INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Although it is commonly believed that loss of motor unit function is secondary to motor neuron cell death or degeneration in motor neuron disease, it is not clear whether this view accurately represents the actual sequence of pathological events. There is, for example, evidence from amyotrophic lateral sclerosis (ALS) that abnormalities of motor terminal synaptic function appear while the motor axon is still capable of producing and conducting action potentials, indications that the motor neuron cell body is alive and supporting peripheral function (Maselli et al. 1993). Additional evidence that actions in the periphery precede motor neuron cell death or degeneration to determine loss of motor unit function derives from studies of transgenic mice that overexpress mutations of the SOD1 enzyme that are linked to familial ALS (Rosen et al. 1993). In these animals, significant motor terminal degeneration precedes the appearance of motor neuron cell death (Frey et al. 2000), and it is a plausible (but not yet established) hypothesis that defects of neuromuscular transmission precede terminal degeneration to cause loss of motor unit function.

Hereditary canine spinal muscular atrophy (HCSMA) is an autosomal dominant, degenerative disorder of motor neurons that shares pathological features with human motor neuron disease (Cork et al. 1982; Pinter et al. 2001a). In HCSMA, it is established that motor unit dysfunction arises before the appearance of motor terminal degeneration and that this dysfunction is an important determinant of weakness in affected animals (Balice-Gordon et al. 2000; Pinter et al. 1995, 1997, 2001a). The most important form of dysfunction is an inability to sustain motor unit force during repetitive activity. Interestingly, this dysfunction occurs to a significant extent in the absence of any electromyographic evidence of denervation that is routinely used to diagnose motor neuron disease in humans (Pinter et al. 2001a; Stalberg 1982; Stalberg and Sanders 1992).

We showed in the previous paper that nerve-evoked synaptic currents are reduced at the endplates of muscles from HCSMA homozygotes that contain failing motor units (Rich et al. 2002). These observations help explain certain aspects of motor unit dysfunction found in HCSMA but provide little insight into the mechanisms that might underlie the neurotransmission defect. The present study provides a quantal analysis of synaptic release in the MG muscle of HCSMA homozygotes. Using voltage-clamp recordings and deconvolution analysis to examine synaptic release kinetics, we show that homozygote motor terminals feature an unusual combination of significantly reduced quantal content with normal levels of release depression during high-frequency activation and that the reduction of endplate currents (EPCs) in HCSMA homozygotes most likely reflects a decreased availability of releasable quanta in motor terminals. A portion of these results have been reported in abstract form (Pinter et al. 2001b).

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

A total of 10 dogs were used in this study; 5 were obtained from the HCSMA breeding colony maintained at Emory University and showed motor symptoms typical of HCSMA homozygotes (Pinter et al. 2001a). The other 5 animals were purpose bred and obtained from vendors to provide genetically normal controls for data comparison. All work reported in this paper was approved the Emory University Institutional Animal Use and Care Committee.

Surgical preparation

All dogs were initially anesthetized with 35-40 mg/kg iv pentobarbital sodium. 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. Animals remained unparalyzed for the duration of the experiments.

Muscle fiber samples and EPC recording

Procedures for obtaining muscle fiber samples for EPC recording and for obtaining voltage-clamp records are described in Rich et al. (2002). For most EPC recordings, holding potentials were 45 mV but, in two initial experiments, 60 mV was used. During several recordings, the holding potential was varied to obtain current-voltage relationships, which were used to adjust the measured amplitudes of final records to various holding potentials. Final EPC records were averages of 16-30 sweeps collected at 0.5 Hz. Various properties were measured from averaged EPCs including amplitude, time-to-peak, and half-width. The decay phases of averaged EPCs were fitted with single exponentials to derive decay time constants. In all cases, the starting point for these fits was located at the 50% of maximum amplitude point on the EPC decay. Quantal content was determined as the ratio of averaged EPC and miniature EPC (mEPC, see following text) amplitudes.

Following sampling of EPCs at low frequency, the nerve was stimulated with trains of 10 stimuli at 50 Hz. A series of 5-20 trains was delivered, each separated by an interval of 2 s. An average of these trains was performed and used to quantify facilitation or depression by normalizing each EPC amplitude in the averaged train to the amplitude of the first EPC. Correlations between the first EPC amplitude in each train and the train sequence number were used to determine whether any trends for longer term facilitation or depression occurred during the time interval over which 50-Hz trains were delivered. Additional treatments of the data derived from 50-Hz trains are described under RESULTS.

