Journal of Neurophysiology

Error message

Notice: PHP Error: Undefined index: custom_texts in highwire_highwire_corrections_content_type_render() (line 33 of /opt/sites/jnl-jn/drupal-highwire/releases/20151124215058/modules/highwire/plugins/content_types/

Reduced Endplate Currents Underlie Motor Unit Dysfunction in Canine Motor Neuron Disease

Mark M. Rich, Robert. F. Waldeck, Linda C. Cork, Rita J. Balice-Gordon, Robert E. W. Fyffe, Xueyong Wang, Timothy C. Cope, Martin J. Pinter


Hereditary canine spinal muscular atrophy (HCSMA) is an autosomal dominant degenerative disorder of motor neurons. In homozygous animals, motor units produce decreased force output and fail during repetitive activity. Previous studies suggest that decreased efficacy of neuromuscular transmission underlies these abnormalities. To examine this, we recorded muscle fiber endplate currents (EPCs) and found reduced amplitudes and increased failures during nerve stimulation in homozygotes compared with wild-type controls. Comparison of EPC amplitudes with muscle fiber current thresholds indicate that many EPCs from homozygotes fall below threshold for activating muscle fibers but can be raised above threshold following potentiation. To determine whether axonal abnormalities might play a role in causing motor unit dysfunction, we examined the postnatal maturation of axonal conduction velocity in relation to the appearance of tetanic failure. We also examined intracellularly labeled motor neurons for evidence of axonal neurofilament accumulations, which are found in many instances of motor neuron disease including HCSMA. Despite the appearance of tetanic failure between 90 and 120 days, average motor axon conduction velocity increased with age in homozygotes and achieved adult levels. Normal correlations between motor neuron properties (including conduction velocity) and motor unit properties were also observed. Labeled proximal motor axons of several motor neurons that supplied failing motor units exhibited little or no evidence of axonal swellings. We conclude that decreased release of transmitter from motor terminals underlies motor unit dysfunction in HCSMA and that the mechanisms determining the maturation of axonal conduction velocity and the pattern of correlation between motor neuron and motor unit properties do not contribute to the appearance or evolution of motor unit dysfunction.


The motor neuron diseases are a collection of complex degenerative disorders that have in common motor neuron cell death and degeneration (Mitsumoto et al. 1998;Rowland 1994). Most animal model studies thus focus on understanding the mechanisms that underlie motor neuron cell death and degeneration and define the pathogenesis in terms of these phenomena. Motivating this approach is the hope that, by inhibiting or preventing cell death and/or degeneration, loss of motor unit function will be inhibited and disease progress slowed. In several instances, however, inhibiting cell death in animal models has not halted disease progression (Klivenyi et al. 1999; Li et al. 2000; Sagot et al. 1995, 1996), and effective therapy remains elusive.

In studying a canine model of motor neuron disease [hereditary canine spinal muscular atrophy (HCSMA)], we have used a different approach that focuses instead on defining how motor unit function is initially lost (Pinter et al. 2001a). One reason for this approach is concern that motor neuron cell death and degeneration are endstage phenomena that may not necessarily share mechanisms that provoke loss of motor unit function. An important advantage of our approach is that pathological changes observed in axons, motor terminals, or cell bodies can be related to the causes of weakness (i.e., motor unit dysfunction) to achieve a functionally based understanding of the pathogenesis.

HCSMA is an autosomal dominant, degenerative disorder of motor neurons that shares pathological features with human motor neuron disease (Cork et al. 1982a; Pinter et al. 2001a). Motor unit dysfunction appears to be the most important cause of weakness in HCSMA. In homozygous animals, this dysfunction appears first in slow motor units and manifests as reduced motor unit force output and as an inability to sustain motor unit force output during repetitive activity, termed tetanic failure (Pinter et al. 1995,2001a).

The cellular mechanisms that give rise to motor unit dysfunction in HCSMA are unknown. The dependency of tetanic failure on stimulus frequency and the ability of 4-aminopyridine (4AP) to increase motor unit force output suggest that defective or insufficient release of ACh from motor terminals is involved (Pinter et al. 1997,2001a) but, thus far, direct evidence has not been obtained. We recently found that motor unit tetanic failure in the medial gastrocnemius (MG) muscle of HCSMA homozygotes is not associated with atrophy of the neuromuscular junction (NMJ) or evidence of motor terminal degeneration (Balice-Gordon et al. 2000). This also suggests that motor unit failure may result from functional deficits at the NMJ but it is not clear whether these deficits are localized to the NMJ or are part of a more generalized pattern of motor neuron failure. Evidence of axonal involvement in HCSMA suggests that axonal abnormalities might contribute directly or indirectly to motor unit or neuromuscular transmission failure. Findings include evidence that motor axon size is reduced in HCSMA homozygotes (Cork et al. 1989) and evidence of decreased mRNA for the light neurofilament protein (Muma and Cork 1993). In addition, large accumulations of maloriented neurofilaments are found in some proximal motor axons in HCSMA (Cork et al. 1982b, 1988). Similar aggregations are commonly observed in other forms of motor neuron disease (Hirano 1991; Williamson et al. 1996). These aggregations in particular and abnormalities of the cytoskeletal components of axons in general have received much emphasis as contributing factors mediating the pathogenesis of motor neuron disease (Julien 1995), but when they appear relative to loss of motor unit function is not known.

To gain further insight into the pathogenesis of HCSMA, we examined the status of neuromuscular synaptic transmission in HCSMA homozygotes by recording endplate currents (EPC) in response to nerve stimulation. In this paper, we also investigated axonal function and structural properties during the evolution of motor unit dysfunction. In particular, we studied the postnatal time course of conduction velocity maturation in HCSMA homozygotes in relation to the onset of motor unit dysfunction and the relationships between axonal conduction velocity and motor unit mechanical properties. In addition, we intracellularly labeled several motor neurons in one homozygote after characterizing motor unit dysfunction and examined directly for the presence of swellings in proximal motor axons. Our results demonstrate that EPCs are reduced at homozygote NMJs to the extent that the ability to generate muscle fiber action potentials is likely compromised. This reduction of amplitude is associated with increased EPC failure in response to nerve stimulation. Axonal conduction velocity appears to mature normally during the evolution of motor unit dysfunction, and axonal neurofilament aggregates appear to play no role in either the onset or evolution of motor unit dysfunction in HCSMA. The following paper presents a quantal analysis of neuromuscular transmission in HCSMA homozygotes and provides evidence for a presynaptic deficit of transmitter release at HCSMA homozygote motor terminals (Rich et al. 2002). A portion of these results have been reported in abstract form (Pinter et al. 2001b).


