Abstract
Hereditary neuropathy with liability to pressure palsy (HNPP) is associated with a heterozygous 1.5 megabase deletion on chromosome 17 that includes the peripheral myelin protein (PMP) genePMP22. We show that heterozygous PMP22 knock-out mice, which carry only one functional pmp22 allele and thus genetically mimic HNPP closely, display similar morphological and electrophysiological features as observed in HNPP nerves. As reported previously, focal hypermyelinating structures called tomacula, the pathological hallmarks of HNPP, develop progressively in young PMP22+/0 mice. By following the fate of tomacula during aging, we demonstrate now that these mutant animals are also interesting models for examining HNPP disease mechanisms. Subtle electrophysiological abnormalities are detected in PMP22+/0mice >1 year old, and a significant number of abnormally swollen and degenerating tomacula are present. Thinly myelinated axons and supernumerary Schwann cells forming onion bulbs as fingerprints of repeated cycles of demyelination and remyelination are also encountered frequently. Quantitative analyses using electron microscopy on cross sections and light microscopy on single teased nerve fibers suggest that tomacula are intrinsically unstable structures that are prone to degeneration; however, the severity of morphological and electrophysiological abnormalities in PMP22+/0 mice is variable. These combined findings are reminiscent of the disease progression in HNPP and offer a possible explanation about why some HNPP patients develop a chronic motor and sensory neuropathy later in life that resembles demyelinating forms of Charcot-Marie-Tooth disease by both morphological and clinical criteria.
- PMP22
- peripheral myelin protein-22
- peripheral neuropathy
- myelin
- Charcot-Marie-Tooth disease
- CMT
- hereditary neuropathy with liability to pressure palsies
- HNPP
- tomaculous neuropathy
- demyelination
The most common inherited human peripheral neuropathy, Charcot-Marie-Tooth disease (CMT) (Skre, 1974), can be grouped on the basis of genetic, electrophysiological, morphological, and clinical criteria. Molecular genetic analysis reveals that point mutations and aberrant expression of the myelin proteins peripheral myelin protein 22 (PMP22), protein zero (P0), and connexin32 (Cx32) lead to CMT types 1A (CMT1A), 1B (CMT1B), and X1 (CMTX1), respectively (for review, see Suter and Snipes, 1995).
CMT1A and hereditary neuropathy with liability to pressure palsies (HNPP) are two inherited autosomal dominant peripheral neuropathies that are both associated with genomic rearrangements on human chromosome 17 (Lupski et al., 1991; Raeymaekers et al., 1991). Duplication of a 1.5 megabase region spanning chromosome 17p11.2-p12 leads to CMT1A, whereas the reciprocal deletion causes HNPP (Pentao et al., 1992; Chance et al., 1993; Reiter et al., 1996). The PMP22 gene (PMP22) is located within the affected region and is likely to play a crucial role in these neuropathies (Patel et al., 1992). This hypothesis is supported by the finding that specific missense mutations within the PMP22 gene lead either to a CMT1A-like phenotype or the congenital hypomyelinating peripheral neuropathy Déjérine-Sottas Syndrome (DSS) (for review, see Suter and Snipes, 1995). In the mouse, the pmp22 gene is located on chromosome 11, and dysmyelinating neuropathies attributable topmp22 mutations have been found in the natural mouse mutantsTrembler and Trembler-J (Suter et al., 1992a,b).
Biochemically, PMP22 is a hydrophobic 22 kDa glycoprotein component of PNS myelin (Snipes et al., 1992) sharing structural features with proteolipid protein (PLP), the major protein of CNS myelin (Suter et al., 1993). In addition to the likely function of PMP22 as a structural component of myelin, in vitro studies suggest a role for PMP22 in the regulation of cell proliferation and apoptosis (Fabbretti et al., 1995; Zoidl et al., 1995). Consistent with this hypothesis, PMP22 transcripts have been found in various embryonic and adult tissues (Welcher et al., 1991; Baechner et al., 1995; Parmantier et al., 1995). Furthermore, PMP22 belongs to a widely expressed gene family whose members are differentially regulated during the cell cycle (Marvin et al., 1995; Taylor et al., 1995; Ruegg et al., 1996; Taylor and Suter, 1996).