To examine EPC amplitude potentiation, single stimuli were interleaved with trains of 20-30 stimuli at 150 Hz, separated by 2 s. This alternating sequence of trains and single stimuli was continued until the amplitudes of single EPCs reached a maximum as indicated by on-line measurements of EPC amplitude. At this point, the alternating stimulus sequence was stopped and, after a delay of 10 s, single EPCs were again evoked at 2-s intervals to determine the postpotentiation time course of EPC amplitude decay. This stimulus paradigm for studying EPC amplitude potentiation was selected to enable comparison with results of previous studies in which motor unit twitch potentiation was investigated in HCSMA homozygotes using the identical stimulus paradigm (Pinter et al. 1995).

mEPC recording

Spontaneously occurring mEPCs were observed routinely in all animals studied. However, in bathing solutions containing normal calcium (2 mM) and at room temperature, we found that the occurrence frequency of spontaneous mEPCs was too low to allow reliable estimation of spontaneous mEPC frequency and to yield sufficient numbers of events to allow for mEPC averaging. In some instances of recording from HCSMA homozygote muscle fibers, one, two, or no mEPC events could be observed during 2-min recording epochs. We considered using extremely long recording intervals to sample mEPCs, but this would have interfered with the need to sample other data from as many muscle fibers as possible. To increase the yield of mEPCs, we instead relied on the increase of mEPC frequency that occurs following tetanic nerve stimulation (del Castillo and Katz 1954; Lev-Tov and Rahamimoff 1980). This approach has been used previously to study human mEPCs (Cull-Candy et al. 1980). During continuous 1- to 2-min recording epochs (digitized at 20 samples/ms), the muscle nerve was stimulated at 10-s intervals with trains of 50-60 stimuli at 150 Hz, and mEPCs were recovered from the intervals between each train. Initial identification of mEPCs was performed using discrimination software (Ankri et al. 1994). Because of baseline noise, the detected onsets of a considerable number of mEPCs did not appear to correspond to the actual mEPC onset (Fig. 1A). When using the detected onsets as "triggers" for averaging mEPCs, these errors produced a noticeable corruption of the onset and rising phase of the averaged mEPC (Fig. 1C). To correct for this, specially written software that displayed each mEPC along with its detected onset and allowed the user to manually adjust the detected onset to correspond more closely with the mEPC onset as judged visually (Fig. 1B) was used. Averaged mEPC records were then obtained using these adjusted onsets as the trigger. Averaged mEPCs were subsequently analyzed identically to averaged EPCs (see above).

View larger version (22K): [in this window] [in a new window]   Fig. 1. Endplate current (EPC) and miniature EPC (mEPC) signal processing. All records were obtained from the same homozygote neuromuscular junction. A: superimposed records of 49 mEPCs detected by software during a 2-min recording in which trains of stimuli were delivered to increase mEPC frequency. All mEPCs occurred between stimulus trains. Detected mEPC onsets occur 1 ms after the beginning of each record. Because of noise and imperfect selection criteria, detected and mEPC onsets do not correspond in many records (arrow). B: detected onsets of mEPC records shown in A were shifted to correspond to the best visual estimate of the actual mEPC onset. Onsets occur 1 ms after the beginning of each record. C: average mEPC records using detected and realigned onsets as "triggers". Lower records show averages of mEPCs in A and B, and upper records show SDs. Realignment reduced the variance of the rising phase, reduced time to peak and increased the amplitude of the average. D: mEPC fit obtained using Eqs. 1 and 2. The average record of C (realigned) is superimposed on the best fit, demonstrating that the fit defines the mEPC contours well. E: EPC average before and after stimulus artifact removal. A sum of 2 exponentials was fitted to data between points 1 and 2 on the rising phase of the stimulus artifact. The fitted function was then extrapolated to the end of the averaged EPC and subtracted from the averaged EPC beginning at point 1. This produced a zeroed baseline beginning at point 1, which was then extended to the beginning of the average. As judged by the complete overlap of averaged records with and without the artifact, this subtraction has little effect on the shape of the EPC but clarified the EPC onset (arrow). F: EPC deconvolution. As described in text, EPC averages with artifacts removed were deconvolved with fitted mEPCs. F illustrates the results of deconvolution of the averaged EPC of E with the fitted mEPC in D. The deconvolution product is called the "release function" in the text. Shown is the release function, which was integrated to yield estimates of total quantal release during EPCs that took account for asynchronous release (integral). The inset illustrates the release function prior to Gaussian filtering as described in text. Scale bars: A-D, 1 nA and 1 ms; E, 2 nA, 1 ms; F, 1 ms.

In several experiments (3 homozygote and 1 normal), we attempted to raise the frequency of spontaneous mEPC occurrence by increasing the potassium concentration in the bathing fluid (to 12 mM). Only spontaneous mEPCs could be recorded during these experiments since the ability to evoke EPCs by nerve stimulation was quickly blocked by the raised potassium concentration.

Deconvolution analysis

Deconvolution analysis provides the means to determine the rate at which quantal events (represented by the average mEPC) must occur in time to account for the features (amplitude, shape, etc) of the average EPC (Borges et al. 1995; Diamond and Jahr 1995; Van der Kloot 1988a,b). An example of deconvolution of the averaged EPC in Fig. 1E with the average mEPC of Fig. 1D is shown in Fig. 1F. The deconvolution product has units of quanta/ms and, in this study, is called the synaptic release function. We performed deconvolution analysis for two main reasons. First, measurements of the synaptic release function (e.g., time to peak, half-width, and decay rates) enabled a quantitative comparison of synaptic release kinetics between homozygote and normal animals. Second, integration of the synaptic release function yielded a convenient estimate of quantal release (called total quantal release) that takes into account asynchronous release of quanta.