Animals used in this study were obtained from the HCSMA breeding colony or from vendors providing purpose-bred normal animals. All work reported in this paper was approved by Institutional Animal Use and Care Committees.

Homozygous dogs can be identified by 6–8 wk after birth by the appearance of weakness in the tail muscles. In this study, data obtained from homozygotes have been compared with data from adult normal members of the HCSMA pedigree, from symptomless younger animals from the HCSMA colony, and from genetically normal, control animals. We use the term “symptomless” because at ages of less than about 1 yr, HCSMA heterozygotes cannot be readily distinguished from phenotypically normal HCSMA animals. Heterozygotes lack motor symptoms throughout the postnatal interval when homozygotes can be studied (2–6 mo) and begin developing signs about 10 mo after birth (Cork et al. 1982a).

Surgical preparation


One collection of HCSMA homozygotes and symptomless animals from the HCSMA pedigree was used in terminal experiments involving measurement of motor unit force and motor neuron electrical properties or intracellular staining of motor neurons. All dogs were initially anesthetized with 35–40 mg/kg iv sodium pentobarbital. Supplemental doses were administered during the experiment via an intravenous cannula to maintain an absence of withdrawal and corneal reflexes. A tracheal cannula was inserted to maintain a patent airway and to provide for monitoring of end-tidal CO2. Blood pressure was continuously monitored via an arterial cannula. Rectal temperature was monitored and maintained at 37 to 38°C with a heating pad and infrared lamps.

The left MG muscle and its nerve were exposed, and the muscle and its tendon were dissected free of the accompanying lateral gastrocnemius muscle and surrounding connective tissue as much as was safely possible and prepared for attachment to a strain gauge for motor unit force recording. The strain gauge was capable of detecting minimal forces of about 100–200 mg. All force recording was performed using a servo-operated device that maintained muscle preload at 100 g. The MG nerve was freed of surrounding tissue and mounted on a pair of bipolar stimulating/recording electrodes. Fine stainless steel wires (2–4 pairs) were inserted into the MG muscle and connected to high-gain differential amplifiers for recording of motor unit electromyographic (EMG) activity. A laminectomy was performed to expose the lumbosacral spinal cord. The animal was then mounted in a frame that immobilized the vertebral column and the left hindlimb. Exposed tissues were covered with warm mineral oil or Vaseline.

At the conclusion of surgical preparations, all animals received bilateral pneumothorax and were mechanically respirated to minimize respiratory-associated movements during intracellular recording. All animals remained unparalyzed.


Muscle fiber samples for EPC or muscle fiber voltage recording were obtained from a separate group of HCSMA homozygotes and normal animals. For these studies, animals were anesthetized and monitored as described above. Either the left or right MG muscle was freed as completely as possible from surrounding connective tissue so that the muscle could be elevated, allowing an elongated metal support platform to be placed against the deep surface of the muscle. Incisions from the superficial to the deep muscle surface were made orthogonal to the medial–lateral muscle dimension following the paths of tendinous inscriptions visible on the muscle surface. In this way, strips of muscle fibers were obtained that extended the length of the muscle, were about 5 mm in thickness, and contained fibers running uninterrupted between the deep and superficial tendinous surfaces of the muscle. Smaller samples were blocked from these strips and placed into Sylgard-lined dishes through which oxygenated solution flowed. Unless otherwise noted, the composition of the bathing solution was (in mM) 118 NaCl, 3.5 KCl, 2.0 CaCl2, 0.7 MgSO4, 26.2 NaHCO3, 1.7 NaH2PO4, 5.5 glucose (pH 7.3–7.4, bubbled with 95% O2–5% CO2).

Recording and staining procedures


Single MG motor neurons or motor axons were identified by antidromic activation following electrical stimulation of the MG nerve using an intensity 5× threshold for the appearance of whole-muscle EMG. Intracellular recording from ventral root motor axons was performed as described previously (Cope and Clark 1991). Intracellular records were obtained using micropipettes filled with 3 M KCl and connected to a conventional amplifier. Following impalement of an MG motor neuron soma or motor axon, data were collected for measuring motor axon conduction velocity. Conduction velocities were determined by dividing the conduction distance between the spinal cord (or ventral root impalement site) and the peripheral nerve stimulation site (measured postmortem) by the time interval between the peripheral nerve stimulation onset and the arrival of an antidromic action potential. In the case of somatic impalement, data for measuring several motor neuron electrical properties such as afterhyperpolarization were obtained as described previously for cat (Pinter et al. 1991).

To determine whether motor axon conduction velocities are slower proximally within the spinal cord or ventral root, we compared in several experiments conduction velocities measured from 20 MG motor neurons impaled within the spinal cord with conduction velocities determined by impaling 20 MG ventral root axons. Conduction velocities measured from motor neuron soma recordings are called “proximal” and include conduction through intraspinal and ventral root portions of the motor axon. Conduction velocities measured from ventral root axonal impalements are termed “distal” and do not include conduction through intraspinal and about half of the ventral root portions of the motor axon.


Single motor units were activated using 0.5-ms depolarizing current pulses delivered through the micropipette. Motor unit twitch data (twitch contraction force and time-to-peak) were collected first. Fused tetanic contractions were collected for stimulus trains of 50 to 200 Hz. Some HCSMA homozygote motor units fail to sustain fused contractions during tetanic stimulation, a phenomenon called tetanic failure (Pinter et al. 1995). Tetanic failure was measured as the difference between unit peak force and that present at the end of the stimulus train, normalized to the peak force and expressed as a percentage. Units showing a difference of 7% or greater are considered to exhibit tetanic failure. This percentage exceeds by 2% the estimated maximum variability of force baseline recordings. For units exhibiting failure, maximum tetanic force was measured as the peak force achieved during the axon or cell body stimulus train. Note that tetanic failure reflects the inability of a motor unit to maintain electrical activity of muscle fibers during an individual tetanic activation and does not reflect motor unit fatigue determined by muscle fiber fatigue as defined by Burke (1981).