To clarify the function of PMP22, the pmp22 gene was disrupted using gene-targeting techniques in mice (Adlkofer et al., 1995). From the results obtained in human genetics, it was anticipated that heterozygous PMP22 knock-out (PMP22+/0) mice would potentially mimic HNPP if haploinsufficiency of the PMP22 locus alone is responsible for developing the disease. Thus, it was assumed that PMP22+/0 mice might represent an accurate model for studying disease mechanisms in this disorder.
The first symptoms of HNPP are usually observed in adolescence, manifested as a neuropathy that is characterized by recurrent pressure palsies precipitated by minor trauma to peripheral nerves (De Jong, 1947; Davies, 1954; Earl et al., 1964; Staal et al., 1965). Pathological changes in HNPP sural nerve biopsies include tomacula and, rarely, segmental demyelination (Meier and Moll, 1982; Dyck et al., 1993; Amato et al., 1996). Although the histopathological features of HNPP and CMT are distinctly different, they share overlapping clinical features that may confuse the diagnosis of the two disorders (Windebank, 1993; Roa et al., 1995).
In this study, we compare the reported pathogenesis of HNPP to PMP22+/0 mice. These animals exhibit similar myelin abnormalities as described in HNPP. Moreover, we observed morphological and electrophysiological changes in PMP22+/0 mice with progressing age that are comparable to the disease course in humans. Finally, we show that tomacula are likely to represent a labile prestage to demyelination, thus providing a possible explanation for the overlapping clinical features in CMT and HNPP.
MATERIALS AND METHODS
Animals and determination of genotype. Mice used in all experiments were obtained from our own breeding strain Agouti SV129 EV/C57BL/6. Genotypes of PMP22+/0 and wild-type mice were assessed by Southern blot analysis of genomic DNA from tail biopsies (Adlkofer et al., 1995).
Electrophysiology. We studied 9 PMP22+/0 mice at age 375–426 d and 14 PMP22+/+ mice at age 358–408 d, using previously described techniques (Adlkofer et al., 1995; Magyar et al., 1996). In brief, mice were anesthetized with a neuroleptic/analgesic combination (Hypnorm; Janssen, Beerse, Belgium). Body temperature as measured under the abdomen with a thermistor was allowed to equilibrate at 33.5–34.5°C under a heating lamp before measurements using a Medelec 92a electromyograph (Medelec, Surrey, UK) were started. Steel needle electrodes (DlSA 13L60 and 64) were placed subcutaneously. For facial nerve studies, stimulating electrodes were placed in the left outer ear canal (active electrode) and behind the ear (inactive electrode). A pair of steel recording electrodes (Picker, Munich, Germany) was placed in the whisker muscles, with the active electrode ipsilateral to the stimulation side. For sciatic nerve studies, a pair of stimulating electrodes (DISA 13L64) was placed in the left sciatic notch and 2 cm laterally (proximal stimulation). A second pair of stimulating electrodes (Picker) was inserted subcutaneously along the tibial nerve just above the ankle (“distal stimulation”). Recording electrodes (Picker) were placed in the skin close to the hallux and between digits 2 and 3 of the left foot. Near-nerve localization of the stimulating cathode was ascertained by determining the lowest possible threshold current to elicit a motor action potential, and studies were performed with a current 50% higher than that needed to elicit maximal stimulation. All potentials were immediately recorded on recording paper, and data calculations from printed records were double-checked. No systematic errors in latency or amplitude determinations could be detected. Latencies and amplitudes of the facial and sciatic compound muscle action potentials (M-response) were determined. In addition, we measured the F-wave elicitability and latency in the sciatic nerve. Mixed afferent sciatic nerve potential latencies were recorded at the sciatic notch with averaging of potentials after stimulation at the distal stimulation site. “Mixed afferent” denotes that the compound nerve action potentials included orthodromically stimulated sensory fibers and antidromically stimulated motor fibers. Motor and mixed afferent nerve conductance velocities (NCVs) were calculated from the sciatic nerve latency measurements. The distance between stimulation sites was measured with a caliper on the slightly bent limb of the mouse.