To accomplish deconvolution, averaged mEPC records required additional processing since mEPCs and EPCs were initially digitized at different rates. A fitting strategy was used that enabled interpolation of mEPC sample points sufficient to equal the sampling frequency used to record EPCs. Each averaged mEPC was first fitted with the following equation using least-squares methods

(1)

where t = time. The a parameters were provided by the fitting routine and

(2)

An example of a fitted mEPC in which f(t) = t² is shown in Fig. 1D. The decision between use of the different f(t)s was ultimately based on visual evaluation of the quality of fit on the mEPC rising phase, since the apparent quality of this fit and the effect of the different f(t)s on minimizing the mean squared error of the fit did not necessarily correspond. This presumably arises because most of the sample points used to fit the above equation are not located on the mEPC rising phase yet carry the most weight in determining mean-squared error. We found that for the majority of averaged mEPCs, f(t) = t² was needed to achieve the best rising phase fit. The final interpolated mEPC was adjusted to have zero latency.

The average EPC also received additional processing. In these experiments, the stimulating electrode was located close (in space) to the recording electrodes, so stimulus artifacts were routinely present close (in time) to the EPC onset (Fig. 1E). Since deconvolution is basically a high-pass operation, the presence of these artifacts caused oscillations that obscured the release function onset. To avoid this, the artifact was removed from the averaged EPC in two steps before deconvolution was performed (Fig. 1E). First, an interval of the artifact decay preceding the EPC onset was fit with a sum of two exponentials. The fitted function was evaluated for the duration of the EPC from the beginning of the selected interval and subtracted from the remainder of the EPC. This produced a zero-baseline preceding the EPC but left the artifact preceding this baseline in place. The second processing step replaced the remaining artifact with a zero-baseline to which was added zero-mean Gaussian noise with a variance determined from the original baseline.

A point-by-point deconvolution of the EPC was performed using an approximation of the Wiener filter (Castleman 1996; Press et al. 1992), which is expressed as (Parker 1997)

(3)

where
are, respectively, the Fourier transforms of the release function, the average EPC, and the average mEPC, and

(4)

where ² is the original baseline noise variance and n is a multiplying factor. For all deconvolutions, we used n = 4 which provided minimal preliminary filtering of high-frequency signal components. The Fourier transform of the release function obtained in this way was inverse transformed. Typically, release functions at this point exhibited considerable high-frequency noise (Fig. 1F, inset) and so were convolved with a Gaussian filter, which used a SD of 10. Comparison of the inset of Fig. 1F with the larger record illustrates the effects of this final filtering.

Measured features of the synaptic release function included maximum amplitude, time-to-peak, and half-width. To obtain an estimate of total release expressed in quantal units, the release function was integrated from its onset to a point where the function returned to baseline. Individual EPCs in averaged 50-Hz trains were also deconvolved by extracting each EPC and deconvolving it with the averaged mEPC. This operation assumes that mEPC amplitude and shape remain constant during 10-pulse, 50-Hz stimulus trains.

STATISTICS. Data from groups of individual animals were tested for phenotype-related differences (i.e., HCSMA homozygote versus normal) using nested ANOVA (Neter et al. 1990). Unless noted otherwise, P values presented in the text or figure legends refer to the results of nested ANOVA. In some comparisons, differences between distributions were tested using a two-sample Kolomogorov-Smirnov test, and differences between mean values were tested using regular ANOVA. Unless noted otherwise, all mean values are shown ± SE.

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

mEPC amplitude and time course are normal in HCSMA homozygotes

In the companion paper, we showed that EPC amplitudes in HCSMA homozygotes are reduced relative to genetically normal controls and that this reduction is associated with failure of nerve-evoked EPCs (Rich et al. 2002). Additional analysis indicated that the reduction of EPC amplitude was sufficient to decrease the likelihood of muscle fiber action potential generation at many HCSMA motor endplates (Rich et al. 2002). These results indicate the existence of a presynaptic defect of ACh release but provide no detailed insight concerning the defective mechanisms. To pursue this issue, we first compared mEPCs recorded in MG muscle fibers from HCSMA homozygotes and from normal control animals and found that mEPCs were similar in both groups. Figure 2A illustrates mean values for mEPC amplitude for all animals included in this study. While mEPC amplitudes appeared slightly greater among homozygotes, a nested ANOVA showed that there were no significant differences of mEPC amplitude between homozygotes and genetically normal animals (P > 0.05). Because EPC amplitudes differ significantly between homozygotes and normal animals (Rich et al. 2002), the similarity of mEPC amplitude indicates that quantal content differs between these groups. This difference is illustrated in Fig. 2B and was significant (P < 0.01). Approximately one-third (57/174) of EPCs recorded at homozygote neuromuscular junction (NMJs) exhibited failure to occur after each nerve stimulus which averaged about 16% (±2%) of trials (range 3-74%). To determine whether the difference in quantal content between homozygote and normal animals was due to this failure, we excluded all homozygote EPCs exhibiting failure from a nested ANOVA and found that a significant decrease of homozygote quantal content remained (P < 0.01).

