|Home | About | Journals | Submit | Contact Us | Français|
Mutations in peripheral myelin protein 22 (PMP22) result in the most common form of Charcot-Marie-Tooth (CMT) disease, CMT1A. This hereditary peripheral neuropathy is characterized by dysmyelination of peripheral nerves, reduced nerve conduction velocity, and muscle weakness. A PMP22 point mutation in L16P (leucine 16 to proline) underlies a form of human CMT1A as well as the Trembler-J mouse model of CMT1A. Homozygote Trembler-J mice (TrJ) die early postnatally, fail to make peripheral myelin, and, therefore, are more similar to patients with congenital hypomyelinating neuropathy than those with CMT1A. Because recent studies of inherited neuropathies in humans and mice have demonstrated that dysfunction and degeneration of neuromuscular synapses or junctions (NMJs) often precede impairments in axonal conduction, we examined the structure and function of NMJs in TrJ mice. Although synapses appeared to be normally innervated even in end-stage TrJ mice, the growth and maturation of the NMJs were altered. In addition, the amplitudes of nerve-evoked muscle endplate potentials were reduced and there was transmission failure during sustained nerve stimulation. These results suggest that the severe congenital hypomyelinating neuropathy that characterizes TrJ mice results in structural and functional deficits of the developing NMJ.
Charcot-Marie-Tooth disease (CMT) is the most frequently inherited peripheral neuropathy, occurring in 1 of 2500 people in the United States (1). Mutations underlying CMT directly affect motor and sensory neurons (Type 2), or indirectly impair their function by causing damage to Schwann cells (Type 1) (2, 3). The principal clinical feature of CMT is muscle weakness and atrophy, which begins in the distal limb and spreads proximally (4). Physiologically, patients with CMT1 but not those with CMT2 exhibit a reduction of motor nerve conduction velocity (MCV), consistent with myelin disruption; however, the clinical manifestations of CMT1 more closely correlate with functional evidence of muscle denervation than with reduced MCVs (5). Consistent with this idea, animal models of CMT1 exhibit a loss of large-diameter axons and impaired axonal transport even before impairment of MCV (6). In addition to these signs of axonal dysfunction, neuromuscular junctions (NMJs) show evidence of denervation in CMT1 mutant mice (7, 8). Therefore, peripheral dysfunction of motor nerve terminals may represent an important early step in the pathophysiology of CMT1.
CMT Type 1A (CMT1A) is the most common form of all CMTs and is characterized by mutations or other genetic disruptions of peripheral myelin protein 22 (PMP22), a 22-kd tetraspan transmembrane glycoprotein located within compact myelin. Histologically, peripheral nerves exhibit signs of demyelination and/or dysmyelination, depending on the PMP22 genotype. For example, replacement of glycine at 150 by aspartic acid (G150D) in the Trembler mouse (Tr) is associated with demyelinated axons with a compensatory, ongoing increase in Schwann cell proliferation; by contrast, substitution of leucine 16 with proline (L16P) in the Trembler-J mouse (TrJ) is associated with dysmyelination with only a transient increase in dividing Schwann cells (9–11). The L16P PMP22 allele, which also underlies a form of human CMT1A (12), is semidominant in TrJ mice and triggers moderate dysmyelination in heterozygotes and severe hypomyelination and early postnatal lethality (~postnatal day 15; P15) in homozygotes (13). This mutation disrupts the formation and stability of myelin in part by increasing the lysosomal degradation of PMP22 protein (14).
The difference in myelination (hypomyelination vs dysmyelination) and prognosis (shortened vs normal longevity) between homozygotic and heterozygotic TrJ mice, respectively, suggests that homozygotes (TrJ mice) may more closely represent a model of congenital hypomyelinating neuropathy, which also exhibits severe hypomyelination and a more severe neurological prognosis than that occurs in patients with CMT1A (15, 16). In particular, some cases of congenital hypomyelinating neuropathy present with defects in respiration, swallowing, and early death (17). Respiratory failure may be caused by central or peripheral mechanisms, such as impaired respiratory drive (18), or peripheral dysfunction or degeneration of NMJs (19). Recently, it was reported that TrJ heterozygote mice, similar to other animal models of CMT1, show axonal degeneration and denervated NMJs (20, 21). However, whether the more severely affected TrJ homozygotes exhibit degeneration of motor axons or NMJs has not been determined. Additionally, although PMP22 is expressed by nonmyelinating Schwann cells of the nerve (22), whether this pattern includes the nonmyelinating subtype at the NMJ (terminal/perisynaptic Schwann cells [T/PSCs]) has not been examined. Because T/PSCs are required to maintain the NMJ (23), toxicity of the TrJ PMP22 mutation within these cells could lead to noncell autonomous dysfunction of the NMJ.