Tissue preservation and electron microscopy. Electron and light microscopy of quadriceps nerve of transcardially perfused mice (4% paraformaldehyde/2% glutaraldehyde in 0.1 mcacodylate buffer, pH 7.5) were performed as described (Martini et al., 1990; Fruttiger et al., 1995). For light microscopy, semi-thin sections were obtained from the same embedded nerve, cut in 2 μm sections, and stained with alkaline toluidine blue.
Teased nerve fiber preparation. The quadriceps branch of the femoral nerve was used for single axon fiber preparation. Femoral nerves were dissected from transcardially perfused mice, and their connective tissue sheath was removed and preteased into small bundles, followed by osmification (2% OsO4 in 0.1 mcacodylate buffer) for 2 hr, dehydration in an ascending acetone series, and immersion in Spurr medium overnight at 4°C. Single fibers were obtained by teasing in embedding medium. Light microscopy was performed on a Zeiss Axiophot using phase-contrast optics (Martini et al., 1995).
Immunohistochemistry. Nerve fibers were dissected from femoral nerve of mice intracardially perfused with 4% paraformaldehyde in 0.1 m cacodylate buffer, pH 7.4. Dissected fibers were incubated in 6.6 nm rhodamine-conjugated phalloidin (Molecular Probes Europe BV, Leiden, Netherlands) in 1% Triton X-100/PBS for 1 hr at room temperature. The nerves were rinsed in PBS, teased in Citifluor antifadent AF-1 (Chemical Laboratory University of Kent, Canterbury, Kent, UK) on an uncoated glass slide, coverslipped, and examined by fluorescence microscopy using a Zeiss Axiophot.
Statistical analyses. For statistical analysis of morphological features, five animals were used for each age group. Between 80 and 100 internodal segments were counted from each animal and included in the teased nerve fiber statistics. Cross sections were obtained at the level where the quadriceps nerve branches out of the femoralis nerve. Approximately 100 axon/Schwann cell units were counted and analyzed for each animal. The data were analyzed by the nonparametric Mann–Whitney U test using Statview II software for morphology analysis and Instat software for electrophysiology (Graph Pad, San Diego, CA). A p value ≤ 0.05 is considered to be significant.
RESULTS
Electrophysiology of aged PMP22+/0 mice
We have shown previously that no electrophysiological abnormalities are detectable in 10-week-old PMP22+/0 mice (Adlkofer et al., 1995); however, 12- to 14 month-old PMP22+/0 mice exhibited significantly reduced M-response amplitudes in the sciatic nerve as compared with age-matched control mice, whereas motor and mixed afferent latencies, F-wave latencies, duration of M- and F-responses, and NCVs were not significantly changed (Table 1).
Electrophysiological findings in 12- to 14-month-oldPMP22+/0 and wild-type mice (mean ± SD)
Morphological analysis by light microscopy
Only a few hypermyelinated axons are observed in the quadriceps nerves of 24-d-old PMP22+/0 mice, whereas the number of focally hypermyelinated axons is significantly increased in 10 week-old PMP22+/0 animals (Adlkofer et al., 1995). To evaluate further age-related morphological changes in peripheral nerves of PMP22+/0 mice, 2 μm semi-thin sections of quadriceps nerves of 5-, 10-, and 15 month-old wild-type and PMP22+/0mice were stained with alkaline toluidine blue (Fig. 1). In 5- and 10 month-old mutant animals, myelin sheaths of increased thickness compared with the caliber of the corresponding axons were observed regularly (Fig. 1b, d). Hypermyelination was still seen frequently in 15-month-old PMP22+/0 mice, but in addition, axons with abnormally thin myelin surrounded by redundant Schwann cells were found. Such onion bulb structures are indicative of demyelination and remyelination processes affecting specific axon–Schwann cell units and were often observed in clusters (Fig. 1f). Interestingly, the saphenous nerve was much less affected in PMP22+/0 mice of all ages examined. This sensory cutaneous branch of the femoral nerve appeared almost normal in young mutant animals, and only some signs of hypermyelination were found at older ages (data not shown). No obvious peripheral nerve abnormalities were observed in wild-type littermates (Fig.1a, c, e).