Muscle samples obtained as described above were dissected further under a microscope into small strips in which nerve fiber bundles could be seen to pass among the muscle fibers. The strips were then stretched and pinned to the bottom of Sylgard-lined dishes. To prevent contractions, muscle fibers were crushed at sites approximately 2–3 mm on either side of the endplate region, which was reliably located in the middle of the muscle sample. The crushing produced muscle fibers with resting potentials in the range of −30 to −40 mV. Contractions generally were blocked for 2–3 h following the crush, after which new crushes were attempted or the muscle sample was replaced. These observations are similar to those made previously during EPC recording in crushed rat fibers (Argentieri et al. 1992). A concentric bipolar stimulating electrode (center pole diameter 250 μm) was used to stimulate the muscle nerve bundles with 0.1-ms duration constant current pulses. The output of the stimulator was capacitively coupled to the electrode to prevent passage of DC polarizing currents and to produce a bipolar-shaped stimulus current (Guyton and Hambrecht 1974).

Following pinning, muscle strips were stained with 10 μM 4-Di-2ASP to allow for visualization of endplates and muscle fibers (Magrassi et al. 1987; Rich and Lichtman 1989) and imaged with an upright epifluorescence microscope (Leica DMR). Once the endplate band was identified, a pair of glass micropipettes, filled with 3 M KCl, was used for separate voltage recording and current passage. The EPCs evoked by nerve stimulation were voltage-clamped using conventional voltage-clamp equipment (Axon Instruments, Foster City, CA). All EPC recording occurred within 250 μm of endplates, which provided full voltage control over the EPC amplitude as demonstrated by linear changes of EPC amplitude produced by varying the holding potential (data not shown). The electrode filling solution also contained 10 ng/ml sulforhodamine to enable fluorescence visualization of the electrode tips. Electrode resistances were generally about 10 MΩ. Holding potentials were usually −45 mV, but in several experiments −60 mV was used. In several instances multiple holding potentials were used to obtain current-voltage relationships, which were used to adjust the measured amplitudes of final records to different holding potentials. Final EPC records were averages of 16–30 sweeps collected at 0.5 Hz. To obtain records of potentiated EPCs, a nerve stimulus paradigm was used that was similar to the paradigm used previously to obtain records of potentiated motor unit twitch contractions in HCSMA homozygotes (Pinter et al. 1995). Trains of 20–30 nerve stimuli (150 Hz) were interleaved with single pulse stimuli at a rate of one stimulus event (single pulse or train) per 2 s until the EPC amplitude evoked by single pulse stimuli had reached a maximum. Under these in vitro conditions, this generally required four to eight stimulus trains.

In several experiments, current thresholds for action potential initiation were obtained from MG muscle fibers. Muscle fibers were not crushed for this recording, and dantrolene (6 mg/ml) was used to inhibit contractions. A pair of micropipette electrodes, filled as described above but with higher resistances (>20 MΩ), was used for separate voltage recording and current passage. In all cases, current injections were located within 250 μm of endplates. Current thresholds were determined using variable amplitude depolarizing waveforms that mimicked the shape of an EPC (time-to-peak, 1 ms; half-width, 3 ms; final time constant, 2.5 ms) and were applied to the current electrode by a D/A converter. In each examined fiber, threshold was determined by slowly increasing the amplitude of this waveform until an action potential occurred. Data were collected only from fibers with resting potentials more negative than about −70 mV.

All muscle fiber recording of EPCs and action potentials was performed at room temperature.


In one homozygous animal aged 144 days, MG motor neurons were intracellularly stained with the dye Neurobiotin. Following characterization of motor unit properties, depolarizing current pulses (400 ms duration, 5–10 nA, 1 Hz for 10–30 min) were used to iontophoretically inject 4% Neurobiotin (Vector Laboratories, Burlingame, CA) in 2.5 M KCl and 10 mM HEPES, pH 7.4. Following a 2-h delay from the final injection to allow for dye diffusion, the animal was transcardially perfused with PBS, followed by 4% paraformaldehyde in PBS. Motor neurons were injected with a minimum rostral–caudal spacing of about 200 μm.


Spinal cords were postfixed for 4 h and rinsed in PBS for 8–12 h, and vibratome sections (50–100 μm) were collected into PBS. Sections were permeabilized in 100% MeOH at 20°C, rinsed in PBS, and incubated in 50 μg/ml fluorescein-conjugated streptavidin (Molecular Probes, Eugene, OR) with shaking at 4°C overnight. After rinsing in PBS, sections were mounted on slides in an antifading medium (VectaShield, Vector Laboratories), and slides were stored at 4°C in the dark. Sections were imaged and analyzed using a confocal microscope (TCS-4D system, Leica, Eagle, PA) and interactive software. All confocal images illustrated are single-plane projections of a multiple-plane stack of images.

Following Neurobiotin staining and confocal microscopic analysis, coverslips were removed, sections rinsed in PBS, blocked in 2% BSA, 0.1% Triton X-100 in PBS, and immunostained using anti-phosphorylated neurofilament antibody (SMI34, 1:500 dilution, Sternberger Monoclonals, Baltimore, MD) followed by rhodamine-conjugated donkey anti-mouse IgG1 secondary antibody (Jackson Immunological). After rinsing, sections were recoverslipped in VectaShield, and sections were reimaged using confocal microscopy.


Following all imaging of fluorescence-labeled cells and processes, sections were processed for Neurobiotin visualization with peroxidase to enable detailed reconstruction under brightfield illumination and to avoid possible loss of structural detail because of photobleaching by prolonged fluorescent illumination. This was accomplished by removing the coverslips and processing the sections (on the slide) with an ABC kit (Vector Laboratories) and visualizing the peroxidase with 0.02 mg/ml diaminobenzidine (Sigma) and 0.01% H2O2 in 0.05 M Tris. The reaction was enhanced by adding 0.01% nickel ammonium sulfate (Alvarez et al. 1997).