Together with previous reports demonstrating that in some animal models of motor neuron disease, peripheral denervation of NMJs occurs before or in the absence of motor neuron death (24–26), these observations raise the possibility that homozygote TrJ mice die early postnatally because of defects in neuromuscular synaptic function or maintenance. In the present study, we examined the anatomical and physiological characteristics of the NMJ in TrJ mice and showed that although neuromuscular innervation is maintained in multiple muscle subtypes, some abnormal structural and functional features of the NMJ are observed. We discuss these deficits in the context of human congenital hypomyelinating neuropathies.
Trembler-J heterozygote mice were purchased from The Jackson Laboratory (Bar Harbor, ME) and then backcrossed 3 times into a Balb/C background before mating to themselves to generate homozygous TrJ mice. This approach was taken because heterozygote females in this strain were better mothers to homozygote pups (personal communication, L. Notterpek). Wild-type (WT), heterozygote and homozygote mice were genotyped for the L16P point mutation by a combined PCR and BanI restriction digest protocol as previously described (14). Homozygote mice become easily identifiable between P8–P10, growing less quickly than WT and never surviving beyond 17 days of age. In contrast, heterozygote mice begin to shake near P20–P25. For these studies, we exclusively characterized WT and homozygote TrJ (heretofore referred to as TrJ) mice. TrJ and WT mice were killed at “end stage,” or when they were no longer able to right themselves (between P13 and P16). Motor performance was measured by noting the day at which TrJ mice could be overtly distinguished by virtue of their impaired coordination and shakiness. Because we failed to note any difference between male and female mice in our analyses, they were pooled. Animals were either cervically dislocated (for electrophysiological studies) or anesthetized with a mixture of ketamine and xylazine and then transcardially perfused with 4% paraformaldehyde (immunohistochemistry) or with a mixture of glutaraldehyde and paraformaldehyde in sodium cacodylate buffer (electron microscopy). The University of Nevada, Reno Institutional Animal Care and Use Committee approved the experimental use of these animals.
Diaphragm muscles from paraformaldehyde-perfused animals were dissected, rinsed in phosphate-buffered saline (PBS) blocked in 0.1 M glycine for 1hour at room temperature (RT), rinsed again, and then incubated as whole mounts in primary antibody solution overnight at 4°C in PBS containing 10% fetal bovine serum and 1% triton-X. Primary antibodies were rabbit-anti-synaptophysin ([Syp] Santa Cruz Biotechnology, Santa Cruz, CA) or rabbit-anti-S100 (Dako, Carpinteria, CA). The next day, muscles were rinsed and incubated for 1hour at RT in the dark with AlexaFluor 488-conjugated donkey-anti-rabbit antibodies (Jackson ImmunoResearch, West Grove, PA) together with AlexaFluor594-conjugated α-bungarotoxin ([BTX] Biotium, Hayward, CA). The extensor digitorum longus (EDL), soleus, tibialis anterior (TA), and erector spinae muscles were individually dissected from perfused animals, postfixed in 4% paraformaldehyde at 4°C overnight, immersed in 30% sucrose, embedded in 3:2 sucrose:OCT cryoprotectant (Tissue-Tek, Sakura, Torrance, CA), and then snap-frozen in isomethylbutane. Sections were cut at 20- or 40-µm thickness and mounted on Superfrost™ slides pretreated with 0.01% poly-D-lysine and stained with the following antibodies: Syp, S100, goat-anti-vesicular acetylcholine transporter (VAChT) (EMD Millipore, Temecula, CA), rabbit-anti-muscle-specific kinase (kindly provided by L. Mei, University of Georgia). Some slides were incubated at the time of secondary antibody application with AlexaFluor 647-conjugated FasII (kindly provided by W. Thompson, Texas A&M), which binds and labels acetylcholinesterase.