Alkaline toluidine blue staining of quadriceps nerve cross sections of wild-type and PMP22+/0 mice of different ages. Two micrometer cross sections of the quadriceps nerve show no myelin abnormalities in 5-, 10-, and 15-month-old control animals (a, c, e). In PMP22+/0 littermates, myelinated fibers with abnormal thick myelin sheaths are visible in transverse sections of quadriceps nerves (arrowheads) (b, d, f). In 15-month-old mutants, onion bulb formation is observed (arrows) (f). Scale bar, 25 μm.
Electron microscopy analysis
To confirm the data obtained by light microscopy, ultrastructural analysis was performed using electron microscopy. Various forms of hypermyelinated axons were observed in 5-, 10- and 15-month-old PMP22+/0 mice. One class of hypermyelination seems to be caused by excessive wrappings of the myelin sheath around the axon resulting in an abnormal number of myelin lamellae (Fig.2b). Another form is characterized by internal or external wrapping of redundant myelin loops (Fig.2d). Such myelin figures may arise initially from infoldings of myelin leading to frequently encountered myelin islands in mutant nerves (Fig. 2c). Similar figures are occasionally also seen in wild-type nerves (data not shown), but they were much more prominent in mutant animals. To account for this fact in the quantitative analysis, we used the criterion that only myelin infolds that were extended to more than half of the axon diameter were considered (Fig. 3).
Ultrastructure of abnormal myelin in PMP22+/0 mice. Cross sections from the quadriceps nerve of a wild-type mouse (a), a 10-month-old PMP22+/0 mouse showing hypermyelination attributable to excessive wrapping of the myelin sheath (b), a 10-month-old PMP22+/0 animal illustrating the invagination of the myelin sheath as a potential start of focal hypermyelination (c), a 5-month-old PMP22+/0mouse showing intermyelin infolds forming a hypermyelin structure (d), a 10-month-old PMP22+/0 mouse showing a hypermyelin structure with displaced axon (e), a 10-month-old PMP22+/0 mouse showing degenerating hypermyelin (f), and a 15-month-old PMP22+/0 mouse showing thinly myelinated axons with concentric Schwann cell processes (g,arrowheads) and basal laminae (h,arrow) forming onion bulbs. Axons (A), compact myelin (m), and degenerating myelin (d) are marked. Scale bar, 2.5 μm.
Quantitation of prominent infoldings of the myelin sheath in 10-month-old PMP22+/0 mice. Significant numbers of myelin infolds that are extended to more than half of the axon diameter are observed on cross sections of quadriceps nerves of 10-month-old PMP22+/0 mice (n = 5). In contrast, such prominent structures were not observed in wild-type littermates (n = 2; **p ≤ 0.01; one-sided U test).
In PMP22+/0 mice older than 10 months, additional forms of myelin degeneration were observed. Hypermyelinated structures were often nonconcentrically arranged around the axon, and the axon appeared compressed (Fig. 2e). Similar to the findings in HNPP biopsies, a characteristic splitting of the major dense line and vacuolation of myelin leading to an intramyelin edema was frequently seen (Fig. 2f) (Madrid and Bradley, 1975). These features may be interpreted as early signs of myelin degeneration. Moreover, Schwann cells with phagocytized myelin debris present in the cytoplasm were observed (not shown). Some myelin sheaths looked abnormally thin, and such profiles were surrounded by Schwann cells and Schwann cell processes forming cellular or basal lamina-type onion bulbs (Fig. 2g,h). Large caliber axons devoid of any myelin were not detected (data not shown). In general, increased myelin splitting after routine specimen preparation was commonly observed in PMP22-deficient animals, possibly indicating a weakened myelin structure (Fig. 2; data not shown).