Data from groups of animals were tested for phenotype-related differences (i.e., HCSMA homozygote versus normal) using nested ANOVA (Neter et al. 1990). Parallelism among linear regression slopes for individual experiments (i.e., individual animals) was tested using analysis of covariance. Phenotype-related differences in they-intercepts of linear regressions from individual experiments were tested using nested analysis of covariance. In some cases, differences between distributions were tested using the Kolomogorov–Smirnov test. All mean values are shown ± SE.



The results of previous studies suggest that motor unit dysfunction in HCSMA homozygotes is based on defective neuromuscular transmission (Pinter et al. 1995, 1997). To examine this possibility, we compared EPCs obtained from MG muscle fibers of 5 HCSMA homozygotes aged 118–144 days (average 132 days) with EPCs obtained from 5 genetically normal animals aged 146–167 days (average 152 days). All 5 of the examined homozygotes exhibited extensive weakness and atrophy of proximal musculature typical of HCSMA (Cork et al. 1982a; Pinter et al. 2001a). As shown in Fig.1, mean EPC amplitudes from homozygotes (−45 mV holding potential) were all less than the lowest mean EPC amplitude from control MG muscles. A nested ANOVA showed that mean EPC amplitudes from homozygotes were significantly less than normal (P < 0.01). We considered the possibility that this difference might be related to the age differences between the homozygote and normal populations. To test this, we performed a separate nested ANOVA on the group of data aged 143–148 days (which includes 2 homozygotes and 2 normal animals) and found that EPC amplitudes were significantly different (P < 0.01) despite the nearly identical ages.

Fig. 1.

Homozygote endplate currents (EPCs) exhibit decreased amplitudes. Bar chart shows average values of EPC current (±SE) for 5 homozygotes and 5 normal animals. Data bars are arranged in the order of ascending age (left to right) and each age is shown on the abscissa. A minimum of 13 EPCs were recorded in each experiment. Measured EPC amplitudes were adjusted to a holding potential of −45 mV.

Muscle fiber diameters measured during recording were on average smaller among homozygotes (30.4 μm ± 1.5, n = 5 homozygotes versus 38.2 μm ± 1.7, n = 5 normal;P = 0.01, nested ANOVA). Only weak correlations were observed between EPC amplitude and fiber diameter within experiments or within phenotype (homozygote, Pearson r = 0.09; normal,r = 0.16; P > 0.05). Thus differences in EPC amplitudes between homozygotes and normal animals are independent of differences in fiber sizes. Muscle fiber size differences between homozygotes and normal animals may be related to the age differences noted above or perhaps to the inability of HCSMA homozygotes to achieve the mobility of normal animals.

Associated with reduced EPC amplitude in homozygotes was the appearance of EPC failure in response to nerve stimulation at 2 Hz. While no EPC failures were encountered in 112 recordings from five normal animals, an average of 35% (range 14–62%) of EPC recordings in homozygotes (173 recordings) showed failure. Among individual EPC recordings that exhibited failure, failure occurred in about 15% of trials (±2.3%, range: 11–24%, n = 5 homozygotes). Overall, more than half (54%) of homozygote EPCs with mean amplitudes of 8 nA or less (−45 mV holding potential) exhibited failure in some stimulus trials.


An important question concerns whether homozygote EPC amplitudes are reduced in relation to threshold currents for action potential generation in muscle fibers. Such a relative reduction could help explain why motor unit forces of HCSMA homozygotes are reduced (Pinter et al. 1995). Since muscle fiber current threshold could not be obtained from crushed muscle fibers used to record EPCs, we compared the measured distributions of EPC amplitude with the distributions of current threshold obtained from uncrushed MG muscle fibers. In a subset of homozygous and normal animals, we measured MG muscle fiber action potential thresholds to current waveforms designed to simulate the shape of EPCs. A bar chart illustrating mean current thresholds sampled from ≥12 muscle fibers in two homozygotes and two normal animals is shown in Fig.2 A. A nested ANOVA indicated that no significant differences existed among these means that could be related to phenotype (homozygote versus normal, P > 0.05). No significant phenotype-related differences in resting membrane potential were found (P > 0.05), but there was a clear indication that differences in the resting potential present when the threshold was determined influenced the current threshold measurements (Fig. 2 B); the homozygote exhibiting the lowest average current threshold also possessed the most depolarized average resting potential. Further analysis of these data showed no significant differences in linear regression parameters relating threshold current (dependent variable) to resting potential among all examined animals (analysis of covariance for slopes and nested analysis of covariance for y-intercepts; both tests, P > 0.05). Data from homozygotes and normal animals were thus pooled to determine overall linear regression parameters. These parameters were used to adjust the measured values of current threshold to a uniform resting potential of −80 mV, which appears to be approximately the normal muscle fiber resting potential in dog muscle fibers as in other species (Rich et al. 1998). These procedures produced a mean value for MG muscle fiber current threshold of 50 nA and a SD of 16 nA.

Fig. 2.

Muscle fiber current thresholds and resting potentials.A: bar chart illustrating mean values (±SE) of medial gastrocnemius (MG) muscle fiber threshold to intracellularly injected currents for hereditary canine spinal muscular atrophy (HCSMA) homozygotes and genetically normal control animals. Data include samples of ≥12 current thresholds from each animal. B: plot of muscle fiber current threshold versus resting membrane potential for HCSMA homozygotes and genetically normal control animals. Data were obtained from 2 homozygotes and 2 control animals. Symbols defined by inset.

To provide a standardized basis for comparing the derived current threshold distribution with measured EPC amplitudes, averaged linear regression parameters obtained from several samples of the relationship between EPC amplitude and holding potential were used to adjust measured EPC amplitudes to a holding potential of −80 mV. Figure3 A illustrates the distributions of these adjusted EPC amplitudes for five HCSMA homozygotes and five normal animals; data for each animal are shown in separate vertical lanes. To set a reasonable lower limit for the range of expected current thresholds at −80 mV, we selected a value of 2 SDs below the average current threshold (18 nA, see above). As shown in Fig. 3 A, almost all EPC amplitudes from normal animals are greater than this value, consistent with our observation that control animals exhibit no motor unit force failure (Pinter et al. 1995). In contrast, Fig. 3 A demonstrates that a substantial fraction of the EPC amplitude data from each homozygous animal is located below this threshold line.