Images of postsynaptic α-BTX-labeled nicotinic acetylcholine receptor (nAChR) clusters (referred to as NMJ) were acquired with the Olympus Ix81 confocal microscope and Fluoview 1000 software using a 63x objective. Using Volumetry G8c, a custom software program developed by author GWH, confocal stacks were rotated for alignment to an orthogonal axis and individual NMJs were isolated from the stack. A normalized threshold, based on a signal-to-noise ratio, was applied to each endplate stack. Maximum fluorescence-intensity maps were created, and depth coloring was used to assess surface curvature and depth. A bounding cube around each NMJ was used to calculate overall NMJ dimensions, and a subvoxel volume routine was used to calculate NMJ volume.
Individual NMJ depth maps were further processed for perforation analysis. Only NMJs located primarily on the top and bottom surface of muscle fibers were analyzed (depth <25μm) to minimize the effect of reduced z-axis resolution in any NMJ located on the side of muscle fibers. Perforations present in isolated NMJs were excluded from analysis if they were less than 0.22μm2 (10 pixels). The size and volume of each perforation were measured.
Semithin sections (1μm) of end-stage TrJ and control diaphragm muscles were cut using an UltraCut ultramicrotome (Type 706201, Leica Microsystems, Vienna, Austria) and stained with toluidine blue. These muscle sections were imaged with a Zeiss Axioskop 2 plus light microscope using a 100x objective (NA 1.25) and a Zeiss Axiocam 105 color camera. Diameters of individual diaphragm muscle fibers were measured utilizing the Feret’s diameter measurement tool in ImageJ (http://rsb.info.nih.gov/ij/index.html), which computes the longest distance between any 2 points along a region of interest’s boundary, and is the most sensitive measure to differences in sectioning angle. Myofiber area measurements were calculated using the Area measurement tool in ImageJ, which uses values and coordinates of neighboring pixels within each selected object or muscle fiber.
Diaphragm muscles from end-stage TrJ and WT mice were dissected and pinned on a Sylgard-coated dish containing oxygenated (95% O2 and 5% CO2) Krebs-Ringer’s solution (121mM NaCl, 5mM KCl, 2mM CaCl2, 1mM MgCl2, 1mM NaH2P04, 12mM NaHCO3, and 11mM glucose), pH 7.3, at RT. After 30minutes of perfusion, the phrenic nerve was drawn into a suction electrode and stimulated for 0.1ms with supramaximal square waves from an S88 stimulator together with a stimulation isolation unit (Grass Technologies, Middleton, WI). Sharp intracellular recording electrodes were made from borosilicate (1mm OD, 0.5mm ID; Sutter Instrument Co, Novato, CA) on a P-97 Flaming-Brown puller of approximately 60 MΩ, which were backfilled with 3M KCl. Signals were amplified and digitized at 2KHz by an AxoClamp 900A amplifier and Digidata 1550 and recorded by pClamp 10 software (Molecular Devices, Sunnyvale, CA). Correct positioning of microelectrodes at the motor endplate of the costal diaphragm was confirmed at the beginning of an experiment by electrophysiological measures (ie, rise-to-peak or 10%–90% rise times of miniature endplate potentials [mEPPs] less than 2ms), as well as by post hoc BTX labeling. mEPP decay times were calculated by measuring the descent time from peak to half amplitude. Endplate potentials (EPPs) were measured after treatment with 2.5 µM Nav1.4 antagonist μ-conotoxin GIIIb ( Peptides International, Louisville, KY). EPPs were normalized to a resting membrane potential of−70mV and the acetylcholine (ACh) reversal potential was assumed to be 0mV (28). EPPs were then corrected for the effects of nonlinear summation (29) using the formula: corrected EPP = EPP/(1-f [EPP/E]), where f=0.8 and E = the difference between the resting and ACh reversal potentials. The number of quanta released in response to a nerve impulse was measured by the direct method by dividing the mean amplitude of normalized and corrected EPPs by the mean amplitude of mEPPs. Only muscle cells with resting membrane potentials between −60 and −75mV were included for analysis. For 40-Hz trains, each value represents the average of 3 EPPs taken at a time after the onset of stimulation. Stimulation episodes were separated by 20minutes to allow recovery.