View larger version (20K): [in this window] [in a new window]   Fig. 2. Mean medial gastrocnemius (MG) quantal content is reduced in hereditary canine spinal muscular atrophy (HCSMA) homozygotes. A: average mEPC amplitudes (±SE) recorded from MG muscle fibers in HCSMA homozygotes and in genetically normal controls. Eight of 10 experiments used 45 mV holding potential while the remainder used 60 mV. All mEPC amplitudes were adjusted to a holding potential of 45 mV before averaging. Bar heights represent averages obtained from 12 to 34 (average = 25) recordings per experiment. Individual mEPC amplitudes were averages of 3 to 140 detected events (mean 25.7 ± 1.3). No significant differences in mEPC amplitude were detected between homozygote and normal animals. B: average quantal contents (±SE) for all experiments. Quantal content was determined as the average EPC amplitude divided by the average mEPC amplitude. Bar heights represent averages obtained from 12 to 34 (average = 24) recordings per experiment. Quantal content is significantly reduced at MG motor terminals in HCSMA homozygotes.

We also compared mEPC and EPC shape parameters between homozygotes and normal animals and found these to be similar. A comparison of EPC and mEPC time-to-peak is shown in Fig. 3. In this case, there was a large overlap among the data from the two groups with no significant differences evident in either measure of time-to-peak (P > 0.05). Other EPC and mEPC properties that were measured included decay time constant, duration at half-maximal amplitude (half-width), and 10-90% rise time (data not shown). A nested ANOVA on each of these parameters showed in each case that no significant differences existed between HCSMA homozygotes and normal animals (P > 0.05). The similarity of EPC and mEPC shapes suggests that synaptic release kinetics do not differ between HCSMA homozygotes and normal animals. Consistent with this, deconvolution analysis showed that no phenotype-related differences were present among several properties of synaptic release functions, including time-to-peak, half-width duration, and decay time constants (P > 0.05, data not shown). Both maximum release rate and total quantal release were, however, significantly decreased among homozygotes (P < 0.01, nested ANOVA), consistent with a decreased quantal content.

View larger version (24K): [in this window] [in a new window]   Fig. 3. EPC and mEPC shapes from HCSMA homozygotes and normal animals are similar. Plot of EPC time-to-peak versus mEPC time-to-peak from MG muscle fibers in HCSMA homozygotes and in genetically normal controls. The line drawn through the data represents unity slope. Note that EPC times-to-peak are generally longer than mEPC times-to-peak and that no difference in the scatter is evident for either homozygote or normal data. No significant differences between homozygote and normal were observed for these data as well as for decay time constants, 10-90% rise times, and half-maximal durations (data not shown).

These data demonstrate that the difference in EPC amplitude observed between HCSMA homozygotes and normal animals (Rich et al. 2002) is due to a reduced quantal content. Since mEPC amplitudes and decay rates do not differ, we conclude that other factors that could influence EPC and mEPC amplitudes and shapes are also unlikely to differ between homozygotes and normal. These factors include transport and packaging of ACh into synaptic vesicles, ACh receptor density, and AChase kinetic properties. The similarities of EPC and mEPC shapes and the results of EPC deconvolution analysis show that the timing properties of synaptic release do not differ between normal and homozygote motor terminals. These data thus implicate release of a reduced number of otherwise normal quanta from motor terminals as the basis for motor unit dysfunction in HCSMA.

mEPC frequency

As noted under METHODS, spontaneous mEPCs appeared infrequently during continuous recording in both homozygotes and normal animals to the extent that obtaining reliable estimates of mEPC occurrence frequency was impractical under standard recording conditions. In several experiments (3 homozygote, 1 normal), we attempted to circumvent this problem by increasing the potassium concentration in the bathing fluid. In 10 fibers from each experiment, we were able to obtain a sufficient number of spontaneously occurring mEPCs over 2- to 5-min recording epochs to enable comparison of mEPC frequencies. Figure 4 illustrates cumulative probability histograms from each experiment, and it may be seen that the largest median value for mEPC occurrence among the homozygote experiments is less than the lowest value found in the normal experiment. Since only one normal experiment was available, it was not possible to judge the significance of this difference using a nested ANOVA. As an alternative approach, we first tested for significant differences among the mean frequencies from homozygote experiments using a standard ANOVA. We found that no significant (P > 0.05, ANOVA) differences were present, so all the homozygote data were pooled into one distribution. A comparison between this distribution and the distribution of normal data showed that mEPC frequencies in the pooled homozygote data were significantly less than normal (P < 0.01, Kolomogorov-Smirnov test). This comparison suggests that, in addition to reduced quantal content, spontaneous mEPC frequency is lowered at homozygote motor terminals.

View larger version (25K): [in this window] [in a new window]   Fig. 4. Spontaneous mEPC frequencies in elevated potassium. Cumulative probability histograms for 4 experiments (3 homozygote and 1 normal) in which the frequency of spontaneous mEPC occurrence was measured during 2- to 5-min continuous recordings in elevated (12 mM) potassium. Homozygote mEPC frequencies appear to be shifted to the left of the normal sample.