Statistical analysis of myelin abnormalities in quadriceps nerve cross sections
Different forms of abnormal myelin were quantitated to assess disease progression by comparing 5-, 10-, and 15-month-old PMP22+/0 mice. Five animals of each age group were analyzed, and generally a relatively high variability was observed (Fig. 4).
Quantitation of myelin abnormalities of PMP22+/0 mice as deduced from electron microscopy analysis. Unaffected myelinated axon–Schwann cell units are significantly reduced in 5-month-old PMP22+/0 mice, whereas hypermyelin structures are frequent. Degenerating tomacula and onion bulb formation were observed only rarely at this age. In 10-month-old PMP22+/0 mice, degenerating tomacula and onion bulb formation occurred but did not reach statistical significance; however, all three myelin abnormalities were significantly increased in 15-month-old PMP22+/0 mice in comparison with wild-type littermates (*p ≤ 0.05; one-sided Utest).
In 5-month-old PMP22+/0 mice, 79.2 ± 12.5% (mean ± SD) of the myelinated axon–Schwann cell units appeared normal, and 24.6 ± 8.7% showed definitive signs of hypermyelination. Degenerating hypermyelin and onion bulb structures were barely detectable (≤1%). In 10-month-old mutant mice, 15.7 ± 7.7% of the myelinated axon–Schwann cells units were hypermyelinated, and 3.2 ± 2.5% hypermyelin structures with clear signs of degeneration were found. Some onion bulbs were present, but the variability between different animals was high (3.7 ± 3.4%); in one out of five mice analyzed, no onion bulbs were detectable. In 15-month-old animals, only 60.1 ± 17.1% of the myelinated axon–Schwann cell units were not affected, and 20.8 ± 9.8% appeared hypermyelinated. Degenerating hypermyelin structures (6.6 ± 3%) and onion bulbs (10.9 ± 7.7%) were often seen.
No significant differences in the number of hypermyelinated structures could be detected when the three age groups were compared (Fig.5). Degenerating hypermyelin structures and onion bulb formation were first found at the age of 10 months. The prevalence of both structures was significantly increased in 15-month-old PMP22+/0 mice in comparison with 5-month-old mice, but interanimal variations were high, reminiscent of the clinical variability observed in older HNPP patients.
Age-related progression of myelin abnormalities in PMP22+/0 mice deduced from electron microscopy analysis. The number of hypermyelin structures does not change significantly during aging of PMP22+/0 mice, whereas degenerating hypermyelin and onion bulb formation were significantly increased in 15-month-old PMP22+/0 mice in comparison with 5-month-old mutant mice (**p ≤ 0.01; one-sidedU test).
Teased fiber analysis
Teased nerve fibers of sural nerve biopsies have been used as a diagnostic method for hereditary motor and sensory neuropathies (Madrid and Bradley, 1975). To compare the observations made in HNPP with the PMP22+/0 mouse model, we used the same experimental strategy to analyze affected internodal segments. Qualitative analysis of teased quadriceps nerve fiber preparations of 5-month-old PMP22+/0 displayed focal hypermyelination at both inter- and paranodal regions (Fig. 6b,c), but paranodal tomacula appeared more frequently. Multiple tomacula within the same internodal segment and displaced myelin thickenings also were often found (Fig. 6c). In 10-month-old mutant mice, the presence of many very thick tomacula became obvious (Fig.6d). In accordance with the corresponding electron microscopy analysis, these expanded structures were interpreted as the early onset of myelin degeneration of a former tomaculum (Fig.2f). In 15-month-old PMP22+/0 mice, thick tomacula were still a prominent feature, but segmental demyelination was also found at this age (Fig.6e,f). Demyelinated internodal segments contained increased numbers of cell nuclei, presumably representing supernumerary Schwann cells forming onion bulbs (Fig. 6f).