Fig. 3.

Homozygote EPCs are reduced relative to muscle fiber current threshold.A and B: distributions of EPC amplitudes adjusted to a holding potential of −80 mV using regression parameters derived from measured intravenous current-voltage relationships. Data from individual animals are plotted in separate vertical lanes and are arranged in the order of ascending age (left to right) as in Fig. 1. Closed and open symbols designate homozygote and control data, respectively. A: unpotentiated EPC amplitudes recorded soon after muscle fiber impalement. B: EPC amplitudes following potentiation using a stimulus paradigm defined undermethods. Horizontal lines drawn through the data ofA and B designate the estimated average value of MG muscle fiber current threshold, adjusted to a resting potential of −80 mV as described in the text, less 2× the SD of the estimated distribution for homozygote and normal animals. About 98% of all current thresholds are thus located above this line. Details concerning the selection of this line are provided in the text.C: plot demonstrates the fraction of EPCs above average MG muscle fiber current threshold minus 2× SDs before and after EPC amplitude potentiation. A significant fraction of unpotentiated EPCs are below threshold in homozygotes but potentiation increases the suprathreshold fraction.

We showed previously that both motor unit twitch force and EMG in HCSMA homozygotes can be dramatically increased immediately following trains of high-frequency stimulation (Pinter et al. 1995). The basis for this effect is likely due to a transient potentiation of EPC amplitude that enables activation of additional muscle fibers. To test this, we measured EPC amplitudes following application of the same stimulus paradigm used previously to demonstrate the presence of posttetanic potentiation of motor unit twitch force and EMG in HCSMA homozygotes (see methods). These data are plotted in Fig.3 B and demonstrate that, following potentiation, an increased fraction of EPC amplitudes obtained from homozygotes is above the threshold line. The percentage of EPC amplitude data above threshold in unpotentiated and potentiated states is summarized in Fig.3 C for homozygotes and normal animals.

These results indicate that, before potentiation, many motor terminals in HCSMA homozygotes are unlikely to produce sufficient current to enable MG muscle fiber action potential generation and that a large increase in the fraction of suprathreshold currents occurs following potentiation. Our analysis also indicates that muscle fiber current thresholds do not differ between HCSMA homozygotes and normal animals. It thus appears that decreased EPC amplitude is an important determinant of motor unit dysfunction in HCSMA. The observation that EPC failures in response to nerve stimulation are associated with EPC amplitude reductions in homozygotes suggests that motor unit dysfunction involves a presynaptic deficit of ACh release. An analysis of EPC quantal properties provided in the following paper supports this by showing that EPC quantal content at HCSMA homozygous motor terminals is significantly reduced relative to normal (Rich et al. 2002).


A previous study found that the size of ventral root motor axons obtained from forelimb spinal segments was reduced in HCSMA homozygotes relative to age-matched controls (Cork et al. 1989), and it was suggested that this might reflect growth arrest, axonal atrophy, or both. If decreased motor axon size occurs in association with the appearance of motor unit failure, then the reductions of synaptic transmission described above (which underlie motor unit failure) may reflect disruptions of axonal function rather than specific changes of synaptic function. The ideal approach for examining this possibility would be to study individual axon size in relation to the properties of single motor units. However, there are a variety of experimental difficulties associated with this approach, including the need for separate labeling of individual motor axons, since motor unit properties are studied in individual units functionally isolated by intracellular impalement of motor axons or cell bodies. As an alternative, we examined homozygote motor axon conduction velocity, since this is known to depend importantly on fiber size (Moore et al. 1978; Waxman 1980) and is readily measurable. To determine possible associations between changes of axon function and motor unit failure, we assessed when motor unit failure first appears in HCSMA MG motor units in relation to the time course of postnatal development of conduction velocity in MG motor axons.

Figure 4 A shows a plot of the fraction of sampled MG motor units that exhibit tetanic failure versus age for 25 homozygotes and symptomless animals. With one exception, the MG muscles of homozygotes older than about 120 days all possessed motor units that did not sustain force during repetitive activity. A considerable amount of interanimal variability in the fraction of failing MG motor units is evident in Fig. 4 A. Part of this variability is likely due to unavoidable sampling variations among animals. Part may also be due to varying expression of the disorder at the motor unit level for reasons that remain unidentified.

Fig. 4.

Average MG conduction velocity attains adult values in homozygotes.A: fraction of MG motor units in individual animals exhibiting tetanic failure during 60 pulse, 100 Hz axon or motor neuron stimulation train plotted against postnatal age. Tetanic failure is quantified by the taking the difference between peak motor unit force and force present at the end of the motor neuron tetanic stimulus train, expressed as a fraction of the peak force (see Pinter et al. 1995 for further details). Units showing a difference of 7% or greater are considered to exhibit tetanic failure. Symbols identified by inset. B: plot of average MG axonal conduction velocity from individual animals versus postnatal age. Each mean value is based on a sample of ≥6 conduction velocities and is shown ±SE. Symbols as in A.

The time course of postnatal development of mean conduction velocity is shown in Fig. 4 B. Although an increased scatter of points may occur among older homozygotes (filled symbols, aged >120 days), average conduction velocity increased in a linear manner postnatally in HCSMA homozygotes. For homozygotes (60–188 days) and symptomless animals (60–140 days), mean conduction velocity and age were well correlated (Pearson r = 0.91, homozygotes;r = 0.92, symptomless; P < 0.01 both cases). Analysis of covariance showed that the slopes and intercepts of the fitted regression lines did not differ significantly (P > 0.05). The average values of about 85 m/s for both the oldest homozygotes (>180 days) and for two adult symptomless dogs from the HCSMA pedigree (aged >1 yr) agree well with reported values of MG motor axon conduction velocity in normal adult dogs (Lee and Bowen 1970).