Synaptic release during repetitive activity

An important component of motor unit dysfunction in HCSMA homozygotes involves an inability to sustain force output during repetitive activation (Pinter et al. 1995, 1997). We were thus interested in comparing synaptic release properties between homozygote and normal motor terminals during high-frequency nerve stimulation. Figure 5 illustrates examples of averaged EPCs recorded during 10-pulse, 50-Hz nerve stimulus trains from a normal MG NMJ (Fig. 5A) and from a homozygote MG NMJ (Fig. 5B). Surprisingly, homozygote EPCs appeared to depress during the train to approximately the same extent as at the normal EPCs, despite the fact that the quantal content of the homozygote EPCs was about fivefold lower when measured at low frequency (0.5 Hz). To quantify release behavior during 50-Hz trains, we first deconvolved each EPC in averaged trains and integrated the resulting synaptic release functions to obtain the total quantal release for each EPC. The total quantal release for each EPC was then normalized to the value obtained for the first stimulus in the train and expressed as a percentage. Figure 5C illustrates the results of this analysis for the release functions of the second and final (10^th) stimuli of the train for all homozygote and normal synapses. Although there was a small but significant tendency for total release to the second stimulus to be less depressed than normal at homozygote NMJs (P = 0.02, nested ANOVA), no significant differences of depression were found at the end of the train (10^th stimulus, P > 0.05). To determine whether any facilitation or depression accrued during the sequences of 50-Hz trains used to obtain averaged records, linear correlations were calculated between the total release to the first stimulus in each train and the train number. Among five normal experiments, four showed incidence of significant correlations (P < 0.01) in an average of about 9% of recorded NMJs (range 6-11%). Of nine significant correlations (of 101 recordings), six were positive. Among homozygote experiments, three of five showed significant correlations (all positive) in an average of about 5% of recorded NMJs (range 4-6%, 111 total recordings). The overall low incidence of significant correlation between total release to the first stimulus and train sequence number indicates that longer term buildups of depression or facilitation did not occur during 50-Hz train averaging. These data also indicate that recovery from depression that occurred during each train was essentially complete by the time the next train in the sequence was initiated (2-s intervals) at both homozygote and normal motor synapses.

View larger version (29K): [in this window] [in a new window]   Fig. 5. Homozygote and normal synapses exhibit similar release depression during repetitive activity. Panel A illustrates a train of EPCs obtained during 50 Hz nerve stimulation from a normal MG synapse while panel B illustrates a similar train from a homozygote synapse. Despite the significantly lower quantal content of the homozygote synapse, both synapses exhibited similar depression during the repetitive stimulation. Panel C summarizes comparisons of EPC behavior during 10 pulse, 50 Hz stimulus trains between homozygote and normal synapses. Bar heights represent mean (±SE) total synaptic release in response to the 2nd and 10th stimuli, normalized to the total release of the first stimulus and expressed as a percentage. In comparison with normal synapses, homozygote synapses exhibit slightly less average depression of release at the 2nd stimuli but exhibit similar levels of depression by the 10th stimulus.

To further analyze synaptic release during repetitive activity, we used an approach similar to that used previously to estimate the size of the readily releasable synaptic vesicle pool in brain stem synapses (Schneggenburger et al. 1999). For this analysis, it is assumed that the initial depression of release during trains is due to a transient decrease in the number of releasable quanta and that recovery from this depression is negligible during the stimulus train. Cumulative release curves were obtained as a running sum of total release from each stimulus in averaged 50-Hz trains as shown in Fig. 6A for the records illustrated in Fig. 5, A and B. Lines were fitted to the final five points of each cumulative release curve and extrapolated to zero, yielding an estimate of the number of initially available quanta. The slopes of these lines estimate the release per stimulus in the steady-state. The fractions of initially available quanta accomplished by release to the first stimulus and by release in the steady-state were then calculated. The results of this analysis are summarized in Fig. 6, B and C. Although homozygote NMJs exhibited a large and significant (P < 0.01) reduction of initially available quanta relative to normal (Fig. 6B), consistent with a lowered quantal content at low-frequency stimulation, the fraction of initially available quanta released by the first stimulus and in the steady-state did not differ significantly (P > 0.05) between homozygote and normal NMJs. These fractions are related to the probability of release, so this result indicates that the probability of synaptic release does not differ significantly between homozygote and normal NMJs.

View larger version (19K): [in this window] [in a new window]   Fig. 6. Homozygote synapses release similar fractions of a reduced supply of initially available quanta during repetitive activity compared with normal synapses. A: cumulative release curves for the 50 Hz trains illustrated in Fig. 6A and B. The curves are running sums of total quantal release to each train stimulus obtained by integration of each release function in the train. The dashed straight lines are linear fits to the final 5 points of each cumulative release curve. When extrapolated to a hypothetical stimulus located at time 0, the y-axis intercepts provide an estimate of the initially available quanta assuming that EPC amplitude depression is due to a transient decrease in the number of releasable quanta and that recovery from depression is negligible during the train (Schneggenburger et al. 1999). The slopes of these lines provide an estimate of the total quantal release per stimulus after the steady state is reached, i.e., when release depression reaches a steady level. B: a comparison of mean values (±SE) for initially available quanta for 50 Hz stimulus trains and shows that homozygote synapses on average possessed fewer initially available quanta. C: average values of total quantal release for the first stimulus and for steady state estimates expressed as a percentage of initially available quanta. In this comparison, homozygote and normal synapses showed no significant differences suggesting that the mechanisms mobilizing quanta during repetitive activation are similar.