Morphology of teased nerve fibers from wild-type and PMP22+/0 animals of different ages. Quadriceps nerve fibers in control animals (a). Para- and internodal tomacula in 5-month-old PMP22+/0 animals (b, c, large arrowheads). In 10-month-old mutant animals, enlarged tomacula presumably undergoing degeneration are observed (d, large arrow). Demyelinated internodal segments are seen in 10-month-old PMP22+/0 mice (e), but their number is increased significantly in 15-month-old mutants (f, arrowsindicate cell nuclei). Nodes of Ranvier are marked with small arrowheads (b–f). Scale bars (shown in f): a, c, d, f, 50 μm; (shown in e): b, e, 50 μm.
Interestingly, teased nerve fibers of 10- and 15-month-old wild-type mice displayed a regular array of Schmidt-Lanterman incisures by light microscopy analysis (Fig. 7a), which was further highlighted by phalloidin staining (Gould et al., 1995) (Fig.7c). In contrast, these regions of uncompacted myelin were severely disorganized in PMP22+/0 littermates, and instead, a “twisted” network of putative cytoplasmic channels was suggested along the long axis of the axon (Fig. 7b,d).
Irregular cytoplasmic inclusions in PMP22+/0 mice. By teased nerve fiber analysis, Schmidt-Lanterman incisures are regularly arranged in wild-type mice (a, c), whereas an unusual distribution of cytoplasmic channels is observed in mutant mice (b, d). Representative fibers are shown and phalloidin staining was used to visualize the distribution of cytoplasm (c, d). Schmidt-Lanterman incisures are marked with a star. Scale bar, 25 μm.
Statistical analysis of teased quadriceps nerve fibers
Because of its limited resolution in the longitudinal axis, the analysis of cross sections does not allow detailed quantitative investigations into the progression of the nerve pathology from hypermyelination to demyelination in aging animals. To achieve the latter goal, we selected teased single fibers with abnormal myelin structures affecting nearly every internodal segment for quantitative analysis (Fig. 8). Approximately 100 affected internodal segments from each animal were counted, and the abnormalities in myelin structure were subdivided into tomacula, degenerating tomacula, and demyelination. For this analysis, internodal segments with one or more focal hypermyelin structures were classified as “tomacula” (Fig.8). Segments containing degenerating tomacula (defined as strongly enlarged, irregular structures) appear as “degenerating tomacula,” regardless of the presence of additional normal-shaped tomacula. Finally, internodal segments that showed established signs of demyelination were grouped under the heading “demyelination.” Five animals were analyzed for each age group. Of the affected internodal segments in 5-month-old PMP22+/0 mice, 88.8 ± 4.1% (mean ± SD) were associated with one or multiple tomacula (Fig.8). In 10- and 15-month-old mutant animals, tomaculous segments decreased significantly to 73 ± 10.4% and 46 ± 11.7%, respectively. Degenerating tomacula were rarely detected in 5-month-old PMP22+/0 animals (3.8 ± 3.1%), whereas in 10-month-old animals, 16.9 ± 13.6% of affected internodal segments contained degenerating tomacula; in 15-month-old animals, this percentage was increased significantly to 35.5 ± 9.7%. Limited demyelination was observed in young mutant animals, whereas in 15-month-old mice, 14.2 ± 5.5% of internodal segments showed clear signs of demyelination. Interestingly, in PMP22+/0animals of all ages, mostly large caliber axons, presumably representing mainly motoneurons, were affected by myelin degeneration (data not shown).
Progression of myelin abnormalities in affected internodal segments in teased nerve fibers of PMP22+/0mice. The number of tomaculous internodes decreases significantly with age. In contrast, an increase in degenerating tomacula is observed in 10- and 15-month-old PMP22+/0 mice but reaches significance only at 15 months. The observed increase in onion bulb formation is significant at 15 months (*p ≤ 0.05, **p ≤ 0.01; one-sided Utest).