End-stage TrJ and WT mice were transcardially perfused after making a small incision in the diaphragm (precordial region), with rinse buffer (0.1 M sodium cacodylate) followed by fix buffer (1.5% glutaraldehyde, 2% paraformaldehyde in 0.1 M sodium cacodylate). The intact and straight segment of the phrenic nerve (for proximal axon area and axon/Schwann cell counts), as well as the endplate-enriched medial portion of the costal diaphragm (for distal axon area and diameter as well as NMJ structural analysis) were each dissected and incubated in fixative at 4°C overnight and then in rinse for several hours at 4°C. Nerve and muscle samples were then postfixed in 2% osmium tetroxide for 30minutes, dehydrated in a graded series of ethanol dehydrations, incubated in propylene oxide, embedded in Spurr’s resin, oriented to generate transverse sections of axons or myofibers, respectively, and polymerized at 60°C overnight. Ultrathin sections were cut at 90μm and stained with uranyl acetate followed by lead citrate. Sections were photographed or digitized using a Phillips CM 10 transmission electron microscope equipped with a Gatan BioScan digital imaging system. For intramuscular phrenic nerve branch axon analysis, we cut the central portion of the NMJ-enriched strip to focus on the central-most branch (ie, the most central branch along the dorsoventral axis). For axon/Schwann cell counts, the number of axons and the number of Schwann cells with nuclei in the plane of the section were quantified in proximal phrenic nerve transverse sections.
Differences between TrJ and WT mice were assessed by unpaired Student t-tests assuming equal variance. For physiological studies, data were generated from 3 or more cells per animal, n=3 or greater. For studies of NMJ volume and maturation in which the total number of examined NMJs was different in each genotype, no less than 10 NMJs per diaphragm (volume) or 4 NMJs per diaphragm (maturation) and 3 diaphragms per genotype were included for each comparison. A p<0.05 was considered significant.
We first examined motor incoordination and lifespan of homozygotic TrJ mice backcrossed at least 3 times into the Balb/C locus. TrJ mice were indistinguishable from heterozygote and WT controls until ~P7. At this age, TrJ mice began to appear uncoordinated and moved unsteadily and shakily, reflected as poor motor performance (Fig. 1A). Next, we examined lifespan in TrJ mice. We defined end stage as the day at which animals could not right themselves in less than 30seconds after being placed on their backs, although mice sometimes died even before this end-stage diagnostic. TrJ mice began to die as early as P11, most reaching end stage between P13 and P15; no mice lived beyond P17 (Fig. 1B). Near the end of their lifespans, TrJ mice occasionally exhibited seizure-like episodes, after which they appeared lifeless for up to several minutes, before finally regaining vigor. At end stage, TrJ mice weighed less than their WT littermates (8.8 ± 0.31 vs 7.8 ± 0.47g, p<0.05, WT vs TrJ mice, n=4). When we examined the wet weight of the rectus femoris, we also observed a small but significant difference between genotypes (6.32 ± 0.23 vs 5.58 ± 0.33mg, p<0.05, WT vs TrJ mice, n=5).
In order to determine whether TrJ mice die early postnatally because of motor axon or nerve terminal degeneration, we examined the area and ultrastructure of axons in the phrenic nerve, which is almost exclusively composed of motor axons innervating the diaphragm muscle. We found that in the phrenic nerve of end-stage TrJ mice (although motor axons were severely hypomyelinated), there were no overt signs of Schwann cell pathology or axonal degeneration, such as debris or axon swellings (Fig. 1C). The area of these proximal phrenic motor axons was also unaffected (4.66 ± 1.53 vs 4.29 ± 2.13μm2, p=0.41, WT vs TrJ mice, at least 10 axons per nerve, n=3). However, more distal, intramuscular phrenic nerve branches were smaller in TrJ than in WT mice (area, 6.42 ± 3.74 vs 3.96 ± 2.26μm2, p<0.01; diameter, 2.16 ± 0.3 vs 2.76 ± 0.4μm, p<0.005, WT vs TrJ mice, n=70 axons from n=3 (WT), and 46 axons from n=3 [TrJ] mice). When we quantified the ratio of Schwann cell nuclei per axon, we observed an increase (36.7 ± 5.25 vs 49.8 ± 6.8 glial cells/100 phrenic nerve axons, p<0.05, WT vs TrJ mice, n=4; Fig. 1C), similar to a previous report of sciatic axons (11). Together, these data suggest that while phrenic motor axons fail to exhibit overt pathological signs of degeneration in end-stage TrJ mice, the distal segments of these axons are atrophic.