To determine whether focal changes of conduction velocity might exist in proximal motor axons, we compared MG motor axon conduction velocities obtained from intracellular recordings made in the cell body with recordings made in ventral root motor axons in two homozygotes, both of which exhibited failing MG motor units (homozygote 1, 90% failing units; homozygote 2, 50%). In both cases, no differences in average conduction velocity obtained from each recording site were observed (Fig. 5; n = 20,P > 0.05, t-test). The greater mean values for homozygote 1 are most likely due its greater age (homozygote 1, 180 days; homozygote 2, 165 days). To confirm that these results are representative of normal, we studied two symptomless HCSMA animals of similar age and obtained an identical result (Fig. 5). These comparisons indicate that the conduction velocities of proximal motor axons (located between the soma and ventral root) do not differ significantly from the conduction velocities of more distal parts of the motor axons in HCSMA homozygotes that have failing MG motor units.

Fig. 5.

Motor axon conduction velocities measured from intracellular recordings made in the soma and in the ventral root do not differ in homozygotes. Panel shows mean values (±SE) of motor axon conduction velocity for each of 2 homozygotes and 2 symptomless animals. Each mean represents the value obtained from 20 samples of conduction velocity. In each comparison, mean conduction velocity did not differ between the proximal (soma) and distal (ventral root) recording sites (P > 0.05, t-test).

Overall, these data demonstrate that the development of axonal conduction velocity does not differ between HCSMA homozygotes and symptomless individuals and that the mechanisms producing reduction of synaptic currents and motor unit dysfunction in HCSMA homozygotes do not interfere with and act independently of mechanisms producing postnatal maturation of motor axonal conduction velocity.


In HCSMA and in other forms of motor neuron disease, swellings appear in proximal motor axons that are filled with misaligned neurofilaments and can achieve sizes as large as motor neuron soma (Cork et al. 1982b; Hirano 1991). To determine whether proximal motor axon swellings are associated with the occurrence of motor unit tetanic failure, we intracellularly labeled four MG motor neurons in one HCSMA homozygote (aged 144 days) after characterizing motor unit mechanical properties for each motor neuron. Three of four motor neurons were stained sufficiently well to enable visualization of the entire motor axon within the spinal cord, and all three of these cells innervated failing motor units. In two of the failing units, swellings were not visible in the labeled motor axons. Figure6 A illustrates EMG and force records from one of these motor units and demonstrates failure to sustain force output during repetitive activation (tetanic failure). Figure 6 B shows a low-magnification montage of several serial sections of the labeled soma, dendrites, and axon from the motor neuron innervating this unit and illustrates the absence of any swellings along the course of the axon from its origin to the ventral white matter. At higher magnification (Fig. 6 C), the axon could be seen originating from a primary dendrite and extending into the parenchyma without discontinuities or enlargements in its proximal portion. The functional properties of the motor unit and the structural properties of the motor neuron soma and axon from another failing unit were nearly identical (not illustrated).

Fig. 6.

Tetanic failure occurs in the absence of proximal axonal swellings.A: motor unit tetanic failure. Unit innervated by an MG motor neuron from a 144-day HCSMA homozygote. Top signal: motor unit EMG. Lower signal: motor unit force. Before injection of Neurobiotin, the motor neuron shown in B was stimulated with a train of 60 suprathreshold depolarizing pulses at 150 Hz (train duration indicated by horizontal bar) while recording force from the MG muscle. The simultaneous failure of EMG signals and force output represents tetanic failure (Pinter et al. 1995, 1997).B: low-magnification image of Neurobiotin-labeled MG motor neuron innervating motor unit in A. This image was constructed by overlaying and aligning images from several serial sections. Black arrows indicate course of motor axon through spinal gray matter. Note the absence of any obvious distensions along the axon. Dotted line indicates gray matter border. C: higher-magnification view of proximal motor axon. Black arrows indicate course of motor axon. The axon presents a smooth profile along its course and can be seen emerging from a proximal dendrite. Scale bar:A, 100 μm; B, 50 μm.

The third labeled motor neuron exhibited small swellings in the proximal motor axon and showed the most extensive motor unit failure observed thus far in HCSMA. In this case, only the first two or three stimuli of a train of 60 depolarizing pulses (100 Hz) delivered to the motor neuron soma produced motor unit EMG potentials and almost no unit force was observed (Fig.7 A). The motor axon exhibited a smooth profile along its length except at the most proximal locations, where small swellings could be observed at low magnification (Fig. 7 B). At higher magnification, confocal imaging showed two distensions with outer diameters slightly larger than flanking axonal regions (Fig. 7 C, white arrows). These distensions are small compared with the sizes of axonal swellings that have been observed previously in the spinal cords of HCSMA homozygotes (Cork et al. 1982b, 1988), and we considered the possibility that this particular homozygote might lack large accumulations of neurofilaments in the spinal gray matter. As shown in Fig. 8, however, we verified the presence of abnormal accumulations of neurofilaments in adjacent sections of this spinal cord. These swellings immunostained positively for phosphorylated neurofilament (SMI31), and many were considerably larger than the apparent distensions observed in the failing motor neuron of Fig. 7. This suggests that, if the distensions seen in Fig.7 C are in fact accumulations of neurofilaments, they are most likely at an early stage in their development.

Fig. 7.

Motor unit exhibits virtually complete failure while proximal motor axon shows small distensions. A: motor unit failure in a 144-day HCSMA homozygote was nearly complete as indicated by the presence of only the first and second electromyographic (EMG) responses to the 60 pulse, 150 Hz train (top signal, indicated by arrows) and negligible force production (lower signal). Train duration indicated by horizontal bar. B: low-magnification image of the Neurobiotin-labeled MG motor neuron innervating motor unit inA. This image was constructed as described for Fig. 6. Black arrows indicate course of motor axon through spinal gray matter. White arrow indicates presence of small swellings on most proximal part of motor axon. Dotted line indicates gray matter border.C: higher-magnification confocal image of the proximal motor axon as it emerges from a proximal dendrite. This image is a single-plane projection of 20 individual confocal planes that are separated by 1.35 μm and span the entire thickness of the axon. This demonstrates that the distensions (arrows) are only slightly larger than the flanking axon. Scale bar: A, 100 μm;B, 10 μm.

Fig. 8.