DISCUSSION
In this report, we have characterized the morphological and electrophysiological alterations in peripheral nerves of transgenic mice carrying only one functional pmp22 allele. Genetically, these mutant mice are models for the inherited human peripheral neuropathy HNPP, which is associated with a heterozygous deletion that includes the pmp22 locus (Suter and Snipes, 1995). Recent results indicate that the observed reduced PMP22 gene dosage is also reflected by a comparable reduction in the PMP22 protein concentration in the myelin sheath of PNS nerves of HNPP patients (Vallat et al., 1996). In complementation to the detailed descriptions of the clinical, electrophysiological, and pathological phenotype of HNPP (Meier and Moll, 1982), the availability of PMP22+/0mice allowed us to follow the particular features of a genetically comparable disease in an animal model and to analyze disease progression in a defined manner.
Clinically, HNPP is characterized as a recurrent polyneuropathy in which acute focal paralysis may be precipitated by minor trauma to PNS nerves. The clinical progression and severity of the disease display high variability, and in specific cases, a chronic peripheral neuropathy may develop that is clinically indistinguishable from CMT (Windebank, 1993). In agreement with these findings, we have observed definitive signs of behavioral abnormalities indicative of a potential chronic peripheral neuropathy, such as unusual cramping of the hindlimbs, only occasionally and exclusively in aged PMP22+/0 mice (data not shown). These observations are consistent with our electrophysiological analysis, which revealed significantly reduced amplitudes of the motor response in the foot muscles of PMP22+/0 mice at 12–14 months of age. In the absence of electrophysiological evidence of demyelination in our mice and in view of the histological findings, an M-response amplitude reduction indicates conduction failure only in the most severely affected fibers. In HNPP patients, persistent conduction block has been reported by several groups (Magistris and Roth, 1985; Sellman and Mayer, 1987). Motor amplitudes were abnormally low in 20–77% of nerves from HNPP patients with deletion of one PMP22 allele (Gouider et al., 1995). In contrast, electrophysiological signs of demyelination are mild (Meier and Moll, 1982; Oh, 1993; Gouider et al., 1995). A follow-up analysis in older mice is currently underway to investigate whether demyelination will become electrophysiologically more evident later in life. Furthermore, given the slight but statistically not significant decrease of NCV in PMP22+/0mice, it will be important to examine the possibility of focal slowing. Interestingly, we did not detect marked F-wave abnormalities in PMP22+/0 mice, which supports our observation that the extent of demyelination is rather mild. In contrast, we found more severe F-wave abnormalities in another animal model of a dysmyelinating neuropathy, the heterozygous P0-deficient mouse, which is affected by a more severe myelin deficiency (Zielasek et al., 1996).
Morphologically, HNPP is characterized as a tomaculous neuropathy with multifocal thickenings of the myelin sheath (Windebank, 1993). Similar to observations in human HNPP nerves, we found that para- and internodal tomacula were present in nearly every internode of large caliber axons in 5-month-old PMP22+/0 mice. PNS nerves of 15-month-old PMP22+/0 mice, however, were affected not only by tomacula formation but also by the prominent presence of enlarged focal tomacula, suggesting degeneration. Furthermore, statistically significant demyelination and onion bulb formation were detectable at this age. Interestingly, focal myelin thickenings are not specific to HNPP, but they are also found in several other forms of peripheral neuropathies, including familial brachial plexus neuropathy, CMT1B, and chronic inflammatory demyelinating neuropathies (Dyck et al., 1993;Windebank, 1993; Thomas et al., 1994), suggesting that tomacula are unstable structures that may predispose to demyelination. This conclusion is supported by our findings in the quantitative analysis of teased peripheral nerve fibers derived from PMP22+/0 mice, which indicates a reduction in the number of tomaculous internodal segments with a concomitant increase of demyelination.
