p11.2-p12 by several independent methods: pulsed-field gel electrophoresis to detect the 500-kb SacII junction fragment specific to the duplication, Southern blot analysis to examine differences in the dosage of polymorphic alleles detected by probes mapping within the duplication, and analysis of polymorphic RM11-GT dinucleotide repeat alleles (three alleles in the presence of the duplication) as described previously⁹. Screening of more than 100 patients with CMT revealed 32 apparently unrelated patients with CMT type 1 who did not have the CMT type 1A duplication (Wise CA, et al.: unpublished data).

Figure 2. Figure 2. Pedigree of a Family (HOU226) with CMT Type 1.

The affected family members were identified by clinical examination and measurements of the motor-nerve conduction velocities of the median and ulnar nerves. The nerve conduction velocities are shown for all family members except Subject 5, who was related by marriage and who had no clinical symptoms of CMT. The (GT)_n dinucleotide repeat alleles at the polymorphic locus RM11-GT on VAW409 (locus D17S122)⁹ were also determined and are listed below the nerve conduction velocities. The (GT)_n genotypes were determined by PCR and scored for the number of visible alleles⁹ (wherein A equals 163 bp, B 161 bp, and C 155 bp). The original report on the analysis of RM11-GT identified seven alleles ranging in size from 165 to 153 bp (A-G); the 163-bp allele was originally designated as allele B, the 161-bp allele as C, and the 155-bp allele as F.⁹

Samples of DNA from these 32 unrelated patients were analyzed for mutations within the coding region of the CMT type 1 candidate gene PMP22. Studies of gene structure revealed that the PMP22 coding region includes four exons (unpublished data), and flanking primers were designed for PCR amplification of individual exons from genomic DNA (Figure 1A). The corresponding PCR products from the 32 unrelated patients were screened for mutations by heteroduplex analysis, wherein PCR products from patients and controls were combined to form duplexes, and base mismatches were detected by the presence of an additional heteroduplex band on high-resolution acrylamide gels²⁹. One patient, Subject 7, was identified as having an apparent base mismatch in coding exon 3 (Figure 1B). Heteroduplex analysis of DNA from family members showed that a mutation in the third coding exon of PMP22 segregated with CMT type 1 (Subjects 4, 6, and 7) (Figure 1B and Figure 2). Heteroduplex analysis of exon 3 in 108 unrelated and unaffected control subjects revealed no apparent mismatch, indicating that the PMP22 mutation detected in the studied family is specific to CMT type 1 and is not a common polymorphism. Furthermore, Subject 4, whose parents were unaffected, had new-onset CMT type 1, which arose simultaneously with an apparently spontaneous mutation in PMP22. A tightly linked marker, RM11-GT,⁹ mapping within 50 kb of PMP22,¹² was used to establish the paternal origin of this apparently new mutation (Figure 2). The results of DNA typing were consistent with paternity, with a probability of 99.98 percent. The clinical onset of CMT type 1 in the pedigree was therefore associated with an apparently spontaneous point mutation of paternal origin in PMP22.

Figure 3. Figure 3. Sequence Determination of the PMP22 Mutation.

Direct sequence determination of biotinylated PCR products corresponding to coding exon 3 was performed. In the PCR reactions, one of the primers was biotinylated at the 5' end. Biotinylated PCR products were bound to streptavidin-conjugated beads, denatured, and washed; the single-stranded biotinylated products that were retained were used as templates for the determination of DNA sequences by standard dideoxynucleotide-chain-termination reactions. The sense and antisense sequences are shown for an unaffected family member (Subject 1) and his affected son (Subject 4). The location of the C-to-G heterozygous point mutation is indicated on the sense (Panel B) and the antisense (Panel D) strands for Subject 4. The C-to-G transversion results in the substitution of cysteine for serine in the 79th codon, as indicated to the right of Panel B. This point mutation does not alter a restriction site.