These results show that, with the exception of the number of initially available quanta, the synaptic release properties of homozygote motor terminals during repetitive activation are strikingly similar to normal. Homozygote synapses exhibit relatively normal levels of net depression during repetitive activation and possess relatively normal abilities for mobilizing quanta and for replenishing the initial but relatively low supply of releasable quanta between epochs of repetitive activation within 2 s.

EPC potentiation

In the accompanying paper, we showed that potentiation of homozygote EPCs following repetitive stimulation could increase their amplitudes such that the probability of activating muscle fibers was increased (Rich et al. 2002). This behavior parallels and appears to explain potentiation of motor unit twitch contraction force that can be demonstrated in HCSMA homozygotes (Pinter et al. 1995). Here, we consider further EPC potentiation at homozygote synapses in the context of quantal release. As described under METHODS, the muscle nerve was alternately stimulated with 150-Hz trains and single stimuli at 2-s intervals until individual EPCs (single stimuli) reached a maximum value. All stimulation was stopped at that point and resumed after a 10-s delay with sampling of single EPC amplitudes at 2-s intervals. When the difference between maximally potentiated and initial EPC amplitudes was expressed as a percentage of the initial EPC amplitude [relative posttetanic potentiation (PTP)], homozygote EPCs exhibited a clear increase of average potentiation relative to EPCs from normal animals (least square group means, 184% for homozygote; 51% for normal; P < 0.01, nested ANOVA). As shown in Fig. 7A, relative PTP was inversely related to total quantal release determined at 0.5 Hz. This indicates that the relative PTP differences between homozygote and normal synapses shown in Fig. 7A are due primarily to the differences in quantal content between the sampled populations. When PTP was instead expressed as the absolute increase of EPC amplitude in terms of quanta (absolute PTP), no significant differences were found between homozygote and normal synapses (least square group means, 8.2 for homozygote; 8.1 for normal; P > 0.05), and no systematic relationships with quantal content were evident in either group (Fig. 7B).

View larger version (20K): [in this window] [in a new window]   Fig. 7. Properties of EPC potentiation. A: inverse relationship between relative PTP and total quantal release for homozygote, and normal synapses. Relative PTP is determined as the difference between the amplitudes of the maximally potentiated EPC and the initial, unpotentiated EPC, normalized to the amplitude of the initial EPC and expressed as a percentage. B: relationship between absolute PTP and total quantal release for homozygote, and normal synapses. Absolute PTP is determined as the difference between the amplitudes of the maximally potentiated EPC and the initial, unpotentiated EPC, normalized to the amplitude of the average mEPC. Note that no clear relationship is evident between absolute PTP and total quantal release.

In 8 of 10 experiments, the rates of decay from the maximally potentiated EPC were compared between homozygote and normal synapses. To accomplish this, single exponentials were fitted to all EPC amplitudes sampled beginning 10 s after the repetitive stimulation was stopped. The generally good quality of these fits is indicated by the significant correlation (r = 0.93, P < 0.01, n = 190) and the near-unity slope (0.95) of a linear regression relating unpotentiated, control EPC amplitudes (sampled at 0.5 Hz) to the EPC amplitudes predicted by extrapolating the exponential fits to the steady-state (data not shown). Comparison of the PTP decay time constants between homozygote and normal synapses yielded no significant differences (P > 0.05). Overall, the decay rate of both homozygote and normal synapses for our method of inducing potentiation is best characterized by a time constant of about 30 s.

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

The results of this study indicate that neuromuscular transmission in HCSMA homozygotes differs from that in normal animals mainly in the supply of releasable quanta. The fact that a variety of other synaptic properties are similar between these populations underscores the specific nature of the release deficit in homozygotes. Several features of synaptic transmission in homozygotes uncovered in this study also help explain how failure at the level of the motor unit arises. Perhaps the most important form of motor unit failure in these animals is the inability to sustain force output during repetitive activity (tetanic failure) because this contributes directly to weakness (Pinter et al. 1995, 2001a). Our results indicate that this failure arises because of the novel combination of abnormally low quantal release and relatively normal levels of release depression during repetitive activity. Apparently, many homozygote motor terminals release sufficient ACh to activate motor unit muscle fibers at the beginning of high-frequency stimulus trains but this decreases as the train continues because release depresses below the amount needed to cause fiber activation (see Fig. 6) (Pinter et al. 1995).