Example of immunostaining for phosphorylated neurofilaments. Section taken from the same homozygous lumbosacral spinal cord that provided the motor neurons of Figs. 6 and 7. Immunostaining for phosphorylated neurofilaments (SMI34) demonstrates the presence of neurofilament accumulations (*) that are as large as nearby motor neuron somas (arrows). Scale bar, 50 μm.

These data demonstrate that failure of motor unit function in HCSMA does not require the presence of neurofilament accumulations in proximal motor axons. These results also suggest that complete motor unit failure can precede the full development of neurofilament accumulations in proximal motor axons.


In normal animals, systematic patterns of correlations exist between and among motor neuron electrical properties and motor unit mechanical properties (Binder et al. 1996; Burke 1981; Pinter 1990). These relationships disappear after axotomy or after neuromuscular transmission is completely eliminated (Pinter et al. 1991; Vanden Noven and Pinter 1989). This functional organization forms an important part of the basis for the normal operation of motor neuron pools (Henning and Lomo 1985) and orderly recruitment of motor units (Cope and Pinter 1995; Zajac 1990). Others have suggested that these correlations may depend on trophic interactions between motor neuron and muscle that are likely mediated by mechanisms such as axonal transport (Mendell et al. 1994; Munson et al. 1997). We reasoned that, if any changes in the mechanisms underlying these correlations occur that provoke or are associated with the appearance of tetanic failure, then these correlations should be not be present or be diminished in data from HCSMA homozygotes. In the present study, we found significant correlations between MG axonal conduction velocity and motor unit mechanical properties in older homozygotes exhibiting tetanic failure as well as in other HCSMA animals exhibiting normal motor unit function. Table 1 summarizes average correlation coefficients for symptomless and homozygous groups for experiments in which five or more motor units were sampled and the fraction of these experiments in which significant (P< 0.05) correlations were observed between conduction velocity, twitch time-to-peak, and peak unit force measured during tetanic activation. An analysis of covariance that only included data from significant correlations showed that no significant differences in regression slopes or intercepts were present within either the homozygous or symptomless groups (P > 0.05).

View this table:
Table 1.

Average correlation coefficients of conduction velocity versus motor unit mechanical properties

Significant correlations were also observed between motor neuron soma afterhyperpolarization (AHP) duration and twitch time-to-peak in homozygotes (aged 80–177 days, r = 0.73,P < 0.01) and in young symptomless animals (aged 87–101 days, r = 0.79, P < 0.01). Analysis of covariance showed that the regression slopes did not differ between the homozygous and symptomless data groups (P> 0.05). These correlations and regressions were determined from data pooled from several experiments because AHP measurements were only made in smaller numbers of cells during individual experiments.

These data demonstrate that the relative incidence and extent of significant correlations between motor axon, motor neuron, and motor unit properties were quite high and independent of HCSMA phenotype. Since all but one older homozygote also exhibited tetanic failure among MG motor units, these data demonstrate that correlations between motor axon and motor unit properties persist despite the presence of motor unit dysfunction.


The results of this study provide several new insights into the pathogenesis of HCSMA at the cellular level. The first concerns the mechanism underlying weakness. Recordings of EPCs and muscle fiber threshold currents demonstrate that EPC amplitudes are reduced relative to EPCs from genetically normal control animals and to the currents needed to cause action potential firing in muscle fibers. A presynaptic deficit is implied by observations that the relative decrease of EPC amplitude in homozygotes is associated with the appearance of failures of EPCs in response to nerve stimulation. The immediate cause of motor unit dysfunction and weakness in HCSMA homozygotes is thus a reduced capability of motor terminals to release sufficient ACh to evoke muscle fiber electrical activity.

The demonstration that a large increase in the fraction of homozygote EPCs capable of evoking muscle fiber action potentials occurs following EPC potentiation helps explain two specific kinds of motor unit dysfunction observed in HCSMA. When first activated by intracellular motor neuron or motor axon stimulation, the twitch force and EMG of many motor units are often negligibly small but can be dramatically increased following tetanic activation of the motor neuron/axon or systemic administration of 4AP (Pinter et al. 1995,1997). These phenomena can now be understood to reflect the existence of EPCs at many motor terminals that are initially subthreshold for muscle fiber activation but become suprathreshold after transmitter release is increased by either tetanic stimulation or 4AP administration.

The second category of information provided by our results concerns the possible role of axonal abnormalities in the appearance of motor unit dysfunction in HCSMA. We examined how axonal conduction velocity matures in HCSMA homozygotes in relation to the appearance of motor unit failure. Motor axon conduction velocity is a functional attribute that depends on multiple factors, but among the most important are axon caliber and myelin thickness (Moore et al. 1978;Waxman 1980). Other studies indicate that cytoskeletal properties such as neurofilament number, density, and phosphorylation status are all important determinants of axon caliber and that myelinating cells influence these properties and perhaps neurofilament movement via slow axonal transport (deWaegh et al. 1992;Hsieh et al. 1994; Julien 1995;Williamson et al. 1996). Our results show that average MG motor axon conduction velocity in HCSMA homozygotes increases postnatally and reaches adult values by 180–190 days despite the earlier appearance of motor unit dysfunction (Fig. 4 B). The data also indicate that the rates of conduction velocity maturation do not differ between homozygotes and symptomless animals and that proximal and distal motor axon conduction velocities do not differ between these groups. These results show that axonal conduction velocity continues to develop normally in HCSMA homozygotes despite the appearance of motor unit failure. Because of the dependency of conduction velocity on multiple factors, complicated interpretations of these results are possible. However, the simplest view is that the large variety of mechanisms responsible for axonal maturation remains unaffected by and is independent of the mechanisms that give rise to reduced EPC amplitudes and motor unit dysfunction in HCSMA homozygotes.