Figure 4. Figure 4. Point Mutations in PMP22 Homologues Affecting the Well-Conserved Putative Transmembrane Domains of PMP22.

The broken arrow indicates the location of the serine-to-cysteine substitution in the second putative transmembrane (TM) domain of human PMP22, which is associated with CMT type 1. The solid arrows point to the previously identified substitutions in murine Pmp22 of trembler (substitution of aspartic acid for glycine at position 150 in putative TM 4)²⁶ and trembler^J (substitution of proline for leucine at position 16 in putative TM 1)²⁷. Dashes under the amino acids in human PMP22 represent identical residues in the rat and mouse homologues. Individual amino acid residues that are not conserved between PMP22 homologues in humans, rats, and mice are indicated. The replacement of serine at the 79th residue of human PMP22 by alanine in mouse Pmp22 occurs between two structurally equivalent amino acids (S and A) whose exchange is not likely to perturb the native protein structure³⁰. In contrast, the CMT type 1 point mutation causes a substitution at the 79th position between the nonequivalent amino acids serine and cysteine (S79C),³⁰ leading to the disease phenotype.

The DNA sequence of PCR products from individual family members was determined. This process identified a C-to-G transversion, resulting in a missense point mutation in the PMP22 coding region. Figure 3 shows the DNA sequence corresponding to the third coding exon of key family members (Subjects 1 and 4). The mutant and wild-type alleles were identified in heterozygous Subject 4, whereas only the wild-type allele was seen in his unaffected father (Subject 1). The same point mutation was detected in Subjects 4, 6, and 7, but was absent in unaffected family members (Subjects 1 and 2) (Figure 3). These data on DNA sequences, together with the verification that Subject 1, who was unaffected, was the father of Subject 4, who was affected, confirmed the spontaneous origin of this PMP22 mutation associated with CMT type 1. Determination of the sequence of four coding exons of Subject 4 did not identify any additional mutations (data not shown). The C-to-G point mutation corresponds to the substitution of cysteine for serine in the 79th codon (S79C) of PMP22 (Figure 3). This substitution occurs in the second putative transmembrane domain of PMP22 (Figure 4).

The identification of a spontaneous point mutation in the PMP22 gene that arises simultaneously with CMT type 1 and segregates with the disease phenotype in one family provides proof of the direct role of the PMP22 candidate gene in the disease process of CMT type 1A. This finding clearly demonstrates that two alternative mechanisms -- a point mutation in PMP22 and DNA duplication of a 1.5-Mb region on 17p11.2-p12 that includes PMP22 -- can give rise to the same disease phenotype. The overall similarity of the clinical and electrophysiologic findings in patients with CMT type 1A who have either the DNA duplication or the PMP22 point mutation means that molecular analysis is the only definitive way to distinguish between the two classes of patients.

Although the stably inherited DNA duplication³¹ is found in the majority of patients and families with CMT,^6,9-17 approximately 30 percent of the patients we studied with slowed nerve conduction velocities did not have the duplication (Wise CA, et al.: unpublished data). The low frequency of mutation of the PMP22 gene observed among the 32 patients with CMT type 1 but without the duplication may be explained by (1) the absence of linkage data, which leaves open the possibility that some of these patients may have mutations at other genetic loci not on chromosome 17p^5,7,32; (2) the analysis of only the coding region and not the regulatory regions of PMP22; and (3) possible limits in the ability of heteroduplex analysis to detect mutations.