In order to evaluate the most distal component of the motor neuron, the NMJ, we wholemount immunostained the diaphragm muscle with antibodies against the presynaptic protein synaptophysin and the postsynaptic marker BTX, which labels nAChR clusters at the NMJ. Similar to control, diaphragm from end-stage TrJ mice exhibited a normal gross pattern of innervation, as indicated by the number of BTX-labeled nAChR clusters exhibiting apposed, synaptophysin-immunoreactive nerve terminals (Fig. 2A). We failed to detect denervated nAChRs, terminal sprouts, or other markers of denervation. We did, however, note that the size of BTX-labeled nAChR clusters appeared smaller in TrJ than in WT mice (see below). Since motor innervation of specific muscle subtypes often develops differently and responds differentially to disease and injury (30), we also examined NMJs in cryosections of muscles containing distinct fiber subtypes or originating from distinct embryological masses, even though overt differentiation of fiber subtypes on the basis of myosin heavy-chain expression does not reportedly occur until the third postnatal week, well after the death of TrJ mice (31). Innervation of the soleus (Type I or oxidative fiber), the EDL (Type IIB or fast glycolytic fiber), and the TA (mixed) muscles, similar to the diaphragm (mixed), was unaffected in TrJ mice (Fig. 2A). Innervation of the erector spinae muscles (epaxial muscle group), similar to the diaphragm (hypaxial group) and leg muscles (appendicular group), was unaffected (data not shown). In order to ensure that synaptophysin antibody labeling accurately represented presynaptic nerve terminals, we immunostained sections of these muscles with an antibody against another presynaptic marker, the VAChT. Similar to results with synaptophysin antibodies, VAChT antibodies labeled the presynaptic terminals of all NMJs in each of the muscles examined (Fig. 2B).
We then examined whether T/PSCs at the NMJ were affected at end-stage TrJ mice. S100-immunoreactive Schwann cells (1–2) were localized to the NMJ in both genotypes (Fig. 3A, B), consistent with previous studies (32). We failed to observe S100-positive bridges between NMJs (33), further indicating a lack of peripheral degenerative changes in these mice. Finally, we examined postsynaptic markers such as muscle-specific kinase and acetylcholinesterase, and, similar to pre- and perisynaptic markers, were unable to observe a difference between TrJ and WT mice (Fig. 3B). There was no fragmentation in the pattern of BTX labeling, providing further evidence that these synapses are not in the process of degeneration. Together, these studies show that neuromuscular synapses of end-stage TrJ mice are innervated and contain a normal tripartite complement of cell-specific markers.
We next analyzed the size of individual postsynaptic nAChR clusters because they appeared smaller in end-stage TrJ mice. The height (parallel to long axis of muscle) and volume from single NMJs were extracted and compared between muscles of TrJ and control mice (Fig. 4A). We observed a significant reduction in the volume of BTX-labeled nAChR clusters in the diaphragm and in the soleus and TA but not the EDL (soleus volume, 329.27 ± 87.61 vs 265.41 ± 67.40 µm3, *p < 0.005, WT vs TrJ mice, 65 NMJs from n=3 [WT] and 80 NMJs from n=3 [TrJ]; EDL volume, 330.83 ± 86.49 vs 314.91 ± 93.70 µm3, p=0.56, WT vs TrJ mice, 46 NMJs from n=3 [WT] and 31 NMJs from n=3 [TrJ]; TA volume, 553.31 ± 103.21 vs 315.80 ± 66.69 µm3, **p < 0.001; WT vs TrJ mice, 46 NMJs from n=3 [WT] and 60 NMJs from n=3 [TrJ]; diaphragm volume, 1265.72 ± 265.11 vs 605.58 ± 87.70 µm3, **p < 0.001, WT vs TrJ mice, 55 NMJs from n=3 [WT] and 101 NMJs from n=3 [TrJ]; Fig. 4B–D). In order to determine whether this reduction in NMJ size was due to a concomitant decrease of muscle size, we stained semithin sections of transverse sections of diaphragm with toluidine blue (Fig. 4E). Both the average cross-sectional area and Feret’s diameter were significantly reduced in TrJ vs WT mice (area, 272.7 ± 78.6 vs 244.7 ± 69.8μm2, p<0.05, area; 24.66 ± 5.2 vs 22.12 ± 3.7μm, Feret’s; p<0.005, WT vs TrJ mice, 70 cells per animal, n=3). The relative increase in NMJ volume between WT vs TrJ mice was 2.1-fold, whereas the corresponding increases in overall weight, muscle wet weight, muscle fiber diameter, and muscle fiber area were all between 1.1-fold and 1.3-fold, suggesting that the reduction of NMJ size was largely not due to these factors.