Our results also provide a basis for understanding the effects of 4-aminopyridine (4AP) on motor unit performance in HCSMA homozygotes. 4AP increases ACh release from motor terminals by indirectly increasing entering Ca²⁺ (Thesleff 1980). In HCSMA homozygotes, systemically administered 4AP produces an increase of motor unit twitch force and increased peak force during repetitive stimulation. Despite these increases, tetanic failure persists (Pinter et al. 1997). Since EPCs at homozygote motor terminals appear to depress during repetitive activity despite decreased quantal contents (Fig. 6), any increase of quantal content produced by 4AP may simply exaggerate this depression and thus prevent meaningful improvement of motor unit performance. Accomplishing such improvement will clearly require a better understanding of the mechanisms that underlie the release deficits.

Many release properties are normal at homozygote synapses

Our findings demonstrate a number of similarities of synaptic function between homozygotes and normal animals. The fact that the amplitudes and time courses of mEPCs are similar (Figs. 2 and 3) indicates that factors such as ACh receptor channel kinetics, receptor density, ACh packaging and transport into vesicles, and AChE activity are unlikely to play contributing roles in determining neurotransmission deficits in HCSMA homozygotes (Salpeter 1987).

The time course of synaptic release also appears to be similar between homozygote and normal synapses. Supporting this are the observations that no significant differences between homozygote and normal synapses were found in the time courses of EPCs or of synaptic release functions. These results suggest that the dynamic process that discharges releasable quanta is normal at homozygote motor terminals. This process includes the steps between the Ca²⁺ trigger event and the final fusion of vesicles with the presynaptic membrane and deposit of ACh into the synaptic cleft. In addition, the observations that both release depression (Fig. 5) and the fractions of the estimated initially available quanta released during 50-Hz stimulus trains (Fig. 6) are similar between homozygote and normal motor terminals suggest that the dynamics of the release process are also relatively normal during repetitive activation of homozygote motor terminals. The absence of any significant depression or facilitation of the first EPC in 50-Hz trains during closely spaced (2-s) repetitions of the trains indicates that homozygote motor terminals possess the ability to replenish quanta and to recover as rapidly as normal from high-frequency release depression.

Possible mechanisms underlying decreased release at homozygote synapses

In some instances, homozygote EPCs fail to appear following each nerve stimulus. This raises the possibility that failure of action potential invasion of motor terminals may play a role in decreasing overall release. Our analysis demonstrates, however, that a significantly reduced quantal output exists among homozygote motor terminals that do not exhibit trial-to-trial failure. It remains possible that reduced synaptic output arises because action potentials fail to propagate into all homozygote motor terminal branches (partial terminal invasion). We consider this to be an unlikely explanation for several reasons. First, evidence indicates that, in mouse, action potential propagation into motor terminal branches is passive (Brigant and Mallart 1982; Mallart and Brigant 1982). Although we lack direct evidence, we are unaware of any reason why canine and mouse motor terminals should differ in this regard. In fact, comparisons indicate that normal canine MG motor terminals are smaller and feature fewer branches (Balice-Gordon et al. 2000) than mouse gastrocnemius motor terminals (Wigston 1990) and so would presumably provide a more favorable substrate for passive action potential propagation. Second, examination of motor terminals in homozygote MG muscles has provided no evidence of morphological changes (such as branch atrophy) that might hinder passive propagation (Balice-Gordon et al. 2000).

Another mechanism that could account for decreased release in homozygotes is decreased Ca²⁺ entry following action potential invasion of motor terminals. This could explain the lower quantal content at homozygote motor terminals as well as the increased incidence of EPC failure in response to nerve stimulation (Rich et al. 2002). Although our results do not provide direct evidence against this possibility, several aspects of our observations are not consistent with lowered Ca²⁺ entry. Lowered Ca²⁺ entry is associated with decreased probability of release (del Castillo and Katz 1954), but the analysis of deconvolved EPCs during 50-Hz stimulus trains (Fig. 6) suggests that the probability of release at homozygote motor terminals does not differ from normal despite a lowered quantal content. When quantal content is reduced by lowering extracellular Ca²⁺ levels, EPC amplitude facilitation is usually observed during repetitive activation (Zucker 1973). Homozygote EPCs, however, invariably exhibit EPC depression during 50-Hz stimulus trains. We have considered the possibility that the release properties of dog motor terminals during repetitive activation might differ from those of other species (i.e., frog or mouse) that have been studied in more detail. In preliminary studies, however, we have observed that normal dog motor terminals bathed in lowered extracellular Ca²⁺ exhibit the expected facilitation during 50-Hz trains when quantal contents are lowered to match those observed in homozygotes (unpublished results). On the basis of these observations, it thus seems unlikely that the differences between normal and homozygote synapses in the number of initially available quanta or in quantal content are due to differences in the probability of release or Ca²⁺ entry at motor terminals.