The mechanism by which the CMT type 1A duplication causes the disease is best explained by the gene-dosage hypothesis, whereby the duplication of a CMT gene leads to increased levels of the gene product and ultimately results in the disease phenotype^9,17,18. Several groups have proposed PMP22 as a dosage-sensitive CMT type 1A candidate gene^21-24. In support of this hypothesis, recent evidence appears to rule out the loss of gene function or decreased levels of gene product as the basis for CMT type 1A. It was demonstrated that DNA deletion of a 1.5-Mb region on 17p11.2-p12, which includes PMP22, is associated with hereditary neuropathy with liability to pressure palsies (HNPP)³³. In contrast to CMT type 1A, HNPP (also called “bulb diggers' palsy”) is characterized by pressure-induced palsies that remit gradually and by sensory- and motor-nerve conduction velocities that are normal to mildly reduced^33,34. The alternative model of gene interruption at the junction of the CMT type 1A duplication is also unlikely to be the cause of the disease, since we demonstrated that a patient with 17p trisomy and a more extensive duplication of DNA had decreased motor-nerve conduction velocities characteristic of CMT type 1¹⁸. Three additional patients with 17p trisomy and different cytogenetic breakpoints have been found to have reduced nerve conduction velocities, lending further support to the gene-dosage model^19,20 (and unpublished data). The collective data are consistent with the hypothesis that increased expression of the peripheral-nerve-specific PMP22 gene causes the clinical and electrophysiologic phenotype of CMT type 1^18-24.

Point mutations causing amino acid substitutions in integral membrane proteins have been documented to cause autosomal dominant diseases, which now include CMT. In hyperkalemic periodic paralysis, missense mutations affecting transmembrane domains of the SCN4A skeletal-muscle sodium-channel gene product were identified, which presumably alter channel properties and result in disease^35,36. The related disorder paramyotonia congenita is associated with different point mutations in SCN4A^37,38. In autosomal dominant retinitis pigmentosa, individual point mutations collectively affect every transmembrane domain of rhodopsin³⁹. These rhodopsin missense mutations segregate in an autosomal dominant manner,³⁹ in contrast to the null mutations that are associated with autosomal recessive retinitis pigmentosa⁴⁰. The PMP22 point mutation described here leads to a serine-to-cysteine substitution (S79C) in the fourth putative transmembrane domain of the integral membrane protein PMP22⁴¹ (Figure 4). This S79C substitution could alter protein structure and function through aberrant disulfide linkage. The previously identified point mutations in the murine Pmp22 gene of the trembler (G150D)²⁶ and trembler^j (L16P)²⁷ mouse models of CMT type 1 are located in evolutionarily conserved putative transmembrane domains²¹ (Figure 4). We proposed on the basis of its structure that PMP22 could function as a pore or channel protein or have a part in membrane adhesion^21,25-27. Both hypotheses could accommodate the pathophysiologic mechanisms resulting from either PMP22 point mutation or 1.5-Mb duplication including the PMP22 gene.

To explain how a duplication encompassing PMP22 or a point mutation of PMP22 could result in CMT type 1A, we hypothesize that either mechanism could disrupt the sensitive macromolecular stoichiometry of the peripheral-nerve membrane. The PMP22 point mutation may constitute a dominant gain-of-function allele, leading to increased or altered function of PMP22, which could presumably mimic the effect of an increased level of gene product resulting from duplication. The involvement of these two mutational mechanisms was also suggested in autosomal dominant retinitis pigmentosa, wherein transgenic mice carrying a human rhodopsin mutation (P23H), as well as mice overexpressing the normal human rhodopsin transgene, had retinal photoreceptor degeneration⁴². On the other hand, the PMP22 point mutation could result in a dominant negative allele. In such a case, the mutation may theoretically lead to the inactivation of a multimeric protein complex, as postulated previously for the CLC-1 skeletal-muscle chloride-channel gene implicated in myotonia congenita⁴³. The elucidation of the biologic function of PMP22 is critical to the unraveling of the exact mechanisms by which the gene duplication and point mutation of PMP22 lead to the same clinical and electrophysiologic phenotype. Understanding the basic pathophysiology may lead to therapies for CMT and other inherited peripheral neuropathies.

Note added in proof: Since this paper was submitted, another report was published that described a PMP22 point mutation in a family with nonduplication CMT type 1A⁴⁴.