Next we performed an ultrastructural analysis of diaphragm NMJs. T/PSCs were present at the synapse in both TrJ and WT mice, and in some cases the cytoplasm of these cells could be seen to terminate partially but not totally into the synaptic gutter (Fig. 5). When we measured the depth of junctional folds in single cross sections of NMJs, we observed a statistically significant reduction in TrJ vs WT mice (individual fold depth, 1.43 ± 0.18 vs 1.03 ± 0.32μm, p<0.05, WT vs TrJ mice, n=5 folds per NMJ, 3 NMJs per animal, 3 animals), consistent with the results obtained above by BTX volume analysis.
The reduction of NMJ size may result from atrophy or delayed maturation. The NMJ undergoes a defined change in structural complexity during the first several weeks of postnatal development (34). For example, during postnatal week 1, NMJs exhibit a plaque-like morphology, characterized by an oval shape, minimal folds, and little or no perforations. During the second and third postnatal weeks, NMJs adopt a more pretzel-like structure, displaying a greater number of invaginations and perforations (35). We could also identify an intermediate morphology where some but not all of the plaques disappeared and were replaced by folds and nascent perforations. Hence, we divided the NMJs of the diaphragm into 3 categories and observed that while the number of NMJs exhibiting intermediate maturation was approximately even between genotypes, the number of immature, plaque-like NMJs was significantly higher in TrJ than in WT mice, and the number of more mature, pretzel-like NMJs was greater in WT than in TrJ mice (Fig. 6B). Next, we performed a detailed analysis that measured the total number of perforations per NMJ, the average area of individual perforations and the percentage of NMJ area that were occupied by perforations. NMJs in control mice exhibited larger perforations than those in TrJ mice (perforation hole number, 2.2 ± 1.08 vs 1.7 ± 0.95 holes/NMJ, p=0.25; perforation size, 3.66 ± 3.46 vs 1.57 ± 1.45μm2, *p < 0.05; perforation area, 5.03 ± 3.87 vs 2.35 ± 2.14% of total NMJ area, *p < 0.05, WT vs TrJ mice, 33 perforations from 15 NMJs from n=3 [WT], and 17 perforations from 13 NMJs from n=3 [TrJ] diaphragms; Fig. 6C). The difference in size and maturation of NMJs between genotypes could also be observed when they were labeled with S100 antibodies marking T/PSCs (Fig. 3B). These results, therefore, favor the idea that the reduced volume of NMJs observed in end-stage TrJ mice results at least in part because of a delay in maturation.