Mice completely deficient of PMP22 (PMP220/0) display retarded myelination followed by prominent focal hypermyelination, and 2-month-old PMP220/0 animals show a CMT1-like phenotype with abundant onion bulb formation and axonal loss (Adlkofer et al., 1995). In comparison, PMP22+/0 mice develop an increasing number of tomacula over the first 10 weeks, and signs of demyelination are not evident before the age of 10–15 months. Such a late onset of myelin deficiency is reminiscent of several other transgenic mouse strains that carry mutations in various myelin genes. In particular, P00/0 mice develop a severe phenotype that is characterized by hypomyelination and degeneration of myelin and axons at young age (Giese et al., 1992), but P0+/0 mice are much less affected and reveal only a mild demyelination that is first detectable at the age of 4 months (Martini et al., 1995). These results suggest that P0+/0 mice provide an accurate animal model for those cases of CMT1B that are associated with a P0 null allele (Warner et al., 1996; Suter, 1997). Furthermore, mutant mice devoid of Cx32 (Cx320/0 or Cx320/y) develop a late-onset demyelinating peripheral neuropathy, in agreement with the strong expression of Cx32 by myelinating Schwann cells, the presence of Cx32 in uncompacted PNS myelin, and frequent mutations found in this X-linked gene in CMTX1 (Paul, 1995; Anzini et al., 1997). Finally, mice deficient in the myelin-associated glycoprotein (MAG0/0) also exhibit a late-onset peripheral nerve demyelination (Fruttiger et al., 1995). These combined findings suggest that the correct gene dosage of several myelin-associated genes is required for the maintenance of the PNS myelin sheath. Moreover, the distinct localizations and different putative functions of the various myelin proteins involved suggest that the classical morphological features of a demyelinating neuropathy can be the consequence of diverse molecular and cellular disease mechanisms.
None of the myelin mutants described so far, however, is characterized by a comparable degree of abundant focal hypermyelination as seen in PMP22-deficient animals. In particular, ageing PMP22+/0 mice provide an excellent model for studying the different stages of tomacula formation and their subsequent degeneration. In these animals, tomacula seem to be stable structures until adulthood, with a late but distinct onset of demyelination. Our results indicate that tomacula can be generated by various mechanisms. One form of tomacula shows a regular structure and seems to be the direct consequence of too many ultrastructurally normal myelin lamellae surrounding the axon. A second class of tomacula may be related to the frequently observed initial invagination of myelin loops, which may lead to the formation of redundant internal loops of the myelin sheath.
Various spontaneous and transgenic rodent strains have revealed that PMP22 is involved in the initial spiralling of myelin in early development, the determination of myelin thickness, and the maintenance of myelin and axons of peripheral nerves (Suter et al., 1992a,b;Adlkofer et al., 1995; Magyar et al., 1996; Sereda et al., 1996).In vitro results further suggest that PMP22 is also involved in the control of Schwann cell proliferation and apoptosis (Fabbretti et al., 1995; Zoidl et al., 1995). Thus, whether the relative instability of tomacula is directly related to the function of PMP22 as a structural component of myelin or reflects a more general role of this protein in Schwann cell physiology remains an intriguing question.
Our results indicate that PMP22+/0 mice closely mimic various aspects of the human hereditary neuropathy HNPP. Thus, these mice are likely to provide an excellent animal model for studying specific aspects of HNPP, in particular disease progression, and the induction and formation of tomacula followed by degeneration. It remains a challenge, however, to understand the reasons that induce a given Schwann cell to focally hypermyelinate an axon on the one hand, while on the other hand the rest of the internode appears quite normally myelinated. It is anticipated that PMP22+/0 mice will not only provide a valuable tool for the search for the cellular disease mechanisms underlying HNPP but that this model will also be helpful in the evaluation and development of specific treatment strategies.
Footnotes
This work was supported by a grant from the Swiss National Science Foundation (U.S.). We thank Corinne Zgraggen for excellent technical assistance, Heide Mayer-Rosa for expert photographical help, Dr. Sara Sancho for stimulating discussions, and Dr. Verdon Taylor for critically reading this manuscript.
Correspondence should be addressed to Dr. Ueli Suter, Institute of Cell Biology, Swiss Federal Institute of Technology, ETH-Hönggerberg, CH-8093 Zürich, Switzerland.