We also considered the possibility that abnormal accumulations of neurofilaments or other focal disruptions of function in proximal motor axons might be associated with or cause motor unit failure. Our intracellular labeling experiment enabled us to examine directly for the first time the proximal axons of diseased motor neurons with known motor unit dysfunction (Figs. 6 and 7). Although the data are from one homozygote, they constitute an existence proof that neither the onset of motor unit dysfunction nor its evolution in HCSMA requires swellings in proximal axons. In the case in which evidence of proximal axonal swellings was observed (Fig. 7), the extent of motor unit failure was nearly complete but the swellings were small compared with the size of accumulations that could be demonstrated in adjacent spinal cord areas (Fig. 8). These results suggest that neurofilament dysfunction as manifested by aggregations in proximal motor axons may not occur until after the motor unit has lost its ability to produce useful force output. Overall, our data suggest that failure in cellular systems that support maturation of the motor axon and mechanisms that produce neurofilament swellings in motor axons do not give rise to reduced EPCs or motor unit tetanic failure in HCSMA. This is consistent with our recent finding that tetanic failure in HCSMA occurs in the absence of evidence for degeneration at neuromuscular junctions (Balice-Gordon et al. 2000).

Our finding that MG axonal conduction velocity matures normally in homozygotes despite the appearance of motor unit failure is in apparent conflict with the study of Cork et al. (1989) who reported that motor axons in HCSMA homozygotes fail to achieve normal size during postnatal development and that some may exhibit atrophy. HCSMA features a stereotyped, spatial–temporal pattern of progression in which the innervation of axial and proximal muscles shows the earliest onset of involvement and is always more progressed than in distally located muscles such as the MG (Balice-Gordon et al. 2000; Pinter et al. 2001a). One factor that may thus have contributed importantly to the results of Cork et al. is that axon size was studied in ventral roots of forelimb spinal segments. These roots most likely included a mixture that included axons supplying neck and rostral paraspinous muscles that are extensively involved at the ages studied. In contrast, the present study focused on a single, distally located muscle that becomes involved later in the course of the disease. Another contributing factor appears to be the presence of an exceptionally large value of axon size for one control animal that probably exaggerated differences between homozygote and normal animals (see Table 1 of Cork et al. 1989). Axonal atrophy and degeneration are components of HCSMA, and it is thus likely that conduction velocities eventually decrease. The present results indicate, however, that the time course of the pathogenesis is such that these events occur after motor unit function is lost because of neurotransmission reductions and that the mechanisms mediating axonal atrophy and degeneration do not play a key role in determining loss of motor unit function.

Correlations between motor neuron and motor unit properties

Significant correlations between MG motor axon and motor neuron properties and motor unit mechanical properties were found in HCSMA homozygotes and were similar to those found in symptomless animals (Table 1). Previous studies have shown that these correlations appear within several weeks after birth in the cat (Bagust et al. 1974) and the same seems likely for the dog. Our data thus demonstrate that motor unit dysfunction occurs in HCSMA homozygotes without disturbing these relationships. Correlations between motor neuron and motor unit properties represent an important part of the functional organization underlying orderly recruitment of motor units (Binder et al. 1996; Burke 1981;Cope and Pinter 1995; Pinter 1990). The presence of these relationships in older HCSMA homozygotes contributes to the sense that MG motor unit recruitment order and usage pattern are likely to be normal, at least at the time when failure begins to appear. The mechanisms that establish and maintain correlations between motor neuron and motor unit properties are not known. Some have suggested that trophic interactions between motor neurons and muscle are responsible at least in part (Mendell et al. 1994;Munson et al. 1997). Following axotomy, correlations among motor neuron properties disappear (Pinter and Vanden Noven 1989). Because these correlations are present among the motor units we sampled, the onset of tetanic failure does not appear to be associated with an axotomy-like response among MG motor neurons in HCSMA. The postnatal increase of average axonal conduction velocity (Fig. 4) also supports this point. By contrast, axonal injury leads to a reduction of conduction velocity (Pinter and Vanden Noven 1989) that is associated with a decrease in axon caliber and in neurofilament protein synthesis (Hoffman et al. 1988). One factor that may inhibit the appearance of axotomy-like changes in motor neurons innervating failing motor units is the persistence of neurotransmission to a few muscle fibers, since axotomy-like changes do not appear in normal motor neurons even when most synaptic transmission with muscle is blocked (Pinter et al. 1991).


Our work in HCSMA demonstrates that diminished neuromuscular synaptic transmission provokes significant or complete loss of motor unit function well before the appearance of axonal neurofilament swellings or evidence of axonal dysfunction, including motor terminal degeneration (Balice-Gordon et al. 2000). Since neurotransmission failure appears to be independent of axonal maturation, it is conceivable that the underlying defect may be specific for mechanisms located in the nerve terminal. It is not yet known whether neurotransmission failure underlies loss of motor unit function in other forms of motor neuron disease, although there is evidence in amyotrophic lateral sclerosis that neuromuscular synaptic dysfunction occurs while motor axons can still conduct action potentials (Maselli et al. 1993). It has also been reported that degeneration of motor terminals occurs well before motor neuron cell death in the SOD1 transgenic model of familial amyotrophic lateral sclerosis (Frey et al. 2000). These observations raise the possibility that inhibition of motor neuron cell death and perhaps even axonal degeneration may not be sufficient to prevent loss of function in motor neuron disease. Indeed, experimental treatments of motor neuron disease that inhibit cell death and axonal degeneration but fail to provide significant and lasting functional benefit might be explained by the persistence of additional but untreated mechanisms that compromise neuromuscular function or structure and thus motor unit function (Klivenyi et al. 1999). This view is consistent with emerging evidence that survival or maintenance of the motor terminal may be regulated differently and independently of mechanisms controlling survival of the axon and cell body (Gillingwater and Ribchester 2001). Further insights into the mechanisms underlying loss of neuromuscular function in motor neuron disease may be necessary to accomplish effective treatment. The following paper provides additional insight into possible mechanisms underlying loss of neuromuscular transmission in HCSMA (Rich et al. 2002).


We thank D. Dewey, D. Smith, and A. Shirley for technical assistance.

This work was supported by National Institute of Neurological Disorders and Stroke Grants NS-31621, NS-07287, NS-31563, and NS-25547.

Present address of R. F. Waldeck: Department of Biology, University of Scranton, Scranton, PA 18510.


  • Address for reprint requests: M. J. Pinter, Department of Physiology, Emory University School of Medicine, Whitehead Bldg., 615 Michael St., Atlanta, GA 30322 (E-mail:mpinter{at}


View Abstract