To determine whether mutant NMJs exhibited a difference in neuromuscular function in addition to the delay in motor endplate maturation, we performed standard electrophysiological analysis of the diaphragm from end-stage TrJ and control mice. Although we failed to detect a difference in the resting membrane potential between genotypes (-70.2 ± 4.0mV vs -68.8 ± 3.0mV, p=0.621, WT vs TrJ mice), when we evoked endplate potentials by stimulating the phrenic nerve in the presence of the Nav1.4 antagonist, μ-conotoxin (conotoxin GIIIb), we observed a significant decrease in the size of the EPP (amplitude, 25.2 ± 0.93 vs 22.9 ± 1.1mV, *p < 0.05, WT vs TrJ mice; Fig. 7A,B). To determine if this reduction was caused by changes in quantal size, we examined the amplitudes and durations of mEPPs, which reflect the postsynaptic response to individual quanta (Fig. 7A, B). Neither of these features was affected (mEPP amplitude, 1.54 ± 0.16 vs 1.56 ± 0.10mV, p=0.862; mEPP rise-to-peak, 1.38 ± 0.62 vs 1.06 ± 0.11ms, p=0.294, mEPP decay time, 2.72 ± 0.44 vs 2.74 ± 0.89ms, p=0.970, WT vs TrJ mice). Quantal content was therefore decreased in TrJ vs WT mice (12.77 ± 0.37 vs 11.70 ± 0.45 quanta/EPP, *p < 0.005, WT vs TrJ mice). Next, we examined the frequency of mEPPs and found that it was significantly reduced in TrJ vs WT mice (1.38 ± 0.19 vs 0.38 ± 0.18 events/s, **p < 0.01, WT vs TrJ mice), suggesting that the lower EPP amplitude obtained from recordings in these mutants reflects a presynaptic deficit (Fig. 7C, D). Next, we examined the response of NMJs to high-frequency nerve stimulation, which induces the depression of ACh release at neuromuscular synapses via presynaptic mechanisms. Moreover, this form of short-term plasticity is also modulated by T/PSCs (36). While NMJs in both genotypes exhibited a quick initial reduction of EPP amplitude, TrJ mice exhibited an exaggerated rundown in synaptic transmission as the period of high-frequency stimulation persisted (Fig. 8A, B). Additionally, high-frequency stimulation failed to trigger EPPs occasionally after ~30s in TrJ mice (Fig. 8C, D). This complete lack of nerve-evoked ACh release was never observed in control mice. Therefore, these data indicate that NMJs in TrJ mice exhibit a presynaptic deficit that causes an enhanced rundown in neuromuscular transmission.
Because homozygous TrJ mice exhibit a congenital lack of myelination and die early in the postnatal period, we tested the hypothesis that this severely shortened lifespan was caused by a failure to maintain neuromuscular synapses, leading to muscle atrophy and paralysis. We failed to find evidence supporting this idea. Across multiple muscle subtypes examined within 1–2 days prior to death, NMJs of TrJ mice contained qualitatively normal expression of markers associated with pre-, peri-, and post-synaptic elements. These findings are in contrast to congenital models of motor neuron disease, such as spinal muscular atrophy, which exhibit anatomical as well as functional evidence of synaptic degeneration at the NMJ (37). They are also different than results obtained from adult, heterozygote TrJ mice as well as other CMT1 mouse models, which exhibit a modest but significant loss of neuromuscular synaptic maintenance (7, 8, 21). Together with the absence of severe muscle atrophy, the lack of peripheral denervation in TrJ mice suggests that the early lethality in this mouse is not caused by a failure to maintain neuromuscular synaptic connections.
The NMJ is abnormal in several muscles of end-stage TrJ mice, however, indicating that the development of this synapse is affected in the absence of myelination. Specifically, the size and complexity of BTX-labeled postsynaptic endplates as well as the depth of postsynaptic junctional folds were reduced in the diaphragm, relative to WT mice. There was no loss of postsynaptic muscle-specific kinase or acetylcholinesterase, however, suggesting that key functional elements of the postsynaptic apparatus were intact. Similarly, the expression of S100 and the number of nuclei affiliated with this protein were similar in WT and TrJ NMJs, indicating that nonmyelinating T/PSCs were not overtly affected. These results instead suggest that the maturation of the NMJ is impaired in TrJ mice. The significance of impaired NMJ maturation is unclear, but a recent report showed that a mouse model of CMT2 exhibits similarly altered NMJ maturation in a subset of muscles that precedes subsequent degeneration (38). Interestingly, similar to NMJs of the diaphragm, NMJs of the soleus and tibialis anterior, but not EDL muscles, were smaller in TrJ than in WT mice. Although these muscles in the adult express different complements of myosin isoforms, with the EDL expressing the highest percentage of fast-fatigable Type IIB fibers, the development of these isoform expression subtypes typically occurs later than the time of lethality of homozygous TrJ mice (31), suggesting that other features underlie the differential sensitivity of these muscle subtypes to NMJ maturation. In addition to the decreased size of NMJs as deduced by quantitative BTX analysis, the distal but not proximal phrenic nerve is reduced in size in TrJ mice. We attribute these changes to impaired growth rather than to atrophy, based on the assessment of NMJ size and maturation discussed above, as well as the lack of overt axonal pathology in either proximal or distal phrenic nerves.