Remaining mechanisms to be considered concern the possibility that homozygote motor terminals feature a decrease in the number of functional synaptic release sites, in the supply of releasable quanta, or a combination of these factors. Both of these mechanisms can explain why homozygote mEPC frequencies are lower than normal in high external potassium, the lower quantal content of homozygote EPCs, the increased failure rate in response to nerve stimulation, and depression of homozygote EPC amplitudes during repetitive activation despite low quantal content. A selective loss of release sites cannot, however, account for other observations. For such losses at homozygote motor terminals, the relative effects associated with depression and PTP would be expected to be unchanged assuming that the remaining release sites function normally. While relative depression of homozygote EPCs during high-frequency trains does not differ significantly from normal (Fig. 5), relative PTP is significantly increased (Fig. 7A). This increase arises because quantal contents are reduced, but the absolute PTP (measured in quanta) is unchanged relative to normal (Fig. 7B).

The close similarity between homozygote and normal synapses in the number of quanta that are recruited during PTP suggests that the supply of vesicles or release sites that can be recruited by PTP is normal in homozygotes. The fundamental mechanism thought to underlie PTP and other forms of synaptic facilitation is an increase of residual Ca²⁺ (Zucker 1999; Zucker and Regehr 2002). The similarity of absolute PTP between homozygote and normal synapses may thus reflect that the increases of residual Ca²⁺ are quantitatively similar and that each synapse population has available a sufficient reserve of recruitable quanta. The results of a preliminary study of an age-matched homozygote indicates the existence of a quantal reserve that far exceeds the average number of quanta recruited during PTP. Total quantal release was determined during 10-s applications of hypertonic saline directly to motor terminals. Release under these conditions is thought to provide a measure of the readily releasable population of vesicles (Rosenmund and Stevens 1996; Stevens and Tsujimoto 1995) from which quanta for PTP are likely to be drawn (Zucker and Regehr 2002). Results from a total of 26 motor terminals provided an average of about 240 quanta released by hypertonic saline (unpublished data), which is considerably larger than the average of about 8 quanta recruited during PTP. Comparable studies of normal motor terminals have not yet been performed.

Other evidence suggests that there are no major decreases of synaptic vesicles in homozygote motor terminals. Immunostaining for the synaptic vesicle proteins SV2 or synaptophysin has failed to reveal obvious differences between labeled MG motor terminals of symptomless, nonhomozygote HCSMA cohorts and homozygote motor terminals from MG muscles that contain failing motor units (and therefore defective synapses) (Balice-Gordon et al. 2000). Thus, if a decrease of available vesicles underlies decreased release in homozygotes, the defect presumably involves vesicle processing. One interesting possibility along these lines involves synapsin. The synapsins are a well-characterized family of proteins that bind synaptic vesicles to an actin network that terminates in the vicinity of release sites (Greengard et al. 1993). Ca²⁺-dependent phosphorylation of synapsins is thought to release vesicles from the actin network and make them available for docking by mechanisms that remain unidentified (Hilfiker et al. 1999; Humeau et al. 2001). Cyclin-dependent kinase 5 (CDK5) is a proline-directed kinase that has been found to be elevated and to possess increased activity in HCSMA (Green et al. 1998). Synapsin I is a substrate of CDK5, and phosphorylation decreases the ability of synapsins to interact with the actin meshwork in nerve terminals but not with the vesicles themselves (Jovanovic et al. 1996). It is thus possible that increased levels of phosphorylation by CDK5 might interfere with the normal actions of synapsin. A possible outcome of this scenario is that the efficiency of the synapsin-mediated mechanism for supplying vesicles might be disrupted, with the consequence that fewer docked vesicles would be available for release. For this idea to work at homozygote motor terminals, which retain a good deal of relatively normal synaptic release properties (see above), one has to imagine that the postulated phosphorylation of synapsin leads to a shift in the average relationship between vesicle release from the actin network and phophorylation by other kinases rather than the more dramatic disruptions of function achieved by genetic manipulations (Rosahl et al. 1995) or by direct blockade of synapsin itself (Pieribone et al. 1995). Clearly, more work will be needed to resolve the mechanisms that lead to reduced quantal output at homozygote motor terminals.

Summary

In considering the results of this study, it is important to keep in mind that HCSMA is a degenerative disorder of motor neurons. It is thus important to consider how these results should be viewed in the context of disease mechanisms. On one hand, the findings at the homozygote motor terminal might reflect a primary disorder of neuromuscular transmission. Alternatively, the results may reflect that the synaptic release process is adversely affected by other defective mechanisms that are the basis of the disease process itself. We favor the latter view, because primary disorders of presynaptic release mechanisms such as Lambert-Eaton Syndrome and botulism, for example, do not progress to degenerative changes in motor axons or to motor neuron cell death, which are features of motor neuron disease. We thus believe that probing the release mechanism to determine how it is affected can provide useful insights into the underlying disease mechanisms. In this sense, the process of synaptic release can serve as a "window" through which to view the disease process. This approach has several advantages. First, much is known about the synaptic release process, both at the cellular and molecular levels, and the information base is rapidly increasing (Cowan et al. 2001). Second, hypotheses generated from animal studies can be tested in tissue obtained from living human victims of motor neuron disease (Maselli et al. 1991). This access has the potential to provide more dynamic information about the disease process as it develops in the human-at the level of the affected cell-than might ever be obtained from the study of pathological material.