Synaptic function is also abnormal in the diaphragm of end-stage TrJ mice. Individual EPPs were smaller and the frequency but not amplitude or duration of mEPPs was reduced in these mice relative to control, suggesting a presynaptic deficit. Whether this deficit precedes the structural postsynaptic alteration is unclear but could be tested by performing a time course of functional and anatomical studies. A more severe deficit in synaptic transmission was observed in response to high-frequency (40Hz) phrenic nerve stimulation. Although the initial reduction to 80% of transmitter release within the first second of such stimulation was similar between genotypes and to previous studies (27, 39), NMJs in TrJ mice exhibited a more pronounced reduction in release as the stimulation continued. Together with the reduced amplitude of the initial EPP, and based on a safety factor at the vertebrate NMJ between 2 and 4 (40), this reduction is predicted to cause neuromuscular transmission failure when the initial control EPP of ~26mV reaches 6.5–13mV, or within 10seconds (vs 45seconds in control), after the onset of stimulation. Even more strikingly, after a period of 30–35seconds of high-frequency stimulation, NMJs in TrJ mice exhibit a complete lack of transmitter release. In contrast, such total failure was never observed in control mice even after a minute of 60-Hz stimulation (data not shown). Together, these results suggest that NMJs in TrJ mice are unable to maintain synaptic transmission in response to high-frequency stimulation. Future studies will be aimed at elucidating the mechanisms of this effect, such as impairments of synaptic vesicle mobilization or synaptic currents. For example, it has been reported that conotoxin GIIIb reduces the potassium current (and thus quantal content) at motor nerve terminals through an action on presynaptic sodium channels (41), an effect that could account for the reduced EPP in TrJ mice.
Although it is tempting to speculate that the failure to release neurotransmitter in response to 40-Hz stimulation contributes to respiratory failure, the present data are too preliminary to support this conclusion. First, while 40-Hz stimulation is well within the phrenic nerve firing rates obtain from ventilated adult rats (42), these frequencies are typically maintained for 1second during an inspiration, followed by an intercycle rest period, resulting in a use- or duty-cycle of ~0.35 (43). This difference may affect the interpretation of our results because less EPP rundown was observed in response to phasic vs continuous stimulation (39). Second, it would be useful to determine whether the differences observed at the endplate translate into functional deficits by measuring respiratory and diaphragm function via plethysmography and isometric tension measurements, respectively. Third, whether failed EPPs result from deficits in ACh release or from impaired nerve conduction remains unclear, but could be addressed by studies of nerve conduction velocity and distal nerve propagation via compound muscle action potentials or extracellular axon segment recording, respectively. Evidence in support of impaired conduction comes from studies of heterozygote TrJ mice, which exhibit significant deficits in motor nerve velocity (44). Together, the results from these studies would provide insight into whether the deficits reported here at the NMJ contribute to the early lethality of homozygous TrJ mice. Alternatively, alterations in central white matter in TrJ mice may contribute to early lethality because PMP22 is expressed, albeit at lower levels, in the CNS (45). Consistent with this idea, reductions in the volume of CNS white matter as well as cognitive impairment have been reported in humans with CMT1A (46).
Together, our studies describe a set of specific structural and functional deficits in the neuromuscular system of TrJ mice, which we propose as an excellent animal model of congenital hypomyelinating neuropathy. Surprisingly, these mice exhibit only modest structural alterations at the NMJ in contrast to the evidence of synaptic degeneration seen in other models of CMT1. Rather, the principal neuromuscular deficit in TrJ mice is functional. Future studies will examine motor nerve conduction velocities in these mice to determine if they are impaired and whether they underlie the peripheral transmission deficits we observed in this study, as would be expected based on the severe hypomyelination of peripheral nerves in these mice. In summary, our data suggest that neurological deficits associated with congenital hypomyelinating neuropathy are likely caused by a functional impairment of synaptic transmission at the NMJ.
We thank Andrea Agarwal and Chris von Bartheld for assistance with the use of the electron microscope, work which was performed in a Core laboratory supported by NIH grant GM103554.